Merge #1699

1699: Bump milli: fix some crashes r=ManyTheFish a=curquiza



Co-authored-by: Clémentine Urquizar <clementine@meilisearch.com>

.dockerignore

@@ -0,0 +1,4 @@
+target
+Dockerfile
+.dockerignore
+.gitignore

.github/ISSUE_TEMPLATE/bug_report.md

@@ -0,0 +1,30 @@
+---
+name: Bug report
+about: Create a report to help us improve
+title: ''
+labels: ''
+assignees: ''
+
+---
+
+**Describe the bug**
+A clear and concise description of what the bug is.
+
+**To Reproduce**
+Steps to reproduce the behavior:
+1. Go to '...'
+2. Click on '....'
+3. Scroll down to '....'
+4. See error
+
+**Expected behavior**
+A clear and concise description of what you expected to happen.
+
+**Screenshots**
+If applicable, add screenshots to help explain your problem.
+
+**MeiliSearch version:** [e.g. v0.20.0]
+
+**Additional context**
+Additional information that may be relevant to the issue.
+[e.g. architecture, device, OS, browser]

.github/ISSUE_TEMPLATE/config.yml

@@ -0,0 +1,10 @@
+contact_links:
+  - name: Feature request
+    url: https://github.com/meilisearch/product/discussions/categories/feedback-feature-proposal
+    about: The feature requests are not managed in this repository, please open a discussion in our dedicated product repository
+  - name: Documentation issue
+    url: https://github.com/meilisearch/documentation/issues/new
+    about: For documentation issues, open an issue or a PR in the documentation repository
+  - name: Support questions & other
+    url: https://github.com/meilisearch/MeiliSearch/discussions/new
+    about: For any other question, open a discussion in this repository

.github/is-latest-release.sh

@@ -0,0 +1,132 @@
+#!/bin/sh
+
+# Checks if the current tag should be the latest (in terms of semver and not of release date).
+# Ex: previous tag -> v0.10.1
+#     new tag -> v0.8.12
+#     The new tag should not be the latest
+#     So it returns "false", the CI should not run for the release v0.8.2
+
+# Used in GHA in publish-docker-latest.yml
+# Returns "true" or "false" (as a string) to be used in the `if` in GHA
+
+# GLOBAL
+GREP_SEMVER_REGEXP='v\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)$' # i.e. v[number].[number].[number]
+
+# FUNCTIONS
+
+# semverParseInto and semverLT from https://github.com/cloudflare/semver_bash/blob/master/semver.sh
+
+# usage: semverParseInto version major minor patch special
+# version: the string version
+# major, minor, patch, special: will be assigned by the function
+semverParseInto() {
+    local RE='[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)'
+    #MAJOR
+    eval $2=`echo $1 | sed -e "s#$RE#\1#"`
+    #MINOR
+    eval $3=`echo $1 | sed -e "s#$RE#\2#"`
+    #MINOR
+    eval $4=`echo $1 | sed -e "s#$RE#\3#"`
+    #SPECIAL
+    eval $5=`echo $1 | sed -e "s#$RE#\4#"`
+}
+
+# usage: semverLT version1 version2
+semverLT() {
+    local MAJOR_A=0
+    local MINOR_A=0
+    local PATCH_A=0
+    local SPECIAL_A=0
+
+    local MAJOR_B=0
+    local MINOR_B=0
+    local PATCH_B=0
+    local SPECIAL_B=0
+
+    semverParseInto $1 MAJOR_A MINOR_A PATCH_A SPECIAL_A
+    semverParseInto $2 MAJOR_B MINOR_B PATCH_B SPECIAL_B
+
+    if [ $MAJOR_A -lt $MAJOR_B ]; then
+        return 0
+    fi
+    if [ $MAJOR_A -le $MAJOR_B ] && [ $MINOR_A -lt $MINOR_B ]; then
+        return 0
+    fi
+    if [ $MAJOR_A -le $MAJOR_B ] && [ $MINOR_A -le $MINOR_B ] && [ $PATCH_A -lt $PATCH_B ]; then
+        return 0
+    fi
+    if [ "_$SPECIAL_A"  == "_" ] && [ "_$SPECIAL_B"  == "_" ] ; then
+        return 1
+    fi
+    if [ "_$SPECIAL_A"  == "_" ] && [ "_$SPECIAL_B"  != "_" ] ; then
+        return 1
+    fi
+    if [ "_$SPECIAL_A"  != "_" ] && [ "_$SPECIAL_B"  == "_" ] ; then
+        return 0
+    fi
+    if [ "_$SPECIAL_A" < "_$SPECIAL_B" ]; then
+        return 0
+    fi
+
+    return 1
+}
+
+# Returns the tag of the latest stable release (in terms of semver and not of release date)
+get_latest() {
+    temp_file='temp_file' # temp_file needed because the grep would start before the download is over
+    curl -s 'https://api.github.com/repos/meilisearch/MeiliSearch/releases' > "$temp_file"
+    releases=$(cat "$temp_file" | \
+        grep -E "tag_name|draft|prerelease" \
+        | tr -d ',"' | cut -d ':' -f2 | tr -d ' ')
+        # Returns a list of [tag_name draft_boolean prerelease_boolean ...]
+        # Ex: v0.10.1 false false v0.9.1-rc.1 false true v0.9.0 false false...
+
+    i=0
+    latest=""
+    current_tag=""
+    for release_info in $releases; do
+        if [ $i -eq 0 ]; then # Cheking tag_name
+            if echo "$release_info" | grep -q "$GREP_SEMVER_REGEXP"; then # If it's not an alpha or beta release
+                current_tag=$release_info
+            else
+                current_tag=""
+            fi
+            i=1
+        elif [ $i -eq 1 ]; then # Checking draft boolean
+            if [ "$release_info" = "true" ]; then
+                current_tag=""
+            fi
+            i=2
+        elif [ $i -eq 2 ]; then # Checking prerelease boolean
+            if [ "$release_info" = "true" ]; then
+                current_tag=""
+            fi
+            i=0
+            if [ "$current_tag" != "" ]; then # If the current_tag is valid
+                if [ "$latest" = "" ]; then # If there is no latest yet
+                    latest="$current_tag"
+                else
+                    semverLT $current_tag $latest # Comparing latest and the current tag
+                    if [ $? -eq 1 ]; then
+                        latest="$current_tag"
+                    fi
+                fi
+            fi
+        fi
+    done
+
+    rm -f "$temp_file"
+    echo $latest
+}
+
+# MAIN
+current_tag="$(echo $GITHUB_REF | tr -d 'refs/tags/')"
+latest="$(get_latest)"
+
+if [ "$current_tag" != "$latest" ]; then
+    # The current release tag is not the latest
+    echo "false"
+else
+    # The current release tag is the latest
+    echo "true"
+fi

.github/release-draft-template.yml

@@ -0,0 +1,13 @@
+name-template: 'v$RESOLVED_VERSION'
+tag-template: 'v$RESOLVED_VERSION'
+version-template: '0.21.0-alpha.$PATCH'
+exclude-labels:
+  - 'skip-changelog'
+template: |
+  ## Changes
+
+  $CHANGES
+no-changes-template: 'Changes are coming soon 😎'
+sort-direction: 'ascending'
+version-resolver:
+  default: patch

.github/workflows/README.md

@@ -0,0 +1,20 @@
+# GitHub Actions Workflow for MeiliSearch
+
+> **Note:**
+
+> - We do not use [cache](https://github.com/actions/cache) yet but we could use it to speed up CI
+
+## Workflow
+
+- On each pull request, we trigger `cargo test`.
+- On each tag, we build:
+    - the tagged Docker image and publish it to Docker Hub
+    - the binaries for MacOS, Ubuntu, and Windows
+    - the Debian package
+- On each stable release (`v*.*.*` tag):
+    - we build the `latest` Docker image and publish it to Docker Hub
+    - we publish the binary to Hombrew and Gemfury
+
+## Problems
+
+- We do not test on Windows because we are unable to make it work, there is a disk space problem.

.github/workflows/coverage.yml

@@ -0,0 +1,34 @@
+---
+on:
+  pull_request:
+    types: [review_requested, ready_for_review]
+
+name: Execute code coverage
+
+jobs:
+  nightly-coverage:
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - uses: actions-rs/toolchain@v1
+        with:
+          toolchain: nightly
+          override: true
+      - uses: actions-rs/cargo@v1
+        with:
+          command: clean
+      - uses: actions-rs/cargo@v1
+        with:
+          command: test
+          args: --all-features --no-fail-fast
+        env:
+          CARGO_INCREMENTAL: "0"
+          RUSTFLAGS: "-Zprofile -Ccodegen-units=1 -Cinline-threshold=0 -Clink-dead-code -Coverflow-checks=off -Cpanic=unwind -Zpanic_abort_tests"
+      - uses: actions-rs/grcov@v0.1
+      - name: Upload coverage to Codecov
+        uses: codecov/codecov-action@v1
+        with:
+          token: ${{ secrets.CODECOV_TOKEN }}
+          file: ${{ steps.coverage.outputs.report }}
+          yml: ./codecov.yml
+          fail_ci_if_error: true

.github/workflows/flaky.yml

@@ -0,0 +1,15 @@
+name: Look for flaky tests
+on:
+  schedule:
+    - cron: "0 12 * * FRI" # every friday at 12:00PM
+
+jobs:
+  flaky:
+    runs-on: ubuntu-18.04
+
+    steps:
+    - uses: actions/checkout@v2
+    - name: Install cargo-flaky
+      run: cargo install cargo-flaky
+    - name: Run cargo flaky 100 times
+      run: cargo flaky -i 100 --release

.github/workflows/publish-binaries.yml

@@ -0,0 +1,62 @@
+on:
+  release:
+    types: [published]
+
+name: Publish binaries to release
+
+jobs:
+  publish:
+    name: Publish for ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-18.04, macos-latest, windows-latest]
+        include:
+          - os: ubuntu-18.04
+            artifact_name: meilisearch
+            asset_name: meilisearch-linux-amd64
+          - os: macos-latest
+            artifact_name: meilisearch
+            asset_name: meilisearch-macos-amd64
+          - os: windows-latest
+            artifact_name: meilisearch.exe
+            asset_name: meilisearch-windows-amd64.exe
+
+    steps:
+    - uses: hecrj/setup-rust-action@master
+      with:
+        rust-version: stable
+    - uses: actions/checkout@v2
+    - name: Build
+      run: cargo build --release --locked
+    - name: Upload binaries to release
+      uses: svenstaro/upload-release-action@v1-release
+      with:
+        repo_token: ${{ secrets.PUBLISH_TOKEN }}
+        file: target/release/${{ matrix.artifact_name }}
+        asset_name: ${{ matrix.asset_name }}
+        tag: ${{ github.ref }}
+
+  publish-armv8:
+    name: Publish for ARMv8
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - uses: uraimo/run-on-arch-action@v1.0.7
+        id: runcmd
+        with:
+          architecture: aarch64 # aka ARMv8
+          distribution: ubuntu18.04
+          run: |
+            apt update
+            apt install -y curl gcc make
+            curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --profile minimal --default-toolchain stable
+            source $HOME/.cargo/env
+            cargo build --release --locked
+      - name: Upload the binary to release
+        uses: svenstaro/upload-release-action@v1-release
+        with:
+          repo_token: ${{ secrets.PUBLISH_TOKEN }}
+          file: target/release/meilisearch
+          asset_name: meilisearch-linux-armv8
+          tag: ${{ github.ref }}

.github/workflows/publish-deb-brew-pkg.yml

@@ -0,0 +1,39 @@
+name: Publish deb pkg to GitHub release & APT repository & Homebrew
+
+on:
+  release:
+    types: [released]
+
+jobs:
+  debian:
+    name: Publish debian packagge
+    runs-on: ubuntu-18.04
+    steps:
+    - uses: hecrj/setup-rust-action@master
+      with:
+        rust-version: stable
+    - name: Install cargo-deb
+      run: cargo install cargo-deb
+    - uses: actions/checkout@v2
+    - name: Build deb package
+      run: cargo deb -p meilisearch-http -o target/debian/meilisearch.deb
+    - name: Upload debian pkg to release
+      uses: svenstaro/upload-release-action@v1-release
+      with:
+        repo_token: ${{ secrets.GITHUB_TOKEN }}
+        file: target/debian/meilisearch.deb
+        asset_name: meilisearch.deb
+        tag: ${{ github.ref }}
+    - name: Upload debian pkg to apt repository
+      run: curl -F package=@target/debian/meilisearch.deb https://${{ secrets.GEMFURY_PUSH_TOKEN }}@push.fury.io/meilisearch/
+
+  homebrew:
+    name: Bump Homebrew formula
+    runs-on: ubuntu-18.04
+    steps:
+      - name: Create PR to Homebrew
+        uses: mislav/bump-homebrew-formula-action@v1
+        with:
+          formula-name: meilisearch
+        env:
+          COMMITTER_TOKEN: ${{ secrets.HOMEBREW_COMMITTER_TOKEN }}

.github/workflows/publish-docker-latest.yml

@@ -0,0 +1,22 @@
+---
+on:
+  release:
+    types: [released]
+
+name: Publish latest image to Docker Hub
+
+jobs:
+  build:
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - name: Check if current release is latest
+        run: echo "##[set-output name=is_latest;]$(sh .github/is-latest-release.sh)"
+        id: release
+      - name: Publish to Registry
+        if: steps.release.outputs.is_latest == 'true'
+        uses: elgohr/Publish-Docker-Github-Action@master
+        with:
+          name: getmeili/meilisearch
+          username: ${{ secrets.DOCKER_USERNAME }}
+          password: ${{ secrets.DOCKER_PASSWORD }}

.github/workflows/publish-docker-tag.yml

@@ -0,0 +1,22 @@
+---
+on:
+  push:
+    tags:
+      - '*'
+
+name: Publish tagged image to Docker Hub
+
+jobs:
+  build:
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - name: Publish to Registry
+        uses: elgohr/Publish-Docker-Github-Action@master
+        env:
+          COMMIT_SHA: ${{ github.sha }}
+        with:
+          name: getmeili/meilisearch
+          username: ${{ secrets.DOCKER_USERNAME }}
+          password: ${{ secrets.DOCKER_PASSWORD }}
+          tag_names: true

.github/workflows/release-drafter.yml

@@ -0,0 +1,16 @@
+name: Release Drafter
+
+on:
+  push:
+    branches:
+      - main
+
+jobs:
+  update_release_draft:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: release-drafter/release-drafter@v5
+        with:
+          config-name: release-draft-template.yml
+        env:
+          GITHUB_TOKEN: ${{ secrets.RELEASE_DRAFTER_TOKEN }}

.github/workflows/rust.yml

@@ -0,0 +1,86 @@
+name: Rust
+
+on:
+  workflow_dispatch:
+  pull_request:
+  push:
+    # trying and staging branches are for Bors config
+    branches:
+      - trying
+      - staging
+
+env:
+  CARGO_TERM_COLOR: always
+
+jobs:
+  tests:
+    name: Tests on ${{ matrix.os }}
+    runs-on: ${{ matrix.os }}
+    strategy:
+      fail-fast: false
+      matrix:
+        os: [ubuntu-18.04, macos-latest, windows-latest]
+    steps:
+    - uses: actions/checkout@v2
+    - name: Cache dependencies
+      uses: actions/cache@v2
+      with:
+        path: |
+          ~/.cargo
+          ./target
+        key: ${{ matrix.os }}-${{ hashFiles('Cargo.lock') }}
+    - name: Run cargo check without any default features
+      uses: actions-rs/cargo@v1
+      with:
+        command: build
+        args: --locked --release --no-default-features
+    - name: Run cargo test
+      uses: actions-rs/cargo@v1
+      with:
+        command: test
+        args: --locked --release
+
+  clippy:
+    name: Run Clippy
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - name: Cache dependencies
+        uses: actions/cache@v2
+        with:
+          path: |
+            ~/.cargo
+            ./target
+          key: ${{ matrix.os }}-${{ hashFiles('Cargo.lock') }}
+      - uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: stable
+          override: true
+          components: clippy
+      - name: Run cargo clippy
+        uses: actions-rs/cargo@v1
+        with:
+          command: clippy
+          args: --all-targets -- --deny warnings
+
+  fmt:
+    name: Run Rustfmt
+    runs-on: ubuntu-18.04
+    steps:
+      - uses: actions/checkout@v2
+      - name: Cache dependencies
+        uses: actions/cache@v2
+        with:
+          path: |
+            ~/.cargo
+            ./target
+          key: ${{ matrix.os }}-${{ hashFiles('Cargo.lock') }}
+      - uses: actions-rs/toolchain@v1
+        with:
+          profile: minimal
+          toolchain: nightly
+          override: true
+          components: rustfmt
+      - name: Run cargo fmt
+        run: cargo fmt --all -- --check

.gitignore

@@ -1,7 +1,9 @@
-/rocksdb
 /target
-/Cargo.lock
-**/*.rs.bk
 **/*.csv
 **/*.json_lines
-**/*.rdb
+**/*.rs.bk
+/*.mdb
+/query-history.txt
+/data.ms
+/snapshots
+/dumps

.travis.yml

@@ -1,22 +0,0 @@
-language: rust
-
-cache: cargo
-
-branches:
-  only:
-    - master
-
-matrix:
-  fast_finish: true
-  include:
-
-  # Test crates on their minimum Rust versions.
-  - rust: 1.32.0
-    name: "meilidb on 1.32.0"
-    script: ./ci/meilidb.sh
-
-  # Test crates on nightly Rust.
-  - rust: nightly
-    name: "meilidb on nightly"
-    script: ./ci/meilidb.sh
-

CODE_OF_CONDUCT.md

@@ -0,0 +1,76 @@
+# Contributor Covenant Code of Conduct
+
+## Our Pledge
+
+In the interest of fostering an open and welcoming environment, we as
+contributors and maintainers pledge to making participation in our project and
+our community a harassment-free experience for everyone, regardless of age, body
+size, disability, ethnicity, sex characteristics, gender identity and expression,
+level of experience, education, socio-economic status, nationality, personal
+appearance, race, religion, or sexual identity and orientation.
+
+## Our Standards
+
+Examples of behavior that contributes to creating a positive environment
+include:
+
+* Using welcoming and inclusive language
+* Being respectful of differing viewpoints and experiences
+* Gracefully accepting constructive criticism
+* Focusing on what is best for the community
+* Showing empathy towards other community members
+
+Examples of unacceptable behavior by participants include:
+
+* The use of sexualized language or imagery and unwelcome sexual attention or
+ advances
+* Trolling, insulting/derogatory comments, and personal or political attacks
+* Public or private harassment
+* Publishing others' private information, such as a physical or electronic
+ address, without explicit permission
+* Other conduct which could reasonably be considered inappropriate in a
+ professional setting
+
+## Our Responsibilities
+
+Project maintainers are responsible for clarifying the standards of acceptable
+behavior and are expected to take appropriate and fair corrective action in
+response to any instances of unacceptable behavior.
+
+Project maintainers have the right and responsibility to remove, edit, or
+reject comments, commits, code, wiki edits, issues, and other contributions
+that are not aligned to this Code of Conduct, or to ban temporarily or
+permanently any contributor for other behaviors that they deem inappropriate,
+threatening, offensive, or harmful.
+
+## Scope
+
+This Code of Conduct applies both within project spaces and in public spaces
+when an individual is representing the project or its community. Examples of
+representing a project or community include using an official project e-mail
+address, posting via an official social media account, or acting as an appointed
+representative at an online or offline event. Representation of a project may be
+further defined and clarified by project maintainers.
+
+## Enforcement
+
+Instances of abusive, harassing, or otherwise unacceptable behavior may be
+reported by contacting the project team at bonjour@meilisearch.com. All
+complaints will be reviewed and investigated and will result in a response that
+is deemed necessary and appropriate to the circumstances. The project team is
+obligated to maintain confidentiality with regard to the reporter of an incident.
+Further details of specific enforcement policies may be posted separately.
+
+Project maintainers who do not follow or enforce the Code of Conduct in good
+faith may face temporary or permanent repercussions as determined by other
+members of the project's leadership.
+
+## Attribution
+
+This Code of Conduct is adapted from the [Contributor Covenant][homepage], version 1.4,
+available at https://www.contributor-covenant.org/version/1/4/code-of-conduct.html
+
+[homepage]: https://www.contributor-covenant.org
+
+For answers to common questions about this code of conduct, see
+https://www.contributor-covenant.org/faq

CONTRIBUTING.md

@@ -0,0 +1,112 @@
+# Contributing
+
+First, thank you for contributing to MeiliSearch! The goal of this document is to
+provide everything you need to start contributing to MeiliSearch. The
+following TOC is sorted progressively, starting with the basics and
+expanding into more specifics.
+
+<!-- MarkdownTOC autolink="true" style="ordered" indent="   " -->
+
+1. [Assumptions](#assumptions)
+1. [Your First Contribution](#your-first-contribution)
+1. [Change Control](#change-control)
+   1. [Git Branches](#git-branches)
+   1. [Git Commits](#git-commits)
+      1. [Style](#style)
+   1. [Github Pull Requests](#github-pull-requests)
+      1. [Reviews & Approvals](#reviews--approvals)
+      1. [Merge Style](#merge-style)
+   1. [CI](#ci)
+1. [Development](#development)
+   1. [Setup](#setup)
+   1. [Testing](#testing)
+   1. [Benchmarking](#benchmarking--profiling)
+1. [Humans](#humans)
+   1. [Documentation](#documentation)
+   1. [Changelog](#changelog)
+
+<!-- /MarkdownTOC -->
+
+## Assumptions
+
+1. **You're familiar with [Github](https://github.com) and the [pull request](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests/about-pull-requests)
+   workflow.**
+2. **You've read the MeiliSearch [docs](https://docs.meilisearch.com).**
+3. **You know about the [MeiliSearch community](https://docs.meilisearch.com/learn/what_is_meilisearch/contact.html).
+   Please use this for help.**
+
+## Your First Contribution
+
+1. Ensure your change has an issue! Find an
+   [existing issue](https://github.com/meilisearch/meilisearch/issues/) or [open a new issue](https://github.com/meilisearch/meilisearch/issues/new).
+   * This is where you can get a feel if the change will be accepted or not.
+2. Once approved, [fork the MeiliSearch repository](https://help.github.com/en/github/getting-started-with-github/fork-a-repo) in your own
+   Github account.
+3. [Create a new Git branch](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests/creating-and-deleting-branches-within-your-repository)
+4. Review the MeiliSearch [workflow](#workflow) and [development](#development).
+5. Make your changes.
+6. [Submit the branch as a pull request](https://help.github.com/en/github/collaborating-with-issues-and-pull-requests/creating-a-pull-request-from-a-fork) to the main MeiliSearch
+   repo. A MeiliSearch team member should comment and/or review your pull request
+   with a few days. Although, depending on the circumstances, it may take
+   longer.
+
+## Change Control
+
+### Git Branches
+
+_All_ changes must be made in a branch and submitted as [pull requests](#pull-requests).
+MeiliSearch does not adopt any type of branch naming style, but please use something
+descriptive of your changes.
+
+### Git Commits
+
+#### Style
+
+Please ensure your commits are small and focused; they should tell a story of
+your change. This helps reviewers to follow your changes, especially for more
+complex changes.
+
+Familiarise yourself with [How to Write a Git Commit Message](https://chris.beams.io/posts/git-commit/).
+
+### Github Pull Requests
+
+Once your changes are ready you must submit your branch as a pull request.
+
+#### Reviews & Approvals
+
+All pull requests must be reviewed and approved by at least one MeiliSearch team
+member.
+
+#### Merge Style
+
+All pull requests are squashed and merged. We generally discourage large pull
+requests that are over 300-500 lines of diff. If you would like to propose
+a change that is larger we suggest coming onto our chat channel and
+discuss it with one of our engineers. This way we can talk through the
+solution and discuss if a change that large is even needed! This overall
+will produce a quicker response to the change and likely produce code that
+aligns better with our process.
+
+## Development
+
+### Setup
+
+See the [MeiliSearch Docs](https://docs.meilisearch.com/reference/features/installation.html) for how to set up a development environment.
+
+### Benchmarking & Profiling
+
+We do not yet do any benchmarking, nor have we formalised our profiling. If you'd like to work on this please get in touch!
+
+## Humans
+
+After making your change, you'll want to prepare it for MeiliSearch users (mostly humans). This usually entails updating documentation and announcing your feature.
+
+### Documentation
+
+Documentation is very important to MeiliSearch. All contributions that
+alter user-facing behavior MUST include documentation changes. Please see
+[GitHub.com/meilisearch/documentation](https://github.com/meilisearch/documentation) for more info.
+
+### Changelog
+
+Until we have guidelines in place, updating the [`Changelog`](/CHANGELOG.md) is solely the responsibility of MeiliSearch team members.

Cargo.toml

@@ -1,55 +1,8 @@
-[package]
-edition = "2018"
-name = "meilidb"
-version = "0.3.2"
-authors = ["Kerollmops <renault.cle@gmail.com>"]
-
-[dependencies]
-arc-swap = "0.3.7"
-bincode = "1.1.2"
-byteorder = "1.3.1"
-fst = "0.3.3"
-hashbrown = { version = "0.1.8", features = ["serde"] }
-lazy_static = "1.2.0"
-levenshtein_automata = { version = "0.1.1", features = ["fst_automaton"] }
-linked-hash-map = { version = "0.5.1", features = ["serde_impl"] }
-lockfree = "0.5.1"
-log = "0.4.6"
-rayon = "1.0.3"
-sdset = "0.3.1"
-serde = "1.0.88"
-serde_derive = "1.0.88"
-serde_json = { version = "1.0.38", features = ["preserve_order"] }
-size_format = "1.0.2"
-slice-group-by = "0.2.4"
-unidecode = "0.3.0"
-
-[dependencies.toml]
-git = "https://github.com/Kerollmops/toml-rs.git"
-features = ["preserve_order"]
-rev = "0372ba6"
-
-[dependencies.rocksdb]
-git = "https://github.com/pingcap/rust-rocksdb.git"
-rev = "306e201"
-
-[features]
-default = ["simd"]
-i128 = ["bincode/i128", "byteorder/i128"]
-portable = ["rocksdb/portable"]
-simd = ["rocksdb/sse"]
-nightly = ["hashbrown/nightly", "slice-group-by/nightly"]
-
-[dev-dependencies]
-csv = "1.0.5"
-env_logger = "0.6.0"
-jemallocator = "0.1.9"
-quickcheck = "0.8.2"
-rand = "0.6.5"
-rand_xorshift = "0.1.1"
-structopt = "0.2.14"
-tempfile = "3.0.7"
-termcolor = "1.0.4"
+[workspace]
+members = [
+    "meilisearch-http",
+    "meilisearch-error",
+]
 
 [profile.release]
 debug = true

Dockerfile

@@ -0,0 +1,45 @@
+# Compile
+FROM    alpine:3.14 AS compiler
+
+RUN     apk update --quiet
+RUN     apk add curl
+RUN     apk add build-base
+
+RUN     curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y
+
+WORKDIR /meilisearch
+
+COPY    Cargo.lock .
+COPY    Cargo.toml .
+
+COPY    meilisearch-error/Cargo.toml meilisearch-error/
+COPY    meilisearch-http/Cargo.toml meilisearch-http/
+
+ENV     RUSTFLAGS="-C target-feature=-crt-static"
+
+# Create dummy main.rs files for each workspace member to be able to compile all the dependencies
+RUN     find . -type d -name "meilisearch-*" | xargs -I{} sh -c 'mkdir {}/src; echo "fn main() { }" > {}/src/main.rs;'
+# Use `cargo build` instead of `cargo vendor` because we need to not only download but compile dependencies too
+RUN     $HOME/.cargo/bin/cargo build --release
+# Cleanup dummy main.rs files
+RUN     find . -path "*/src/main.rs" -delete
+
+ARG     COMMIT_SHA
+ARG     COMMIT_DATE
+ENV     COMMIT_SHA=${COMMIT_SHA} COMMIT_DATE=${COMMIT_DATE}
+
+COPY    . .
+RUN     $HOME/.cargo/bin/cargo build --release
+
+# Run
+FROM    alpine:3.14
+
+RUN     apk add -q --no-cache libgcc tini
+
+COPY    --from=compiler /meilisearch/target/release/meilisearch .
+
+ENV     MEILI_HTTP_ADDR 0.0.0.0:7700
+EXPOSE  7700/tcp
+
+ENTRYPOINT ["tini", "--"]
+CMD     ./meilisearch

LICENSE

@@ -1,6 +1,6 @@
 MIT License
 
-Copyright (c) 2018 Clément Renault
+Copyright (c) 2019-2021 Meili SAS
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -1,60 +1,201 @@
-# MeiliDB
+<p align="center">
+  <img src="assets/logo.svg" alt="MeiliSearch" width="200" height="200" />
+</p>
 
-[![Build Status](https://travis-ci.org/Kerollmops/MeiliDB.svg?branch=master)](https://travis-ci.org/Kerollmops/MeiliDB)
-[![dependency status](https://deps.rs/repo/github/Kerollmops/MeiliDB/status.svg)](https://deps.rs/repo/github/Kerollmops/MeiliDB)
-[![License](https://img.shields.io/github/license/Kerollmops/MeiliDB.svg)](https://github.com/Kerollmops/MeiliDB)
-[![Rust 1.31+](https://img.shields.io/badge/rust-1.31+-lightgray.svg)](
-https://www.rust-lang.org)
+<h1 align="center">MeiliSearch</h1>
 
-A _full-text search database_ using a key-value store internally.
+<h4 align="center">
+  <a href="https://www.meilisearch.com">Website</a> |
+  <a href="https://roadmap.meilisearch.com/tabs/1-under-consideration">Roadmap</a> |
+  <a href="https://blog.meilisearch.com">Blog</a> |
+  <a href="https://fr.linkedin.com/company/meilisearch">LinkedIn</a> |
+  <a href="https://twitter.com/meilisearch">Twitter</a> |
+  <a href="https://docs.meilisearch.com">Documentation</a> |
+  <a href="https://docs.meilisearch.com/faq/">FAQ</a>
+</h4>
 
-It uses [RocksDB](https://github.com/facebook/rocksdb) as the internal key-value store. The key-value store allows us to handle updates and queries with small memory and CPU overheads.
+<p align="center">
+  <a href="https://github.com/meilisearch/MeiliSearch/actions"><img src="https://github.com/meilisearch/MeiliSearch/workflows/Cargo%20test/badge.svg" alt="Build Status"></a>
+  <a href="https://deps.rs/repo/github/meilisearch/MeiliSearch"><img src="https://deps.rs/repo/github/meilisearch/MeiliSearch/status.svg" alt="Dependency status"></a>
+  <a href="https://github.com/meilisearch/MeiliSearch/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-informational" alt="License"></a>
+  <a href="https://slack.meilisearch.com"><img src="https://img.shields.io/badge/slack-MeiliSearch-blue.svg?logo=slack" alt="Slack"></a>
+  <a href="https://github.com/meilisearch/MeiliSearch/discussions" alt="Discussions"><img src="https://img.shields.io/badge/github-discussions-red" /></a>
+  <a href="https://app.bors.tech/repositories/26457"><img src="https://bors.tech/images/badge_small.svg" alt="Bors enabled"></a>
+</p>
 
-You can [read the deep dive](deep-dive.md) if you want more information on the engine, it describes the whole process of generating updates and handling queries or you can take a look at the [typos and ranking rules](typos-ranking-rules.md) if you want to know the default rules used to sort the documents.
+<p align="center">⚡ Lightning Fast, Ultra Relevant, and Typo-Tolerant Search Engine 🔍</p>
 
-We will be proud if you submit issues and pull requests. You can help to grow this project and start contributing by checking [issues tagged "good-first-issue"](https://github.com/Kerollmops/MeiliDB/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22). It is a good start!
+**MeiliSearch** is a powerful, fast, open-source, easy to use and deploy search engine. Both searching and indexing are highly customizable. Features such as typo-tolerance, filters, and synonyms are provided out-of-the-box.
+For more information about features go to [our documentation](https://docs.meilisearch.com/).
 
-The project is only a library yet. It means that there is no binary provided yet. To get started, you can check the examples wich are made to work with the data located in the `misc/` folder.
+<p align="center">
+  <img src="assets/trumen-fast.gif" alt="Web interface gif" />
+</p>
 
-MeiliDB will be a binary in a near future so you will be able to use it as a database out-of-the-box. We should be able to query it using a [to-be-defined](https://github.com/Kerollmops/MeiliDB/issues/38) protocol. This is our current goal, [see the milestones](https://github.com/Kerollmops/MeiliDB/milestones). In the end, the binary will be a bunch of network protocols and wrappers around the library - which will also be published on [crates.io](https://crates.io). Both the binary and the library will follow the same update cycle.
+## ✨ Features
+* Search-as-you-type experience (answers < 50 milliseconds)
+* Full-text search
+* Typo tolerant (understands typos and misspelling)
+* Faceted search and filters
+* Supports hanzi (Chinese characters)
+* Supports synonyms
+* Easy to install, deploy, and maintain
+* Whole documents are returned
+* Highly customizable
+* RESTful API
 
+## Getting started
 
+### Deploy the Server
 
-## Performances
-
-With a database composed of _100 353_ documents with _352_ attributes each and _3_ of them indexed.
-So more than _300 000_ fields indexed for _35 million_ stored we can handle more than _2.8k req/sec_ with an average response time of _9 ms_ on an Intel i7-7700 (8) @ 4.2GHz.
-
-Requests are made using [wrk](https://github.com/wg/wrk) and scripted to simulate real users queries.
-
-```
-Running 10s test @ http://localhost:2230
-  2 threads and 25 connections
-  Thread Stats   Avg      Stdev     Max   +/- Stdev
-    Latency     9.52ms    7.61ms  99.25ms   84.58%
-    Req/Sec     1.41k   119.11     1.78k    64.50%
-  28080 requests in 10.01s, 7.42MB read
-Requests/sec:   2806.46
-Transfer/sec:    759.17KB
-```
-
-### Notes
-
-The default Rust allocator has recently been [changed to use the system allocator](https://github.com/rust-lang/rust/pull/51241/).
-We have seen much better performances when [using jemalloc as the global allocator](https://github.com/alexcrichton/jemallocator#documentation).
-
-## Usage and examples
-
-MeiliDB runs with an index like most search engines.
-So to test the library you can create one by indexing a simple csv file.
+#### Homebrew (Mac OS)
 
 ```bash
-cargo run --release --example create-database -- test.mdb misc/kaggle.csv --schema schema-example.toml
+brew update && brew install meilisearch
+meilisearch
 ```
 
-Once the command is executed, the index should be in the `test.mdb` folder. You are now able to run the `query-database` example and play with MeiliDB.
+#### Docker
 
 ```bash
-cargo run --release --example query-database -- test.mdb -n 10 id title
+docker run -p 7700:7700 -v "$(pwd)/data.ms:/data.ms" getmeili/meilisearch
 ```
 
+#### Try MeiliSearch in our Sandbox
+
+Create a MeiliSearch instance in [MeiliSearch Sandbox](https://sandbox.meilisearch.com/). This instance is free, and will be active for 48 hours.
+
+#### Run on Digital Ocean
+
+[![DigitalOcean Marketplace](assets/do-btn-blue.svg)](https://marketplace.digitalocean.com/apps/meilisearch?action=deploy&refcode=7c67bd97e101)
+
+#### Deploy on Platform.sh
+
+<a href="https://console.platform.sh/projects/create-project?template=https://raw.githubusercontent.com/platformsh/template-builder/master/templates/meilisearch/.platform.template.yaml&utm_content=meilisearch&utm_source=github&utm_medium=button&utm_campaign=deploy_on_platform">
+    <img src="https://platform.sh/images/deploy/lg-blue.svg" alt="Deploy on Platform.sh" width="180px" />
+</a>
+
+#### APT (Debian & Ubuntu)
+
+```bash
+echo "deb [trusted=yes] https://apt.fury.io/meilisearch/ /" > /etc/apt/sources.list.d/fury.list
+apt update && apt install meilisearch-http
+meilisearch
+```
+
+#### Download the binary (Linux & Mac OS)
+
+```bash
+curl -L https://install.meilisearch.com | sh
+./meilisearch
+```
+
+#### Compile and run it from sources
+
+If you have the latest stable Rust toolchain installed on your local system, clone the repository and change it to your working directory.
+
+```bash
+git clone https://github.com/meilisearch/MeiliSearch.git
+cd MeiliSearch
+cargo run --release
+```
+
+### Create an Index and Upload Some Documents
+
+Let's create an index! If you need a sample dataset, use [this movie database](https://www.notion.so/meilisearch/A-movies-dataset-to-test-Meili-1cbf7c9cfa4247249c40edfa22d7ca87#b5ae399b81834705ba5420ac70358a65). You can also find it in the `datasets/` directory.
+
+```bash
+curl -L 'https://bit.ly/2PAcw9l' -o movies.json
+```
+
+Now, you're ready to index some data.
+
+```bash
+curl -i -X POST 'http://127.0.0.1:7700/indexes/movies/documents' \
+  --header 'content-type: application/json' \
+  --data-binary @movies.json
+```
+
+### Search for Documents
+
+#### In command line
+
+The search engine is now aware of your documents and can serve those via an HTTP server.
+
+The [`jq` command-line tool](https://stedolan.github.io/jq/) can greatly help you read the server responses.
+
+```bash
+curl 'http://127.0.0.1:7700/indexes/movies/search?q=botman+robin&limit=2' | jq
+```
+
+```json
+{
+  "hits": [
+    {
+      "id": "415",
+      "title": "Batman & Robin",
+      "poster": "https://image.tmdb.org/t/p/w1280/79AYCcxw3kSKbhGpx1LiqaCAbwo.jpg",
+      "overview": "Along with crime-fighting partner Robin and new recruit Batgirl, Batman battles the dual threat of frosty genius Mr. Freeze and homicidal horticulturalist Poison Ivy. Freeze plans to put Gotham City on ice, while Ivy tries to drive a wedge between the dynamic duo.",
+      "release_date": 866768400
+    },
+    {
+      "id": "411736",
+      "title": "Batman: Return of the Caped Crusaders",
+      "poster": "https://image.tmdb.org/t/p/w1280/GW3IyMW5Xgl0cgCN8wu96IlNpD.jpg",
+      "overview": "Adam West and Burt Ward returns to their iconic roles of Batman and Robin. Featuring the voices of Adam West, Burt Ward, and Julie Newmar, the film sees the superheroes going up against classic villains like The Joker, The Riddler, The Penguin and Catwoman, both in Gotham City… and in space.",
+      "release_date": 1475888400
+    }
+  ],
+  "nbHits": 8,
+  "exhaustiveNbHits": false,
+  "query": "botman robin",
+  "limit": 2,
+  "offset": 0,
+  "processingTimeMs": 2
+}
+```
+
+#### Use the Web Interface
+
+We also deliver an **out-of-the-box [web interface](https://github.com/meilisearch/mini-dashboard)** in which you can test MeiliSearch interactively.
+
+You can access the web interface in your web browser at the root of the server. The default URL is [http://127.0.0.1:7700](http://127.0.0.1:7700). All you need to do is open your web browser and enter MeiliSearch’s address to visit it. This will lead you to a web page with a search bar that will allow you to search in the selected index.
+
+| [See the gif above](#demo)
+
+## Documentation
+
+Now that your MeiliSearch server is up and running, you can learn more about how to tune your search engine in [the documentation](https://docs.meilisearch.com).
+
+## Contributing
+
+Hey! We're glad you're thinking about contributing to MeiliSearch! Feel free to pick an [issue labeled as `good first issue`](https://github.com/meilisearch/MeiliSearch/issues?q=is%3Aissue+is%3Aopen+label%3A%22good+first+issue%22), and to ask any question you need. Some points might not be clear and we are available to help you!
+
+Also, we recommend following the [CONTRIBUTING](./CONTRIBUTING.md) to create your PR.
+
+## Core engine and tokenizer
+
+The code in this repository is only concerned with managing multiple indexes, handling the update store, and exposing an HTTP API.
+
+Search and indexation are the domain of our core engine, [`milli`](https://github.com/meilisearch/milli), while tokenization is handled by [our `tokenizer` library](https://github.com/meilisearch/tokenizer/).
+## Telemetry
+
+MeiliSearch collects anonymous data regarding general usage.
+This helps us better understand developers' usage of MeiliSearch features.
+
+To see what information we're retrieving, please see the complete list [on the dedicated issue](https://github.com/meilisearch/MeiliSearch/issues/720).
+
+This program is optional, you can disable these analytics by using the `MEILI_NO_ANALYTICS` env variable.
+
+## Feature request
+
+The feature requests are not managed in this repository. Please visit our [dedicated repository](https://github.com/meilisearch/product) to see our work about the MeiliSearch product.
+
+If you have a feature request or any feedback about an existing feature, please open [a discussion](https://github.com/meilisearch/product/discussions).
+Also, feel free to participate in the current discussions, we are looking forward to reading your comments.
+
+## 💌 Contact
+
+Please visit [this page](https://docs.meilisearch.com/learn/what_is_meilisearch/contact.html#contact-us).
+
+MeiliSearch is developed by [Meili](https://www.meilisearch.com), a young company. To know more about us, you can [read our blog](https://blog.meilisearch.com). Any suggestion or feedback is highly appreciated. Thank you for your support!

assets/do-btn-blue.svg

@@ -0,0 +1,23 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<svg width="200px" height="42px" viewBox="0 0 200 42" version="1.1" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink">
+    <!-- Generator: Sketch 52.5 (67469) - http://www.bohemiancoding.com/sketch -->
+    <title>do-btn-blue</title>
+    <desc>Created with Sketch.</desc>
+    <g id="Page-1" stroke="none" stroke-width="1" fill="none" fill-rule="evenodd">
+        <g id="Partner-welcome-kit-Copy-3" transform="translate(-651.000000, -762.000000)">
+            <g id="do-btn-blue" transform="translate(651.000000, 763.000000)">
+                <rect id="Rectangle-Copy" fill="#0069FF" x="0" y="0" width="200" height="40" rx="6"></rect>
+                <path d="M45,0 L45,40" id="Line-2" stroke="#FFFFFF" stroke-linecap="square"></path>
+                <g id="DO_Logo_horizontal_blue-Copy" transform="translate(13.000000, 11.000000)" fill="#FFFFFF">
+                    <path d="M10.0098493,20 L10.0098493,16.1262429 C14.12457,16.1262429 17.2897398,12.0548452 15.7269372,7.74627862 C15.1334679,6.14538921 13.8674,4.86072487 12.2650328,4.28756693 C7.952489,2.72620566 3.87733294,5.88845634 3.87733294,9.99938223 C3.87733294,9.99938223 3.87733294,9.99938223 3.87733294,9.99938223 L0,9.99938223 C0,3.45747613 6.3303395,-1.64165309 13.1948014,0.492866119 C16.2017127,1.42177726 18.57559,3.81322933 19.5053586,6.79760341 C21.6418482,13.6754986 16.5577943,20 10.0098493,20 Z" id="XMLID_49_"></path>
+                    <polygon id="XMLID_47_" points="9.56521739 15.6521739 6.08695652 15.6521739 6.08695652 12.173913 6.08695652 12.173913 9.56521739 12.173913 9.56521739 12.173913"></polygon>
+                    <polygon id="XMLID_46_" points="6.08695652 19.1304348 3.47826087 19.1304348 3.47826087 19.1304348 3.47826087 16.5217391 6.08695652 16.5217391"></polygon>
+                    <polygon id="XMLID_45_" points="3.47826087 16.5217391 0.869565217 16.5217391 0.869565217 16.5217391 0.869565217 13.9130435 0.869565217 13.9130435 3.47826087 13.9130435 3.47826087 13.9130435"></polygon>
+                </g>
+                <text id="Create-a-Droplet-Copy" font-family="Sailec-Medium, Sailec" font-size="16" font-weight="400" fill="#FFFFFF">
+                    <tspan x="58" y="26">Create a Droplet</tspan>
+                </text>
+            </g>
+        </g>
+    </g>
+</svg>

assets/logo.svg

@@ -0,0 +1,17 @@
+<svg width="360" height="360" viewBox="0 0 360 360" fill="none" xmlns="http://www.w3.org/2000/svg">
+<g id="logo_main">
+<rect id="Rectangle" x="107.333" y="0.150146" width="274.315" height="274.315" rx="98.8334" transform="rotate(23 107.333 0.150146)" fill="url(#paint0_linear)"/>
+<path id="Rectangle_2" fill-rule="evenodd" clip-rule="evenodd" d="M61.3296 230.199C46.2224 194.608 38.6688 176.813 38.208 160.329C37.5286 136.025 47.0175 112.539 64.3891 95.5282C76.1718 83.9904 93.9669 76.4368 129.557 61.3296C165.147 46.2224 182.943 38.6688 199.427 38.208C223.731 37.5286 247.217 47.0175 264.228 64.3891C275.766 76.1718 283.319 93.9669 298.426 129.557C313.534 165.147 321.087 182.943 321.548 199.427C322.227 223.731 312.738 247.217 295.367 264.228C283.584 275.766 265.789 283.319 230.199 298.426C194.608 313.534 176.813 321.087 160.329 321.548C136.025 322.227 112.539 312.738 95.5282 295.367C83.9903 283.584 76.4368 265.789 61.3296 230.199Z" fill="url(#paint1_linear)"/>
+<path id="m" fill-rule="evenodd" clip-rule="evenodd" d="M219.568 130.748C242.363 130.748 259.263 147.451 259.263 174.569V229.001H227.232V179.678C227.232 166.119 220.747 159.634 210.136 159.634C205.223 159.634 200.311 161.796 195.595 167.494C195.791 169.852 195.988 172.21 195.988 174.569V229.001H164.154V179.678C164.154 166.119 157.472 159.634 147.057 159.634C142.145 159.634 137.429 161.992 132.712 168.084V229.001H100.878V133.695H132.712V139.394C139.197 133.892 145.878 130.748 156.49 130.748C168.477 130.748 178.695 135.267 185.769 143.52C195.791 134.678 205.42 130.748 219.568 130.748Z" fill="white"/>
+</g>
+<defs>
+<linearGradient id="paint0_linear" x1="-13.6248" y1="129.208" x2="244.49" y2="403.522" gradientUnits="userSpaceOnUse">
+<stop stop-color="#E41359"/>
+<stop offset="1" stop-color="#F23C79"/>
+</linearGradient>
+<linearGradient id="paint1_linear" x1="11.0088" y1="111.65" x2="111.65" y2="348.747" gradientUnits="userSpaceOnUse">
+<stop stop-color="#24222F"/>
+<stop offset="1" stop-color="#2B2937"/>
+</linearGradient>
+</defs>
+</svg>

bors.toml

@@ -0,0 +1,9 @@
+status = [
+    'Tests on ubuntu-18.04',
+    'Tests on macos-latest',
+    'Tests on windows-latest',
+    'Run Clippy',
+    'Run Rustfmt'
+]
+# 3 hours timeout
+timeout-sec = 10800

ci/meilidb.sh

@@ -1,15 +0,0 @@
-#!/bin/bash
-
-cd "$(dirname "$0")"/..
-set -ex
-
-export RUSTFLAGS="-D warnings"
-
-cargo check --no-default-features
-cargo check --bins --examples --tests
-cargo test
-
-if [[ "$TRAVIS_RUST_VERSION" == "nightly" ]]; then
-    cargo check --no-default-features --features nightly
-    cargo test --features nightly
-fi

datasets/movies/README.md

@@ -0,0 +1 @@
+_datas in movies.json are from https://www.themoviedb.org/_

deep-dive.md

@@ -1,140 +0,0 @@
-# A deep dive in MeiliDB
-
-On the 9 of december 2018.
-
-MeiliDB is a full text search engine based on a final state transducer named [fst](https://github.com/BurntSushi/fst) and a key-value store named [RocksDB](https://github.com/facebook/rocksdb). The goal of a search engine is to store data and to respond to queries as accurate and fast as possible. To achieve this it must save the data as an [inverted index](https://en.wikipedia.org/wiki/Inverted_index).
-
-
-
-<!-- MarkdownTOC autolink="true" -->
-
-- [Where is the data stored?](#where-is-the-data-stored)
-- [What does the key-value store contains?](#what-does-the-key-value-store-contains)
-    - [The blob type](#the-blob-type)
-        - [A final state transducer](#a-final-state-transducer)
-        - [Document indexes](#document-indexes)
-        - [Document ids](#document-ids)
-    - [The schema](#the-schema)
-    - [Document attributes](#document-attributes)
-- [How is an update handled?](#how-is-an-update-handled)
-    - [The merge operation is CPU consuming](#the-merge-operation-is-cpu-consuming)
-- [How is a request processed?](#how-is-a-request-processed)
-    - [Query lexemes](#query-lexemes)
-    - [Automatons and query index](#automatons-and-query-index)
-    - [Sort by criteria](#sort-by-criteria)
-    - [Retrieve original documents](#retrieve-original-documents)
-
-<!-- /MarkdownTOC -->
-
-## Where is the data stored?
-
-MeiliDB is entirely backed by a key-value store like any good database (i.e. Postgres, MySQL). This brings a great flexibility in the way documents can be stored and updates handled along time.
-
-[RocksDB brings some](https://rocksdb.org/blog/2015/02/27/write-batch-with-index.html) of the [A.C.I.D. properties](https://en.wikipedia.org/wiki/ACID_(computer_science)) to help us be sure the saved data is consistent, for example we use SST files and the key-value store ability to load them in one time to manage updates.
-
-Note that the SST file have the same restriction as the fst, it needs its keys to be added in order at creation.
-
-
-
-## What does the key-value store contains?
-
-It contain the blob, the schema and the documents stored attributes.
-
-### The blob type
-
-[The Blob type](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/blob/mod.rs#L16-L19) is a data structure that indicate if an update is a positive or a negative one. In the case where the update is considered positive, the blob will contain [an fst map and the document indexes](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/blob/positive/blob.rs#L15-L18) associated. In the other case it will only contain [all the document ids](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/blob/negative/blob.rs#L12-L14) that must be considered removed.
-
-The Blob type [is stored under the "*data-index*" entry](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/update/positive/update.rs#L497-L499) and marked as [a merge operation](https://github.com/facebook/rocksdb/wiki/Merge-Operator-Implementation) in the key-value store.
-
-#### A final state transducer
-
-_...also abbreviated fst_
-
-This is the first entry point of the engine, you can read more about how it work with the beautiful blog post of @BurntSushi, [Index 1,600,000,000 Keys with Automata and Rust](https://blog.burntsushi.net/transducers/).
-
-To make it short it is a powerful way to store all the words that are present in the indexed documents. You construct it by giving it all the words you want to index associated with a value that, for the moment, can only be an `u64`. When you want to search in it you can provide any automaton you want, in MeiliDB [a custom levenshtein automaton](https://github.com/tantivy-search/levenshtein-automata/) is used.
-
-Note that the number under each word is auto-incremental, each new word have a new number that is greater than the prevous one.
-
-Another powerful feature of `fst` is that it can nearly avoid using RAM and be streamed to disk for example, the problem is that the keys must be always added in lexicographic order, so you must sort them before, for the moment MeiliDB uses a [BTreeMap](https://github.com/Kerollmops/raptor-rs/blob/8abdb0a228e2808fe1814a6a0641a4b72d158579/src/metadata/doc_indexes.rs#L107-L112).
-
-#### Document indexes
-
-As it has been specified, the `fst` can only store a number corresponding to a word, an `u64`, but the goal of the search engine is to retrieve a match in a document when a query is made. You want it to return some sort of position in an attribute in a document, an information about where the given word match.
-
-To make it possible, a custom data structure has been developed, the document indexes is composed of two arrays, the ranges array and all the docindexes corresponding to a given range, each range identify the word number. The [DocIndexes](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/data/doc_indexes.rs#L23) type is designed to be streamed when constructed, consumming a minimum amount of ram like the fst. Another advantage is that the slices are accessible in `O(1)` when you know the word associated number.
-
-#### Document ids
-
-This is a simple ordered list of all documents ids which must be considered deleted. It is used with [the sdset library](https://docs.rs/sdset/0.3.0/sdset/duo/struct.DifferenceByKey.html), the docindexes and the `DifferenceByKey` operation builder when merging blobs.
-
-When a blob represent a negative update it only contains this simple slice of deleted documents ids.
-
-### The schema
-
-The schema is a data struture that represents which documents attributes should be stored and which should be indexed. It is stored under the "_data-schema_" entry and given to MeiliDB only at the creation.
-
-Each document attribute is associated to a unique 32 bit number named `SchemaAttr`.
-
-In the future this schema type could be given along with updates and probably be different from the original, the database could be able to handled this document structure and reindex it.
-
-### Document attributes
-
-When the engine handle a query the result that the requester want is a document, not only the [match](https://github.com/Kerollmops/MeiliDB/blob/fc2cdf92596fc002ce278e3aa8718640ac44724d/src/lib.rs#L51-L79) associated to it, fields of the original document must be returned too.
-
-So MeiliDB again uses the power of the underlying key-value store and save the documents attributes marked as _STORE_. The key is prefixed by "_doc_" followed by the 64 bit document id in bytes and the schema attribute number in bytes corresponding to the document attribute stored.
-
-When a document field is saved in the key-value store its value is binary encoded using the [bincode](https://docs.rs/bincode/) library, so a document must be serializable using serde.
-
-
-
-## How is an update handled?
-
-First of all an update in MeiliDB is nothing more than [a RocksDB SST file](https://github.com/facebook/rocksdb/wiki/Creating-and-Ingesting-SST-files). It contains the blob and all the documents attributes binary encoded like described above. Note that the blob is stored under the "_data-index_" key marked as [a merge operation](https://github.com/facebook/rocksdb/wiki/Merge-Operator-Implementation).
-
-### The merge operation is CPU consuming
-
-When [the database ingest an update](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/mod.rs#L108-L145) it gives the SST file to the underlying RocksDB, once it has ingested it there is a "_data-index_" entry available, we can request it but the key-value store will call a function before, a merge operation is performed.
-
-This merge operation is done on multiple blobs as you have understood and will compute a [PositiveBlob](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/blob/positive/blob.rs#L15), this type contains the fst and document indexes structures allowing us to search for documents. This two data structures can be considered as the inverted index.
-
-The computation time of this merge is important, RocksDB doesn't keep the previous merged result, it will call our merge operation each time until it decided to do a compaction. So [we must force this compaction earlier](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/mod.rs#L129-L131) when we receive an update to reduce this cost.
-
-This way when we request the "_data-index_" value it will gives us the previously merged positive blob without any other merge overhead.
-
-
-
-## How is a request processed?
-
-Now that we have our "_data-index_" we are able to return results based on a query. In the MeiliDB universe a query is a string.
-
-### Query lexemes
-
-The first step to be able to call the underlying structures is to split the query in words, for that we use a [custom tokenizer](https://github.com/Kerollmops/MeiliDB/blob/fc2cdf92596fc002ce278e3aa8718640ac44724d/src/tokenizer/mod.rs) that is not finished for the moment, [there is an open issue](https://github.com/Kerollmops/MeiliDB/issues/3). Note that a tokenizer is specialized for a human language, this is the hard part.
-
-### Automatons and query index
-
-So to query the fst we need an automaton, in MeiliDB we use a [levenshtein automaton](https://en.wikipedia.org/wiki/Levenshtein_automaton), this automaton is constructed using a string and a maximum distance. According to the [Algolia's blog post](https://blog.algolia.com/inside-the-algolia-engine-part-3-query-processing/#algolia%e2%80%99s-way-of-searching-for-alternatives) we [created the DFAs](https://github.com/Kerollmops/MeiliDB/blob/fc2cdf92596fc002ce278e3aa8718640ac44724d/src/automaton.rs#L62-L75) with different settings.
-
-Thanks to the power of the fst library [it is possible to union multiple automatons](https://docs.rs/fst/0.3.2/fst/map/struct.OpBuilder.html#method.union) on the same fst map, it will allow us to know which [automaton returns a word according to its index](https://github.com/Kerollmops/MeiliDB/blob/fc2cdf92596fc002ce278e3aa8718640ac44724d/src/metadata/ops.rs#L111). The `Stream` is able to return all the numbers associated to the words. We use these numbers to find the whole list of `DocIndexes` associated and do the union set operation.
-
-With all these informations it is possible [to reconstruct a list of all the DocIndexes associated](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/rank/query_builder.rs#L62-L99) with the words queried.
-
-### Sort by criteria
-
-Now that we are able to get a big list of [DocIndexes](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/lib.rs#L21-L36) it is not enough to sort them by criteria, we need more informations like the levenshtein distance or the fact that a query word match exactly the word stored in the fst. So [we stuff it a little bit](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/rank/query_builder.rs#L86-L93), and aggregate all these [Matches](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/lib.rs#L47-L74) for each document. This way it will be easy to sort a simple vector of document using a bunch of functions.
-
-With this big list of documents and associated matches [we are able to sort only the part of the slice that we want](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/rank/query_builder.rs#L108-L119) using bucket sorting. [Each criterion](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/rank/criterion/mod.rs#L75-L87) is evaluated on each subslice without copy, thanks to [GroupByMut](https://github.com/Kerollmops/group-by/blob/cab857bae01463dbd0edb99b0e0d7f3624e6c6f5/src/lib.rs#L180-L185) which, I hope [will soon be merged](https://github.com/rust-lang/rfcs/pull/2477).
-
-Note that it is possible to customize the criteria used by using the `QueryBuilder::with_criteria` constructor, this way you can implement some custom ranking based on the document attributes using the appropriate structure and the `retrieve_document` method.
-
-### Retrieve original documents
-
-The [DatabaseView](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/database_view.rs#L18-L24) structure that you must have created to be able to query the database have [two functions](https://github.com/Kerollmops/MeiliDB/blob/550dc1e99224e386516877450320f694947332d4/src/database/database_view.rs#L60-L76) that allows you to retrieve a full (or not) document according to the schema you specified at creation time (i.e. the _STORED_ attributes).
-
-As you can see, these functions force the created type `T` to implement [the serde Deserialize trait](https://docs.rs/serde/1.0.81/serde/trait.Deserialize.html), MeiliDB will use the `bincode::deserialise` function for each attribute to construct your type and return it to you.
-
-
-
-At this point, MeiliDB work is over 🎉
-

download-latest.sh

@@ -0,0 +1,200 @@
+#!/bin/sh
+
+# COLORS
+RED='\033[31m'
+GREEN='\033[32m'
+DEFAULT='\033[0m'
+
+# GLOBALS
+GREP_SEMVER_REGEXP='v\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)$' # i.e. v[number].[number].[number]
+
+# FUNCTIONS
+
+# semverParseInto and semverLT from https://github.com/cloudflare/semver_bash/blob/master/semver.sh
+
+# usage: semverParseInto version major minor patch special
+# version: the string version
+# major, minor, patch, special: will be assigned by the function
+semverParseInto() {
+    local RE='[^0-9]*\([0-9]*\)[.]\([0-9]*\)[.]\([0-9]*\)\([0-9A-Za-z-]*\)'
+    #MAJOR
+    eval $2=`echo $1 | sed -e "s#$RE#\1#"`
+    #MINOR
+    eval $3=`echo $1 | sed -e "s#$RE#\2#"`
+    #PATCH
+    eval $4=`echo $1 | sed -e "s#$RE#\3#"`
+    #SPECIAL
+    eval $5=`echo $1 | sed -e "s#$RE#\4#"`
+}
+
+# usage: semverLT version1 version2
+semverLT() {
+    local MAJOR_A=0
+    local MINOR_A=0
+    local PATCH_A=0
+    local SPECIAL_A=0
+
+    local MAJOR_B=0
+    local MINOR_B=0
+    local PATCH_B=0
+    local SPECIAL_B=0
+
+    semverParseInto $1 MAJOR_A MINOR_A PATCH_A SPECIAL_A
+    semverParseInto $2 MAJOR_B MINOR_B PATCH_B SPECIAL_B
+
+    if [ $MAJOR_A -lt $MAJOR_B ]; then
+        return 0
+    fi
+    if [ $MAJOR_A -le $MAJOR_B ] && [ $MINOR_A -lt $MINOR_B ]; then
+        return 0
+    fi
+    if [ $MAJOR_A -le $MAJOR_B ] && [ $MINOR_A -le $MINOR_B ] && [ $PATCH_A -lt $PATCH_B ]; then
+        return 0
+    fi
+    if [ "_$SPECIAL_A"  == "_" ] && [ "_$SPECIAL_B"  == "_" ] ; then
+        return 1
+    fi
+    if [ "_$SPECIAL_A"  == "_" ] && [ "_$SPECIAL_B"  != "_" ] ; then
+        return 1
+    fi
+    if [ "_$SPECIAL_A"  != "_" ] && [ "_$SPECIAL_B"  == "_" ] ; then
+        return 0
+    fi
+    if [ "_$SPECIAL_A" < "_$SPECIAL_B" ]; then
+        return 0
+    fi
+
+    return 1
+}
+
+# Returns the tag of the latest stable release (in terms of semver and not of release date)
+get_latest() {
+    temp_file='temp_file' # temp_file needed because the grep would start before the download is over
+    curl -s 'https://api.github.com/repos/meilisearch/MeiliSearch/releases' > "$temp_file" || return 1
+    releases=$(cat "$temp_file" | \
+        grep -E "tag_name|draft|prerelease" \
+        | tr -d ',"' | cut -d ':' -f2 | tr -d ' ')
+        # Returns a list of [tag_name draft_boolean prerelease_boolean ...]
+        # Ex: v0.10.1 false false v0.9.1-rc.1 false true v0.9.0 false false...
+
+    i=0
+    latest=""
+    current_tag=""
+    for release_info in $releases; do
+        if [ $i -eq 0 ]; then # Cheking tag_name
+            if echo "$release_info" | grep -q "$GREP_SEMVER_REGEXP"; then # If it's not an alpha or beta release
+                current_tag=$release_info
+            else
+                current_tag=""
+            fi
+            i=1
+        elif [ $i -eq 1 ]; then # Checking draft boolean
+            if [ "$release_info" = "true" ]; then
+                current_tag=""
+            fi
+            i=2
+        elif [ $i -eq 2 ]; then # Checking prerelease boolean
+            if [ "$release_info" = "true" ]; then
+                current_tag=""
+            fi
+            i=0
+            if [ "$current_tag" != "" ]; then # If the current_tag is valid
+                if [ "$latest" = "" ]; then # If there is no latest yet
+                    latest="$current_tag"
+                else
+                    semverLT $current_tag $latest # Comparing latest and the current tag
+                    if [ $? -eq 1 ]; then
+                        latest="$current_tag"
+                    fi
+                fi
+            fi
+        fi
+    done
+
+    rm -f "$temp_file"
+    echo $latest
+}
+
+# Gets the OS by setting the $os variable
+# Returns 0 in case of success, 1 otherwise.
+get_os() {
+    os_name=$(uname -s)
+    case "$os_name" in
+    'Darwin')
+        os='macos'
+        ;;
+    'Linux')
+        os='linux'
+        ;;
+	'MINGW'*)
+        os='windows'
+        ;;
+    *)
+        return 1
+    esac
+    return 0
+}
+
+# Gets the architecture by setting the $archi variable
+# Returns 0 in case of success, 1 otherwise.
+get_archi() {
+    architecture=$(uname -m)
+    case "$architecture" in
+    'x86_64' | 'amd64')
+        archi='amd64'
+        ;;
+    'aarch64')
+        archi='armv8'
+        ;;
+    *)
+        return 1
+    esac
+    return 0
+}
+
+success_usage() {
+    printf "$GREEN%s\n$DEFAULT" "MeiliSearch binary successfully downloaded as '$BINARY_NAME' file."
+    echo ''
+    echo 'Run it:'
+    echo '    $ ./meilisearch'
+    echo 'Usage:'
+    echo '    $ ./meilisearch --help'
+}
+
+failure_usage() {
+    printf "$RED%s\n$DEFAULT" 'ERROR: MeiliSearch binary is not available for your OS distribution or your architecture yet.'
+    echo ''
+    echo 'However, you can easily compile the binary from the source files.'
+    echo 'Follow the steps at the page ("Source" tab): https://docs.meilisearch.com/learn/getting_started/installation.html'
+}
+
+# MAIN
+latest="$(get_latest)"
+
+if ! get_os; then
+    failure_usage
+    exit 1
+fi
+
+if ! get_archi; then
+    failure_usage
+    exit 1
+fi
+
+echo "Downloading MeiliSearch binary $latest for $os, architecture $archi..."
+case "$os" in
+    'windows')
+        release_file="meilisearch-$os-$archi.exe"
+		BINARY_NAME='meilisearch.exe'
+
+        ;;
+	*) 
+		release_file="meilisearch-$os-$archi"
+		BINARY_NAME='meilisearch'
+
+esac
+link="https://github.com/meilisearch/MeiliSearch/releases/download/$latest/$release_file"
+curl -OL "$link"
+mv "$release_file" "$BINARY_NAME"
+chmod 744 "$BINARY_NAME"
+success_usage

examples/create-database.rs

@@ -1,137 +0,0 @@
-#[global_allocator]
-static ALLOC: jemallocator::Jemalloc = jemallocator::Jemalloc;
-
-use std::collections::{HashMap, HashSet};
-use std::io::{self, BufRead, BufReader};
-use std::path::{Path, PathBuf};
-use std::time::Instant;
-use std::error::Error;
-use std::borrow::Cow;
-use std::fs::File;
-
-use serde_derive::{Serialize, Deserialize};
-use structopt::StructOpt;
-
-use meilidb::database::{Database, Schema};
-use meilidb::tokenizer::DefaultBuilder;
-
-#[derive(Debug, StructOpt)]
-pub struct Opt {
-    /// The destination where the database must be created.
-    #[structopt(parse(from_os_str))]
-    pub database_path: PathBuf,
-
-    /// The csv file to index.
-    #[structopt(parse(from_os_str))]
-    pub csv_data_path: PathBuf,
-
-    /// The path to the schema.
-    #[structopt(long = "schema", parse(from_os_str))]
-    pub schema_path: PathBuf,
-
-    /// The path to the list of stop words (one by line).
-    #[structopt(long = "stop-words", parse(from_os_str))]
-    pub stop_words_path: Option<PathBuf>,
-
-    #[structopt(long = "update-group-size")]
-    pub update_group_size: Option<usize>,
-}
-
-#[derive(Serialize, Deserialize)]
-struct Document<'a> (
-    #[serde(borrow)]
-    HashMap<Cow<'a, str>, Cow<'a, str>>
-);
-
-fn index(
-    schema: Schema,
-    database_path: &Path,
-    csv_data_path: &Path,
-    update_group_size: Option<usize>,
-    stop_words: &HashSet<String>,
-) -> Result<Database, Box<Error>>
-{
-    let database = Database::create(database_path)?;
-
-    database.create_index("default", &schema)?;
-
-    let mut rdr = csv::Reader::from_path(csv_data_path)?;
-    let mut raw_record = csv::StringRecord::new();
-    let headers = rdr.headers()?.clone();
-
-    let mut i = 0;
-    let mut end_of_file = false;
-
-    while !end_of_file {
-        let tokenizer_builder = DefaultBuilder::new();
-        let mut update = database.start_update("default")?;
-
-        loop {
-            end_of_file = !rdr.read_record(&mut raw_record)?;
-            if end_of_file { break }
-
-            let document: Document = match raw_record.deserialize(Some(&headers)) {
-                Ok(document) => document,
-                Err(e) => {
-                    eprintln!("{:?}", e);
-                    continue;
-                }
-            };
-
-            update.update_document(&document, &tokenizer_builder, &stop_words)?;
-
-            print!("\rindexing document {}", i);
-            i += 1;
-
-            if let Some(group_size) = update_group_size {
-                if i % group_size == 0 { break }
-            }
-        }
-
-        println!();
-
-        println!("committing update...");
-        database.commit_update(update)?;
-    }
-
-    Ok(database)
-}
-
-fn retrieve_stop_words(path: &Path) -> io::Result<HashSet<String>> {
-    let f = File::open(path)?;
-    let reader = BufReader::new(f);
-    let mut words = HashSet::new();
-
-    for line in reader.lines() {
-        let line = line?;
-        let word = line.trim().to_string();
-        words.insert(word);
-    }
-
-    Ok(words)
-}
-
-fn main() -> Result<(), Box<Error>> {
-    let _ = env_logger::init();
-    let opt = Opt::from_args();
-
-    let schema = {
-        let file = File::open(&opt.schema_path)?;
-        Schema::from_toml(file)?
-    };
-
-    let stop_words = match opt.stop_words_path {
-        Some(ref path) => retrieve_stop_words(path)?,
-        None           => HashSet::new(),
-    };
-
-    let start = Instant::now();
-    let result = index(schema, &opt.database_path, &opt.csv_data_path, opt.update_group_size, &stop_words);
-
-    if let Err(e) = result {
-        return Err(e.into())
-    }
-
-    println!("database created in {:.2?} at: {:?}", start.elapsed(), opt.database_path);
-    Ok(())
-}

examples/query-database.rs

@@ -1,210 +0,0 @@
-#[global_allocator]
-static ALLOC: jemallocator::Jemalloc = jemallocator::Jemalloc;
-
-use std::collections::btree_map::{BTreeMap, Entry};
-use std::iter::FromIterator;
-use std::io::{self, Write};
-use std::time::Instant;
-use std::path::PathBuf;
-use std::error::Error;
-
-use hashbrown::{HashMap, HashSet};
-use termcolor::{Color, ColorChoice, ColorSpec, StandardStream, WriteColor};
-use structopt::StructOpt;
-
-use meilidb::database::schema::SchemaAttr;
-use meilidb::database::Database;
-use meilidb::Match;
-
-#[derive(Debug, StructOpt)]
-pub struct Opt {
-    /// The destination where the database must be created
-    #[structopt(parse(from_os_str))]
-    pub database_path: PathBuf,
-
-    /// Fields that must be displayed.
-    pub displayed_fields: Vec<String>,
-
-    /// The number of returned results
-    #[structopt(short = "n", long = "number-results", default_value = "10")]
-    pub number_results: usize,
-
-    /// The number of characters before and after the first match
-    #[structopt(short = "C", long = "context", default_value = "35")]
-    pub char_context: usize,
-}
-
-type Document = HashMap<String, String>;
-
-fn display_highlights(text: &str, ranges: &[usize]) -> io::Result<()> {
-    let mut stdout = StandardStream::stdout(ColorChoice::Always);
-    let mut highlighted = false;
-
-    for range in ranges.windows(2) {
-        let [start, end] = match range { [start, end] => [*start, *end], _ => unreachable!() };
-        if highlighted {
-            stdout.set_color(ColorSpec::new().set_fg(Some(Color::Yellow)))?;
-        }
-        write!(&mut stdout, "{}", &text[start..end])?;
-        stdout.reset()?;
-        highlighted = !highlighted;
-    }
-
-    Ok(())
-}
-
-fn char_to_byte_range(index: usize, length: usize, text: &str) -> (usize, usize) {
-    let mut byte_index = 0;
-    let mut byte_length = 0;
-
-    for (n, (i, c)) in text.char_indices().enumerate() {
-        if n == index {
-            byte_index = i;
-        }
-
-        if n + 1 == index + length {
-            byte_length = i - byte_index + c.len_utf8();
-            break;
-        }
-    }
-
-    (byte_index, byte_length)
-}
-
-fn create_highlight_areas(text: &str, matches: &[Match]) -> Vec<usize> {
-    let mut byte_indexes = BTreeMap::new();
-
-    for match_ in matches {
-        let char_index = match_.char_index as usize;
-        let char_length = match_.char_length as usize;
-        let (byte_index, byte_length) = char_to_byte_range(char_index, char_length, text);
-
-        match byte_indexes.entry(byte_index) {
-            Entry::Vacant(entry) => { entry.insert(byte_length); },
-            Entry::Occupied(mut entry) => {
-                if *entry.get() < byte_length {
-                    entry.insert(byte_length);
-                }
-            },
-        }
-    }
-
-    let mut title_areas = Vec::new();
-    title_areas.push(0);
-    for (byte_index, length) in byte_indexes {
-        title_areas.push(byte_index);
-        title_areas.push(byte_index + length);
-    }
-    title_areas.push(text.len());
-    title_areas.sort_unstable();
-    title_areas
-}
-
-/// note: matches must have been sorted by `char_index` and `char_length` before being passed.
-///
-/// ```no_run
-/// matches.sort_unstable_by_key(|m| (m.char_index, m.char_length));
-///
-/// let matches = matches.matches.iter().filter(|m| SchemaAttr::new(m.attribute) == attr).cloned();
-///
-/// let (text, matches) = crop_text(&text, matches, 35);
-/// ```
-fn crop_text(
-    text: &str,
-    matches: impl IntoIterator<Item=Match>,
-    context: usize,
-) -> (String, Vec<Match>)
-{
-    let mut matches = matches.into_iter().peekable();
-
-    let char_index = matches.peek().map(|m| m.char_index as usize).unwrap_or(0);
-    let start = char_index.saturating_sub(context);
-    let text = text.chars().skip(start).take(context * 2).collect();
-
-    let matches = matches
-        .take_while(|m| {
-            (m.char_index as usize) + (m.char_length as usize) <= start + (context * 2)
-        })
-        .map(|match_| {
-            Match { char_index: match_.char_index - start as u32, ..match_ }
-        })
-        .collect();
-
-    (text, matches)
-}
-
-fn main() -> Result<(), Box<Error>> {
-    let _ = env_logger::init();
-    let opt = Opt::from_args();
-
-    let start = Instant::now();
-    let database = Database::open(&opt.database_path)?;
-    println!("database prepared for you in {:.2?}", start.elapsed());
-
-    let mut buffer = String::new();
-    let input = io::stdin();
-
-    loop {
-        print!("Searching for: ");
-        io::stdout().flush()?;
-
-        if input.read_line(&mut buffer)? == 0 { break }
-        let query = buffer.trim_end_matches('\n');
-
-        let view = database.view("default")?;
-        let schema = view.schema();
-
-        let start = Instant::now();
-
-        let builder = view.query_builder();
-        let documents = builder.query(query, 0..opt.number_results);
-
-        let number_of_documents = documents.len();
-        for mut doc in documents {
-
-            doc.matches.sort_unstable_by_key(|m| (m.char_index, m.char_index));
-
-            match view.document_by_id::<Document>(doc.id) {
-                Ok(document) => {
-                    for name in &opt.displayed_fields {
-                        let attr = match schema.attribute(name) {
-                            Some(attr) => attr,
-                            None => continue,
-                        };
-                        let text = match document.get(name) {
-                            Some(text) => text,
-                            None => continue,
-                        };
-
-                        print!("{}: ", name);
-                        let matches = doc.matches.iter()
-                                        .filter(|m| SchemaAttr::new(m.attribute) == attr)
-                                        .cloned();
-                        let (text, matches) = crop_text(&text, matches, opt.char_context);
-                        let areas = create_highlight_areas(&text, &matches);
-                        display_highlights(&text, &areas)?;
-                        println!();
-                    }
-                },
-                Err(e) => eprintln!("{}", e),
-            }
-
-            let mut matching_attributes = HashSet::new();
-            for _match in doc.matches {
-                let attr = SchemaAttr::new(_match.attribute);
-                let name = schema.attribute_name(attr);
-                matching_attributes.insert(name);
-            }
-
-            let matching_attributes = Vec::from_iter(matching_attributes);
-            println!("matching in: {:?}", matching_attributes);
-
-            println!();
-        }
-
-        eprintln!("===== Found {} results in {:.2?} =====", number_of_documents, start.elapsed());
-        buffer.clear();
-    }
-
-    Ok(())
-}

examples/schema-example.toml

@@ -1,19 +0,0 @@
-# This schema has been generated ...
-# The order in which the attributes are declared is important,
-# it specify the attribute xxx...
-
-identifier = "id"
-
-[attributes.id]
-stored = true
-
-[attributes.title]
-stored = true
-indexed = true
-
-[attributes.description]
-stored = true
-indexed = true
-
-[attributes.image]
-stored = true

meilisearch-error/Cargo.toml

@@ -0,0 +1,8 @@
+[package]
+name = "meilisearch-error"
+version = "0.22.0"
+authors = ["marin <postma.marin@protonmail.com>"]
+edition = "2018"
+
+[dependencies]
+actix-http = "=3.0.0-beta.6"

meilisearch-error/src/lib.rs

@@ -0,0 +1,203 @@
+use std::fmt;
+
+use actix_http::http::StatusCode;
+
+pub trait ErrorCode: std::error::Error {
+    fn error_code(&self) -> Code;
+
+    /// returns the HTTP status code ascociated with the error
+    fn http_status(&self) -> StatusCode {
+        self.error_code().http()
+    }
+
+    /// returns the doc url ascociated with the error
+    fn error_url(&self) -> String {
+        self.error_code().url()
+    }
+
+    /// returns error name, used as error code
+    fn error_name(&self) -> String {
+        self.error_code().name()
+    }
+
+    /// return the error type
+    fn error_type(&self) -> String {
+        self.error_code().type_()
+    }
+}
+
+#[allow(clippy::enum_variant_names)]
+enum ErrorType {
+    InternalError,
+    InvalidRequestError,
+    AuthenticationError,
+}
+
+impl fmt::Display for ErrorType {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        use ErrorType::*;
+
+        match self {
+            InternalError => write!(f, "internal_error"),
+            InvalidRequestError => write!(f, "invalid_request_error"),
+            AuthenticationError => write!(f, "authentication_error"),
+        }
+    }
+}
+
+pub enum Code {
+    // index related error
+    CreateIndex,
+    IndexAlreadyExists,
+    IndexNotFound,
+    InvalidIndexUid,
+    OpenIndex,
+
+    // invalid state error
+    InvalidState,
+    MissingPrimaryKey,
+    PrimaryKeyAlreadyPresent,
+
+    MaxFieldsLimitExceeded,
+    MissingDocumentId,
+
+    Facet,
+    Filter,
+    Sort,
+
+    BadParameter,
+    BadRequest,
+    DocumentNotFound,
+    Internal,
+    InvalidToken,
+    MissingAuthorizationHeader,
+    NotFound,
+    PayloadTooLarge,
+    RetrieveDocument,
+    SearchDocuments,
+    UnsupportedMediaType,
+
+    DumpAlreadyInProgress,
+    DumpProcessFailed,
+}
+
+impl Code {
+    /// ascociate a `Code` variant to the actual ErrCode
+    fn err_code(&self) -> ErrCode {
+        use Code::*;
+
+        match self {
+            // index related errors
+            // create index is thrown on internal error while creating an index.
+            CreateIndex => ErrCode::internal("index_creation_failed", StatusCode::BAD_REQUEST),
+            IndexAlreadyExists => ErrCode::invalid("index_already_exists", StatusCode::BAD_REQUEST),
+            // thrown when requesting an unexisting index
+            IndexNotFound => ErrCode::invalid("index_not_found", StatusCode::NOT_FOUND),
+            InvalidIndexUid => ErrCode::invalid("invalid_index_uid", StatusCode::BAD_REQUEST),
+            OpenIndex => {
+                ErrCode::internal("index_not_accessible", StatusCode::INTERNAL_SERVER_ERROR)
+            }
+
+            // invalid state error
+            InvalidState => ErrCode::internal("invalid_state", StatusCode::INTERNAL_SERVER_ERROR),
+            // thrown when no primary key has been set
+            MissingPrimaryKey => ErrCode::invalid("missing_primary_key", StatusCode::BAD_REQUEST),
+            // error thrown when trying to set an already existing primary key
+            PrimaryKeyAlreadyPresent => {
+                ErrCode::invalid("primary_key_already_present", StatusCode::BAD_REQUEST)
+            }
+
+            // invalid document
+            MaxFieldsLimitExceeded => {
+                ErrCode::invalid("max_fields_limit_exceeded", StatusCode::BAD_REQUEST)
+            }
+            MissingDocumentId => ErrCode::invalid("missing_document_id", StatusCode::BAD_REQUEST),
+
+            // error related to facets
+            Facet => ErrCode::invalid("invalid_facet", StatusCode::BAD_REQUEST),
+            // error related to filters
+            Filter => ErrCode::invalid("invalid_filter", StatusCode::BAD_REQUEST),
+            // error related to sorts
+            Sort => ErrCode::invalid("invalid_sort", StatusCode::BAD_REQUEST),
+
+            BadParameter => ErrCode::invalid("bad_parameter", StatusCode::BAD_REQUEST),
+            BadRequest => ErrCode::invalid("bad_request", StatusCode::BAD_REQUEST),
+            DocumentNotFound => ErrCode::invalid("document_not_found", StatusCode::NOT_FOUND),
+            Internal => ErrCode::internal("internal", StatusCode::INTERNAL_SERVER_ERROR),
+            InvalidToken => ErrCode::authentication("invalid_token", StatusCode::FORBIDDEN),
+            MissingAuthorizationHeader => {
+                ErrCode::authentication("missing_authorization_header", StatusCode::UNAUTHORIZED)
+            }
+            NotFound => ErrCode::invalid("not_found", StatusCode::NOT_FOUND),
+            PayloadTooLarge => ErrCode::invalid("payload_too_large", StatusCode::PAYLOAD_TOO_LARGE),
+            RetrieveDocument => {
+                ErrCode::internal("unretrievable_document", StatusCode::BAD_REQUEST)
+            }
+            SearchDocuments => ErrCode::internal("search_error", StatusCode::BAD_REQUEST),
+            UnsupportedMediaType => {
+                ErrCode::invalid("unsupported_media_type", StatusCode::UNSUPPORTED_MEDIA_TYPE)
+            }
+
+            // error related to dump
+            DumpAlreadyInProgress => {
+                ErrCode::invalid("dump_already_in_progress", StatusCode::CONFLICT)
+            }
+            DumpProcessFailed => {
+                ErrCode::internal("dump_process_failed", StatusCode::INTERNAL_SERVER_ERROR)
+            }
+        }
+    }
+
+    /// return the HTTP status code ascociated with the `Code`
+    fn http(&self) -> StatusCode {
+        self.err_code().status_code
+    }
+
+    /// return error name, used as error code
+    fn name(&self) -> String {
+        self.err_code().error_name.to_string()
+    }
+
+    /// return the error type
+    fn type_(&self) -> String {
+        self.err_code().error_type.to_string()
+    }
+
+    /// return the doc url ascociated with the error
+    fn url(&self) -> String {
+        format!("https://docs.meilisearch.com/errors#{}", self.name())
+    }
+}
+
+/// Internal structure providing a convenient way to create error codes
+struct ErrCode {
+    status_code: StatusCode,
+    error_type: ErrorType,
+    error_name: &'static str,
+}
+
+impl ErrCode {
+    fn authentication(error_name: &'static str, status_code: StatusCode) -> ErrCode {
+        ErrCode {
+            status_code,
+            error_name,
+            error_type: ErrorType::AuthenticationError,
+        }
+    }
+
+    fn internal(error_name: &'static str, status_code: StatusCode) -> ErrCode {
+        ErrCode {
+            status_code,
+            error_name,
+            error_type: ErrorType::InternalError,
+        }
+    }
+
+    fn invalid(error_name: &'static str, status_code: StatusCode) -> ErrCode {
+        ErrCode {
+            status_code,
+            error_name,
+            error_type: ErrorType::InvalidRequestError,
+        }
+    }
+}

meilisearch-http/Cargo.toml

@@ -0,0 +1,109 @@
+[package]
+authors = ["Quentin de Quelen <quentin@dequelen.me>", "Clément Renault <clement@meilisearch.com>"]
+description = "MeiliSearch HTTP server"
+edition = "2018"
+license = "MIT"
+name = "meilisearch-http"
+version = "0.22.0"
+
+[[bin]]
+name = "meilisearch"
+path = "src/main.rs"
+
+[build-dependencies]
+actix-web-static-files = { git = "https://github.com/MarinPostma/actix-web-static-files.git", rev = "6db8c3e", optional = true }
+anyhow = { version = "*", optional = true }
+cargo_toml = { version = "0.9.0", optional = true }
+hex = { version = "0.4.3", optional = true }
+reqwest = { version = "0.11.3", features = ["blocking", "rustls-tls"], default-features = false, optional = true }
+sha-1 = { version = "0.9.4", optional = true }
+tempfile = { version = "3.1.0", optional = true }
+vergen = { version = "5.1.15", default-features = false, features = ["git"] }
+zip = { version = "0.5.12", optional = true }
+
+[dependencies]
+actix-cors = { git = "https://github.com/MarinPostma/actix-extras.git", rev = "2dac1a4"}
+actix-http = { version = "=3.0.0-beta.6" }
+actix-service = "2.0.0"
+actix-web = { version = "=4.0.0-beta.6", features = ["rustls"] }
+actix-web-static-files = { git = "https://github.com/MarinPostma/actix-web-static-files.git", rev = "6db8c3e", optional = true }
+anyhow = "1.0.36"
+async-stream = "0.3.0"
+async-trait = "0.1.42"
+arc-swap = "1.2.0"
+byte-unit = { version = "4.0.9", default-features = false, features = ["std"] }
+bytes = "0.6.0"
+chrono = { version = "0.4.19", features = ["serde"] }
+crossbeam-channel = "0.5.0"
+either = "1.6.1"
+env_logger = "0.8.2"
+flate2 = "1.0.19"
+fst = "0.4.5"
+futures = "0.3.7"
+futures-util = "0.3.8"
+heed = { git = "https://github.com/Kerollmops/heed", tag = "v0.12.1" }
+http = "0.2.1"
+indexmap = { version = "1.3.2", features = ["serde-1"] }
+itertools = "0.10.0"
+log = "0.4.8"
+main_error = "0.1.0"
+meilisearch-error = { path = "../meilisearch-error" }
+meilisearch-tokenizer = { git = "https://github.com/meilisearch/tokenizer.git", tag = "v0.2.5" }
+memmap = "0.7.0"
+milli = { git = "https://github.com/meilisearch/milli.git", tag = "v0.13.1" }
+mime = "0.3.16"
+num_cpus = "1.13.0"
+once_cell = "1.5.2"
+parking_lot = "0.11.1"
+rand = "0.7.3"
+rayon = "1.5.0"
+regex = "1.4.2"
+rustls = "0.19"
+serde = { version = "1.0", features = ["derive"] }
+serde_json = { version = "1.0.59", features = ["preserve_order"] }
+sha2 = "0.9.1"
+siphasher = "0.3.2"
+slice-group-by = "0.2.6"
+structopt = "0.3.20"
+tar = "0.4.29"
+tempfile = "3.1.0"
+thiserror = "1.0.24"
+tokio = { version = "1", features = ["full"] }
+uuid = { version =  "0.8.2", features = ["serde"] }
+walkdir = "2.3.2"
+obkv = "0.2.0"
+pin-project = "1.0.7"
+whoami = { version = "1.1.2", optional = true }
+reqwest = { version = "0.11.3", features = ["json", "rustls-tls"], default-features = false, optional = true }
+serdeval = "0.1.0"
+sysinfo = "0.20.0"
+
+[dev-dependencies]
+actix-rt = "2.1.0"
+assert-json-diff = { branch = "master", git = "https://github.com/qdequele/assert-json-diff" }
+mockall = "0.9.1"
+paste = "1.0.5"
+serde_url_params = "0.2.1"
+tempdir = "0.3.7"
+urlencoding = "1.1.1"
+
+[features]
+mini-dashboard = [
+    "actix-web-static-files",
+    "anyhow",
+    "cargo_toml",
+    "hex",
+    "reqwest",
+    "sha-1",
+    "tempfile",
+    "zip",
+]
+analytics = ["whoami", "reqwest"]
+default = ["analytics", "mini-dashboard"]
+
+[target.'cfg(target_os = "linux")'.dependencies]
+jemallocator = "0.3.2"
+
+[package.metadata.mini-dashboard]
+assets-url = "https://github.com/meilisearch/mini-dashboard/releases/download/v0.1.4/build.zip"
+sha1 = "750e8a8e56cfa61fbf9ead14b08a5f17ad3f3d37"

meilisearch-http/build.rs

@@ -0,0 +1,86 @@
+use vergen::{vergen, Config};
+
+fn main() {
+    if let Err(e) = vergen(Config::default()) {
+        println!("cargo:warning=vergen: {}", e);
+    }
+
+    #[cfg(feature = "mini-dashboard")]
+    mini_dashboard::setup_mini_dashboard().expect("Could not load the mini-dashboard assets");
+}
+
+#[cfg(feature = "mini-dashboard")]
+mod mini_dashboard {
+    use std::env;
+    use std::fs::{create_dir_all, File, OpenOptions};
+    use std::io::{Cursor, Read, Write};
+    use std::path::PathBuf;
+
+    use actix_web_static_files::resource_dir;
+    use anyhow::Context;
+    use cargo_toml::Manifest;
+    use reqwest::blocking::get;
+    use sha1::{Digest, Sha1};
+
+    pub fn setup_mini_dashboard() -> anyhow::Result<()> {
+        let cargo_manifest_dir = PathBuf::from(env::var("CARGO_MANIFEST_DIR").unwrap());
+        let cargo_toml = cargo_manifest_dir.join("Cargo.toml");
+        let out_dir = PathBuf::from(env::var("OUT_DIR").unwrap());
+
+        let sha1_path = out_dir.join(".mini-dashboard.sha1");
+        let dashboard_dir = out_dir.join("mini-dashboard");
+
+        let manifest = Manifest::from_path(cargo_toml).unwrap();
+
+        let meta = &manifest
+            .package
+            .as_ref()
+            .context("package not specified in Cargo.toml")?
+            .metadata
+            .as_ref()
+            .context("no metadata specified in Cargo.toml")?["mini-dashboard"];
+
+        // Check if there already is a dashboard built, and if it is up to date.
+        if sha1_path.exists() && dashboard_dir.exists() {
+            let mut sha1_file = File::open(&sha1_path)?;
+            let mut sha1 = String::new();
+            sha1_file.read_to_string(&mut sha1)?;
+            if sha1 == meta["sha1"].as_str().unwrap() {
+                // Nothing to do.
+                return Ok(());
+            }
+        }
+
+        let url = meta["assets-url"].as_str().unwrap();
+
+        let dashboard_assets_bytes = get(url)?.bytes()?;
+
+        let mut hasher = Sha1::new();
+        hasher.update(&dashboard_assets_bytes);
+        let sha1 = hex::encode(hasher.finalize());
+
+        assert_eq!(
+            meta["sha1"].as_str().unwrap(),
+            sha1,
+            "Downloaded mini-dashboard shasum differs from the one specified in the Cargo.toml"
+        );
+
+        create_dir_all(&dashboard_dir)?;
+        let cursor = Cursor::new(&dashboard_assets_bytes);
+        let mut zip = zip::read::ZipArchive::new(cursor)?;
+        zip.extract(&dashboard_dir)?;
+        resource_dir(&dashboard_dir).build()?;
+
+        // Write the sha1 for the dashboard back to file.
+        let mut file = OpenOptions::new()
+            .write(true)
+            .create(true)
+            .truncate(true)
+            .open(sha1_path)?;
+
+        file.write_all(sha1.as_bytes())?;
+        file.flush()?;
+
+        Ok(())
+    }
+}

meilisearch-http/src/analytics.rs

@@ -0,0 +1,126 @@
+use std::hash::{Hash, Hasher};
+use std::time::{Duration, Instant, SystemTime, UNIX_EPOCH};
+
+use log::debug;
+use serde::Serialize;
+use siphasher::sip::SipHasher;
+
+use crate::Data;
+use crate::Opt;
+
+const AMPLITUDE_API_KEY: &str = "f7fba398780e06d8fe6666a9be7e3d47";
+
+#[derive(Debug, Serialize)]
+struct EventProperties {
+    database_size: u64,
+    last_update_timestamp: Option<i64>, //timestamp
+    number_of_documents: Vec<u64>,
+}
+
+impl EventProperties {
+    async fn from(data: Data) -> anyhow::Result<EventProperties> {
+        let stats = data.index_controller.get_all_stats().await?;
+
+        let database_size = stats.database_size;
+        let last_update_timestamp = stats.last_update.map(|u| u.timestamp());
+        let number_of_documents = stats
+            .indexes
+            .values()
+            .map(|index| index.number_of_documents)
+            .collect();
+
+        Ok(EventProperties {
+            database_size,
+            last_update_timestamp,
+            number_of_documents,
+        })
+    }
+}
+
+#[derive(Debug, Serialize)]
+struct UserProperties<'a> {
+    env: &'a str,
+    start_since_days: u64,
+    user_email: Option<String>,
+    server_provider: Option<String>,
+}
+
+#[derive(Debug, Serialize)]
+struct Event<'a> {
+    user_id: &'a str,
+    event_type: &'a str,
+    device_id: &'a str,
+    time: u64,
+    app_version: &'a str,
+    user_properties: UserProperties<'a>,
+    event_properties: Option<EventProperties>,
+}
+
+#[derive(Debug, Serialize)]
+struct AmplitudeRequest<'a> {
+    api_key: &'a str,
+    events: Vec<Event<'a>>,
+}
+
+pub async fn analytics_sender(data: Data, opt: Opt) {
+    let username = whoami::username();
+    let hostname = whoami::hostname();
+    let platform = whoami::platform();
+
+    let uid = username + &hostname + &platform.to_string();
+
+    let mut hasher = SipHasher::new();
+    uid.hash(&mut hasher);
+    let hash = hasher.finish();
+
+    let uid = format!("{:X}", hash);
+    let platform = platform.to_string();
+    let first_start = Instant::now();
+
+    loop {
+        let n = SystemTime::now().duration_since(UNIX_EPOCH).unwrap();
+        let user_id = &uid;
+        let device_id = &platform;
+        let time = n.as_secs();
+        let event_type = "runtime_tick";
+        let elapsed_since_start = first_start.elapsed().as_secs() / 86_400; // One day
+        let event_properties = EventProperties::from(data.clone()).await.ok();
+        let app_version = env!("CARGO_PKG_VERSION").to_string();
+        let app_version = app_version.as_str();
+        let user_email = std::env::var("MEILI_USER_EMAIL").ok();
+        let server_provider = std::env::var("MEILI_SERVER_PROVIDER").ok();
+        let user_properties = UserProperties {
+            env: &opt.env,
+            start_since_days: elapsed_since_start,
+            user_email,
+            server_provider,
+        };
+
+        let event = Event {
+            user_id,
+            event_type,
+            device_id,
+            time,
+            app_version,
+            user_properties,
+            event_properties,
+        };
+
+        let request = AmplitudeRequest {
+            api_key: AMPLITUDE_API_KEY,
+            events: vec![event],
+        };
+
+        let response = reqwest::Client::new()
+            .post("https://api2.amplitude.com/2/httpapi")
+            .timeout(Duration::from_secs(60)) // 1 minute max
+            .json(&request)
+            .send()
+            .await;
+        if let Err(e) = response {
+            debug!("Unsuccessful call to Amplitude: {}", e);
+        }
+
+        tokio::time::sleep(Duration::from_secs(3600)).await;
+    }
+}

meilisearch-http/src/data/mod.rs

@@ -0,0 +1,133 @@
+use std::ops::Deref;
+use std::sync::Arc;
+
+use sha2::Digest;
+
+use crate::index::{Checked, Settings};
+use crate::index_controller::{
+    error::Result, DumpInfo, IndexController, IndexMetadata, IndexSettings, IndexStats, Stats,
+};
+use crate::option::Opt;
+
+pub mod search;
+mod updates;
+
+#[derive(Clone)]
+pub struct Data {
+    inner: Arc<DataInner>,
+}
+
+impl Deref for Data {
+    type Target = DataInner;
+
+    fn deref(&self) -> &Self::Target {
+        &self.inner
+    }
+}
+
+pub struct DataInner {
+    pub index_controller: IndexController,
+    pub api_keys: ApiKeys,
+    options: Opt,
+}
+
+#[derive(Clone)]
+pub struct ApiKeys {
+    pub public: Option<String>,
+    pub private: Option<String>,
+    pub master: Option<String>,
+}
+
+impl ApiKeys {
+    pub fn generate_missing_api_keys(&mut self) {
+        if let Some(master_key) = &self.master {
+            if self.private.is_none() {
+                let key = format!("{}-private", master_key);
+                let sha = sha2::Sha256::digest(key.as_bytes());
+                self.private = Some(format!("{:x}", sha));
+            }
+            if self.public.is_none() {
+                let key = format!("{}-public", master_key);
+                let sha = sha2::Sha256::digest(key.as_bytes());
+                self.public = Some(format!("{:x}", sha));
+            }
+        }
+    }
+}
+
+impl Data {
+    pub fn new(options: Opt) -> anyhow::Result<Data> {
+        let path = options.db_path.clone();
+
+        let index_controller = IndexController::new(&path, &options)?;
+
+        let mut api_keys = ApiKeys {
+            master: options.clone().master_key,
+            private: None,
+            public: None,
+        };
+
+        api_keys.generate_missing_api_keys();
+
+        let inner = DataInner {
+            index_controller,
+            api_keys,
+            options,
+        };
+        let inner = Arc::new(inner);
+
+        Ok(Data { inner })
+    }
+
+    pub async fn settings(&self, uid: String) -> Result<Settings<Checked>> {
+        self.index_controller.settings(uid).await
+    }
+
+    pub async fn list_indexes(&self) -> Result<Vec<IndexMetadata>> {
+        self.index_controller.list_indexes().await
+    }
+
+    pub async fn index(&self, uid: String) -> Result<IndexMetadata> {
+        self.index_controller.get_index(uid).await
+    }
+
+    pub async fn create_index(
+        &self,
+        uid: String,
+        primary_key: Option<String>,
+    ) -> Result<IndexMetadata> {
+        let settings = IndexSettings {
+            uid: Some(uid),
+            primary_key,
+        };
+
+        let meta = self.index_controller.create_index(settings).await?;
+        Ok(meta)
+    }
+
+    pub async fn get_index_stats(&self, uid: String) -> Result<IndexStats> {
+        Ok(self.index_controller.get_index_stats(uid).await?)
+    }
+
+    pub async fn get_all_stats(&self) -> Result<Stats> {
+        Ok(self.index_controller.get_all_stats().await?)
+    }
+
+    pub async fn create_dump(&self) -> Result<DumpInfo> {
+        Ok(self.index_controller.create_dump().await?)
+    }
+
+    pub async fn dump_status(&self, uid: String) -> Result<DumpInfo> {
+        Ok(self.index_controller.dump_info(uid).await?)
+    }
+
+    #[inline]
+    pub fn http_payload_size_limit(&self) -> usize {
+        self.options.http_payload_size_limit.get_bytes() as usize
+    }
+
+    #[inline]
+    pub fn api_keys(&self) -> &ApiKeys {
+        &self.api_keys
+    }
+}

meilisearch-http/src/data/search.rs

@@ -0,0 +1,34 @@
+use serde_json::{Map, Value};
+
+use super::Data;
+use crate::index::{SearchQuery, SearchResult};
+use crate::index_controller::error::Result;
+
+impl Data {
+    pub async fn search(&self, index: String, search_query: SearchQuery) -> Result<SearchResult> {
+        self.index_controller.search(index, search_query).await
+    }
+
+    pub async fn retrieve_documents(
+        &self,
+        index: String,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Vec<Map<String, Value>>> {
+        self.index_controller
+            .documents(index, offset, limit, attributes_to_retrieve)
+            .await
+    }
+
+    pub async fn retrieve_document(
+        &self,
+        index: String,
+        document_id: String,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Map<String, Value>> {
+        self.index_controller
+            .document(index, document_id, attributes_to_retrieve)
+            .await
+    }
+}

meilisearch-http/src/data/updates.rs

@@ -0,0 +1,80 @@
+use milli::update::{IndexDocumentsMethod, UpdateFormat};
+
+use crate::extractors::payload::Payload;
+use crate::index::{Checked, Settings};
+use crate::index_controller::{error::Result, IndexMetadata, IndexSettings, UpdateStatus};
+use crate::Data;
+
+impl Data {
+    pub async fn add_documents(
+        &self,
+        index: String,
+        method: IndexDocumentsMethod,
+        format: UpdateFormat,
+        stream: Payload,
+        primary_key: Option<String>,
+    ) -> Result<UpdateStatus> {
+        let update_status = self
+            .index_controller
+            .add_documents(index, method, format, stream, primary_key)
+            .await?;
+        Ok(update_status)
+    }
+
+    pub async fn update_settings(
+        &self,
+        index: String,
+        settings: Settings<Checked>,
+        create: bool,
+    ) -> Result<UpdateStatus> {
+        let update = self
+            .index_controller
+            .update_settings(index, settings, create)
+            .await?;
+        Ok(update)
+    }
+
+    pub async fn clear_documents(&self, index: String) -> Result<UpdateStatus> {
+        let update = self.index_controller.clear_documents(index).await?;
+        Ok(update)
+    }
+
+    pub async fn delete_documents(
+        &self,
+        index: String,
+        document_ids: Vec<String>,
+    ) -> Result<UpdateStatus> {
+        let update = self
+            .index_controller
+            .delete_documents(index, document_ids)
+            .await?;
+        Ok(update)
+    }
+
+    pub async fn delete_index(&self, index: String) -> Result<()> {
+        self.index_controller.delete_index(index).await?;
+        Ok(())
+    }
+
+    pub async fn get_update_status(&self, index: String, uid: u64) -> Result<UpdateStatus> {
+        self.index_controller.update_status(index, uid).await
+    }
+
+    pub async fn get_updates_status(&self, index: String) -> Result<Vec<UpdateStatus>> {
+        self.index_controller.all_update_status(index).await
+    }
+
+    pub async fn update_index(
+        &self,
+        uid: String,
+        primary_key: Option<String>,
+        new_uid: Option<String>,
+    ) -> Result<IndexMetadata> {
+        let settings = IndexSettings {
+            uid: new_uid,
+            primary_key,
+        };
+
+        self.index_controller.update_index(uid, settings).await
+    }
+}

meilisearch-http/src/error.rs

@@ -0,0 +1,171 @@
+use std::error::Error;
+use std::fmt;
+
+use actix_web as aweb;
+use actix_web::body::Body;
+use actix_web::dev::BaseHttpResponseBuilder;
+use actix_web::http::StatusCode;
+use aweb::error::{JsonPayloadError, QueryPayloadError};
+use meilisearch_error::{Code, ErrorCode};
+use milli::UserError;
+use serde::{Deserialize, Serialize};
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct ResponseError {
+    #[serde(skip)]
+    code: StatusCode,
+    message: String,
+    error_code: String,
+    error_type: String,
+    error_link: String,
+}
+
+impl fmt::Display for ResponseError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.message.fmt(f)
+    }
+}
+
+impl<T> From<T> for ResponseError
+where
+    T: ErrorCode,
+{
+    fn from(other: T) -> Self {
+        Self {
+            code: other.http_status(),
+            message: other.to_string(),
+            error_code: other.error_name(),
+            error_type: other.error_type(),
+            error_link: other.error_url(),
+        }
+    }
+}
+
+impl aweb::error::ResponseError for ResponseError {
+    fn error_response(&self) -> aweb::BaseHttpResponse<Body> {
+        let json = serde_json::to_vec(self).unwrap();
+        BaseHttpResponseBuilder::new(self.status_code())
+            .content_type("application/json")
+            .body(json)
+    }
+
+    fn status_code(&self) -> StatusCode {
+        self.code
+    }
+}
+
+macro_rules! internal_error {
+    ($target:ty : $($other:path), *) => {
+        $(
+            impl From<$other> for $target {
+                fn from(other: $other) -> Self {
+                    Self::Internal(Box::new(other))
+                }
+            }
+        )*
+    }
+}
+
+#[derive(Debug)]
+pub struct MilliError<'a>(pub &'a milli::Error);
+
+impl Error for MilliError<'_> {}
+
+impl fmt::Display for MilliError<'_> {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+impl ErrorCode for MilliError<'_> {
+    fn error_code(&self) -> Code {
+        match self.0 {
+            milli::Error::InternalError(_) => Code::Internal,
+            milli::Error::IoError(_) => Code::Internal,
+            milli::Error::UserError(ref error) => {
+                match error {
+                    // TODO: wait for spec for new error codes.
+                    UserError::Csv(_)
+                    | UserError::SerdeJson(_)
+                    | UserError::MaxDatabaseSizeReached
+                    | UserError::InvalidCriterionName { .. }
+                    | UserError::InvalidDocumentId { .. }
+                    | UserError::InvalidStoreFile
+                    | UserError::NoSpaceLeftOnDevice
+                    | UserError::InvalidAscDescSyntax { .. }
+                    | UserError::DocumentLimitReached => Code::Internal,
+                    UserError::AttributeLimitReached => Code::MaxFieldsLimitExceeded,
+                    UserError::InvalidFilter(_) => Code::Filter,
+                    UserError::InvalidFilterAttribute(_) => Code::Filter,
+                    UserError::InvalidSortName { .. } => Code::Sort,
+                    UserError::MissingDocumentId { .. } => Code::MissingDocumentId,
+                    UserError::MissingPrimaryKey => Code::MissingPrimaryKey,
+                    UserError::PrimaryKeyCannotBeChanged => Code::PrimaryKeyAlreadyPresent,
+                    UserError::PrimaryKeyCannotBeReset => Code::PrimaryKeyAlreadyPresent,
+                    UserError::SortRankingRuleMissing => Code::Sort,
+                    UserError::UnknownInternalDocumentId { .. } => Code::DocumentNotFound,
+                    UserError::InvalidFacetsDistribution { .. } => Code::BadRequest,
+                    UserError::InvalidSortableAttribute { .. } => Code::Sort,
+                }
+            }
+        }
+    }
+}
+
+impl fmt::Display for PayloadError {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        match self {
+            PayloadError::Json(e) => e.fmt(f),
+            PayloadError::Query(e) => e.fmt(f),
+        }
+    }
+}
+
+#[derive(Debug)]
+pub enum PayloadError {
+    Json(JsonPayloadError),
+    Query(QueryPayloadError),
+}
+
+impl Error for PayloadError {}
+
+impl ErrorCode for PayloadError {
+    fn error_code(&self) -> Code {
+        match self {
+            PayloadError::Json(err) => match err {
+                JsonPayloadError::Overflow => Code::PayloadTooLarge,
+                JsonPayloadError::ContentType => Code::UnsupportedMediaType,
+                JsonPayloadError::Payload(aweb::error::PayloadError::Overflow) => {
+                    Code::PayloadTooLarge
+                }
+                JsonPayloadError::Deserialize(_) | JsonPayloadError::Payload(_) => Code::BadRequest,
+                JsonPayloadError::Serialize(_) => Code::Internal,
+                _ => Code::Internal,
+            },
+            PayloadError::Query(err) => match err {
+                QueryPayloadError::Deserialize(_) => Code::BadRequest,
+                _ => Code::Internal,
+            },
+        }
+    }
+}
+
+impl From<JsonPayloadError> for PayloadError {
+    fn from(other: JsonPayloadError) -> Self {
+        Self::Json(other)
+    }
+}
+
+impl From<QueryPayloadError> for PayloadError {
+    fn from(other: QueryPayloadError) -> Self {
+        Self::Query(other)
+    }
+}
+
+pub fn payload_error_handler<E>(err: E) -> ResponseError
+where
+    E: Into<PayloadError>,
+{
+    err.into().into()
+}

meilisearch-http/src/extractors/authentication/error.rs

@@ -0,0 +1,25 @@
+use meilisearch_error::{Code, ErrorCode};
+
+#[derive(Debug, thiserror::Error)]
+pub enum AuthenticationError {
+    #[error("You must have an authorization token")]
+    MissingAuthorizationHeader,
+    #[error("Invalid API key")]
+    InvalidToken(String),
+    // Triggered on configuration error.
+    #[error("Irretrievable state")]
+    IrretrievableState,
+    #[error("Unknown authentication policy")]
+    UnknownPolicy,
+}
+
+impl ErrorCode for AuthenticationError {
+    fn error_code(&self) -> Code {
+        match self {
+            AuthenticationError::MissingAuthorizationHeader => Code::MissingAuthorizationHeader,
+            AuthenticationError::InvalidToken(_) => Code::InvalidToken,
+            AuthenticationError::IrretrievableState => Code::Internal,
+            AuthenticationError::UnknownPolicy => Code::Internal,
+        }
+    }
+}

meilisearch-http/src/extractors/authentication/mod.rs

@@ -0,0 +1,182 @@
+mod error;
+
+use std::any::{Any, TypeId};
+use std::collections::HashMap;
+use std::marker::PhantomData;
+use std::ops::Deref;
+
+use actix_web::FromRequest;
+use futures::future::err;
+use futures::future::{ok, Ready};
+
+use crate::error::ResponseError;
+use error::AuthenticationError;
+
+macro_rules! create_policies {
+    ($($name:ident), *) => {
+        pub mod policies {
+            use std::collections::HashSet;
+            use crate::extractors::authentication::Policy;
+
+            $(
+                #[derive(Debug, Default)]
+                pub struct $name {
+                    inner: HashSet<Vec<u8>>
+                }
+
+                impl $name {
+                    pub fn new() -> Self {
+                        Self { inner: HashSet::new() }
+                    }
+
+                    pub fn add(&mut self, token: Vec<u8>) {
+                        self.inner.insert(token);
+                    }
+                }
+
+                impl Policy for $name {
+                    fn authenticate(&self, token: &[u8]) -> bool {
+                        self.inner.contains(token)
+                    }
+                }
+            )*
+        }
+    };
+}
+
+create_policies!(Public, Private, Admin);
+
+/// Instanciate a `Policies`, filled with the given policies.
+macro_rules! init_policies {
+    ($($name:ident), *) => {
+        {
+            let mut policies = crate::extractors::authentication::Policies::new();
+            $(
+                let policy = $name::new();
+                policies.insert(policy);
+            )*
+            policies
+        }
+    };
+}
+
+/// Adds user to all specified policies.
+macro_rules! create_users {
+    ($policies:ident, $($user:expr => { $($policy:ty), * }), *) => {
+        {
+            $(
+                $(
+                    $policies.get_mut::<$policy>().map(|p| p.add($user.to_owned()));
+                )*
+            )*
+        }
+    };
+}
+
+pub struct GuardedData<T, D> {
+    data: D,
+    _marker: PhantomData<T>,
+}
+
+impl<T, D> Deref for GuardedData<T, D> {
+    type Target = D;
+
+    fn deref(&self) -> &Self::Target {
+        &self.data
+    }
+}
+
+pub trait Policy {
+    fn authenticate(&self, token: &[u8]) -> bool;
+}
+
+#[derive(Debug)]
+pub struct Policies {
+    inner: HashMap<TypeId, Box<dyn Any>>,
+}
+
+impl Policies {
+    pub fn new() -> Self {
+        Self {
+            inner: HashMap::new(),
+        }
+    }
+
+    pub fn insert<S: Policy + 'static>(&mut self, policy: S) {
+        self.inner.insert(TypeId::of::<S>(), Box::new(policy));
+    }
+
+    pub fn get<S: Policy + 'static>(&self) -> Option<&S> {
+        self.inner
+            .get(&TypeId::of::<S>())
+            .and_then(|p| p.downcast_ref::<S>())
+    }
+
+    pub fn get_mut<S: Policy + 'static>(&mut self) -> Option<&mut S> {
+        self.inner
+            .get_mut(&TypeId::of::<S>())
+            .and_then(|p| p.downcast_mut::<S>())
+    }
+}
+
+impl Default for Policies {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+pub enum AuthConfig {
+    NoAuth,
+    Auth(Policies),
+}
+
+impl Default for AuthConfig {
+    fn default() -> Self {
+        Self::NoAuth
+    }
+}
+
+impl<P: Policy + 'static, D: 'static + Clone> FromRequest for GuardedData<P, D> {
+    type Config = AuthConfig;
+
+    type Error = ResponseError;
+
+    type Future = Ready<Result<Self, Self::Error>>;
+
+    fn from_request(
+        req: &actix_web::HttpRequest,
+        _payload: &mut actix_http::Payload,
+    ) -> Self::Future {
+        match req.app_data::<Self::Config>() {
+            Some(config) => match config {
+                AuthConfig::NoAuth => match req.app_data::<D>().cloned() {
+                    Some(data) => ok(Self {
+                        data,
+                        _marker: PhantomData,
+                    }),
+                    None => err(AuthenticationError::IrretrievableState.into()),
+                },
+                AuthConfig::Auth(policies) => match policies.get::<P>() {
+                    Some(policy) => match req.headers().get("x-meili-api-key") {
+                        Some(token) => {
+                            if policy.authenticate(token.as_bytes()) {
+                                match req.app_data::<D>().cloned() {
+                                    Some(data) => ok(Self {
+                                        data,
+                                        _marker: PhantomData,
+                                    }),
+                                    None => err(AuthenticationError::IrretrievableState.into()),
+                                }
+                            } else {
+                                err(AuthenticationError::InvalidToken(String::from("hello")).into())
+                            }
+                        }
+                        None => err(AuthenticationError::MissingAuthorizationHeader.into()),
+                    },
+                    None => err(AuthenticationError::UnknownPolicy.into()),
+                },
+            },
+            None => err(AuthenticationError::IrretrievableState.into()),
+        }
+    }
+}

meilisearch-http/src/extractors/mod.rs

@@ -0,0 +1,3 @@
+pub mod payload;
+#[macro_use]
+pub mod authentication;

meilisearch-http/src/extractors/payload.rs

@@ -0,0 +1,69 @@
+use std::pin::Pin;
+use std::task::{Context, Poll};
+
+use actix_http::error::PayloadError;
+use actix_web::{dev, web, FromRequest, HttpRequest};
+use futures::future::{ready, Ready};
+use futures::Stream;
+
+pub struct Payload {
+    payload: dev::Payload,
+    limit: usize,
+}
+
+pub struct PayloadConfig {
+    limit: usize,
+}
+
+impl PayloadConfig {
+    pub fn new(limit: usize) -> Self {
+        Self { limit }
+    }
+}
+
+impl Default for PayloadConfig {
+    fn default() -> Self {
+        Self { limit: 256 * 1024 }
+    }
+}
+
+impl FromRequest for Payload {
+    type Config = PayloadConfig;
+
+    type Error = PayloadError;
+
+    type Future = Ready<Result<Payload, Self::Error>>;
+
+    #[inline]
+    fn from_request(req: &HttpRequest, payload: &mut dev::Payload) -> Self::Future {
+        let limit = req
+            .app_data::<PayloadConfig>()
+            .map(|c| c.limit)
+            .unwrap_or(Self::Config::default().limit);
+        ready(Ok(Payload {
+            payload: payload.take(),
+            limit,
+        }))
+    }
+}
+
+impl Stream for Payload {
+    type Item = Result<web::Bytes, PayloadError>;
+
+    #[inline]
+    fn poll_next(mut self: Pin<&mut Self>, cx: &mut Context<'_>) -> Poll<Option<Self::Item>> {
+        match Pin::new(&mut self.payload).poll_next(cx) {
+            Poll::Ready(Some(result)) => match result {
+                Ok(bytes) => match self.limit.checked_sub(bytes.len()) {
+                    Some(new_limit) => {
+                        self.limit = new_limit;
+                        Poll::Ready(Some(Ok(bytes)))
+                    }
+                    None => Poll::Ready(Some(Err(PayloadError::Overflow))),
+                },
+                x => Poll::Ready(Some(x)),
+            },
+            otherwise => otherwise,
+        }
+    }
+}

meilisearch-http/src/helpers/compression.rs

@@ -0,0 +1,26 @@
+use std::fs::{create_dir_all, File};
+use std::io::Write;
+use std::path::Path;
+
+use flate2::{read::GzDecoder, write::GzEncoder, Compression};
+use tar::{Archive, Builder};
+
+pub fn to_tar_gz(src: impl AsRef<Path>, dest: impl AsRef<Path>) -> anyhow::Result<()> {
+    let mut f = File::create(dest)?;
+    let gz_encoder = GzEncoder::new(&mut f, Compression::default());
+    let mut tar_encoder = Builder::new(gz_encoder);
+    tar_encoder.append_dir_all(".", src)?;
+    let gz_encoder = tar_encoder.into_inner()?;
+    gz_encoder.finish()?;
+    f.flush()?;
+    Ok(())
+}
+
+pub fn from_tar_gz(src: impl AsRef<Path>, dest: impl AsRef<Path>) -> anyhow::Result<()> {
+    let f = File::open(&src)?;
+    let gz = GzDecoder::new(f);
+    let mut ar = Archive::new(gz);
+    create_dir_all(&dest)?;
+    ar.unpack(&dest)?;
+    Ok(())
+}

meilisearch-http/src/helpers/env.rs

@@ -0,0 +1,16 @@
+use walkdir::WalkDir;
+
+pub trait EnvSizer {
+    fn size(&self) -> u64;
+}
+
+impl EnvSizer for heed::Env {
+    fn size(&self) -> u64 {
+        WalkDir::new(self.path())
+            .into_iter()
+            .filter_map(|entry| entry.ok())
+            .filter_map(|entry| entry.metadata().ok())
+            .filter(|metadata| metadata.is_file())
+            .fold(0, |acc, m| acc + m.len())
+    }
+}

meilisearch-http/src/helpers/mod.rs

@@ -0,0 +1,4 @@
+pub mod compression;
+mod env;
+
+pub use env::EnvSizer;

meilisearch-http/src/index/dump.rs

@@ -0,0 +1,170 @@
+use std::fs::{create_dir_all, File};
+use std::io::{BufRead, BufReader, Write};
+use std::path::Path;
+use std::sync::Arc;
+
+use anyhow::{bail, Context};
+use heed::RoTxn;
+use indexmap::IndexMap;
+use milli::update::{IndexDocumentsMethod, UpdateFormat::JsonStream};
+use serde::{Deserialize, Serialize};
+use serde_json::Value;
+
+use crate::index_controller::{asc_ranking_rule, desc_ranking_rule};
+use crate::option::IndexerOpts;
+
+use super::error::Result;
+use super::{update_handler::UpdateHandler, Index, Settings, Unchecked};
+
+#[derive(Serialize, Deserialize)]
+struct DumpMeta {
+    settings: Settings<Unchecked>,
+    primary_key: Option<String>,
+}
+
+const META_FILE_NAME: &str = "meta.json";
+const DATA_FILE_NAME: &str = "documents.jsonl";
+
+impl Index {
+    pub fn dump(&self, path: impl AsRef<Path>) -> Result<()> {
+        // acquire write txn make sure any ongoing write is finished before we start.
+        let txn = self.env.write_txn()?;
+
+        self.dump_documents(&txn, &path)?;
+        self.dump_meta(&txn, &path)?;
+
+        Ok(())
+    }
+
+    fn dump_documents(&self, txn: &RoTxn, path: impl AsRef<Path>) -> Result<()> {
+        let document_file_path = path.as_ref().join(DATA_FILE_NAME);
+        let mut document_file = File::create(&document_file_path)?;
+
+        let documents = self.all_documents(txn)?;
+        let fields_ids_map = self.fields_ids_map(txn)?;
+
+        // dump documents
+        let mut json_map = IndexMap::new();
+        for document in documents {
+            let (_, reader) = document?;
+
+            for (fid, bytes) in reader.iter() {
+                if let Some(name) = fields_ids_map.name(fid) {
+                    json_map.insert(name, serde_json::from_slice::<serde_json::Value>(bytes)?);
+                }
+            }
+
+            serde_json::to_writer(&mut document_file, &json_map)?;
+            document_file.write_all(b"\n")?;
+
+            json_map.clear();
+        }
+
+        Ok(())
+    }
+
+    fn dump_meta(&self, txn: &RoTxn, path: impl AsRef<Path>) -> Result<()> {
+        let meta_file_path = path.as_ref().join(META_FILE_NAME);
+        let mut meta_file = File::create(&meta_file_path)?;
+
+        let settings = self.settings_txn(txn)?.into_unchecked();
+        let primary_key = self.primary_key(txn)?.map(String::from);
+        let meta = DumpMeta {
+            settings,
+            primary_key,
+        };
+
+        serde_json::to_writer(&mut meta_file, &meta)?;
+
+        Ok(())
+    }
+
+    pub fn load_dump(
+        src: impl AsRef<Path>,
+        dst: impl AsRef<Path>,
+        size: usize,
+        indexing_options: &IndexerOpts,
+    ) -> anyhow::Result<()> {
+        let dir_name = src
+            .as_ref()
+            .file_name()
+            .with_context(|| format!("invalid dump index: {}", src.as_ref().display()))?;
+
+        let dst_dir_path = dst.as_ref().join("indexes").join(dir_name);
+        create_dir_all(&dst_dir_path)?;
+
+        let meta_path = src.as_ref().join(META_FILE_NAME);
+        let mut meta_file = File::open(meta_path)?;
+
+        // We first deserialize the dump meta into a serde_json::Value and change
+        // the custom ranking rules settings from the old format to the new format.
+        let mut meta: Value = serde_json::from_reader(&mut meta_file)?;
+        if let Some(ranking_rules) = meta.pointer_mut("/settings/rankingRules") {
+            convert_custom_ranking_rules(ranking_rules);
+        }
+
+        // Then we serialize it back into a vec to deserialize it
+        // into a `DumpMeta` struct with the newly patched `rankingRules` format.
+        let patched_meta = serde_json::to_vec(&meta)?;
+
+        let DumpMeta {
+            settings,
+            primary_key,
+        } = serde_json::from_slice(&patched_meta)?;
+        let settings = settings.check();
+        let index = Self::open(&dst_dir_path, size)?;
+        let mut txn = index.write_txn()?;
+
+        let handler = UpdateHandler::new(indexing_options)?;
+
+        index.update_settings_txn(&mut txn, &settings, handler.update_builder(0))?;
+
+        let document_file_path = src.as_ref().join(DATA_FILE_NAME);
+        let reader = File::open(&document_file_path)?;
+        let mut reader = BufReader::new(reader);
+        reader.fill_buf()?;
+        // If the document file is empty, we don't perform the document addition, to prevent
+        // a primary key error to be thrown.
+        if !reader.buffer().is_empty() {
+            index.update_documents_txn(
+                &mut txn,
+                JsonStream,
+                IndexDocumentsMethod::UpdateDocuments,
+                Some(reader),
+                handler.update_builder(0),
+                primary_key.as_deref(),
+            )?;
+        }
+
+        txn.commit()?;
+
+        match Arc::try_unwrap(index.0) {
+            Ok(inner) => inner.prepare_for_closing().wait(),
+            Err(_) => bail!("Could not close index properly."),
+        }
+
+        Ok(())
+    }
+}
+
+/// Converts the ranking rules from the format `asc(_)`, `desc(_)` to the format `_:asc`, `_:desc`.
+///
+/// This is done for compatibility reasons, and to avoid a new dump version,
+/// since the new syntax was introduced soon after the new dump version.
+fn convert_custom_ranking_rules(ranking_rules: &mut Value) {
+    *ranking_rules = match ranking_rules.take() {
+        Value::Array(values) => values
+            .into_iter()
+            .filter_map(|value| match value {
+                Value::String(s) if s.starts_with("asc") => asc_ranking_rule(&s)
+                    .map(|f| format!("{}:asc", f))
+                    .map(Value::String),
+                Value::String(s) if s.starts_with("desc") => desc_ranking_rule(&s)
+                    .map(|f| format!("{}:desc", f))
+                    .map(Value::String),
+                otherwise => Some(otherwise),
+            })
+            .collect(),
+        otherwise => otherwise,
+    }
+}

meilisearch-http/src/index/error.rs

@@ -0,0 +1,52 @@
+use std::error::Error;
+
+use meilisearch_error::{Code, ErrorCode};
+use serde_json::Value;
+
+use crate::error::MilliError;
+
+pub type Result<T> = std::result::Result<T, IndexError>;
+
+#[derive(Debug, thiserror::Error)]
+pub enum IndexError {
+    #[error("Internal error: {0}")]
+    Internal(Box<dyn Error + Send + Sync + 'static>),
+    #[error("Document with id {0} not found.")]
+    DocumentNotFound(String),
+    #[error("{0}")]
+    Facet(#[from] FacetError),
+    #[error("{0}")]
+    Milli(#[from] milli::Error),
+}
+
+internal_error!(
+    IndexError: std::io::Error,
+    heed::Error,
+    fst::Error,
+    serde_json::Error
+);
+
+impl ErrorCode for IndexError {
+    fn error_code(&self) -> Code {
+        match self {
+            IndexError::Internal(_) => Code::Internal,
+            IndexError::DocumentNotFound(_) => Code::DocumentNotFound,
+            IndexError::Facet(e) => e.error_code(),
+            IndexError::Milli(e) => MilliError(e).error_code(),
+        }
+    }
+}
+
+#[derive(Debug, thiserror::Error)]
+pub enum FacetError {
+    #[error("Invalid facet expression, expected {}, found: {1}", .0.join(", "))]
+    InvalidExpression(&'static [&'static str], Value),
+}
+
+impl ErrorCode for FacetError {
+    fn error_code(&self) -> Code {
+        match self {
+            FacetError::InvalidExpression(_, _) => Code::Facet,
+        }
+    }
+}

meilisearch-http/src/index/mod.rs

@@ -0,0 +1,198 @@
+use std::collections::{BTreeSet, HashSet};
+use std::fs::create_dir_all;
+use std::marker::PhantomData;
+use std::ops::Deref;
+use std::path::Path;
+use std::sync::Arc;
+
+use heed::{EnvOpenOptions, RoTxn};
+use milli::update::Setting;
+use milli::{obkv_to_json, FieldId};
+use serde_json::{Map, Value};
+
+use error::Result;
+pub use search::{default_crop_length, SearchQuery, SearchResult, DEFAULT_SEARCH_LIMIT};
+pub use updates::{Checked, Facets, Settings, Unchecked};
+
+use crate::helpers::EnvSizer;
+
+use self::error::IndexError;
+
+pub mod error;
+pub mod update_handler;
+
+mod dump;
+mod search;
+mod updates;
+
+pub type Document = Map<String, Value>;
+
+#[derive(Clone)]
+pub struct Index(pub Arc<milli::Index>);
+
+impl Deref for Index {
+    type Target = milli::Index;
+
+    fn deref(&self) -> &Self::Target {
+        self.0.as_ref()
+    }
+}
+
+impl Index {
+    pub fn open(path: impl AsRef<Path>, size: usize) -> Result<Self> {
+        create_dir_all(&path)?;
+        let mut options = EnvOpenOptions::new();
+        options.map_size(size);
+        let index = milli::Index::new(options, &path)?;
+        Ok(Index(Arc::new(index)))
+    }
+
+    pub fn settings(&self) -> Result<Settings<Checked>> {
+        let txn = self.read_txn()?;
+        self.settings_txn(&txn)
+    }
+
+    pub fn settings_txn(&self, txn: &RoTxn) -> Result<Settings<Checked>> {
+        let displayed_attributes = self
+            .displayed_fields(txn)?
+            .map(|fields| fields.into_iter().map(String::from).collect());
+
+        let searchable_attributes = self
+            .searchable_fields(txn)?
+            .map(|fields| fields.into_iter().map(String::from).collect());
+
+        let filterable_attributes = self.filterable_fields(txn)?.into_iter().collect();
+
+        let sortable_attributes = self.sortable_fields(txn)?.into_iter().collect();
+
+        let criteria = self
+            .criteria(txn)?
+            .into_iter()
+            .map(|c| c.to_string())
+            .collect();
+
+        let stop_words = self
+            .stop_words(txn)?
+            .map(|stop_words| -> Result<BTreeSet<_>> {
+                Ok(stop_words.stream().into_strs()?.into_iter().collect())
+            })
+            .transpose()?
+            .unwrap_or_else(BTreeSet::new);
+        let distinct_field = self.distinct_field(txn)?.map(String::from);
+
+        // in milli each word in the synonyms map were split on their separator. Since we lost
+        // this information we are going to put space between words.
+        let synonyms = self
+            .synonyms(txn)?
+            .iter()
+            .map(|(key, values)| {
+                (
+                    key.join(" "),
+                    values.iter().map(|value| value.join(" ")).collect(),
+                )
+            })
+            .collect();
+
+        Ok(Settings {
+            displayed_attributes: match displayed_attributes {
+                Some(attrs) => Setting::Set(attrs),
+                None => Setting::Reset,
+            },
+            searchable_attributes: match searchable_attributes {
+                Some(attrs) => Setting::Set(attrs),
+                None => Setting::Reset,
+            },
+            filterable_attributes: Setting::Set(filterable_attributes),
+            sortable_attributes: Setting::Set(sortable_attributes),
+            ranking_rules: Setting::Set(criteria),
+            stop_words: Setting::Set(stop_words),
+            distinct_attribute: match distinct_field {
+                Some(field) => Setting::Set(field),
+                None => Setting::Reset,
+            },
+            synonyms: Setting::Set(synonyms),
+            _kind: PhantomData,
+        })
+    }
+
+    pub fn retrieve_documents<S: AsRef<str>>(
+        &self,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<S>>,
+    ) -> Result<Vec<Map<String, Value>>> {
+        let txn = self.read_txn()?;
+
+        let fields_ids_map = self.fields_ids_map(&txn)?;
+        let fields_to_display =
+            self.fields_to_display(&txn, &attributes_to_retrieve, &fields_ids_map)?;
+
+        let iter = self.documents.range(&txn, &(..))?.skip(offset).take(limit);
+
+        let mut documents = Vec::new();
+
+        for entry in iter {
+            let (_id, obkv) = entry?;
+            let object = obkv_to_json(&fields_to_display, &fields_ids_map, obkv)?;
+            documents.push(object);
+        }
+
+        Ok(documents)
+    }
+
+    pub fn retrieve_document<S: AsRef<str>>(
+        &self,
+        doc_id: String,
+        attributes_to_retrieve: Option<Vec<S>>,
+    ) -> Result<Map<String, Value>> {
+        let txn = self.read_txn()?;
+
+        let fields_ids_map = self.fields_ids_map(&txn)?;
+
+        let fields_to_display =
+            self.fields_to_display(&txn, &attributes_to_retrieve, &fields_ids_map)?;
+
+        let internal_id = self
+            .external_documents_ids(&txn)?
+            .get(doc_id.as_bytes())
+            .ok_or_else(|| IndexError::DocumentNotFound(doc_id.clone()))?;
+
+        let document = self
+            .documents(&txn, std::iter::once(internal_id))?
+            .into_iter()
+            .next()
+            .map(|(_, d)| d)
+            .ok_or(IndexError::DocumentNotFound(doc_id))?;
+
+        let document = obkv_to_json(&fields_to_display, &fields_ids_map, document)?;
+
+        Ok(document)
+    }
+
+    pub fn size(&self) -> u64 {
+        self.env.size()
+    }
+
+    fn fields_to_display<S: AsRef<str>>(
+        &self,
+        txn: &heed::RoTxn,
+        attributes_to_retrieve: &Option<Vec<S>>,
+        fields_ids_map: &milli::FieldsIdsMap,
+    ) -> Result<Vec<FieldId>> {
+        let mut displayed_fields_ids = match self.displayed_fields_ids(txn)? {
+            Some(ids) => ids.into_iter().collect::<Vec<_>>(),
+            None => fields_ids_map.iter().map(|(id, _)| id).collect(),
+        };
+
+        let attributes_to_retrieve_ids = match attributes_to_retrieve {
+            Some(attrs) => attrs
+                .iter()
+                .filter_map(|f| fields_ids_map.id(f.as_ref()))
+                .collect::<HashSet<_>>(),
+            None => fields_ids_map.iter().map(|(id, _)| id).collect(),
+        };
+
+        displayed_fields_ids.retain(|fid| attributes_to_retrieve_ids.contains(fid));
+        Ok(displayed_fields_ids)
+    }
+}

meilisearch-http/src/index/update_handler.rs

@@ -0,0 +1,89 @@
+use std::fs::File;
+
+use crate::index::Index;
+use milli::update::UpdateBuilder;
+use milli::CompressionType;
+use rayon::ThreadPool;
+
+use crate::index_controller::UpdateMeta;
+use crate::index_controller::{Failed, Processed, Processing};
+use crate::option::IndexerOpts;
+
+pub struct UpdateHandler {
+    max_nb_chunks: Option<usize>,
+    chunk_compression_level: Option<u32>,
+    thread_pool: ThreadPool,
+    log_frequency: usize,
+    max_memory: Option<usize>,
+    chunk_compression_type: CompressionType,
+}
+
+impl UpdateHandler {
+    pub fn new(opt: &IndexerOpts) -> anyhow::Result<Self> {
+        let thread_pool = rayon::ThreadPoolBuilder::new()
+            .num_threads(opt.indexing_jobs.unwrap_or(num_cpus::get() / 2))
+            .build()?;
+
+        Ok(Self {
+            max_nb_chunks: opt.max_nb_chunks,
+            chunk_compression_level: opt.chunk_compression_level,
+            thread_pool,
+            log_frequency: opt.log_every_n,
+            max_memory: opt.max_memory.map(|m| m.get_bytes() as usize),
+            chunk_compression_type: opt.chunk_compression_type,
+        })
+    }
+
+    pub fn update_builder(&self, update_id: u64) -> UpdateBuilder {
+        // We prepare the update by using the update builder.
+        let mut update_builder = UpdateBuilder::new(update_id);
+        if let Some(max_nb_chunks) = self.max_nb_chunks {
+            update_builder.max_nb_chunks(max_nb_chunks);
+        }
+        if let Some(chunk_compression_level) = self.chunk_compression_level {
+            update_builder.chunk_compression_level(chunk_compression_level);
+        }
+        update_builder.thread_pool(&self.thread_pool);
+        update_builder.log_every_n(self.log_frequency);
+        if let Some(max_memory) = self.max_memory {
+            update_builder.max_memory(max_memory);
+        }
+        update_builder.chunk_compression_type(self.chunk_compression_type);
+        update_builder
+    }
+
+    pub fn handle_update(
+        &self,
+        meta: Processing,
+        content: Option<File>,
+        index: Index,
+    ) -> Result<Processed, Failed> {
+        use UpdateMeta::*;
+
+        let update_id = meta.id();
+
+        let update_builder = self.update_builder(update_id);
+
+        let result = match meta.meta() {
+            DocumentsAddition {
+                method,
+                format,
+                primary_key,
+            } => index.update_documents(
+                *format,
+                *method,
+                content,
+                update_builder,
+                primary_key.as_deref(),
+            ),
+            ClearDocuments => index.clear_documents(update_builder),
+            DeleteDocuments { ids } => index.delete_documents(ids, update_builder),
+            Settings(settings) => index.update_settings(&settings.clone().check(), update_builder),
+        };
+
+        match result {
+            Ok(result) => Ok(meta.process(result)),
+            Err(e) => Err(meta.fail(e.into())),
+        }
+    }
+}

meilisearch-http/src/index/updates.rs

@@ -0,0 +1,383 @@
+use std::collections::{BTreeMap, BTreeSet, HashSet};
+use std::io;
+use std::marker::PhantomData;
+use std::num::NonZeroUsize;
+
+use flate2::read::GzDecoder;
+use log::{debug, info, trace};
+use milli::update::{IndexDocumentsMethod, Setting, UpdateBuilder, UpdateFormat};
+use serde::{Deserialize, Serialize, Serializer};
+
+use crate::index_controller::UpdateResult;
+
+use super::error::Result;
+use super::Index;
+
+fn serialize_with_wildcard<S>(
+    field: &Setting<Vec<String>>,
+    s: S,
+) -> std::result::Result<S::Ok, S::Error>
+where
+    S: Serializer,
+{
+    let wildcard = vec!["*".to_string()];
+    match field {
+        Setting::Set(value) => Some(value),
+        Setting::Reset => Some(&wildcard),
+        Setting::NotSet => None,
+    }
+    .serialize(s)
+}
+
+#[derive(Clone, Default, Debug, Serialize)]
+pub struct Checked;
+
+#[derive(Clone, Default, Debug, Serialize, Deserialize)]
+pub struct Unchecked;
+
+#[derive(Debug, Clone, Default, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+#[serde(rename_all = "camelCase")]
+#[serde(bound(serialize = "T: Serialize", deserialize = "T: Deserialize<'static>"))]
+pub struct Settings<T> {
+    #[serde(
+        default,
+        serialize_with = "serialize_with_wildcard",
+        skip_serializing_if = "Setting::is_not_set"
+    )]
+    pub displayed_attributes: Setting<Vec<String>>,
+
+    #[serde(
+        default,
+        serialize_with = "serialize_with_wildcard",
+        skip_serializing_if = "Setting::is_not_set"
+    )]
+    pub searchable_attributes: Setting<Vec<String>>,
+
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub filterable_attributes: Setting<BTreeSet<String>>,
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub sortable_attributes: Setting<BTreeSet<String>>,
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub ranking_rules: Setting<Vec<String>>,
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub stop_words: Setting<BTreeSet<String>>,
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub synonyms: Setting<BTreeMap<String, Vec<String>>>,
+    #[serde(default, skip_serializing_if = "Setting::is_not_set")]
+    pub distinct_attribute: Setting<String>,
+
+    #[serde(skip)]
+    pub _kind: PhantomData<T>,
+}
+
+impl Settings<Checked> {
+    pub fn cleared() -> Settings<Checked> {
+        Settings {
+            displayed_attributes: Setting::Reset,
+            searchable_attributes: Setting::Reset,
+            filterable_attributes: Setting::Reset,
+            sortable_attributes: Setting::Reset,
+            ranking_rules: Setting::Reset,
+            stop_words: Setting::Reset,
+            synonyms: Setting::Reset,
+            distinct_attribute: Setting::Reset,
+            _kind: PhantomData,
+        }
+    }
+
+    pub fn into_unchecked(self) -> Settings<Unchecked> {
+        let Self {
+            displayed_attributes,
+            searchable_attributes,
+            filterable_attributes,
+            sortable_attributes,
+            ranking_rules,
+            stop_words,
+            synonyms,
+            distinct_attribute,
+            ..
+        } = self;
+
+        Settings {
+            displayed_attributes,
+            searchable_attributes,
+            filterable_attributes,
+            sortable_attributes,
+            ranking_rules,
+            stop_words,
+            synonyms,
+            distinct_attribute,
+            _kind: PhantomData,
+        }
+    }
+}
+
+impl Settings<Unchecked> {
+    pub fn check(self) -> Settings<Checked> {
+        let displayed_attributes = match self.displayed_attributes {
+            Setting::Set(fields) => {
+                if fields.iter().any(|f| f == "*") {
+                    Setting::Reset
+                } else {
+                    Setting::Set(fields)
+                }
+            }
+            otherwise => otherwise,
+        };
+
+        let searchable_attributes = match self.searchable_attributes {
+            Setting::Set(fields) => {
+                if fields.iter().any(|f| f == "*") {
+                    Setting::Reset
+                } else {
+                    Setting::Set(fields)
+                }
+            }
+            otherwise => otherwise,
+        };
+
+        Settings {
+            displayed_attributes,
+            searchable_attributes,
+            filterable_attributes: self.filterable_attributes,
+            sortable_attributes: self.sortable_attributes,
+            ranking_rules: self.ranking_rules,
+            stop_words: self.stop_words,
+            synonyms: self.synonyms,
+            distinct_attribute: self.distinct_attribute,
+            _kind: PhantomData,
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(deny_unknown_fields)]
+#[serde(rename_all = "camelCase")]
+pub struct Facets {
+    pub level_group_size: Option<NonZeroUsize>,
+    pub min_level_size: Option<NonZeroUsize>,
+}
+
+impl Index {
+    pub fn update_documents(
+        &self,
+        format: UpdateFormat,
+        method: IndexDocumentsMethod,
+        content: Option<impl io::Read>,
+        update_builder: UpdateBuilder,
+        primary_key: Option<&str>,
+    ) -> Result<UpdateResult> {
+        let mut txn = self.write_txn()?;
+        let result = self.update_documents_txn(
+            &mut txn,
+            format,
+            method,
+            content,
+            update_builder,
+            primary_key,
+        )?;
+        txn.commit()?;
+        Ok(result)
+    }
+
+    pub fn update_documents_txn<'a, 'b>(
+        &'a self,
+        txn: &mut heed::RwTxn<'a, 'b>,
+        format: UpdateFormat,
+        method: IndexDocumentsMethod,
+        content: Option<impl io::Read>,
+        update_builder: UpdateBuilder,
+        primary_key: Option<&str>,
+    ) -> Result<UpdateResult> {
+        trace!("performing document addition");
+
+        // Set the primary key if not set already, ignore if already set.
+        if let (None, Some(primary_key)) = (self.primary_key(txn)?, primary_key) {
+            let mut builder = UpdateBuilder::new(0).settings(txn, self);
+            builder.set_primary_key(primary_key.to_string());
+            builder.execute(|_, _| ())?;
+        }
+
+        let mut builder = update_builder.index_documents(txn, self);
+        builder.update_format(format);
+        builder.index_documents_method(method);
+
+        let indexing_callback =
+            |indexing_step, update_id| debug!("update {}: {:?}", update_id, indexing_step);
+
+        let gzipped = false;
+        let addition = match content {
+            Some(content) if gzipped => {
+                builder.execute(GzDecoder::new(content), indexing_callback)?
+            }
+            Some(content) => builder.execute(content, indexing_callback)?,
+            None => builder.execute(std::io::empty(), indexing_callback)?,
+        };
+
+        info!("document addition done: {:?}", addition);
+
+        Ok(UpdateResult::DocumentsAddition(addition))
+    }
+
+    pub fn clear_documents(&self, update_builder: UpdateBuilder) -> Result<UpdateResult> {
+        // We must use the write transaction of the update here.
+        let mut wtxn = self.write_txn()?;
+        let builder = update_builder.clear_documents(&mut wtxn, self);
+
+        let _count = builder.execute()?;
+
+        wtxn.commit()
+            .and(Ok(UpdateResult::Other))
+            .map_err(Into::into)
+    }
+
+    pub fn update_settings_txn<'a, 'b>(
+        &'a self,
+        txn: &mut heed::RwTxn<'a, 'b>,
+        settings: &Settings<Checked>,
+        update_builder: UpdateBuilder,
+    ) -> Result<UpdateResult> {
+        // We must use the write transaction of the update here.
+        let mut builder = update_builder.settings(txn, self);
+
+        match settings.searchable_attributes {
+            Setting::Set(ref names) => builder.set_searchable_fields(names.clone()),
+            Setting::Reset => builder.reset_searchable_fields(),
+            Setting::NotSet => (),
+        }
+
+        match settings.displayed_attributes {
+            Setting::Set(ref names) => builder.set_displayed_fields(names.clone()),
+            Setting::Reset => builder.reset_displayed_fields(),
+            Setting::NotSet => (),
+        }
+
+        match settings.filterable_attributes {
+            Setting::Set(ref facets) => {
+                builder.set_filterable_fields(facets.clone().into_iter().collect())
+            }
+            Setting::Reset => builder.reset_filterable_fields(),
+            Setting::NotSet => (),
+        }
+
+        match settings.sortable_attributes {
+            Setting::Set(ref fields) => {
+                builder.set_sortable_fields(fields.iter().cloned().collect())
+            }
+            Setting::Reset => {
+                // TODO we must use the reset_sortable_fields in a futur PR.
+                builder.set_sortable_fields(HashSet::new())
+            }
+            Setting::NotSet => (),
+        }
+
+        match settings.ranking_rules {
+            Setting::Set(ref criteria) => builder.set_criteria(criteria.clone()),
+            Setting::Reset => builder.reset_criteria(),
+            Setting::NotSet => (),
+        }
+
+        match settings.stop_words {
+            Setting::Set(ref stop_words) => builder.set_stop_words(stop_words.clone()),
+            Setting::Reset => builder.reset_stop_words(),
+            Setting::NotSet => (),
+        }
+
+        match settings.synonyms {
+            Setting::Set(ref synonyms) => {
+                builder.set_synonyms(synonyms.clone().into_iter().collect())
+            }
+            Setting::Reset => builder.reset_synonyms(),
+            Setting::NotSet => (),
+        }
+
+        match settings.distinct_attribute {
+            Setting::Set(ref attr) => builder.set_distinct_field(attr.clone()),
+            Setting::Reset => builder.reset_distinct_field(),
+            Setting::NotSet => (),
+        }
+
+        builder.execute(|indexing_step, update_id| {
+            debug!("update {}: {:?}", update_id, indexing_step)
+        })?;
+
+        Ok(UpdateResult::Other)
+    }
+
+    pub fn update_settings(
+        &self,
+        settings: &Settings<Checked>,
+        update_builder: UpdateBuilder,
+    ) -> Result<UpdateResult> {
+        let mut txn = self.write_txn()?;
+        let result = self.update_settings_txn(&mut txn, settings, update_builder)?;
+        txn.commit()?;
+        Ok(result)
+    }
+
+    pub fn delete_documents(
+        &self,
+        document_ids: &[String],
+        update_builder: UpdateBuilder,
+    ) -> Result<UpdateResult> {
+        let mut txn = self.write_txn()?;
+        let mut builder = update_builder.delete_documents(&mut txn, self)?;
+
+        // We ignore unexisting document ids
+        document_ids.iter().for_each(|id| {
+            builder.delete_external_id(id);
+        });
+
+        let deleted = builder.execute()?;
+        txn.commit()
+            .and(Ok(UpdateResult::DocumentDeletion { deleted }))
+            .map_err(Into::into)
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn test_setting_check() {
+        // test no changes
+        let settings = Settings {
+            displayed_attributes: Setting::Set(vec![String::from("hello")]),
+            searchable_attributes: Setting::Set(vec![String::from("hello")]),
+            filterable_attributes: Setting::NotSet,
+            sortable_attributes: Setting::NotSet,
+            ranking_rules: Setting::NotSet,
+            stop_words: Setting::NotSet,
+            synonyms: Setting::NotSet,
+            distinct_attribute: Setting::NotSet,
+            _kind: PhantomData::<Unchecked>,
+        };
+
+        let checked = settings.clone().check();
+        assert_eq!(settings.displayed_attributes, checked.displayed_attributes);
+        assert_eq!(
+            settings.searchable_attributes,
+            checked.searchable_attributes
+        );
+
+        // test wildcard
+        // test no changes
+        let settings = Settings {
+            displayed_attributes: Setting::Set(vec![String::from("*")]),
+            searchable_attributes: Setting::Set(vec![String::from("hello"), String::from("*")]),
+            filterable_attributes: Setting::NotSet,
+            sortable_attributes: Setting::NotSet,
+            ranking_rules: Setting::NotSet,
+            stop_words: Setting::NotSet,
+            synonyms: Setting::NotSet,
+            distinct_attribute: Setting::NotSet,
+            _kind: PhantomData::<Unchecked>,
+        };
+
+        let checked = settings.check();
+        assert_eq!(checked.displayed_attributes, Setting::Reset);
+        assert_eq!(checked.searchable_attributes, Setting::Reset);
+    }
+}

meilisearch-http/src/index_controller/dump_actor/actor.rs

@@ -0,0 +1,157 @@
+use std::collections::HashMap;
+use std::path::{Path, PathBuf};
+use std::sync::Arc;
+
+use async_stream::stream;
+use chrono::Utc;
+use futures::{lock::Mutex, stream::StreamExt};
+use log::{error, trace};
+use tokio::sync::{mpsc, oneshot, RwLock};
+use update_actor::UpdateActorHandle;
+use uuid_resolver::UuidResolverHandle;
+
+use super::error::{DumpActorError, Result};
+use super::{DumpInfo, DumpMsg, DumpStatus, DumpTask};
+use crate::index_controller::{update_actor, uuid_resolver};
+
+pub const CONCURRENT_DUMP_MSG: usize = 10;
+
+pub struct DumpActor<UuidResolver, Update> {
+    inbox: Option<mpsc::Receiver<DumpMsg>>,
+    uuid_resolver: UuidResolver,
+    update: Update,
+    dump_path: PathBuf,
+    lock: Arc<Mutex<()>>,
+    dump_infos: Arc<RwLock<HashMap<String, DumpInfo>>>,
+    update_db_size: usize,
+    index_db_size: usize,
+}
+
+/// Generate uid from creation date
+fn generate_uid() -> String {
+    Utc::now().format("%Y%m%d-%H%M%S%3f").to_string()
+}
+
+impl<UuidResolver, Update> DumpActor<UuidResolver, Update>
+where
+    UuidResolver: UuidResolverHandle + Send + Sync + Clone + 'static,
+    Update: UpdateActorHandle + Send + Sync + Clone + 'static,
+{
+    pub fn new(
+        inbox: mpsc::Receiver<DumpMsg>,
+        uuid_resolver: UuidResolver,
+        update: Update,
+        dump_path: impl AsRef<Path>,
+        index_db_size: usize,
+        update_db_size: usize,
+    ) -> Self {
+        let dump_infos = Arc::new(RwLock::new(HashMap::new()));
+        let lock = Arc::new(Mutex::new(()));
+        Self {
+            inbox: Some(inbox),
+            uuid_resolver,
+            update,
+            dump_path: dump_path.as_ref().into(),
+            dump_infos,
+            lock,
+            index_db_size,
+            update_db_size,
+        }
+    }
+
+    pub async fn run(mut self) {
+        trace!("Started dump actor.");
+
+        let mut inbox = self
+            .inbox
+            .take()
+            .expect("Dump Actor must have a inbox at this point.");
+
+        let stream = stream! {
+            loop {
+                match inbox.recv().await {
+                    Some(msg) => yield msg,
+                    None => break,
+                }
+            }
+        };
+
+        stream
+            .for_each_concurrent(Some(CONCURRENT_DUMP_MSG), |msg| self.handle_message(msg))
+            .await;
+
+        error!("Dump actor stopped.");
+    }
+
+    async fn handle_message(&self, msg: DumpMsg) {
+        use DumpMsg::*;
+
+        match msg {
+            CreateDump { ret } => {
+                let _ = self.handle_create_dump(ret).await;
+            }
+            DumpInfo { ret, uid } => {
+                let _ = ret.send(self.handle_dump_info(uid).await);
+            }
+        }
+    }
+
+    async fn handle_create_dump(&self, ret: oneshot::Sender<Result<DumpInfo>>) {
+        let uid = generate_uid();
+        let info = DumpInfo::new(uid.clone(), DumpStatus::InProgress);
+
+        let _lock = match self.lock.try_lock() {
+            Some(lock) => lock,
+            None => {
+                ret.send(Err(DumpActorError::DumpAlreadyRunning))
+                    .expect("Dump actor is dead");
+                return;
+            }
+        };
+
+        self.dump_infos
+            .write()
+            .await
+            .insert(uid.clone(), info.clone());
+
+        ret.send(Ok(info)).expect("Dump actor is dead");
+
+        let task = DumpTask {
+            path: self.dump_path.clone(),
+            uuid_resolver: self.uuid_resolver.clone(),
+            update_handle: self.update.clone(),
+            uid: uid.clone(),
+            update_db_size: self.update_db_size,
+            index_db_size: self.index_db_size,
+        };
+
+        let task_result = tokio::task::spawn(task.run()).await;
+
+        let mut dump_infos = self.dump_infos.write().await;
+        let dump_infos = dump_infos
+            .get_mut(&uid)
+            .expect("dump entry deleted while lock was acquired");
+
+        match task_result {
+            Ok(Ok(())) => {
+                dump_infos.done();
+                trace!("Dump succeed");
+            }
+            Ok(Err(e)) => {
+                dump_infos.with_error(e.to_string());
+                error!("Dump failed: {}", e);
+            }
+            Err(_) => {
+                dump_infos.with_error("Unexpected error while performing dump.".to_string());
+                error!("Dump panicked. Dump status set to failed");
+            }
+        };
+    }
+
+    async fn handle_dump_info(&self, uid: String) -> Result<DumpInfo> {
+        match self.dump_infos.read().await.get(&uid) {
+            Some(info) => Ok(info.clone()),
+            _ => Err(DumpActorError::DumpDoesNotExist(uid)),
+        }
+    }
+}

meilisearch-http/src/index_controller/dump_actor/error.rs

@@ -0,0 +1,52 @@
+use meilisearch_error::{Code, ErrorCode};
+
+use crate::index_controller::update_actor::error::UpdateActorError;
+use crate::index_controller::uuid_resolver::error::UuidResolverError;
+
+pub type Result<T> = std::result::Result<T, DumpActorError>;
+
+#[derive(thiserror::Error, Debug)]
+pub enum DumpActorError {
+    #[error("Another dump is already in progress")]
+    DumpAlreadyRunning,
+    #[error("Dump `{0}` not found")]
+    DumpDoesNotExist(String),
+    #[error("Internal error: {0}")]
+    Internal(Box<dyn std::error::Error + Send + Sync + 'static>),
+    #[error("{0}")]
+    UuidResolver(#[from] UuidResolverError),
+    #[error("{0}")]
+    UpdateActor(#[from] UpdateActorError),
+}
+
+macro_rules! internal_error {
+    ($($other:path), *) => {
+        $(
+            impl From<$other> for DumpActorError {
+                fn from(other: $other) -> Self {
+                    Self::Internal(Box::new(other))
+                }
+            }
+        )*
+    }
+}
+
+internal_error!(
+    heed::Error,
+    std::io::Error,
+    tokio::task::JoinError,
+    serde_json::error::Error,
+    tempfile::PersistError
+);
+
+impl ErrorCode for DumpActorError {
+    fn error_code(&self) -> Code {
+        match self {
+            DumpActorError::DumpAlreadyRunning => Code::DumpAlreadyInProgress,
+            DumpActorError::DumpDoesNotExist(_) => Code::NotFound,
+            DumpActorError::Internal(_) => Code::Internal,
+            DumpActorError::UuidResolver(e) => e.error_code(),
+            DumpActorError::UpdateActor(e) => e.error_code(),
+        }
+    }
+}

meilisearch-http/src/index_controller/dump_actor/handle_impl.rs

@@ -0,0 +1,53 @@
+use std::path::Path;
+
+use actix_web::web::Bytes;
+use tokio::sync::{mpsc, oneshot};
+
+use super::error::Result;
+use super::{DumpActor, DumpActorHandle, DumpInfo, DumpMsg};
+
+#[derive(Clone)]
+pub struct DumpActorHandleImpl {
+    sender: mpsc::Sender<DumpMsg>,
+}
+
+#[async_trait::async_trait]
+impl DumpActorHandle for DumpActorHandleImpl {
+    async fn create_dump(&self) -> Result<DumpInfo> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = DumpMsg::CreateDump { ret };
+        let _ = self.sender.send(msg).await;
+        receiver.await.expect("IndexActor has been killed")
+    }
+
+    async fn dump_info(&self, uid: String) -> Result<DumpInfo> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = DumpMsg::DumpInfo { ret, uid };
+        let _ = self.sender.send(msg).await;
+        receiver.await.expect("IndexActor has been killed")
+    }
+}
+
+impl DumpActorHandleImpl {
+    pub fn new(
+        path: impl AsRef<Path>,
+        uuid_resolver: crate::index_controller::uuid_resolver::UuidResolverHandleImpl,
+        update: crate::index_controller::update_actor::UpdateActorHandleImpl<Bytes>,
+        index_db_size: usize,
+        update_db_size: usize,
+    ) -> anyhow::Result<Self> {
+        let (sender, receiver) = mpsc::channel(10);
+        let actor = DumpActor::new(
+            receiver,
+            uuid_resolver,
+            update,
+            path,
+            index_db_size,
+            update_db_size,
+        );
+
+        tokio::task::spawn(actor.run());
+
+        Ok(Self { sender })
+    }
+}

meilisearch-http/src/index_controller/dump_actor/loaders/mod.rs

@@ -0,0 +1,2 @@
+pub mod v1;
+pub mod v2;

meilisearch-http/src/index_controller/dump_actor/loaders/v1.rs

@@ -0,0 +1,228 @@
+use std::collections::{BTreeMap, BTreeSet};
+use std::fs::{create_dir_all, File};
+use std::io::BufRead;
+use std::marker::PhantomData;
+use std::path::Path;
+use std::sync::Arc;
+
+use heed::EnvOpenOptions;
+use log::{error, info, warn};
+use milli::update::{IndexDocumentsMethod, Setting, UpdateFormat};
+use serde::{Deserialize, Deserializer, Serialize};
+use uuid::Uuid;
+
+use crate::index_controller::{self, uuid_resolver::HeedUuidStore, IndexMetadata};
+use crate::index_controller::{asc_ranking_rule, desc_ranking_rule};
+use crate::{
+    index::{update_handler::UpdateHandler, Index, Unchecked},
+    option::IndexerOpts,
+};
+
+#[derive(Serialize, Deserialize, Debug)]
+#[serde(rename_all = "camelCase")]
+pub struct MetadataV1 {
+    db_version: String,
+    indexes: Vec<IndexMetadata>,
+}
+
+impl MetadataV1 {
+    pub fn load_dump(
+        self,
+        src: impl AsRef<Path>,
+        dst: impl AsRef<Path>,
+        size: usize,
+        indexer_options: &IndexerOpts,
+    ) -> anyhow::Result<()> {
+        info!(
+            "Loading dump, dump database version: {}, dump version: V1",
+            self.db_version
+        );
+
+        let uuid_store = HeedUuidStore::new(&dst)?;
+        for index in self.indexes {
+            let uuid = Uuid::new_v4();
+            uuid_store.insert(index.uid.clone(), uuid)?;
+            let src = src.as_ref().join(index.uid);
+            load_index(
+                &src,
+                &dst,
+                uuid,
+                index.meta.primary_key.as_deref(),
+                size,
+                indexer_options,
+            )?;
+        }
+
+        Ok(())
+    }
+}
+
+pub fn deserialize_some<'de, T, D>(deserializer: D) -> std::result::Result<Option<T>, D::Error>
+where
+    T: Deserialize<'de>,
+    D: Deserializer<'de>,
+{
+    Deserialize::deserialize(deserializer).map(Some)
+}
+
+// These are the settings used in legacy meilisearch (<v0.21.0).
+#[derive(Default, Clone, Serialize, Deserialize, Debug)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+struct Settings {
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub ranking_rules: Option<Option<Vec<String>>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub distinct_attribute: Option<Option<String>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub searchable_attributes: Option<Option<Vec<String>>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub displayed_attributes: Option<Option<BTreeSet<String>>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub stop_words: Option<Option<BTreeSet<String>>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub synonyms: Option<Option<BTreeMap<String, Vec<String>>>>,
+    #[serde(default, deserialize_with = "deserialize_some")]
+    pub attributes_for_faceting: Option<Option<Vec<String>>>,
+}
+
+fn load_index(
+    src: impl AsRef<Path>,
+    dst: impl AsRef<Path>,
+    uuid: Uuid,
+    primary_key: Option<&str>,
+    size: usize,
+    indexer_options: &IndexerOpts,
+) -> anyhow::Result<()> {
+    let index_path = dst.as_ref().join(&format!("indexes/index-{}", uuid));
+
+    create_dir_all(&index_path)?;
+    let mut options = EnvOpenOptions::new();
+    options.map_size(size);
+    let index = milli::Index::new(options, index_path)?;
+    let index = Index(Arc::new(index));
+
+    // extract `settings.json` file and import content
+    let settings = import_settings(&src)?;
+    let settings: index_controller::Settings<Unchecked> = settings.into();
+
+    let mut txn = index.write_txn()?;
+
+    let handler = UpdateHandler::new(indexer_options)?;
+
+    index.update_settings_txn(&mut txn, &settings.check(), handler.update_builder(0))?;
+
+    let file = File::open(&src.as_ref().join("documents.jsonl"))?;
+    let mut reader = std::io::BufReader::new(file);
+    reader.fill_buf()?;
+    if !reader.buffer().is_empty() {
+        index.update_documents_txn(
+            &mut txn,
+            UpdateFormat::JsonStream,
+            IndexDocumentsMethod::ReplaceDocuments,
+            Some(reader),
+            handler.update_builder(0),
+            primary_key,
+        )?;
+    }
+
+    txn.commit()?;
+
+    // Finaly, we extract the original milli::Index and close it
+    Arc::try_unwrap(index.0)
+        .map_err(|_e| "Couldn't close the index properly")
+        .unwrap()
+        .prepare_for_closing()
+        .wait();
+
+    // Updates are ignored in dumps V1.
+
+    Ok(())
+}
+
+/// we need to **always** be able to convert the old settings to the settings currently being used
+impl From<Settings> for index_controller::Settings<Unchecked> {
+    fn from(settings: Settings) -> Self {
+        Self {
+            distinct_attribute: match settings.distinct_attribute {
+                Some(Some(attr)) => Setting::Set(attr),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            // we need to convert the old `Vec<String>` into a `BTreeSet<String>`
+            displayed_attributes: match settings.displayed_attributes {
+                Some(Some(attrs)) => Setting::Set(attrs.into_iter().collect()),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            searchable_attributes: match settings.searchable_attributes {
+                Some(Some(attrs)) => Setting::Set(attrs),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            filterable_attributes: match settings.attributes_for_faceting {
+                Some(Some(attrs)) => Setting::Set(attrs.into_iter().collect()),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            sortable_attributes: Setting::NotSet,
+            ranking_rules: match settings.ranking_rules {
+                Some(Some(ranking_rules)) => Setting::Set(ranking_rules.into_iter().filter_map(|criterion| {
+                    match criterion.as_str() {
+                        "words" | "typo" | "proximity" | "attribute" | "exactness" => Some(criterion),
+                        s if s.starts_with("asc") => asc_ranking_rule(s).map(|f| format!("{}:asc", f)),
+                        s if s.starts_with("desc") => desc_ranking_rule(s).map(|f| format!("{}:desc", f)),
+                        "wordsPosition" => {
+                            warn!("The criteria `attribute` and `wordsPosition` have been merged \
+                                into a single criterion `attribute` so `wordsPositon` will be \
+                                ignored");
+                            None
+                        }
+                        s => {
+                            error!("Unknown criterion found in the dump: `{}`, it will be ignored", s);
+                            None
+                        }
+                    }
+                }).collect()),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            // we need to convert the old `Vec<String>` into a `BTreeSet<String>`
+            stop_words: match settings.stop_words {
+                Some(Some(stop_words)) => Setting::Set(stop_words.into_iter().collect()),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            // we need to convert the old `Vec<String>` into a `BTreeMap<String>`
+            synonyms: match settings.synonyms {
+                Some(Some(synonyms)) => Setting::Set(synonyms.into_iter().collect()),
+                Some(None) => Setting::Reset,
+                None => Setting::NotSet
+            },
+            _kind: PhantomData,
+        }
+    }
+}
+
+/// Extract Settings from `settings.json` file present at provided `dir_path`
+fn import_settings(dir_path: impl AsRef<Path>) -> anyhow::Result<Settings> {
+    let path = dir_path.as_ref().join("settings.json");
+    let file = File::open(path)?;
+    let reader = std::io::BufReader::new(file);
+    let metadata = serde_json::from_reader(reader)?;
+
+    Ok(metadata)
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn settings_format_regression() {
+        let settings = Settings::default();
+        assert_eq!(
+            r##"{"rankingRules":null,"distinctAttribute":null,"searchableAttributes":null,"displayedAttributes":null,"stopWords":null,"synonyms":null,"attributesForFaceting":null}"##,
+            serde_json::to_string(&settings).unwrap()
+        );
+    }
+}

meilisearch-http/src/index_controller/dump_actor/loaders/v2.rs

@@ -0,0 +1,59 @@
+use std::path::Path;
+
+use chrono::{DateTime, Utc};
+use log::info;
+use serde::{Deserialize, Serialize};
+
+use crate::index::Index;
+use crate::index_controller::{update_actor::UpdateStore, uuid_resolver::HeedUuidStore};
+use crate::option::IndexerOpts;
+
+#[derive(Serialize, Deserialize, Debug)]
+#[serde(rename_all = "camelCase")]
+pub struct MetadataV2 {
+    db_version: String,
+    index_db_size: usize,
+    update_db_size: usize,
+    dump_date: DateTime<Utc>,
+}
+
+impl MetadataV2 {
+    pub fn new(index_db_size: usize, update_db_size: usize) -> Self {
+        Self {
+            db_version: env!("CARGO_PKG_VERSION").to_string(),
+            index_db_size,
+            update_db_size,
+            dump_date: Utc::now(),
+        }
+    }
+
+    pub fn load_dump(
+        self,
+        src: impl AsRef<Path>,
+        dst: impl AsRef<Path>,
+        index_db_size: usize,
+        update_db_size: usize,
+        indexing_options: &IndexerOpts,
+    ) -> anyhow::Result<()> {
+        info!(
+            "Loading dump from {}, dump database version: {}, dump version: V2",
+            self.dump_date, self.db_version
+        );
+
+        info!("Loading index database.");
+        HeedUuidStore::load_dump(src.as_ref(), &dst)?;
+
+        info!("Loading updates.");
+        UpdateStore::load_dump(&src, &dst, update_db_size)?;
+
+        info!("Loading indexes.");
+        let indexes_path = src.as_ref().join("indexes");
+        let indexes = indexes_path.read_dir()?;
+        for index in indexes {
+            let index = index?;
+            Index::load_dump(&index.path(), &dst, index_db_size, indexing_options)?;
+        }
+
+        Ok(())
+    }
+}

meilisearch-http/src/index_controller/dump_actor/message.rs

@@ -0,0 +1,14 @@
+use tokio::sync::oneshot;
+
+use super::error::Result;
+use super::DumpInfo;
+
+pub enum DumpMsg {
+    CreateDump {
+        ret: oneshot::Sender<Result<DumpInfo>>,
+    },
+    DumpInfo {
+        uid: String,
+        ret: oneshot::Sender<Result<DumpInfo>>,
+    },
+}

meilisearch-http/src/index_controller/dump_actor/mod.rs

@@ -0,0 +1,203 @@
+use std::fs::File;
+use std::path::{Path, PathBuf};
+
+use anyhow::Context;
+use chrono::{DateTime, Utc};
+use log::{info, trace, warn};
+#[cfg(test)]
+use mockall::automock;
+use serde::{Deserialize, Serialize};
+use tokio::fs::create_dir_all;
+
+use loaders::v1::MetadataV1;
+use loaders::v2::MetadataV2;
+
+pub use actor::DumpActor;
+pub use handle_impl::*;
+pub use message::DumpMsg;
+
+use super::{update_actor::UpdateActorHandle, uuid_resolver::UuidResolverHandle};
+use crate::index_controller::dump_actor::error::DumpActorError;
+use crate::{helpers::compression, option::IndexerOpts};
+use error::Result;
+
+mod actor;
+pub mod error;
+mod handle_impl;
+mod loaders;
+mod message;
+
+const META_FILE_NAME: &str = "metadata.json";
+
+#[async_trait::async_trait]
+#[cfg_attr(test, automock)]
+pub trait DumpActorHandle {
+    /// Start the creation of a dump
+    /// Implementation: [handle_impl::DumpActorHandleImpl::create_dump]
+    async fn create_dump(&self) -> Result<DumpInfo>;
+
+    /// Return the status of an already created dump
+    /// Implementation: [handle_impl::DumpActorHandleImpl::dump_info]
+    async fn dump_info(&self, uid: String) -> Result<DumpInfo>;
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+#[serde(tag = "dumpVersion")]
+pub enum Metadata {
+    V1(MetadataV1),
+    V2(MetadataV2),
+}
+
+impl Metadata {
+    pub fn new_v2(index_db_size: usize, update_db_size: usize) -> Self {
+        let meta = MetadataV2::new(index_db_size, update_db_size);
+        Self::V2(meta)
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize, PartialEq, Clone)]
+#[serde(rename_all = "snake_case")]
+pub enum DumpStatus {
+    Done,
+    InProgress,
+    Failed,
+}
+
+#[derive(Debug, Serialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct DumpInfo {
+    pub uid: String,
+    pub status: DumpStatus,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub error: Option<String>,
+    started_at: DateTime<Utc>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    finished_at: Option<DateTime<Utc>>,
+}
+
+impl DumpInfo {
+    pub fn new(uid: String, status: DumpStatus) -> Self {
+        Self {
+            uid,
+            status,
+            error: None,
+            started_at: Utc::now(),
+            finished_at: None,
+        }
+    }
+
+    pub fn with_error(&mut self, error: String) {
+        self.status = DumpStatus::Failed;
+        self.finished_at = Some(Utc::now());
+        self.error = Some(error);
+    }
+
+    pub fn done(&mut self) {
+        self.finished_at = Some(Utc::now());
+        self.status = DumpStatus::Done;
+    }
+
+    pub fn dump_already_in_progress(&self) -> bool {
+        self.status == DumpStatus::InProgress
+    }
+}
+
+pub fn load_dump(
+    dst_path: impl AsRef<Path>,
+    src_path: impl AsRef<Path>,
+    index_db_size: usize,
+    update_db_size: usize,
+    indexer_opts: &IndexerOpts,
+) -> anyhow::Result<()> {
+    let tmp_src = tempfile::tempdir_in(".")?;
+    let tmp_src_path = tmp_src.path();
+
+    compression::from_tar_gz(&src_path, tmp_src_path)?;
+
+    let meta_path = tmp_src_path.join(META_FILE_NAME);
+    let mut meta_file = File::open(&meta_path)?;
+    let meta: Metadata = serde_json::from_reader(&mut meta_file)?;
+
+    let dst_dir = dst_path
+        .as_ref()
+        .parent()
+        .with_context(|| format!("Invalid db path: {}", dst_path.as_ref().display()))?;
+
+    let tmp_dst = tempfile::tempdir_in(dst_dir)?;
+
+    match meta {
+        Metadata::V1(meta) => {
+            meta.load_dump(&tmp_src_path, tmp_dst.path(), index_db_size, indexer_opts)?
+        }
+        Metadata::V2(meta) => meta.load_dump(
+            &tmp_src_path,
+            tmp_dst.path(),
+            index_db_size,
+            update_db_size,
+            indexer_opts,
+        )?,
+    }
+    // Persist and atomically rename the db
+    let persisted_dump = tmp_dst.into_path();
+    if dst_path.as_ref().exists() {
+        warn!("Overwriting database at {}", dst_path.as_ref().display());
+        std::fs::remove_dir_all(&dst_path)?;
+    }
+
+    std::fs::rename(&persisted_dump, &dst_path)?;
+
+    Ok(())
+}
+
+struct DumpTask<U, P> {
+    path: PathBuf,
+    uuid_resolver: U,
+    update_handle: P,
+    uid: String,
+    update_db_size: usize,
+    index_db_size: usize,
+}
+
+impl<U, P> DumpTask<U, P>
+where
+    U: UuidResolverHandle + Send + Sync + Clone + 'static,
+    P: UpdateActorHandle + Send + Sync + Clone + 'static,
+{
+    async fn run(self) -> Result<()> {
+        trace!("Performing dump.");
+
+        create_dir_all(&self.path).await?;
+
+        let path_clone = self.path.clone();
+        let temp_dump_dir =
+            tokio::task::spawn_blocking(|| tempfile::TempDir::new_in(path_clone)).await??;
+        let temp_dump_path = temp_dump_dir.path().to_owned();
+
+        let meta = Metadata::new_v2(self.index_db_size, self.update_db_size);
+        let meta_path = temp_dump_path.join(META_FILE_NAME);
+        let mut meta_file = File::create(&meta_path)?;
+        serde_json::to_writer(&mut meta_file, &meta)?;
+
+        let uuids = self.uuid_resolver.dump(temp_dump_path.clone()).await?;
+
+        self.update_handle
+            .dump(uuids, temp_dump_path.clone())
+            .await?;
+
+        let dump_path = tokio::task::spawn_blocking(move || -> Result<PathBuf> {
+            let temp_dump_file = tempfile::NamedTempFile::new_in(&self.path)?;
+            compression::to_tar_gz(temp_dump_path, temp_dump_file.path())
+                .map_err(|e| DumpActorError::Internal(e.into()))?;
+
+            let dump_path = self.path.join(self.uid).with_extension("dump");
+            temp_dump_file.persist(&dump_path)?;
+
+            Ok(dump_path)
+        })
+        .await??;
+
+        info!("Created dump in {:?}.", dump_path);
+
+        Ok(())
+    }
+}

meilisearch-http/src/index_controller/error.rs

@@ -0,0 +1,40 @@
+use meilisearch_error::Code;
+use meilisearch_error::ErrorCode;
+
+use crate::index::error::IndexError;
+
+use super::dump_actor::error::DumpActorError;
+use super::index_actor::error::IndexActorError;
+use super::update_actor::error::UpdateActorError;
+use super::uuid_resolver::error::UuidResolverError;
+
+pub type Result<T> = std::result::Result<T, IndexControllerError>;
+
+#[derive(Debug, thiserror::Error)]
+pub enum IndexControllerError {
+    #[error("Index creation must have an uid")]
+    MissingUid,
+    #[error("{0}")]
+    Uuid(#[from] UuidResolverError),
+    #[error("{0}")]
+    IndexActor(#[from] IndexActorError),
+    #[error("{0}")]
+    UpdateActor(#[from] UpdateActorError),
+    #[error("{0}")]
+    DumpActor(#[from] DumpActorError),
+    #[error("{0}")]
+    IndexError(#[from] IndexError),
+}
+
+impl ErrorCode for IndexControllerError {
+    fn error_code(&self) -> Code {
+        match self {
+            IndexControllerError::MissingUid => Code::BadRequest,
+            IndexControllerError::Uuid(e) => e.error_code(),
+            IndexControllerError::IndexActor(e) => e.error_code(),
+            IndexControllerError::UpdateActor(e) => e.error_code(),
+            IndexControllerError::DumpActor(e) => e.error_code(),
+            IndexControllerError::IndexError(e) => e.error_code(),
+        }
+    }
+}

meilisearch-http/src/index_controller/index_actor/actor.rs

@@ -0,0 +1,351 @@
+use std::fs::File;
+use std::path::PathBuf;
+use std::sync::Arc;
+
+use async_stream::stream;
+use futures::stream::StreamExt;
+use heed::CompactionOption;
+use log::debug;
+use milli::update::UpdateBuilder;
+use tokio::task::spawn_blocking;
+use tokio::{fs, sync::mpsc};
+use uuid::Uuid;
+
+use crate::index::{
+    update_handler::UpdateHandler, Checked, Document, SearchQuery, SearchResult, Settings,
+};
+use crate::index_controller::{
+    get_arc_ownership_blocking, Failed, IndexStats, Processed, Processing,
+};
+use crate::option::IndexerOpts;
+
+use super::error::{IndexActorError, Result};
+use super::{IndexMeta, IndexMsg, IndexSettings, IndexStore};
+
+pub const CONCURRENT_INDEX_MSG: usize = 10;
+
+pub struct IndexActor<S> {
+    receiver: Option<mpsc::Receiver<IndexMsg>>,
+    update_handler: Arc<UpdateHandler>,
+    store: S,
+}
+
+impl<S: IndexStore + Sync + Send> IndexActor<S> {
+    pub fn new(
+        receiver: mpsc::Receiver<IndexMsg>,
+        store: S,
+        options: &IndexerOpts,
+    ) -> anyhow::Result<Self> {
+        let update_handler = UpdateHandler::new(options)?;
+        let update_handler = Arc::new(update_handler);
+        let receiver = Some(receiver);
+        Ok(Self {
+            receiver,
+            update_handler,
+            store,
+        })
+    }
+
+    /// `run` poll the write_receiver and read_receiver concurrently, but while messages send
+    /// through the read channel are processed concurrently, the messages sent through the write
+    /// channel are processed one at a time.
+    pub async fn run(mut self) {
+        let mut receiver = self
+            .receiver
+            .take()
+            .expect("Index Actor must have a inbox at this point.");
+
+        let stream = stream! {
+            loop {
+                match receiver.recv().await {
+                    Some(msg) => yield msg,
+                    None => break,
+                }
+            }
+        };
+
+        stream
+            .for_each_concurrent(Some(CONCURRENT_INDEX_MSG), |msg| self.handle_message(msg))
+            .await;
+    }
+
+    async fn handle_message(&self, msg: IndexMsg) {
+        use IndexMsg::*;
+        match msg {
+            CreateIndex {
+                uuid,
+                primary_key,
+                ret,
+            } => {
+                let _ = ret.send(self.handle_create_index(uuid, primary_key).await);
+            }
+            Update {
+                ret,
+                meta,
+                data,
+                uuid,
+            } => {
+                let _ = ret.send(self.handle_update(uuid, meta, data).await);
+            }
+            Search { ret, query, uuid } => {
+                let _ = ret.send(self.handle_search(uuid, query).await);
+            }
+            Settings { ret, uuid } => {
+                let _ = ret.send(self.handle_settings(uuid).await);
+            }
+            Documents {
+                ret,
+                uuid,
+                attributes_to_retrieve,
+                offset,
+                limit,
+            } => {
+                let _ = ret.send(
+                    self.handle_fetch_documents(uuid, offset, limit, attributes_to_retrieve)
+                        .await,
+                );
+            }
+            Document {
+                uuid,
+                attributes_to_retrieve,
+                doc_id,
+                ret,
+            } => {
+                let _ = ret.send(
+                    self.handle_fetch_document(uuid, doc_id, attributes_to_retrieve)
+                        .await,
+                );
+            }
+            Delete { uuid, ret } => {
+                let _ = ret.send(self.handle_delete(uuid).await);
+            }
+            GetMeta { uuid, ret } => {
+                let _ = ret.send(self.handle_get_meta(uuid).await);
+            }
+            UpdateIndex {
+                uuid,
+                index_settings,
+                ret,
+            } => {
+                let _ = ret.send(self.handle_update_index(uuid, index_settings).await);
+            }
+            Snapshot { uuid, path, ret } => {
+                let _ = ret.send(self.handle_snapshot(uuid, path).await);
+            }
+            Dump { uuid, path, ret } => {
+                let _ = ret.send(self.handle_dump(uuid, path).await);
+            }
+            GetStats { uuid, ret } => {
+                let _ = ret.send(self.handle_get_stats(uuid).await);
+            }
+        }
+    }
+
+    async fn handle_search(&self, uuid: Uuid, query: SearchQuery) -> Result<SearchResult> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+        let result = spawn_blocking(move || index.perform_search(query)).await??;
+        Ok(result)
+    }
+
+    async fn handle_create_index(
+        &self,
+        uuid: Uuid,
+        primary_key: Option<String>,
+    ) -> Result<IndexMeta> {
+        let index = self.store.create(uuid, primary_key).await?;
+        let meta = spawn_blocking(move || IndexMeta::new(&index)).await??;
+        Ok(meta)
+    }
+
+    async fn handle_update(
+        &self,
+        uuid: Uuid,
+        meta: Processing,
+        data: Option<File>,
+    ) -> Result<std::result::Result<Processed, Failed>> {
+        debug!("Processing update {}", meta.id());
+        let update_handler = self.update_handler.clone();
+        let index = match self.store.get(uuid).await? {
+            Some(index) => index,
+            None => self.store.create(uuid, None).await?,
+        };
+
+        Ok(spawn_blocking(move || update_handler.handle_update(meta, data, index)).await?)
+    }
+
+    async fn handle_settings(&self, uuid: Uuid) -> Result<Settings<Checked>> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+        let result = spawn_blocking(move || index.settings()).await??;
+        Ok(result)
+    }
+
+    async fn handle_fetch_documents(
+        &self,
+        uuid: Uuid,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Vec<Document>> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+        let result =
+            spawn_blocking(move || index.retrieve_documents(offset, limit, attributes_to_retrieve))
+                .await??;
+
+        Ok(result)
+    }
+
+    async fn handle_fetch_document(
+        &self,
+        uuid: Uuid,
+        doc_id: String,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Document> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+
+        let result =
+            spawn_blocking(move || index.retrieve_document(doc_id, attributes_to_retrieve))
+                .await??;
+
+        Ok(result)
+    }
+
+    async fn handle_delete(&self, uuid: Uuid) -> Result<()> {
+        let index = self.store.delete(uuid).await?;
+
+        if let Some(index) = index {
+            tokio::task::spawn(async move {
+                let index = index.0;
+                let store = get_arc_ownership_blocking(index).await;
+                spawn_blocking(move || {
+                    store.prepare_for_closing().wait();
+                    debug!("Index closed");
+                });
+            });
+        }
+
+        Ok(())
+    }
+
+    async fn handle_get_meta(&self, uuid: Uuid) -> Result<IndexMeta> {
+        match self.store.get(uuid).await? {
+            Some(index) => {
+                let meta = spawn_blocking(move || IndexMeta::new(&index)).await??;
+                Ok(meta)
+            }
+            None => Err(IndexActorError::UnexistingIndex),
+        }
+    }
+
+    async fn handle_update_index(
+        &self,
+        uuid: Uuid,
+        index_settings: IndexSettings,
+    ) -> Result<IndexMeta> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+
+        let result = spawn_blocking(move || match index_settings.primary_key {
+            Some(primary_key) => {
+                let mut txn = index.write_txn()?;
+                if index.primary_key(&txn)?.is_some() {
+                    return Err(IndexActorError::ExistingPrimaryKey);
+                }
+                let mut builder = UpdateBuilder::new(0).settings(&mut txn, &index);
+                builder.set_primary_key(primary_key);
+                builder.execute(|_, _| ())?;
+                let meta = IndexMeta::new_txn(&index, &txn)?;
+                txn.commit()?;
+                Ok(meta)
+            }
+            None => {
+                let meta = IndexMeta::new(&index)?;
+                Ok(meta)
+            }
+        })
+        .await??;
+
+        Ok(result)
+    }
+
+    async fn handle_snapshot(&self, uuid: Uuid, mut path: PathBuf) -> Result<()> {
+        use tokio::fs::create_dir_all;
+
+        path.push("indexes");
+        create_dir_all(&path).await?;
+
+        if let Some(index) = self.store.get(uuid).await? {
+            let mut index_path = path.join(format!("index-{}", uuid));
+
+            create_dir_all(&index_path).await?;
+
+            index_path.push("data.mdb");
+            spawn_blocking(move || -> Result<()> {
+                // Get write txn to wait for ongoing write transaction before snapshot.
+                let _txn = index.write_txn()?;
+                index
+                    .env
+                    .copy_to_path(index_path, CompactionOption::Enabled)?;
+                Ok(())
+            })
+            .await??;
+        }
+
+        Ok(())
+    }
+
+    /// Create a `documents.jsonl` and a `settings.json` in `path/uid/` with a dump of all the
+    /// documents and all the settings.
+    async fn handle_dump(&self, uuid: Uuid, path: PathBuf) -> Result<()> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+
+        let path = path.join(format!("indexes/index-{}/", uuid));
+        fs::create_dir_all(&path).await?;
+
+        tokio::task::spawn_blocking(move || index.dump(path)).await??;
+
+        Ok(())
+    }
+
+    async fn handle_get_stats(&self, uuid: Uuid) -> Result<IndexStats> {
+        let index = self
+            .store
+            .get(uuid)
+            .await?
+            .ok_or(IndexActorError::UnexistingIndex)?;
+
+        spawn_blocking(move || {
+            let rtxn = index.read_txn()?;
+
+            Ok(IndexStats {
+                size: index.size(),
+                number_of_documents: index.number_of_documents(&rtxn)?,
+                is_indexing: None,
+                field_distribution: index.field_distribution(&rtxn)?,
+            })
+        })
+        .await?
+    }
+}

meilisearch-http/src/index_controller/index_actor/error.rs

@@ -0,0 +1,48 @@
+use meilisearch_error::{Code, ErrorCode};
+
+use crate::{error::MilliError, index::error::IndexError};
+
+pub type Result<T> = std::result::Result<T, IndexActorError>;
+
+#[derive(thiserror::Error, Debug)]
+pub enum IndexActorError {
+    #[error("{0}")]
+    IndexError(#[from] IndexError),
+    #[error("Index already exists")]
+    IndexAlreadyExists,
+    #[error("Index not found")]
+    UnexistingIndex,
+    #[error("A primary key is already present. It's impossible to update it")]
+    ExistingPrimaryKey,
+    #[error("Internal Error: {0}")]
+    Internal(Box<dyn std::error::Error + Send + Sync + 'static>),
+    #[error("{0}")]
+    Milli(#[from] milli::Error),
+}
+
+macro_rules! internal_error {
+    ($($other:path), *) => {
+        $(
+            impl From<$other> for IndexActorError {
+                fn from(other: $other) -> Self {
+                    Self::Internal(Box::new(other))
+                }
+            }
+        )*
+    }
+}
+
+internal_error!(heed::Error, tokio::task::JoinError, std::io::Error);
+
+impl ErrorCode for IndexActorError {
+    fn error_code(&self) -> Code {
+        match self {
+            IndexActorError::IndexError(e) => e.error_code(),
+            IndexActorError::IndexAlreadyExists => Code::IndexAlreadyExists,
+            IndexActorError::UnexistingIndex => Code::IndexNotFound,
+            IndexActorError::ExistingPrimaryKey => Code::PrimaryKeyAlreadyPresent,
+            IndexActorError::Internal(_) => Code::Internal,
+            IndexActorError::Milli(e) => MilliError(e).error_code(),
+        }
+    }
+}

meilisearch-http/src/index_controller/index_actor/handle_impl.rs

@@ -0,0 +1,164 @@
+use crate::option::IndexerOpts;
+use std::path::{Path, PathBuf};
+
+use tokio::sync::{mpsc, oneshot};
+use uuid::Uuid;
+
+use crate::{
+    index::Checked,
+    index_controller::{IndexSettings, IndexStats, Processing},
+};
+use crate::{
+    index::{Document, SearchQuery, SearchResult, Settings},
+    index_controller::{Failed, Processed},
+};
+
+use super::error::Result;
+use super::{IndexActor, IndexActorHandle, IndexMeta, IndexMsg, MapIndexStore};
+
+#[derive(Clone)]
+pub struct IndexActorHandleImpl {
+    sender: mpsc::Sender<IndexMsg>,
+}
+
+#[async_trait::async_trait]
+impl IndexActorHandle for IndexActorHandleImpl {
+    async fn create_index(&self, uuid: Uuid, primary_key: Option<String>) -> Result<IndexMeta> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::CreateIndex {
+            ret,
+            uuid,
+            primary_key,
+        };
+        let _ = self.sender.send(msg).await;
+        receiver.await.expect("IndexActor has been killed")
+    }
+
+    async fn update(
+        &self,
+        uuid: Uuid,
+        meta: Processing,
+        data: Option<std::fs::File>,
+    ) -> Result<std::result::Result<Processed, Failed>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Update {
+            ret,
+            meta,
+            data,
+            uuid,
+        };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn search(&self, uuid: Uuid, query: SearchQuery) -> Result<SearchResult> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Search { uuid, query, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn settings(&self, uuid: Uuid) -> Result<Settings<Checked>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Settings { uuid, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn documents(
+        &self,
+        uuid: Uuid,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Vec<Document>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Documents {
+            uuid,
+            ret,
+            offset,
+            attributes_to_retrieve,
+            limit,
+        };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn document(
+        &self,
+        uuid: Uuid,
+        doc_id: String,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Document> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Document {
+            uuid,
+            ret,
+            doc_id,
+            attributes_to_retrieve,
+        };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn delete(&self, uuid: Uuid) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Delete { uuid, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn get_index_meta(&self, uuid: Uuid) -> Result<IndexMeta> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::GetMeta { uuid, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn update_index(&self, uuid: Uuid, index_settings: IndexSettings) -> Result<IndexMeta> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::UpdateIndex {
+            uuid,
+            index_settings,
+            ret,
+        };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn snapshot(&self, uuid: Uuid, path: PathBuf) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Snapshot { uuid, path, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn dump(&self, uuid: Uuid, path: PathBuf) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::Dump { uuid, path, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+
+    async fn get_index_stats(&self, uuid: Uuid) -> Result<IndexStats> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = IndexMsg::GetStats { uuid, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver.await.expect("IndexActor has been killed")?)
+    }
+}
+
+impl IndexActorHandleImpl {
+    pub fn new(
+        path: impl AsRef<Path>,
+        index_size: usize,
+        options: &IndexerOpts,
+    ) -> anyhow::Result<Self> {
+        let (sender, receiver) = mpsc::channel(100);
+
+        let store = MapIndexStore::new(path, index_size);
+        let actor = IndexActor::new(receiver, store, options)?;
+        tokio::task::spawn(actor.run());
+        Ok(Self { sender })
+    }
+}

meilisearch-http/src/index_controller/index_actor/message.rs

@@ -0,0 +1,74 @@
+use std::path::PathBuf;
+
+use tokio::sync::oneshot;
+use uuid::Uuid;
+
+use super::error::Result as IndexResult;
+use crate::index::{Checked, Document, SearchQuery, SearchResult, Settings};
+use crate::index_controller::{Failed, IndexStats, Processed, Processing};
+
+use super::{IndexMeta, IndexSettings};
+
+#[allow(clippy::large_enum_variant)]
+pub enum IndexMsg {
+    CreateIndex {
+        uuid: Uuid,
+        primary_key: Option<String>,
+        ret: oneshot::Sender<IndexResult<IndexMeta>>,
+    },
+    Update {
+        uuid: Uuid,
+        meta: Processing,
+        data: Option<std::fs::File>,
+        ret: oneshot::Sender<IndexResult<Result<Processed, Failed>>>,
+    },
+    Search {
+        uuid: Uuid,
+        query: SearchQuery,
+        ret: oneshot::Sender<IndexResult<SearchResult>>,
+    },
+    Settings {
+        uuid: Uuid,
+        ret: oneshot::Sender<IndexResult<Settings<Checked>>>,
+    },
+    Documents {
+        uuid: Uuid,
+        attributes_to_retrieve: Option<Vec<String>>,
+        offset: usize,
+        limit: usize,
+        ret: oneshot::Sender<IndexResult<Vec<Document>>>,
+    },
+    Document {
+        uuid: Uuid,
+        attributes_to_retrieve: Option<Vec<String>>,
+        doc_id: String,
+        ret: oneshot::Sender<IndexResult<Document>>,
+    },
+    Delete {
+        uuid: Uuid,
+        ret: oneshot::Sender<IndexResult<()>>,
+    },
+    GetMeta {
+        uuid: Uuid,
+        ret: oneshot::Sender<IndexResult<IndexMeta>>,
+    },
+    UpdateIndex {
+        uuid: Uuid,
+        index_settings: IndexSettings,
+        ret: oneshot::Sender<IndexResult<IndexMeta>>,
+    },
+    Snapshot {
+        uuid: Uuid,
+        path: PathBuf,
+        ret: oneshot::Sender<IndexResult<()>>,
+    },
+    Dump {
+        uuid: Uuid,
+        path: PathBuf,
+        ret: oneshot::Sender<IndexResult<()>>,
+    },
+    GetStats {
+        uuid: Uuid,
+        ret: oneshot::Sender<IndexResult<IndexStats>>,
+    },
+}

meilisearch-http/src/index_controller/index_actor/mod.rs

@@ -0,0 +1,169 @@
+use std::fs::File;
+use std::path::PathBuf;
+
+use chrono::{DateTime, Utc};
+#[cfg(test)]
+use mockall::automock;
+use serde::{Deserialize, Serialize};
+use uuid::Uuid;
+
+use actor::IndexActor;
+pub use actor::CONCURRENT_INDEX_MSG;
+pub use handle_impl::IndexActorHandleImpl;
+use message::IndexMsg;
+use store::{IndexStore, MapIndexStore};
+
+use crate::index::{Checked, Document, Index, SearchQuery, SearchResult, Settings};
+use crate::index_controller::{Failed, IndexStats, Processed, Processing};
+use error::Result;
+
+use super::IndexSettings;
+
+mod actor;
+pub mod error;
+mod handle_impl;
+mod message;
+mod store;
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct IndexMeta {
+    created_at: DateTime<Utc>,
+    pub updated_at: DateTime<Utc>,
+    pub primary_key: Option<String>,
+}
+
+impl IndexMeta {
+    fn new(index: &Index) -> Result<Self> {
+        let txn = index.read_txn()?;
+        Self::new_txn(index, &txn)
+    }
+
+    fn new_txn(index: &Index, txn: &heed::RoTxn) -> Result<Self> {
+        let created_at = index.created_at(txn)?;
+        let updated_at = index.updated_at(txn)?;
+        let primary_key = index.primary_key(txn)?.map(String::from);
+        Ok(Self {
+            created_at,
+            updated_at,
+            primary_key,
+        })
+    }
+}
+
+#[async_trait::async_trait]
+#[cfg_attr(test, automock)]
+pub trait IndexActorHandle {
+    async fn create_index(&self, uuid: Uuid, primary_key: Option<String>) -> Result<IndexMeta>;
+    async fn update(
+        &self,
+        uuid: Uuid,
+        meta: Processing,
+        data: Option<File>,
+    ) -> Result<std::result::Result<Processed, Failed>>;
+    async fn search(&self, uuid: Uuid, query: SearchQuery) -> Result<SearchResult>;
+    async fn settings(&self, uuid: Uuid) -> Result<Settings<Checked>>;
+
+    async fn documents(
+        &self,
+        uuid: Uuid,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Vec<Document>>;
+    async fn document(
+        &self,
+        uuid: Uuid,
+        doc_id: String,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Document>;
+    async fn delete(&self, uuid: Uuid) -> Result<()>;
+    async fn get_index_meta(&self, uuid: Uuid) -> Result<IndexMeta>;
+    async fn update_index(&self, uuid: Uuid, index_settings: IndexSettings) -> Result<IndexMeta>;
+    async fn snapshot(&self, uuid: Uuid, path: PathBuf) -> Result<()>;
+    async fn dump(&self, uuid: Uuid, path: PathBuf) -> Result<()>;
+    async fn get_index_stats(&self, uuid: Uuid) -> Result<IndexStats>;
+}
+
+#[cfg(test)]
+mod test {
+    use std::sync::Arc;
+
+    use super::*;
+
+    #[async_trait::async_trait]
+    /// Useful for passing around an `Arc<MockIndexActorHandle>` in tests.
+    impl IndexActorHandle for Arc<MockIndexActorHandle> {
+        async fn create_index(&self, uuid: Uuid, primary_key: Option<String>) -> Result<IndexMeta> {
+            self.as_ref().create_index(uuid, primary_key).await
+        }
+
+        async fn update(
+            &self,
+            uuid: Uuid,
+            meta: Processing,
+            data: Option<std::fs::File>,
+        ) -> Result<std::result::Result<Processed, Failed>> {
+            self.as_ref().update(uuid, meta, data).await
+        }
+
+        async fn search(&self, uuid: Uuid, query: SearchQuery) -> Result<SearchResult> {
+            self.as_ref().search(uuid, query).await
+        }
+
+        async fn settings(&self, uuid: Uuid) -> Result<Settings<Checked>> {
+            self.as_ref().settings(uuid).await
+        }
+
+        async fn documents(
+            &self,
+            uuid: Uuid,
+            offset: usize,
+            limit: usize,
+            attributes_to_retrieve: Option<Vec<String>>,
+        ) -> Result<Vec<Document>> {
+            self.as_ref()
+                .documents(uuid, offset, limit, attributes_to_retrieve)
+                .await
+        }
+
+        async fn document(
+            &self,
+            uuid: Uuid,
+            doc_id: String,
+            attributes_to_retrieve: Option<Vec<String>>,
+        ) -> Result<Document> {
+            self.as_ref()
+                .document(uuid, doc_id, attributes_to_retrieve)
+                .await
+        }
+
+        async fn delete(&self, uuid: Uuid) -> Result<()> {
+            self.as_ref().delete(uuid).await
+        }
+
+        async fn get_index_meta(&self, uuid: Uuid) -> Result<IndexMeta> {
+            self.as_ref().get_index_meta(uuid).await
+        }
+
+        async fn update_index(
+            &self,
+            uuid: Uuid,
+            index_settings: IndexSettings,
+        ) -> Result<IndexMeta> {
+            self.as_ref().update_index(uuid, index_settings).await
+        }
+
+        async fn snapshot(&self, uuid: Uuid, path: PathBuf) -> Result<()> {
+            self.as_ref().snapshot(uuid, path).await
+        }
+
+        async fn dump(&self, uuid: Uuid, path: PathBuf) -> Result<()> {
+            self.as_ref().dump(uuid, path).await
+        }
+
+        async fn get_index_stats(&self, uuid: Uuid) -> Result<IndexStats> {
+            self.as_ref().get_index_stats(uuid).await
+        }
+    }
+}

meilisearch-http/src/index_controller/index_actor/store.rs

@@ -0,0 +1,103 @@
+use std::collections::HashMap;
+use std::path::{Path, PathBuf};
+use std::sync::Arc;
+
+use milli::update::UpdateBuilder;
+use tokio::fs;
+use tokio::sync::RwLock;
+use tokio::task::spawn_blocking;
+use uuid::Uuid;
+
+use super::error::{IndexActorError, Result};
+use crate::index::Index;
+
+type AsyncMap<K, V> = Arc<RwLock<HashMap<K, V>>>;
+
+#[async_trait::async_trait]
+pub trait IndexStore {
+    async fn create(&self, uuid: Uuid, primary_key: Option<String>) -> Result<Index>;
+    async fn get(&self, uuid: Uuid) -> Result<Option<Index>>;
+    async fn delete(&self, uuid: Uuid) -> Result<Option<Index>>;
+}
+
+pub struct MapIndexStore {
+    index_store: AsyncMap<Uuid, Index>,
+    path: PathBuf,
+    index_size: usize,
+}
+
+impl MapIndexStore {
+    pub fn new(path: impl AsRef<Path>, index_size: usize) -> Self {
+        let path = path.as_ref().join("indexes/");
+        let index_store = Arc::new(RwLock::new(HashMap::new()));
+        Self {
+            index_store,
+            path,
+            index_size,
+        }
+    }
+}
+
+#[async_trait::async_trait]
+impl IndexStore for MapIndexStore {
+    async fn create(&self, uuid: Uuid, primary_key: Option<String>) -> Result<Index> {
+        // We need to keep the lock until we are sure the db file has been opened correclty, to
+        // ensure that another db is not created at the same time.
+        let mut lock = self.index_store.write().await;
+
+        if let Some(index) = lock.get(&uuid) {
+            return Ok(index.clone());
+        }
+        let path = self.path.join(format!("index-{}", uuid));
+        if path.exists() {
+            return Err(IndexActorError::IndexAlreadyExists);
+        }
+
+        let index_size = self.index_size;
+        let index = spawn_blocking(move || -> Result<Index> {
+            let index = Index::open(path, index_size)?;
+            if let Some(primary_key) = primary_key {
+                let mut txn = index.write_txn()?;
+
+                let mut builder = UpdateBuilder::new(0).settings(&mut txn, &index);
+                builder.set_primary_key(primary_key);
+                builder.execute(|_, _| ())?;
+
+                txn.commit()?;
+            }
+            Ok(index)
+        })
+        .await??;
+
+        lock.insert(uuid, index.clone());
+
+        Ok(index)
+    }
+
+    async fn get(&self, uuid: Uuid) -> Result<Option<Index>> {
+        let guard = self.index_store.read().await;
+        match guard.get(&uuid) {
+            Some(index) => Ok(Some(index.clone())),
+            None => {
+                // drop the guard here so we can perform the write after without deadlocking;
+                drop(guard);
+                let path = self.path.join(format!("index-{}", uuid));
+                if !path.exists() {
+                    return Ok(None);
+                }
+
+                let index_size = self.index_size;
+                let index = spawn_blocking(move || Index::open(path, index_size)).await??;
+                self.index_store.write().await.insert(uuid, index.clone());
+                Ok(Some(index))
+            }
+        }
+    }
+
+    async fn delete(&self, uuid: Uuid) -> Result<Option<Index>> {
+        let db_path = self.path.join(format!("index-{}", uuid));
+        fs::remove_dir_all(db_path).await?;
+        let index = self.index_store.write().await.remove(&uuid);
+        Ok(index)
+    }
+}

meilisearch-http/src/index_controller/mod.rs

@@ -0,0 +1,456 @@
+use std::collections::BTreeMap;
+use std::path::Path;
+use std::sync::Arc;
+use std::time::Duration;
+
+use actix_web::web::Bytes;
+use chrono::{DateTime, Utc};
+use futures::stream::StreamExt;
+use log::error;
+use log::info;
+use milli::FieldDistribution;
+use serde::{Deserialize, Serialize};
+use tokio::sync::mpsc;
+use tokio::time::sleep;
+use uuid::Uuid;
+
+use dump_actor::DumpActorHandle;
+pub use dump_actor::{DumpInfo, DumpStatus};
+use index_actor::IndexActorHandle;
+use snapshot::{load_snapshot, SnapshotService};
+use update_actor::UpdateActorHandle;
+pub use updates::*;
+use uuid_resolver::{error::UuidResolverError, UuidResolverHandle};
+
+use crate::extractors::payload::Payload;
+use crate::index::{Checked, Document, SearchQuery, SearchResult, Settings};
+use crate::option::Opt;
+use error::Result;
+
+use self::dump_actor::load_dump;
+use self::error::IndexControllerError;
+
+mod dump_actor;
+pub mod error;
+pub mod index_actor;
+mod snapshot;
+mod update_actor;
+mod updates;
+mod uuid_resolver;
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct IndexMetadata {
+    #[serde(skip)]
+    pub uuid: Uuid,
+    pub uid: String,
+    name: String,
+    #[serde(flatten)]
+    pub meta: index_actor::IndexMeta,
+}
+
+#[derive(Clone, Debug)]
+pub struct IndexSettings {
+    pub uid: Option<String>,
+    pub primary_key: Option<String>,
+}
+
+#[derive(Serialize, Debug)]
+#[serde(rename_all = "camelCase")]
+pub struct IndexStats {
+    #[serde(skip)]
+    pub size: u64,
+    pub number_of_documents: u64,
+    /// Whether the current index is performing an update. It is initially `None` when the
+    /// index returns it, since it is the `UpdateStore` that knows what index is currently indexing. It is
+    /// later set to either true or false, we we retrieve the information from the `UpdateStore`
+    pub is_indexing: Option<bool>,
+    pub field_distribution: FieldDistribution,
+}
+
+#[derive(Clone)]
+pub struct IndexController {
+    uuid_resolver: uuid_resolver::UuidResolverHandleImpl,
+    index_handle: index_actor::IndexActorHandleImpl,
+    update_handle: update_actor::UpdateActorHandleImpl<Bytes>,
+    dump_handle: dump_actor::DumpActorHandleImpl,
+}
+
+#[derive(Serialize, Debug)]
+#[serde(rename_all = "camelCase")]
+pub struct Stats {
+    pub database_size: u64,
+    pub last_update: Option<DateTime<Utc>>,
+    pub indexes: BTreeMap<String, IndexStats>,
+}
+
+impl IndexController {
+    pub fn new(path: impl AsRef<Path>, options: &Opt) -> anyhow::Result<Self> {
+        let index_size = options.max_index_size.get_bytes() as usize;
+        let update_store_size = options.max_index_size.get_bytes() as usize;
+
+        if let Some(ref path) = options.import_snapshot {
+            info!("Loading from snapshot {:?}", path);
+            load_snapshot(
+                &options.db_path,
+                path,
+                options.ignore_snapshot_if_db_exists,
+                options.ignore_missing_snapshot,
+            )?;
+        } else if let Some(ref src_path) = options.import_dump {
+            load_dump(
+                &options.db_path,
+                src_path,
+                options.max_index_size.get_bytes() as usize,
+                options.max_udb_size.get_bytes() as usize,
+                &options.indexer_options,
+            )?;
+        }
+
+        std::fs::create_dir_all(&path)?;
+
+        let uuid_resolver = uuid_resolver::UuidResolverHandleImpl::new(&path)?;
+        let index_handle =
+            index_actor::IndexActorHandleImpl::new(&path, index_size, &options.indexer_options)?;
+        let update_handle = update_actor::UpdateActorHandleImpl::new(
+            index_handle.clone(),
+            &path,
+            update_store_size,
+        )?;
+        let dump_handle = dump_actor::DumpActorHandleImpl::new(
+            &options.dumps_dir,
+            uuid_resolver.clone(),
+            update_handle.clone(),
+            options.max_index_size.get_bytes() as usize,
+            options.max_udb_size.get_bytes() as usize,
+        )?;
+
+        if options.schedule_snapshot {
+            let snapshot_service = SnapshotService::new(
+                uuid_resolver.clone(),
+                update_handle.clone(),
+                Duration::from_secs(options.snapshot_interval_sec),
+                options.snapshot_dir.clone(),
+                options
+                    .db_path
+                    .file_name()
+                    .map(|n| n.to_owned().into_string().expect("invalid path"))
+                    .unwrap_or_else(|| String::from("data.ms")),
+            );
+
+            tokio::task::spawn(snapshot_service.run());
+        }
+
+        Ok(Self {
+            uuid_resolver,
+            index_handle,
+            update_handle,
+            dump_handle,
+        })
+    }
+
+    pub async fn add_documents(
+        &self,
+        uid: String,
+        method: milli::update::IndexDocumentsMethod,
+        format: milli::update::UpdateFormat,
+        payload: Payload,
+        primary_key: Option<String>,
+    ) -> Result<UpdateStatus> {
+        let perform_update = |uuid| async move {
+            let meta = UpdateMeta::DocumentsAddition {
+                method,
+                format,
+                primary_key,
+            };
+            let (sender, receiver) = mpsc::channel(10);
+
+            // It is necessary to spawn a local task to send the payload to the update handle to
+            // prevent dead_locking between the update_handle::update that waits for the update to be
+            // registered and the update_actor that waits for the the payload to be sent to it.
+            tokio::task::spawn_local(async move {
+                payload
+                    .for_each(|r| async {
+                        let _ = sender.send(r).await;
+                    })
+                    .await
+            });
+
+            // This must be done *AFTER* spawning the task.
+            self.update_handle.update(meta, receiver, uuid).await
+        };
+
+        match self.uuid_resolver.get(uid).await {
+            Ok(uuid) => Ok(perform_update(uuid).await?),
+            Err(UuidResolverError::UnexistingIndex(name)) => {
+                let uuid = Uuid::new_v4();
+                let status = perform_update(uuid).await?;
+                // ignore if index creation fails now, since it may already have been created
+                let _ = self.index_handle.create_index(uuid, None).await;
+                self.uuid_resolver.insert(name, uuid).await?;
+                Ok(status)
+            }
+            Err(e) => Err(e.into()),
+        }
+    }
+
+    pub async fn clear_documents(&self, uid: String) -> Result<UpdateStatus> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let meta = UpdateMeta::ClearDocuments;
+        let (_, receiver) = mpsc::channel(1);
+        let status = self.update_handle.update(meta, receiver, uuid).await?;
+        Ok(status)
+    }
+
+    pub async fn delete_documents(
+        &self,
+        uid: String,
+        documents: Vec<String>,
+    ) -> Result<UpdateStatus> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let meta = UpdateMeta::DeleteDocuments { ids: documents };
+        let (_, receiver) = mpsc::channel(1);
+        let status = self.update_handle.update(meta, receiver, uuid).await?;
+        Ok(status)
+    }
+
+    pub async fn update_settings(
+        &self,
+        uid: String,
+        settings: Settings<Checked>,
+        create: bool,
+    ) -> Result<UpdateStatus> {
+        let perform_udpate = |uuid| async move {
+            let meta = UpdateMeta::Settings(settings.into_unchecked());
+            // Nothing so send, drop the sender right away, as not to block the update actor.
+            let (_, receiver) = mpsc::channel(1);
+            self.update_handle.update(meta, receiver, uuid).await
+        };
+
+        match self.uuid_resolver.get(uid).await {
+            Ok(uuid) => Ok(perform_udpate(uuid).await?),
+            Err(UuidResolverError::UnexistingIndex(name)) if create => {
+                let uuid = Uuid::new_v4();
+                let status = perform_udpate(uuid).await?;
+                // ignore if index creation fails now, since it may already have been created
+                let _ = self.index_handle.create_index(uuid, None).await;
+                self.uuid_resolver.insert(name, uuid).await?;
+                Ok(status)
+            }
+            Err(e) => Err(e.into()),
+        }
+    }
+
+    pub async fn create_index(&self, index_settings: IndexSettings) -> Result<IndexMetadata> {
+        let IndexSettings { uid, primary_key } = index_settings;
+        let uid = uid.ok_or(IndexControllerError::MissingUid)?;
+        let uuid = Uuid::new_v4();
+        let meta = self.index_handle.create_index(uuid, primary_key).await?;
+        self.uuid_resolver.insert(uid.clone(), uuid).await?;
+        let meta = IndexMetadata {
+            uuid,
+            name: uid.clone(),
+            uid,
+            meta,
+        };
+
+        Ok(meta)
+    }
+
+    pub async fn delete_index(&self, uid: String) -> Result<()> {
+        let uuid = self.uuid_resolver.delete(uid).await?;
+
+        // We remove the index from the resolver synchronously, and effectively perform the index
+        // deletion as a background task.
+        let update_handle = self.update_handle.clone();
+        let index_handle = self.index_handle.clone();
+        tokio::spawn(async move {
+            if let Err(e) = update_handle.delete(uuid).await {
+                error!("Error while deleting index: {}", e);
+            }
+            if let Err(e) = index_handle.delete(uuid).await {
+                error!("Error while deleting index: {}", e);
+            }
+        });
+
+        Ok(())
+    }
+
+    pub async fn update_status(&self, uid: String, id: u64) -> Result<UpdateStatus> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let result = self.update_handle.update_status(uuid, id).await?;
+        Ok(result)
+    }
+
+    pub async fn all_update_status(&self, uid: String) -> Result<Vec<UpdateStatus>> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let result = self.update_handle.get_all_updates_status(uuid).await?;
+        Ok(result)
+    }
+
+    pub async fn list_indexes(&self) -> Result<Vec<IndexMetadata>> {
+        let uuids = self.uuid_resolver.list().await?;
+
+        let mut ret = Vec::new();
+
+        for (uid, uuid) in uuids {
+            let meta = self.index_handle.get_index_meta(uuid).await?;
+            let meta = IndexMetadata {
+                uuid,
+                name: uid.clone(),
+                uid,
+                meta,
+            };
+            ret.push(meta);
+        }
+
+        Ok(ret)
+    }
+
+    pub async fn settings(&self, uid: String) -> Result<Settings<Checked>> {
+        let uuid = self.uuid_resolver.get(uid.clone()).await?;
+        let settings = self.index_handle.settings(uuid).await?;
+        Ok(settings)
+    }
+
+    pub async fn documents(
+        &self,
+        uid: String,
+        offset: usize,
+        limit: usize,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Vec<Document>> {
+        let uuid = self.uuid_resolver.get(uid.clone()).await?;
+        let documents = self
+            .index_handle
+            .documents(uuid, offset, limit, attributes_to_retrieve)
+            .await?;
+        Ok(documents)
+    }
+
+    pub async fn document(
+        &self,
+        uid: String,
+        doc_id: String,
+        attributes_to_retrieve: Option<Vec<String>>,
+    ) -> Result<Document> {
+        let uuid = self.uuid_resolver.get(uid.clone()).await?;
+        let document = self
+            .index_handle
+            .document(uuid, doc_id, attributes_to_retrieve)
+            .await?;
+        Ok(document)
+    }
+
+    pub async fn update_index(
+        &self,
+        uid: String,
+        mut index_settings: IndexSettings,
+    ) -> Result<IndexMetadata> {
+        if index_settings.uid.is_some() {
+            index_settings.uid.take();
+        }
+
+        let uuid = self.uuid_resolver.get(uid.clone()).await?;
+        let meta = self.index_handle.update_index(uuid, index_settings).await?;
+        let meta = IndexMetadata {
+            uuid,
+            name: uid.clone(),
+            uid,
+            meta,
+        };
+        Ok(meta)
+    }
+
+    pub async fn search(&self, uid: String, query: SearchQuery) -> Result<SearchResult> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let result = self.index_handle.search(uuid, query).await?;
+        Ok(result)
+    }
+
+    pub async fn get_index(&self, uid: String) -> Result<IndexMetadata> {
+        let uuid = self.uuid_resolver.get(uid.clone()).await?;
+        let meta = self.index_handle.get_index_meta(uuid).await?;
+        let meta = IndexMetadata {
+            uuid,
+            name: uid.clone(),
+            uid,
+            meta,
+        };
+        Ok(meta)
+    }
+
+    pub async fn get_uuids_size(&self) -> Result<u64> {
+        Ok(self.uuid_resolver.get_size().await?)
+    }
+
+    pub async fn get_index_stats(&self, uid: String) -> Result<IndexStats> {
+        let uuid = self.uuid_resolver.get(uid).await?;
+        let update_infos = self.update_handle.get_info().await?;
+        let mut stats = self.index_handle.get_index_stats(uuid).await?;
+        // Check if the currently indexing update is from out index.
+        stats.is_indexing = Some(Some(uuid) == update_infos.processing);
+        Ok(stats)
+    }
+
+    pub async fn get_all_stats(&self) -> Result<Stats> {
+        let update_infos = self.update_handle.get_info().await?;
+        let mut database_size = self.get_uuids_size().await? + update_infos.size;
+        let mut last_update: Option<DateTime<_>> = None;
+        let mut indexes = BTreeMap::new();
+
+        for index in self.list_indexes().await? {
+            let mut index_stats = self.index_handle.get_index_stats(index.uuid).await?;
+            database_size += index_stats.size;
+
+            last_update = last_update.map_or(Some(index.meta.updated_at), |last| {
+                Some(last.max(index.meta.updated_at))
+            });
+
+            index_stats.is_indexing = Some(Some(index.uuid) == update_infos.processing);
+
+            indexes.insert(index.uid, index_stats);
+        }
+
+        Ok(Stats {
+            database_size,
+            last_update,
+            indexes,
+        })
+    }
+
+    pub async fn create_dump(&self) -> Result<DumpInfo> {
+        Ok(self.dump_handle.create_dump().await?)
+    }
+
+    pub async fn dump_info(&self, uid: String) -> Result<DumpInfo> {
+        Ok(self.dump_handle.dump_info(uid).await?)
+    }
+}
+
+pub async fn get_arc_ownership_blocking<T>(mut item: Arc<T>) -> T {
+    loop {
+        match Arc::try_unwrap(item) {
+            Ok(item) => return item,
+            Err(item_arc) => {
+                item = item_arc;
+                sleep(Duration::from_millis(100)).await;
+                continue;
+            }
+        }
+    }
+}
+
+/// Parses the v1 version of the Asc ranking rules `asc(price)`and returns the field name.
+pub fn asc_ranking_rule(text: &str) -> Option<&str> {
+    text.split_once("asc(")
+        .and_then(|(_, tail)| tail.rsplit_once(")"))
+        .map(|(field, _)| field)
+}
+
+/// Parses the v1 version of the Desc ranking rules `asc(price)`and returns the field name.
+pub fn desc_ranking_rule(text: &str) -> Option<&str> {
+    text.split_once("desc(")
+        .and_then(|(_, tail)| tail.rsplit_once(")"))
+        .map(|(field, _)| field)
+}

meilisearch-http/src/index_controller/snapshot.rs

@@ -0,0 +1,268 @@
+use std::path::{Path, PathBuf};
+use std::time::Duration;
+
+use anyhow::bail;
+use log::{error, info, trace};
+use tokio::fs;
+use tokio::task::spawn_blocking;
+use tokio::time::sleep;
+
+use super::update_actor::UpdateActorHandle;
+use super::uuid_resolver::UuidResolverHandle;
+use crate::helpers::compression;
+
+pub struct SnapshotService<U, R> {
+    uuid_resolver_handle: R,
+    update_handle: U,
+    snapshot_period: Duration,
+    snapshot_path: PathBuf,
+    db_name: String,
+}
+
+impl<U, R> SnapshotService<U, R>
+where
+    U: UpdateActorHandle,
+    R: UuidResolverHandle,
+{
+    pub fn new(
+        uuid_resolver_handle: R,
+        update_handle: U,
+        snapshot_period: Duration,
+        snapshot_path: PathBuf,
+        db_name: String,
+    ) -> Self {
+        Self {
+            uuid_resolver_handle,
+            update_handle,
+            snapshot_period,
+            snapshot_path,
+            db_name,
+        }
+    }
+
+    pub async fn run(self) {
+        info!(
+            "Snapshot scheduled every {}s.",
+            self.snapshot_period.as_secs()
+        );
+        loop {
+            if let Err(e) = self.perform_snapshot().await {
+                error!("Error while performing snapshot: {}", e);
+            }
+            sleep(self.snapshot_period).await;
+        }
+    }
+
+    async fn perform_snapshot(&self) -> anyhow::Result<()> {
+        trace!("Performing snapshot.");
+
+        let snapshot_dir = self.snapshot_path.clone();
+        fs::create_dir_all(&snapshot_dir).await?;
+        let temp_snapshot_dir =
+            spawn_blocking(move || tempfile::tempdir_in(snapshot_dir)).await??;
+        let temp_snapshot_path = temp_snapshot_dir.path().to_owned();
+
+        let uuids = self
+            .uuid_resolver_handle
+            .snapshot(temp_snapshot_path.clone())
+            .await?;
+
+        if uuids.is_empty() {
+            return Ok(());
+        }
+
+        self.update_handle
+            .snapshot(uuids, temp_snapshot_path.clone())
+            .await?;
+        let snapshot_dir = self.snapshot_path.clone();
+        let snapshot_path = self
+            .snapshot_path
+            .join(format!("{}.snapshot", self.db_name));
+        let snapshot_path = spawn_blocking(move || -> anyhow::Result<PathBuf> {
+            let temp_snapshot_file = tempfile::NamedTempFile::new_in(snapshot_dir)?;
+            let temp_snapshot_file_path = temp_snapshot_file.path().to_owned();
+            compression::to_tar_gz(temp_snapshot_path, temp_snapshot_file_path)?;
+            temp_snapshot_file.persist(&snapshot_path)?;
+            Ok(snapshot_path)
+        })
+        .await??;
+
+        trace!("Created snapshot in {:?}.", snapshot_path);
+
+        Ok(())
+    }
+}
+
+pub fn load_snapshot(
+    db_path: impl AsRef<Path>,
+    snapshot_path: impl AsRef<Path>,
+    ignore_snapshot_if_db_exists: bool,
+    ignore_missing_snapshot: bool,
+) -> anyhow::Result<()> {
+    if !db_path.as_ref().exists() && snapshot_path.as_ref().exists() {
+        match compression::from_tar_gz(snapshot_path, &db_path) {
+            Ok(()) => Ok(()),
+            Err(e) => {
+                // clean created db folder
+                std::fs::remove_dir_all(&db_path)?;
+                Err(e)
+            }
+        }
+    } else if db_path.as_ref().exists() && !ignore_snapshot_if_db_exists {
+        bail!(
+            "database already exists at {:?}, try to delete it or rename it",
+            db_path
+                .as_ref()
+                .canonicalize()
+                .unwrap_or_else(|_| db_path.as_ref().to_owned())
+        )
+    } else if !snapshot_path.as_ref().exists() && !ignore_missing_snapshot {
+        bail!(
+            "snapshot doesn't exist at {:?}",
+            snapshot_path
+                .as_ref()
+                .canonicalize()
+                .unwrap_or_else(|_| snapshot_path.as_ref().to_owned())
+        )
+    } else {
+        Ok(())
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use std::iter::FromIterator;
+    use std::{collections::HashSet, sync::Arc};
+
+    use futures::future::{err, ok};
+    use rand::Rng;
+    use tokio::time::timeout;
+    use uuid::Uuid;
+
+    use super::*;
+    use crate::index_controller::index_actor::MockIndexActorHandle;
+    use crate::index_controller::update_actor::{
+        error::UpdateActorError, MockUpdateActorHandle, UpdateActorHandleImpl,
+    };
+    use crate::index_controller::uuid_resolver::{
+        error::UuidResolverError, MockUuidResolverHandle,
+    };
+
+    #[actix_rt::test]
+    async fn test_normal() {
+        let mut rng = rand::thread_rng();
+        let uuids_num: usize = rng.gen_range(5, 10);
+        let uuids = (0..uuids_num)
+            .map(|_| Uuid::new_v4())
+            .collect::<HashSet<_>>();
+
+        let mut uuid_resolver = MockUuidResolverHandle::new();
+        let uuids_clone = uuids.clone();
+        uuid_resolver
+            .expect_snapshot()
+            .times(1)
+            .returning(move |_| Box::pin(ok(uuids_clone.clone())));
+
+        let uuids_clone = uuids.clone();
+        let mut index_handle = MockIndexActorHandle::new();
+        index_handle
+            .expect_snapshot()
+            .withf(move |uuid, _path| uuids_clone.contains(uuid))
+            .times(uuids_num)
+            .returning(move |_, _| Box::pin(ok(())));
+
+        let dir = tempfile::tempdir_in(".").unwrap();
+        let handle = Arc::new(index_handle);
+        let update_handle =
+            UpdateActorHandleImpl::<Vec<u8>>::new(handle.clone(), dir.path(), 4096 * 100).unwrap();
+
+        let snapshot_path = tempfile::tempdir_in(".").unwrap();
+        let snapshot_service = SnapshotService::new(
+            uuid_resolver,
+            update_handle,
+            Duration::from_millis(100),
+            snapshot_path.path().to_owned(),
+            "data.ms".to_string(),
+        );
+
+        snapshot_service.perform_snapshot().await.unwrap();
+    }
+
+    #[actix_rt::test]
+    async fn error_performing_uuid_snapshot() {
+        let mut uuid_resolver = MockUuidResolverHandle::new();
+        uuid_resolver
+            .expect_snapshot()
+            .times(1)
+            // abitrary error
+            .returning(|_| Box::pin(err(UuidResolverError::NameAlreadyExist)));
+
+        let update_handle = MockUpdateActorHandle::new();
+
+        let snapshot_path = tempfile::tempdir_in(".").unwrap();
+        let snapshot_service = SnapshotService::new(
+            uuid_resolver,
+            update_handle,
+            Duration::from_millis(100),
+            snapshot_path.path().to_owned(),
+            "data.ms".to_string(),
+        );
+
+        assert!(snapshot_service.perform_snapshot().await.is_err());
+        // Nothing was written to the file
+        assert!(!snapshot_path.path().join("data.ms.snapshot").exists());
+    }
+
+    #[actix_rt::test]
+    async fn error_performing_index_snapshot() {
+        let uuid = Uuid::new_v4();
+        let mut uuid_resolver = MockUuidResolverHandle::new();
+        uuid_resolver
+            .expect_snapshot()
+            .times(1)
+            .returning(move |_| Box::pin(ok(HashSet::from_iter(Some(uuid)))));
+
+        let mut update_handle = MockUpdateActorHandle::new();
+        update_handle
+            .expect_snapshot()
+            // abitrary error
+            .returning(|_, _| Box::pin(err(UpdateActorError::UnexistingUpdate(0))));
+
+        let snapshot_path = tempfile::tempdir_in(".").unwrap();
+        let snapshot_service = SnapshotService::new(
+            uuid_resolver,
+            update_handle,
+            Duration::from_millis(100),
+            snapshot_path.path().to_owned(),
+            "data.ms".to_string(),
+        );
+
+        assert!(snapshot_service.perform_snapshot().await.is_err());
+        // Nothing was written to the file
+        assert!(!snapshot_path.path().join("data.ms.snapshot").exists());
+    }
+
+    #[actix_rt::test]
+    async fn test_loop() {
+        let mut uuid_resolver = MockUuidResolverHandle::new();
+        uuid_resolver
+            .expect_snapshot()
+            // we expect the funtion to be called between 2 and 3 time in the given interval.
+            .times(2..4)
+            // abitrary error, to short-circuit the function
+            .returning(move |_| Box::pin(err(UuidResolverError::NameAlreadyExist)));
+
+        let update_handle = MockUpdateActorHandle::new();
+
+        let snapshot_path = tempfile::tempdir_in(".").unwrap();
+        let snapshot_service = SnapshotService::new(
+            uuid_resolver,
+            update_handle,
+            Duration::from_millis(100),
+            snapshot_path.path().to_owned(),
+            "data.ms".to_string(),
+        );
+
+        let _ = timeout(Duration::from_millis(300), snapshot_service.run()).await;
+    }
+}

meilisearch-http/src/index_controller/update_actor/actor.rs

@@ -0,0 +1,270 @@
+use std::collections::HashSet;
+use std::io::SeekFrom;
+use std::path::{Path, PathBuf};
+use std::sync::atomic::AtomicBool;
+use std::sync::Arc;
+
+use async_stream::stream;
+use futures::StreamExt;
+use log::trace;
+use serdeval::*;
+use tokio::fs;
+use tokio::io::AsyncWriteExt;
+use tokio::sync::mpsc;
+use uuid::Uuid;
+
+use super::error::{Result, UpdateActorError};
+use super::{PayloadData, UpdateMsg, UpdateStore, UpdateStoreInfo};
+use crate::index_controller::index_actor::IndexActorHandle;
+use crate::index_controller::{UpdateMeta, UpdateStatus};
+
+pub struct UpdateActor<D, I> {
+    path: PathBuf,
+    store: Arc<UpdateStore>,
+    inbox: Option<mpsc::Receiver<UpdateMsg<D>>>,
+    index_handle: I,
+    must_exit: Arc<AtomicBool>,
+}
+
+impl<D, I> UpdateActor<D, I>
+where
+    D: AsRef<[u8]> + Sized + 'static,
+    I: IndexActorHandle + Clone + Send + Sync + 'static,
+{
+    pub fn new(
+        update_db_size: usize,
+        inbox: mpsc::Receiver<UpdateMsg<D>>,
+        path: impl AsRef<Path>,
+        index_handle: I,
+    ) -> anyhow::Result<Self> {
+        let path = path.as_ref().join("updates");
+
+        std::fs::create_dir_all(&path)?;
+
+        let mut options = heed::EnvOpenOptions::new();
+        options.map_size(update_db_size);
+
+        let must_exit = Arc::new(AtomicBool::new(false));
+
+        let store = UpdateStore::open(options, &path, index_handle.clone(), must_exit.clone())?;
+        std::fs::create_dir_all(path.join("update_files"))?;
+        let inbox = Some(inbox);
+        Ok(Self {
+            path,
+            store,
+            inbox,
+            index_handle,
+            must_exit,
+        })
+    }
+
+    pub async fn run(mut self) {
+        use UpdateMsg::*;
+
+        trace!("Started update actor.");
+
+        let mut inbox = self
+            .inbox
+            .take()
+            .expect("A receiver should be present by now.");
+
+        let must_exit = self.must_exit.clone();
+        let stream = stream! {
+            loop {
+                let msg = inbox.recv().await;
+
+                if must_exit.load(std::sync::atomic::Ordering::Relaxed) {
+                    break;
+                }
+
+                match msg {
+                    Some(msg) => yield msg,
+                    None => break,
+                }
+            }
+        };
+
+        stream
+            .for_each_concurrent(Some(10), |msg| async {
+                match msg {
+                    Update {
+                        uuid,
+                        meta,
+                        data,
+                        ret,
+                    } => {
+                        let _ = ret.send(self.handle_update(uuid, meta, data).await);
+                    }
+                    ListUpdates { uuid, ret } => {
+                        let _ = ret.send(self.handle_list_updates(uuid).await);
+                    }
+                    GetUpdate { uuid, ret, id } => {
+                        let _ = ret.send(self.handle_get_update(uuid, id).await);
+                    }
+                    Delete { uuid, ret } => {
+                        let _ = ret.send(self.handle_delete(uuid).await);
+                    }
+                    Snapshot { uuids, path, ret } => {
+                        let _ = ret.send(self.handle_snapshot(uuids, path).await);
+                    }
+                    GetInfo { ret } => {
+                        let _ = ret.send(self.handle_get_info().await);
+                    }
+                    Dump { uuids, path, ret } => {
+                        let _ = ret.send(self.handle_dump(uuids, path).await);
+                    }
+                }
+            })
+            .await;
+    }
+
+    async fn handle_update(
+        &self,
+        uuid: Uuid,
+        meta: UpdateMeta,
+        payload: mpsc::Receiver<PayloadData<D>>,
+    ) -> Result<UpdateStatus> {
+        let file_path = match meta {
+            UpdateMeta::DocumentsAddition { .. } => {
+                let update_file_id = uuid::Uuid::new_v4();
+                let path = self
+                    .path
+                    .join(format!("update_files/update_{}", update_file_id));
+                let mut file = fs::OpenOptions::new()
+                    .read(true)
+                    .write(true)
+                    .create(true)
+                    .open(&path)
+                    .await?;
+
+                async fn write_to_file<D>(
+                    file: &mut fs::File,
+                    mut payload: mpsc::Receiver<PayloadData<D>>,
+                ) -> Result<usize>
+                where
+                    D: AsRef<[u8]> + Sized + 'static,
+                {
+                    let mut file_len = 0;
+
+                    while let Some(bytes) = payload.recv().await {
+                        let bytes = bytes?;
+                        file_len += bytes.as_ref().len();
+                        file.write_all(bytes.as_ref()).await?;
+                    }
+
+                    file.flush().await?;
+
+                    Ok(file_len)
+                }
+
+                let file_len = write_to_file(&mut file, payload).await;
+
+                match file_len {
+                    Ok(len) if len > 0 => {
+                        let file = file.into_std().await;
+                        Some((file, update_file_id))
+                    }
+                    Err(e) => {
+                        fs::remove_file(&path).await?;
+                        return Err(e);
+                    }
+                    _ => {
+                        fs::remove_file(&path).await?;
+                        None
+                    }
+                }
+            }
+            _ => None,
+        };
+
+        let update_store = self.store.clone();
+
+        tokio::task::spawn_blocking(move || {
+            use std::io::{BufReader, Seek};
+
+            // If the payload is empty, ignore the check.
+            let update_uuid = if let Some((mut file, uuid)) = file_path {
+                // set the file back to the beginning
+                file.seek(SeekFrom::Start(0))?;
+                // Check that the json payload is valid:
+                let reader = BufReader::new(&mut file);
+                // Validate that the payload is in the correct format.
+                let _: Seq<Map<Str, Any>> = serde_json::from_reader(reader)
+                    .map_err(|e| UpdateActorError::InvalidPayload(Box::new(e)))?;
+
+                Some(uuid)
+            } else {
+                None
+            };
+
+            // The payload is valid, we can register it to the update store.
+            let status = update_store
+                .register_update(meta, update_uuid, uuid)
+                .map(UpdateStatus::Enqueued)?;
+            Ok(status)
+        })
+        .await?
+    }
+
+    async fn handle_list_updates(&self, uuid: Uuid) -> Result<Vec<UpdateStatus>> {
+        let update_store = self.store.clone();
+        tokio::task::spawn_blocking(move || {
+            let result = update_store.list(uuid)?;
+            Ok(result)
+        })
+        .await?
+    }
+
+    async fn handle_get_update(&self, uuid: Uuid, id: u64) -> Result<UpdateStatus> {
+        let store = self.store.clone();
+        tokio::task::spawn_blocking(move || {
+            let result = store
+                .meta(uuid, id)?
+                .ok_or(UpdateActorError::UnexistingUpdate(id))?;
+            Ok(result)
+        })
+        .await?
+    }
+
+    async fn handle_delete(&self, uuid: Uuid) -> Result<()> {
+        let store = self.store.clone();
+
+        tokio::task::spawn_blocking(move || store.delete_all(uuid)).await??;
+
+        Ok(())
+    }
+
+    async fn handle_snapshot(&self, uuids: HashSet<Uuid>, path: PathBuf) -> Result<()> {
+        let index_handle = self.index_handle.clone();
+        let update_store = self.store.clone();
+
+        tokio::task::spawn_blocking(move || update_store.snapshot(&uuids, &path, index_handle))
+            .await??;
+
+        Ok(())
+    }
+
+    async fn handle_dump(&self, uuids: HashSet<Uuid>, path: PathBuf) -> Result<()> {
+        let index_handle = self.index_handle.clone();
+        let update_store = self.store.clone();
+
+        tokio::task::spawn_blocking(move || -> Result<()> {
+            update_store.dump(&uuids, path.to_path_buf(), index_handle)?;
+            Ok(())
+        })
+        .await??;
+
+        Ok(())
+    }
+
+    async fn handle_get_info(&self) -> Result<UpdateStoreInfo> {
+        let update_store = self.store.clone();
+        let info = tokio::task::spawn_blocking(move || -> Result<UpdateStoreInfo> {
+            let info = update_store.get_info()?;
+            Ok(info)
+        })
+        .await??;
+
+        Ok(info)
+    }
+}

meilisearch-http/src/index_controller/update_actor/error.rs

@@ -0,0 +1,61 @@
+use std::error::Error;
+
+use meilisearch_error::{Code, ErrorCode};
+
+use crate::index_controller::index_actor::error::IndexActorError;
+
+pub type Result<T> = std::result::Result<T, UpdateActorError>;
+
+#[derive(Debug, thiserror::Error)]
+#[allow(clippy::large_enum_variant)]
+pub enum UpdateActorError {
+    #[error("Update {0} not found.")]
+    UnexistingUpdate(u64),
+    #[error("Internal error: {0}")]
+    Internal(Box<dyn Error + Send + Sync + 'static>),
+    #[error("{0}")]
+    IndexActor(#[from] IndexActorError),
+    #[error(
+        "update store was shut down due to a fatal error, please check your logs for more info."
+    )]
+    FatalUpdateStoreError,
+    #[error("{0}")]
+    InvalidPayload(Box<dyn Error + Send + Sync + 'static>),
+    #[error("{0}")]
+    PayloadError(#[from] actix_web::error::PayloadError),
+}
+
+impl<T> From<tokio::sync::mpsc::error::SendError<T>> for UpdateActorError {
+    fn from(_: tokio::sync::mpsc::error::SendError<T>) -> Self {
+        Self::FatalUpdateStoreError
+    }
+}
+
+impl From<tokio::sync::oneshot::error::RecvError> for UpdateActorError {
+    fn from(_: tokio::sync::oneshot::error::RecvError) -> Self {
+        Self::FatalUpdateStoreError
+    }
+}
+
+internal_error!(
+    UpdateActorError: heed::Error,
+    std::io::Error,
+    serde_json::Error,
+    tokio::task::JoinError
+);
+
+impl ErrorCode for UpdateActorError {
+    fn error_code(&self) -> Code {
+        match self {
+            UpdateActorError::UnexistingUpdate(_) => Code::NotFound,
+            UpdateActorError::Internal(_) => Code::Internal,
+            UpdateActorError::IndexActor(e) => e.error_code(),
+            UpdateActorError::FatalUpdateStoreError => Code::Internal,
+            UpdateActorError::InvalidPayload(_) => Code::BadRequest,
+            UpdateActorError::PayloadError(error) => match error {
+                actix_http::error::PayloadError::Overflow => Code::PayloadTooLarge,
+                _ => Code::Internal,
+            },
+        }
+    }
+}

meilisearch-http/src/index_controller/update_actor/handle_impl.rs

@@ -0,0 +1,103 @@
+use std::collections::HashSet;
+use std::path::{Path, PathBuf};
+
+use tokio::sync::{mpsc, oneshot};
+use uuid::Uuid;
+
+use crate::index_controller::{IndexActorHandle, UpdateStatus};
+
+use super::error::Result;
+use super::{PayloadData, UpdateActor, UpdateActorHandle, UpdateMeta, UpdateMsg, UpdateStoreInfo};
+
+#[derive(Clone)]
+pub struct UpdateActorHandleImpl<D> {
+    sender: mpsc::Sender<UpdateMsg<D>>,
+}
+
+impl<D> UpdateActorHandleImpl<D>
+where
+    D: AsRef<[u8]> + Sized + 'static + Sync + Send,
+{
+    pub fn new<I>(
+        index_handle: I,
+        path: impl AsRef<Path>,
+        update_store_size: usize,
+    ) -> anyhow::Result<Self>
+    where
+        I: IndexActorHandle + Clone + Send + Sync + 'static,
+    {
+        let path = path.as_ref().to_owned();
+        let (sender, receiver) = mpsc::channel(100);
+        let actor = UpdateActor::new(update_store_size, receiver, path, index_handle)?;
+
+        tokio::task::spawn(actor.run());
+
+        Ok(Self { sender })
+    }
+}
+
+#[async_trait::async_trait]
+impl<D> UpdateActorHandle for UpdateActorHandleImpl<D>
+where
+    D: AsRef<[u8]> + Sized + 'static + Sync + Send,
+{
+    type Data = D;
+
+    async fn get_all_updates_status(&self, uuid: Uuid) -> Result<Vec<UpdateStatus>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::ListUpdates { uuid, ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+    async fn update_status(&self, uuid: Uuid, id: u64) -> Result<UpdateStatus> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::GetUpdate { uuid, id, ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+
+    async fn delete(&self, uuid: Uuid) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::Delete { uuid, ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+
+    async fn snapshot(&self, uuids: HashSet<Uuid>, path: PathBuf) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::Snapshot { uuids, path, ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+
+    async fn dump(&self, uuids: HashSet<Uuid>, path: PathBuf) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::Dump { uuids, path, ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+
+    async fn get_info(&self) -> Result<UpdateStoreInfo> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::GetInfo { ret };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+
+    async fn update(
+        &self,
+        meta: UpdateMeta,
+        data: mpsc::Receiver<PayloadData<Self::Data>>,
+        uuid: Uuid,
+    ) -> Result<UpdateStatus> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UpdateMsg::Update {
+            uuid,
+            data,
+            meta,
+            ret,
+        };
+        self.sender.send(msg).await?;
+        receiver.await?
+    }
+}

meilisearch-http/src/index_controller/update_actor/message.rs

@@ -0,0 +1,43 @@
+use std::collections::HashSet;
+use std::path::PathBuf;
+
+use tokio::sync::{mpsc, oneshot};
+use uuid::Uuid;
+
+use super::error::Result;
+use super::{PayloadData, UpdateMeta, UpdateStatus, UpdateStoreInfo};
+
+pub enum UpdateMsg<D> {
+    Update {
+        uuid: Uuid,
+        meta: UpdateMeta,
+        data: mpsc::Receiver<PayloadData<D>>,
+        ret: oneshot::Sender<Result<UpdateStatus>>,
+    },
+    ListUpdates {
+        uuid: Uuid,
+        ret: oneshot::Sender<Result<Vec<UpdateStatus>>>,
+    },
+    GetUpdate {
+        uuid: Uuid,
+        ret: oneshot::Sender<Result<UpdateStatus>>,
+        id: u64,
+    },
+    Delete {
+        uuid: Uuid,
+        ret: oneshot::Sender<Result<()>>,
+    },
+    Snapshot {
+        uuids: HashSet<Uuid>,
+        path: PathBuf,
+        ret: oneshot::Sender<Result<()>>,
+    },
+    Dump {
+        uuids: HashSet<Uuid>,
+        path: PathBuf,
+        ret: oneshot::Sender<Result<()>>,
+    },
+    GetInfo {
+        ret: oneshot::Sender<Result<UpdateStoreInfo>>,
+    },
+}

meilisearch-http/src/index_controller/update_actor/mod.rs

@@ -0,0 +1,44 @@
+use std::{collections::HashSet, path::PathBuf};
+
+use actix_http::error::PayloadError;
+use tokio::sync::mpsc;
+use uuid::Uuid;
+
+use crate::index_controller::{UpdateMeta, UpdateStatus};
+
+use actor::UpdateActor;
+use error::Result;
+use message::UpdateMsg;
+
+pub use handle_impl::UpdateActorHandleImpl;
+pub use store::{UpdateStore, UpdateStoreInfo};
+
+mod actor;
+pub mod error;
+mod handle_impl;
+mod message;
+pub mod store;
+
+type PayloadData<D> = std::result::Result<D, PayloadError>;
+
+#[cfg(test)]
+use mockall::automock;
+
+#[async_trait::async_trait]
+#[cfg_attr(test, automock(type Data=Vec<u8>;))]
+pub trait UpdateActorHandle {
+    type Data: AsRef<[u8]> + Sized + 'static + Sync + Send;
+
+    async fn get_all_updates_status(&self, uuid: Uuid) -> Result<Vec<UpdateStatus>>;
+    async fn update_status(&self, uuid: Uuid, id: u64) -> Result<UpdateStatus>;
+    async fn delete(&self, uuid: Uuid) -> Result<()>;
+    async fn snapshot(&self, uuid: HashSet<Uuid>, path: PathBuf) -> Result<()>;
+    async fn dump(&self, uuids: HashSet<Uuid>, path: PathBuf) -> Result<()>;
+    async fn get_info(&self) -> Result<UpdateStoreInfo>;
+    async fn update(
+        &self,
+        meta: UpdateMeta,
+        data: mpsc::Receiver<PayloadData<Self::Data>>,
+        uuid: Uuid,
+    ) -> Result<UpdateStatus>;
+}

meilisearch-http/src/index_controller/update_actor/store/codec.rs

@@ -0,0 +1,86 @@
+use std::{borrow::Cow, convert::TryInto, mem::size_of};
+
+use heed::{BytesDecode, BytesEncode};
+use uuid::Uuid;
+
+pub struct NextIdCodec;
+
+pub enum NextIdKey {
+    Global,
+    Index(Uuid),
+}
+
+impl<'a> BytesEncode<'a> for NextIdCodec {
+    type EItem = NextIdKey;
+
+    fn bytes_encode(item: &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
+        match item {
+            NextIdKey::Global => Some(Cow::Borrowed(b"__global__")),
+            NextIdKey::Index(ref uuid) => Some(Cow::Borrowed(uuid.as_bytes())),
+        }
+    }
+}
+
+pub struct PendingKeyCodec;
+
+impl<'a> BytesEncode<'a> for PendingKeyCodec {
+    type EItem = (u64, Uuid, u64);
+
+    fn bytes_encode((global_id, uuid, update_id): &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
+        let mut bytes = Vec::with_capacity(size_of::<Self::EItem>());
+        bytes.extend_from_slice(&global_id.to_be_bytes());
+        bytes.extend_from_slice(uuid.as_bytes());
+        bytes.extend_from_slice(&update_id.to_be_bytes());
+        Some(Cow::Owned(bytes))
+    }
+}
+
+impl<'a> BytesDecode<'a> for PendingKeyCodec {
+    type DItem = (u64, Uuid, u64);
+
+    fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
+        let global_id_bytes = bytes.get(0..size_of::<u64>())?.try_into().ok()?;
+        let global_id = u64::from_be_bytes(global_id_bytes);
+
+        let uuid_bytes = bytes
+            .get(size_of::<u64>()..(size_of::<u64>() + size_of::<Uuid>()))?
+            .try_into()
+            .ok()?;
+        let uuid = Uuid::from_bytes(uuid_bytes);
+
+        let update_id_bytes = bytes
+            .get((size_of::<u64>() + size_of::<Uuid>())..)?
+            .try_into()
+            .ok()?;
+        let update_id = u64::from_be_bytes(update_id_bytes);
+
+        Some((global_id, uuid, update_id))
+    }
+}
+
+pub struct UpdateKeyCodec;
+
+impl<'a> BytesEncode<'a> for UpdateKeyCodec {
+    type EItem = (Uuid, u64);
+
+    fn bytes_encode((uuid, update_id): &'a Self::EItem) -> Option<Cow<'a, [u8]>> {
+        let mut bytes = Vec::with_capacity(size_of::<Self::EItem>());
+        bytes.extend_from_slice(uuid.as_bytes());
+        bytes.extend_from_slice(&update_id.to_be_bytes());
+        Some(Cow::Owned(bytes))
+    }
+}
+
+impl<'a> BytesDecode<'a> for UpdateKeyCodec {
+    type DItem = (Uuid, u64);
+
+    fn bytes_decode(bytes: &'a [u8]) -> Option<Self::DItem> {
+        let uuid_bytes = bytes.get(0..size_of::<Uuid>())?.try_into().ok()?;
+        let uuid = Uuid::from_bytes(uuid_bytes);
+
+        let update_id_bytes = bytes.get(size_of::<Uuid>()..)?.try_into().ok()?;
+        let update_id = u64::from_be_bytes(update_id_bytes);
+
+        Some((uuid, update_id))
+    }
+}

meilisearch-http/src/index_controller/update_actor/store/dump.rs

@@ -0,0 +1,184 @@
+use std::{
+    collections::HashSet,
+    fs::{create_dir_all, File},
+    io::{BufRead, BufReader, Write},
+    path::{Path, PathBuf},
+};
+
+use heed::{EnvOpenOptions, RoTxn};
+use serde::{Deserialize, Serialize};
+use uuid::Uuid;
+
+use super::{Result, State, UpdateStore};
+use crate::index_controller::{
+    index_actor::IndexActorHandle, update_actor::store::update_uuid_to_file_path, Enqueued,
+    UpdateStatus,
+};
+
+#[derive(Serialize, Deserialize)]
+struct UpdateEntry {
+    uuid: Uuid,
+    update: UpdateStatus,
+}
+
+impl UpdateStore {
+    pub fn dump(
+        &self,
+        uuids: &HashSet<Uuid>,
+        path: PathBuf,
+        handle: impl IndexActorHandle,
+    ) -> Result<()> {
+        let state_lock = self.state.write();
+        state_lock.swap(State::Dumping);
+
+        // txn must *always* be acquired after state lock, or it will dead lock.
+        let txn = self.env.write_txn()?;
+
+        let dump_path = path.join("updates");
+        create_dir_all(&dump_path)?;
+
+        self.dump_updates(&txn, uuids, &dump_path)?;
+
+        let fut = dump_indexes(uuids, handle, &path);
+        tokio::runtime::Handle::current().block_on(fut)?;
+
+        state_lock.swap(State::Idle);
+
+        Ok(())
+    }
+
+    fn dump_updates(
+        &self,
+        txn: &RoTxn,
+        uuids: &HashSet<Uuid>,
+        path: impl AsRef<Path>,
+    ) -> Result<()> {
+        let dump_data_path = path.as_ref().join("data.jsonl");
+        let mut dump_data_file = File::create(dump_data_path)?;
+
+        let update_files_path = path.as_ref().join(super::UPDATE_DIR);
+        create_dir_all(&update_files_path)?;
+
+        self.dump_pending(txn, uuids, &mut dump_data_file, &path)?;
+        self.dump_completed(txn, uuids, &mut dump_data_file)?;
+
+        Ok(())
+    }
+
+    fn dump_pending(
+        &self,
+        txn: &RoTxn,
+        uuids: &HashSet<Uuid>,
+        mut file: &mut File,
+        dst_path: impl AsRef<Path>,
+    ) -> Result<()> {
+        let pendings = self.pending_queue.iter(txn)?.lazily_decode_data();
+
+        for pending in pendings {
+            let ((_, uuid, _), data) = pending?;
+            if uuids.contains(&uuid) {
+                let update = data.decode()?;
+
+                if let Some(ref update_uuid) = update.content {
+                    let src = super::update_uuid_to_file_path(&self.path, *update_uuid);
+                    let dst = super::update_uuid_to_file_path(&dst_path, *update_uuid);
+                    std::fs::copy(src, dst)?;
+                }
+
+                let update_json = UpdateEntry {
+                    uuid,
+                    update: update.into(),
+                };
+
+                serde_json::to_writer(&mut file, &update_json)?;
+                file.write_all(b"\n")?;
+            }
+        }
+
+        Ok(())
+    }
+
+    fn dump_completed(
+        &self,
+        txn: &RoTxn,
+        uuids: &HashSet<Uuid>,
+        mut file: &mut File,
+    ) -> Result<()> {
+        let updates = self.updates.iter(txn)?.lazily_decode_data();
+
+        for update in updates {
+            let ((uuid, _), data) = update?;
+            if uuids.contains(&uuid) {
+                let update = data.decode()?;
+
+                let update_json = UpdateEntry { uuid, update };
+
+                serde_json::to_writer(&mut file, &update_json)?;
+                file.write_all(b"\n")?;
+            }
+        }
+
+        Ok(())
+    }
+
+    pub fn load_dump(
+        src: impl AsRef<Path>,
+        dst: impl AsRef<Path>,
+        db_size: usize,
+    ) -> anyhow::Result<()> {
+        let dst_update_path = dst.as_ref().join("updates/");
+        create_dir_all(&dst_update_path)?;
+
+        let mut options = EnvOpenOptions::new();
+        options.map_size(db_size as usize);
+        let (store, _) = UpdateStore::new(options, &dst_update_path)?;
+
+        let src_update_path = src.as_ref().join("updates");
+        let update_data = File::open(&src_update_path.join("data.jsonl"))?;
+        let mut update_data = BufReader::new(update_data);
+
+        std::fs::create_dir_all(dst_update_path.join("update_files/"))?;
+
+        let mut wtxn = store.env.write_txn()?;
+        let mut line = String::new();
+        loop {
+            match update_data.read_line(&mut line) {
+                Ok(0) => break,
+                Ok(_) => {
+                    let UpdateEntry { uuid, update } = serde_json::from_str(&line)?;
+                    store.register_raw_updates(&mut wtxn, &update, uuid)?;
+
+                    // Copy ascociated update path if it exists
+                    if let UpdateStatus::Enqueued(Enqueued {
+                        content: Some(uuid),
+                        ..
+                    }) = update
+                    {
+                        let src = update_uuid_to_file_path(&src_update_path, uuid);
+                        let dst = update_uuid_to_file_path(&dst_update_path, uuid);
+                        std::fs::copy(src, dst)?;
+                    }
+                }
+                _ => break,
+            }
+
+            line.clear();
+        }
+
+        wtxn.commit()?;
+
+        Ok(())
+    }
+}
+
+async fn dump_indexes(
+    uuids: &HashSet<Uuid>,
+    handle: impl IndexActorHandle,
+    path: impl AsRef<Path>,
+) -> Result<()> {
+    for uuid in uuids {
+        handle.dump(*uuid, path.as_ref().to_owned()).await?;
+    }
+
+    Ok(())
+}

meilisearch-http/src/index_controller/update_actor/store/mod.rs

@@ -0,0 +1,729 @@
+mod codec;
+pub mod dump;
+
+use std::fs::{copy, create_dir_all, remove_file, File};
+use std::path::Path;
+use std::sync::atomic::{AtomicBool, Ordering};
+use std::sync::Arc;
+use std::{
+    collections::{BTreeMap, HashSet},
+    path::PathBuf,
+    time::Duration,
+};
+
+use arc_swap::ArcSwap;
+use futures::StreamExt;
+use heed::types::{ByteSlice, OwnedType, SerdeJson};
+use heed::zerocopy::U64;
+use heed::{CompactionOption, Database, Env, EnvOpenOptions};
+use log::error;
+use parking_lot::{Mutex, MutexGuard};
+use tokio::runtime::Handle;
+use tokio::sync::mpsc;
+use tokio::sync::mpsc::error::TrySendError;
+use tokio::time::timeout;
+use uuid::Uuid;
+
+use codec::*;
+
+use super::error::Result;
+use super::UpdateMeta;
+use crate::helpers::EnvSizer;
+use crate::index_controller::{index_actor::CONCURRENT_INDEX_MSG, updates::*, IndexActorHandle};
+
+#[allow(clippy::upper_case_acronyms)]
+type BEU64 = U64<heed::byteorder::BE>;
+
+const UPDATE_DIR: &str = "update_files";
+
+pub struct UpdateStoreInfo {
+    /// Size of the update store in bytes.
+    pub size: u64,
+    /// Uuid of the currently processing update if it exists
+    pub processing: Option<Uuid>,
+}
+
+/// A data structure that allows concurrent reads AND exactly one writer.
+pub struct StateLock {
+    lock: Mutex<()>,
+    data: ArcSwap<State>,
+}
+
+pub struct StateLockGuard<'a> {
+    _lock: MutexGuard<'a, ()>,
+    state: &'a StateLock,
+}
+
+impl StateLockGuard<'_> {
+    pub fn swap(&self, state: State) -> Arc<State> {
+        self.state.data.swap(Arc::new(state))
+    }
+}
+
+impl StateLock {
+    fn from_state(state: State) -> Self {
+        let lock = Mutex::new(());
+        let data = ArcSwap::from(Arc::new(state));
+        Self { lock, data }
+    }
+
+    pub fn read(&self) -> Arc<State> {
+        self.data.load().clone()
+    }
+
+    pub fn write(&self) -> StateLockGuard {
+        let _lock = self.lock.lock();
+        let state = &self;
+        StateLockGuard { _lock, state }
+    }
+}
+
+#[allow(clippy::large_enum_variant)]
+pub enum State {
+    Idle,
+    Processing(Uuid, Processing),
+    Snapshoting,
+    Dumping,
+}
+
+#[derive(Clone)]
+pub struct UpdateStore {
+    pub env: Env,
+    /// A queue containing the updates to process, ordered by arrival.
+    /// The key are built as follow:
+    /// | global_update_id | index_uuid | update_id |
+    /// |     8-bytes      |  16-bytes  |  8-bytes  |
+    pending_queue: Database<PendingKeyCodec, SerdeJson<Enqueued>>,
+    /// Map indexes to the next available update id. If NextIdKey::Global is queried, then the next
+    /// global update id is returned
+    next_update_id: Database<NextIdCodec, OwnedType<BEU64>>,
+    /// Contains all the performed updates meta, be they failed, aborted, or processed.
+    /// The keys are built as follow:
+    /// |    Uuid  |   id    |
+    /// | 16-bytes | 8-bytes |
+    updates: Database<UpdateKeyCodec, SerdeJson<UpdateStatus>>,
+    /// Indicates the current state of the update store,
+    state: Arc<StateLock>,
+    /// Wake up the loop when a new event occurs.
+    notification_sender: mpsc::Sender<()>,
+    path: PathBuf,
+}
+
+impl UpdateStore {
+    fn new(
+        mut options: EnvOpenOptions,
+        path: impl AsRef<Path>,
+    ) -> anyhow::Result<(Self, mpsc::Receiver<()>)> {
+        options.max_dbs(5);
+
+        let env = options.open(&path)?;
+        let pending_queue = env.create_database(Some("pending-queue"))?;
+        let next_update_id = env.create_database(Some("next-update-id"))?;
+        let updates = env.create_database(Some("updates"))?;
+
+        let state = Arc::new(StateLock::from_state(State::Idle));
+
+        let (notification_sender, notification_receiver) = mpsc::channel(1);
+
+        Ok((
+            Self {
+                env,
+                pending_queue,
+                next_update_id,
+                updates,
+                state,
+                notification_sender,
+                path: path.as_ref().to_owned(),
+            },
+            notification_receiver,
+        ))
+    }
+
+    pub fn open(
+        options: EnvOpenOptions,
+        path: impl AsRef<Path>,
+        index_handle: impl IndexActorHandle + Clone + Sync + Send + 'static,
+        must_exit: Arc<AtomicBool>,
+    ) -> anyhow::Result<Arc<Self>> {
+        let (update_store, mut notification_receiver) = Self::new(options, path)?;
+        let update_store = Arc::new(update_store);
+
+        // Send a first notification to trigger the process.
+        if let Err(TrySendError::Closed(())) = update_store.notification_sender.try_send(()) {
+            panic!("Failed to init update store");
+        }
+
+        // We need a weak reference so we can take ownership on the arc later when we
+        // want to close the index.
+        let duration = Duration::from_secs(10 * 60); // 10 minutes
+        let update_store_weak = Arc::downgrade(&update_store);
+        tokio::task::spawn(async move {
+            // Block and wait for something to process with a timeout. The timeout
+            // function returns a Result and we must just unlock the loop on Result.
+            'outer: while timeout(duration, notification_receiver.recv())
+                .await
+                .map_or(true, |o| o.is_some())
+            {
+                loop {
+                    match update_store_weak.upgrade() {
+                        Some(update_store) => {
+                            let handler = index_handle.clone();
+                            let res = tokio::task::spawn_blocking(move || {
+                                update_store.process_pending_update(handler)
+                            })
+                            .await
+                            .expect("Fatal error processing update.");
+                            match res {
+                                Ok(Some(_)) => (),
+                                Ok(None) => break,
+                                Err(e) => {
+                                    error!("Fatal error while processing an update that requires the update store to shutdown: {}", e);
+                                    must_exit.store(true, Ordering::SeqCst);
+                                    break 'outer;
+                                }
+                            }
+                        }
+                        // the ownership on the arc has been taken, we need to exit.
+                        None => break 'outer,
+                    }
+                }
+            }
+
+            error!("Update store loop exited.");
+        });
+
+        Ok(update_store)
+    }
+
+    /// Returns the next global update id and the next update id for a given `index_uuid`.
+    fn next_update_id(&self, txn: &mut heed::RwTxn, index_uuid: Uuid) -> heed::Result<(u64, u64)> {
+        let global_id = self
+            .next_update_id
+            .get(txn, &NextIdKey::Global)?
+            .map(U64::get)
+            .unwrap_or_default();
+
+        self.next_update_id
+            .put(txn, &NextIdKey::Global, &BEU64::new(global_id + 1))?;
+
+        let update_id = self.next_update_id_raw(txn, index_uuid)?;
+
+        Ok((global_id, update_id))
+    }
+
+    /// Returns the next next update id for a given `index_uuid` without
+    /// incrementing the global update id. This is useful for the dumps.
+    fn next_update_id_raw(&self, txn: &mut heed::RwTxn, index_uuid: Uuid) -> heed::Result<u64> {
+        let update_id = self
+            .next_update_id
+            .get(txn, &NextIdKey::Index(index_uuid))?
+            .map(U64::get)
+            .unwrap_or_default();
+
+        self.next_update_id.put(
+            txn,
+            &NextIdKey::Index(index_uuid),
+            &BEU64::new(update_id + 1),
+        )?;
+
+        Ok(update_id)
+    }
+
+    /// Registers the update content in the pending store and the meta
+    /// into the pending-meta store. Returns the new unique update id.
+    pub fn register_update(
+        &self,
+        meta: UpdateMeta,
+        content: Option<Uuid>,
+        index_uuid: Uuid,
+    ) -> heed::Result<Enqueued> {
+        let mut txn = self.env.write_txn()?;
+
+        let (global_id, update_id) = self.next_update_id(&mut txn, index_uuid)?;
+        let meta = Enqueued::new(meta, update_id, content);
+
+        self.pending_queue
+            .put(&mut txn, &(global_id, index_uuid, update_id), &meta)?;
+
+        txn.commit()?;
+
+        if let Err(TrySendError::Closed(())) = self.notification_sender.try_send(()) {
+            panic!("Update store loop exited");
+        }
+
+        Ok(meta)
+    }
+
+    /// Push already processed update in the UpdateStore without triggering the notification
+    /// process. This is useful for the dumps.
+    pub fn register_raw_updates(
+        &self,
+        wtxn: &mut heed::RwTxn,
+        update: &UpdateStatus,
+        index_uuid: Uuid,
+    ) -> heed::Result<()> {
+        match update {
+            UpdateStatus::Enqueued(enqueued) => {
+                let (global_id, _update_id) = self.next_update_id(wtxn, index_uuid)?;
+                self.pending_queue.remap_key_type::<PendingKeyCodec>().put(
+                    wtxn,
+                    &(global_id, index_uuid, enqueued.id()),
+                    enqueued,
+                )?;
+            }
+            _ => {
+                let _update_id = self.next_update_id_raw(wtxn, index_uuid)?;
+                self.updates.put(wtxn, &(index_uuid, update.id()), update)?;
+            }
+        }
+        Ok(())
+    }
+
+    /// Executes the user provided function on the next pending update (the one with the lowest id).
+    /// This is asynchronous as it let the user process the update with a read-only txn and
+    /// only writing the result meta to the processed-meta store *after* it has been processed.
+    fn process_pending_update(&self, index_handle: impl IndexActorHandle) -> Result<Option<()>> {
+        // Create a read transaction to be able to retrieve the pending update in order.
+        let rtxn = self.env.read_txn()?;
+        let first_meta = self.pending_queue.first(&rtxn)?;
+        drop(rtxn);
+
+        // If there is a pending update we process and only keep
+        // a reader while processing it, not a writer.
+        match first_meta {
+            Some(((global_id, index_uuid, _), mut pending)) => {
+                let content = pending.content.take();
+                let processing = pending.processing();
+                // Acquire the state lock and set the current state to processing.
+                // txn must *always* be acquired after state lock, or it will dead lock.
+                let state = self.state.write();
+                state.swap(State::Processing(index_uuid, processing.clone()));
+
+                let result =
+                    self.perform_update(content, processing, index_handle, index_uuid, global_id);
+
+                state.swap(State::Idle);
+
+                result
+            }
+            None => Ok(None),
+        }
+    }
+
+    fn perform_update(
+        &self,
+        content: Option<Uuid>,
+        processing: Processing,
+        index_handle: impl IndexActorHandle,
+        index_uuid: Uuid,
+        global_id: u64,
+    ) -> Result<Option<()>> {
+        let content_path = content.map(|uuid| update_uuid_to_file_path(&self.path, uuid));
+        let update_id = processing.id();
+
+        let file = match content_path {
+            Some(ref path) => {
+                let file = File::open(path)?;
+                Some(file)
+            }
+            None => None,
+        };
+
+        // Process the pending update using the provided user function.
+        let handle = Handle::current();
+        let result =
+            match handle.block_on(index_handle.update(index_uuid, processing.clone(), file)) {
+                Ok(result) => result,
+                Err(e) => Err(processing.fail(e.into())),
+            };
+
+        // Once the pending update have been successfully processed
+        // we must remove the content from the pending and processing stores and
+        // write the *new* meta to the processed-meta store and commit.
+        let mut wtxn = self.env.write_txn()?;
+        self.pending_queue
+            .delete(&mut wtxn, &(global_id, index_uuid, update_id))?;
+
+        let result = match result {
+            Ok(res) => res.into(),
+            Err(res) => res.into(),
+        };
+
+        self.updates
+            .put(&mut wtxn, &(index_uuid, update_id), &result)?;
+
+        wtxn.commit()?;
+
+        if let Some(ref path) = content_path {
+            remove_file(&path)?;
+        }
+
+        Ok(Some(()))
+    }
+
+    /// List the updates for `index_uuid`.
+    pub fn list(&self, index_uuid: Uuid) -> Result<Vec<UpdateStatus>> {
+        let mut update_list = BTreeMap::<u64, UpdateStatus>::new();
+
+        let txn = self.env.read_txn()?;
+
+        let pendings = self.pending_queue.iter(&txn)?.lazily_decode_data();
+        for entry in pendings {
+            let ((_, uuid, id), pending) = entry?;
+            if uuid == index_uuid {
+                update_list.insert(id, pending.decode()?.into());
+            }
+        }
+
+        let updates = self
+            .updates
+            .remap_key_type::<ByteSlice>()
+            .prefix_iter(&txn, index_uuid.as_bytes())?;
+
+        for entry in updates {
+            let (_, update) = entry?;
+            update_list.insert(update.id(), update);
+        }
+
+        // If the currently processing update is from this index, replace the corresponding pending update with this one.
+        match *self.state.read() {
+            State::Processing(uuid, ref processing) if uuid == index_uuid => {
+                update_list.insert(processing.id(), processing.clone().into());
+            }
+            _ => (),
+        }
+
+        Ok(update_list.into_iter().map(|(_, v)| v).collect())
+    }
+
+    /// Returns the update associated meta or `None` if the update doesn't exist.
+    pub fn meta(&self, index_uuid: Uuid, update_id: u64) -> heed::Result<Option<UpdateStatus>> {
+        // Check if the update is the one currently processing
+        match *self.state.read() {
+            State::Processing(uuid, ref processing)
+                if uuid == index_uuid && processing.id() == update_id =>
+            {
+                return Ok(Some(processing.clone().into()));
+            }
+            _ => (),
+        }
+
+        let txn = self.env.read_txn()?;
+        // Else, check if it is in the updates database:
+        let update = self.updates.get(&txn, &(index_uuid, update_id))?;
+
+        if let Some(update) = update {
+            return Ok(Some(update));
+        }
+
+        // If nothing was found yet, we resolve to iterate over the pending queue.
+        let pendings = self.pending_queue.iter(&txn)?.lazily_decode_data();
+
+        for entry in pendings {
+            let ((_, uuid, id), pending) = entry?;
+            if uuid == index_uuid && id == update_id {
+                return Ok(Some(pending.decode()?.into()));
+            }
+        }
+
+        // No update was found.
+        Ok(None)
+    }
+
+    /// Delete all updates for an index from the update store. If the currently processing update
+    /// is for `index_uuid`, the call will block until the update is terminated.
+    pub fn delete_all(&self, index_uuid: Uuid) -> Result<()> {
+        let mut txn = self.env.write_txn()?;
+        // Contains all the content file paths that we need to be removed if the deletion was successful.
+        let mut uuids_to_remove = Vec::new();
+
+        let mut pendings = self.pending_queue.iter_mut(&mut txn)?.lazily_decode_data();
+
+        while let Some(Ok(((_, uuid, _), pending))) = pendings.next() {
+            if uuid == index_uuid {
+                let mut pending = pending.decode()?;
+                if let Some(update_uuid) = pending.content.take() {
+                    uuids_to_remove.push(update_uuid);
+                }
+
+                // Invariant check: we can only delete the current entry when we don't hold
+                // references to it anymore. This must be done after we have retrieved its content.
+                unsafe {
+                    pendings.del_current()?;
+                }
+            }
+        }
+
+        drop(pendings);
+
+        let mut updates = self
+            .updates
+            .remap_key_type::<ByteSlice>()
+            .prefix_iter_mut(&mut txn, index_uuid.as_bytes())?
+            .lazily_decode_data();
+
+        while let Some(_) = updates.next() {
+            unsafe {
+                updates.del_current()?;
+            }
+        }
+
+        drop(updates);
+
+        txn.commit()?;
+
+        // If the currently processing update is from our index, we wait until it is
+        // finished before returning. This ensure that no write to the index occurs after we delete it.
+        if let State::Processing(uuid, _) = *self.state.read() {
+            if uuid == index_uuid {
+                // wait for a write lock, do nothing with it.
+                self.state.write();
+            }
+        }
+
+        // Finally, remove any outstanding update files. This must be done after waiting for the
+        // last update to ensure that the update files are not deleted before the update needs
+        // them.
+        uuids_to_remove
+            .iter()
+            .map(|uuid| update_uuid_to_file_path(&self.path, *uuid))
+            .for_each(|path| {
+                let _ = remove_file(path);
+            });
+
+        Ok(())
+    }
+
+    pub fn snapshot(
+        &self,
+        uuids: &HashSet<Uuid>,
+        path: impl AsRef<Path>,
+        handle: impl IndexActorHandle + Clone,
+    ) -> Result<()> {
+        let state_lock = self.state.write();
+        state_lock.swap(State::Snapshoting);
+
+        let txn = self.env.write_txn()?;
+
+        let update_path = path.as_ref().join("updates");
+        create_dir_all(&update_path)?;
+
+        // acquire write lock to prevent further writes during snapshot
+        create_dir_all(&update_path)?;
+        let db_path = update_path.join("data.mdb");
+
+        // create db snapshot
+        self.env.copy_to_path(&db_path, CompactionOption::Enabled)?;
+
+        let update_files_path = update_path.join(UPDATE_DIR);
+        create_dir_all(&update_files_path)?;
+
+        let pendings = self.pending_queue.iter(&txn)?.lazily_decode_data();
+
+        for entry in pendings {
+            let ((_, uuid, _), pending) = entry?;
+            if uuids.contains(&uuid) {
+                if let Enqueued {
+                    content: Some(uuid),
+                    ..
+                } = pending.decode()?
+                {
+                    let path = update_uuid_to_file_path(&self.path, uuid);
+                    copy(path, &update_files_path)?;
+                }
+            }
+        }
+
+        let path = &path.as_ref().to_path_buf();
+        let handle = &handle;
+        // Perform the snapshot of each index concurently. Only a third of the capabilities of
+        // the index actor at a time not to put too much pressure on the index actor
+        let mut stream = futures::stream::iter(uuids.iter())
+            .map(move |uuid| handle.snapshot(*uuid, path.clone()))
+            .buffer_unordered(CONCURRENT_INDEX_MSG / 3);
+
+        Handle::current().block_on(async {
+            while let Some(res) = stream.next().await {
+                res?;
+            }
+            Ok(()) as Result<()>
+        })?;
+
+        Ok(())
+    }
+
+    pub fn get_info(&self) -> Result<UpdateStoreInfo> {
+        let mut size = self.env.size();
+        let txn = self.env.read_txn()?;
+        for entry in self.pending_queue.iter(&txn)? {
+            let (_, pending) = entry?;
+            if let Enqueued {
+                content: Some(uuid),
+                ..
+            } = pending
+            {
+                let path = update_uuid_to_file_path(&self.path, uuid);
+                size += File::open(path)?.metadata()?.len();
+            }
+        }
+        let processing = match *self.state.read() {
+            State::Processing(uuid, _) => Some(uuid),
+            _ => None,
+        };
+
+        Ok(UpdateStoreInfo { size, processing })
+    }
+}
+
+fn update_uuid_to_file_path(root: impl AsRef<Path>, uuid: Uuid) -> PathBuf {
+    root.as_ref()
+        .join(UPDATE_DIR)
+        .join(format!("update_{}", uuid))
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use crate::index_controller::{
+        index_actor::{error::IndexActorError, MockIndexActorHandle},
+        UpdateResult,
+    };
+
+    use futures::future::ok;
+
+    #[actix_rt::test]
+    async fn test_next_id() {
+        let dir = tempfile::tempdir_in(".").unwrap();
+        let mut options = EnvOpenOptions::new();
+        let handle = Arc::new(MockIndexActorHandle::new());
+        options.map_size(4096 * 100);
+        let update_store = UpdateStore::open(
+            options,
+            dir.path(),
+            handle,
+            Arc::new(AtomicBool::new(false)),
+        )
+        .unwrap();
+
+        let index1_uuid = Uuid::new_v4();
+        let index2_uuid = Uuid::new_v4();
+
+        let mut txn = update_store.env.write_txn().unwrap();
+        let ids = update_store.next_update_id(&mut txn, index1_uuid).unwrap();
+        txn.commit().unwrap();
+        assert_eq!((0, 0), ids);
+
+        let mut txn = update_store.env.write_txn().unwrap();
+        let ids = update_store.next_update_id(&mut txn, index2_uuid).unwrap();
+        txn.commit().unwrap();
+        assert_eq!((1, 0), ids);
+
+        let mut txn = update_store.env.write_txn().unwrap();
+        let ids = update_store.next_update_id(&mut txn, index1_uuid).unwrap();
+        txn.commit().unwrap();
+        assert_eq!((2, 1), ids);
+    }
+
+    #[actix_rt::test]
+    async fn test_register_update() {
+        let dir = tempfile::tempdir_in(".").unwrap();
+        let mut options = EnvOpenOptions::new();
+        let handle = Arc::new(MockIndexActorHandle::new());
+        options.map_size(4096 * 100);
+        let update_store = UpdateStore::open(
+            options,
+            dir.path(),
+            handle,
+            Arc::new(AtomicBool::new(false)),
+        )
+        .unwrap();
+        let meta = UpdateMeta::ClearDocuments;
+        let uuid = Uuid::new_v4();
+        let store_clone = update_store.clone();
+        tokio::task::spawn_blocking(move || {
+            store_clone.register_update(meta, None, uuid).unwrap();
+        })
+        .await
+        .unwrap();
+
+        let txn = update_store.env.read_txn().unwrap();
+        assert!(update_store
+            .pending_queue
+            .get(&txn, &(0, uuid, 0))
+            .unwrap()
+            .is_some());
+    }
+
+    #[actix_rt::test]
+    async fn test_process_update() {
+        let dir = tempfile::tempdir_in(".").unwrap();
+        let mut handle = MockIndexActorHandle::new();
+
+        handle
+            .expect_update()
+            .times(2)
+            .returning(|_index_uuid, processing, _file| {
+                if processing.id() == 0 {
+                    Box::pin(ok(Ok(processing.process(UpdateResult::Other))))
+                } else {
+                    Box::pin(ok(Err(
+                        processing.fail(IndexActorError::ExistingPrimaryKey.into())
+                    )))
+                }
+            });
+
+        let handle = Arc::new(handle);
+
+        let mut options = EnvOpenOptions::new();
+        options.map_size(4096 * 100);
+        let store = UpdateStore::open(
+            options,
+            dir.path(),
+            handle.clone(),
+            Arc::new(AtomicBool::new(false)),
+        )
+        .unwrap();
+
+        // wait a bit for the event loop exit.
+        tokio::time::sleep(std::time::Duration::from_millis(50)).await;
+
+        let mut txn = store.env.write_txn().unwrap();
+
+        let update = Enqueued::new(UpdateMeta::ClearDocuments, 0, None);
+        let uuid = Uuid::new_v4();
+
+        store
+            .pending_queue
+            .put(&mut txn, &(0, uuid, 0), &update)
+            .unwrap();
+
+        let update = Enqueued::new(UpdateMeta::ClearDocuments, 1, None);
+
+        store
+            .pending_queue
+            .put(&mut txn, &(1, uuid, 1), &update)
+            .unwrap();
+
+        txn.commit().unwrap();
+
+        // Process the pending, and check that it has been moved to the update databases, and
+        // removed from the pending database.
+        let store_clone = store.clone();
+        tokio::task::spawn_blocking(move || {
+            store_clone.process_pending_update(handle.clone()).unwrap();
+            store_clone.process_pending_update(handle).unwrap();
+        })
+        .await
+        .unwrap();
+
+        let txn = store.env.read_txn().unwrap();
+
+        assert!(store.pending_queue.first(&txn).unwrap().is_none());
+        let update = store.updates.get(&txn, &(uuid, 0)).unwrap().unwrap();
+
+        assert!(matches!(update, UpdateStatus::Processed(_)));
+        let update = store.updates.get(&txn, &(uuid, 1)).unwrap().unwrap();
+
+        assert!(matches!(update, UpdateStatus::Failed(_)));
+    }
+}

meilisearch-http/src/index_controller/updates.rs

@@ -0,0 +1,233 @@
+use chrono::{DateTime, Utc};
+use milli::update::{DocumentAdditionResult, IndexDocumentsMethod, UpdateFormat};
+use serde::{Deserialize, Serialize};
+use uuid::Uuid;
+
+use crate::{
+    error::ResponseError,
+    index::{Settings, Unchecked},
+};
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+pub enum UpdateResult {
+    DocumentsAddition(DocumentAdditionResult),
+    DocumentDeletion { deleted: u64 },
+    Other,
+}
+
+#[allow(clippy::large_enum_variant)]
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(tag = "type")]
+pub enum UpdateMeta {
+    DocumentsAddition {
+        method: IndexDocumentsMethod,
+        format: UpdateFormat,
+        primary_key: Option<String>,
+    },
+    ClearDocuments,
+    DeleteDocuments {
+        ids: Vec<String>,
+    },
+    Settings(Settings<Unchecked>),
+}
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct Enqueued {
+    pub update_id: u64,
+    pub meta: UpdateMeta,
+    pub enqueued_at: DateTime<Utc>,
+    pub content: Option<Uuid>,
+}
+
+impl Enqueued {
+    pub fn new(meta: UpdateMeta, update_id: u64, content: Option<Uuid>) -> Self {
+        Self {
+            enqueued_at: Utc::now(),
+            meta,
+            update_id,
+            content,
+        }
+    }
+
+    pub fn processing(self) -> Processing {
+        Processing {
+            from: self,
+            started_processing_at: Utc::now(),
+        }
+    }
+
+    pub fn abort(self) -> Aborted {
+        Aborted {
+            from: self,
+            aborted_at: Utc::now(),
+        }
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        &self.meta
+    }
+
+    pub fn id(&self) -> u64 {
+        self.update_id
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct Processed {
+    pub success: UpdateResult,
+    pub processed_at: DateTime<Utc>,
+    #[serde(flatten)]
+    pub from: Processing,
+}
+
+impl Processed {
+    pub fn id(&self) -> u64 {
+        self.from.id()
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        self.from.meta()
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct Processing {
+    #[serde(flatten)]
+    pub from: Enqueued,
+    pub started_processing_at: DateTime<Utc>,
+}
+
+impl Processing {
+    pub fn id(&self) -> u64 {
+        self.from.id()
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        self.from.meta()
+    }
+
+    pub fn process(self, success: UpdateResult) -> Processed {
+        Processed {
+            success,
+            from: self,
+            processed_at: Utc::now(),
+        }
+    }
+
+    pub fn fail(self, error: ResponseError) -> Failed {
+        Failed {
+            from: self,
+            error,
+            failed_at: Utc::now(),
+        }
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize, Clone)]
+#[serde(rename_all = "camelCase")]
+pub struct Aborted {
+    #[serde(flatten)]
+    from: Enqueued,
+    aborted_at: DateTime<Utc>,
+}
+
+impl Aborted {
+    pub fn id(&self) -> u64 {
+        self.from.id()
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        self.from.meta()
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct Failed {
+    #[serde(flatten)]
+    pub from: Processing,
+    pub error: ResponseError,
+    pub failed_at: DateTime<Utc>,
+}
+
+impl Failed {
+    pub fn id(&self) -> u64 {
+        self.from.id()
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        self.from.meta()
+    }
+}
+
+#[derive(Debug, Serialize, Deserialize)]
+#[serde(tag = "status", rename_all = "camelCase")]
+pub enum UpdateStatus {
+    Processing(Processing),
+    Enqueued(Enqueued),
+    Processed(Processed),
+    Aborted(Aborted),
+    Failed(Failed),
+}
+
+impl UpdateStatus {
+    pub fn id(&self) -> u64 {
+        match self {
+            UpdateStatus::Processing(u) => u.id(),
+            UpdateStatus::Enqueued(u) => u.id(),
+            UpdateStatus::Processed(u) => u.id(),
+            UpdateStatus::Aborted(u) => u.id(),
+            UpdateStatus::Failed(u) => u.id(),
+        }
+    }
+
+    pub fn meta(&self) -> &UpdateMeta {
+        match self {
+            UpdateStatus::Processing(u) => u.meta(),
+            UpdateStatus::Enqueued(u) => u.meta(),
+            UpdateStatus::Processed(u) => u.meta(),
+            UpdateStatus::Aborted(u) => u.meta(),
+            UpdateStatus::Failed(u) => u.meta(),
+        }
+    }
+
+    pub fn processed(&self) -> Option<&Processed> {
+        match self {
+            UpdateStatus::Processed(p) => Some(p),
+            _ => None,
+        }
+    }
+}
+
+impl From<Enqueued> for UpdateStatus {
+    fn from(other: Enqueued) -> Self {
+        Self::Enqueued(other)
+    }
+}
+
+impl From<Aborted> for UpdateStatus {
+    fn from(other: Aborted) -> Self {
+        Self::Aborted(other)
+    }
+}
+
+impl From<Processed> for UpdateStatus {
+    fn from(other: Processed) -> Self {
+        Self::Processed(other)
+    }
+}
+
+impl From<Processing> for UpdateStatus {
+    fn from(other: Processing) -> Self {
+        Self::Processing(other)
+    }
+}
+
+impl From<Failed> for UpdateStatus {
+    fn from(other: Failed) -> Self {
+        Self::Failed(other)
+    }
+}

meilisearch-http/src/index_controller/uuid_resolver/actor.rs

@@ -0,0 +1,98 @@
+use std::{collections::HashSet, path::PathBuf};
+
+use log::{trace, warn};
+use tokio::sync::mpsc;
+use uuid::Uuid;
+
+use super::{error::UuidResolverError, Result, UuidResolveMsg, UuidStore};
+
+pub struct UuidResolverActor<S> {
+    inbox: mpsc::Receiver<UuidResolveMsg>,
+    store: S,
+}
+
+impl<S: UuidStore> UuidResolverActor<S> {
+    pub fn new(inbox: mpsc::Receiver<UuidResolveMsg>, store: S) -> Self {
+        Self { inbox, store }
+    }
+
+    pub async fn run(mut self) {
+        use UuidResolveMsg::*;
+
+        trace!("uuid resolver started");
+
+        loop {
+            match self.inbox.recv().await {
+                Some(Get { uid: name, ret }) => {
+                    let _ = ret.send(self.handle_get(name).await);
+                }
+                Some(Delete { uid: name, ret }) => {
+                    let _ = ret.send(self.handle_delete(name).await);
+                }
+                Some(List { ret }) => {
+                    let _ = ret.send(self.handle_list().await);
+                }
+                Some(Insert { ret, uuid, name }) => {
+                    let _ = ret.send(self.handle_insert(name, uuid).await);
+                }
+                Some(SnapshotRequest { path, ret }) => {
+                    let _ = ret.send(self.handle_snapshot(path).await);
+                }
+                Some(GetSize { ret }) => {
+                    let _ = ret.send(self.handle_get_size().await);
+                }
+                Some(DumpRequest { path, ret }) => {
+                    let _ = ret.send(self.handle_dump(path).await);
+                }
+                // all senders have been dropped, need to quit.
+                None => break,
+            }
+        }
+
+        warn!("exiting uuid resolver loop");
+    }
+
+    async fn handle_get(&self, uid: String) -> Result<Uuid> {
+        self.store
+            .get_uuid(uid.clone())
+            .await?
+            .ok_or(UuidResolverError::UnexistingIndex(uid))
+    }
+
+    async fn handle_delete(&self, uid: String) -> Result<Uuid> {
+        self.store
+            .delete(uid.clone())
+            .await?
+            .ok_or(UuidResolverError::UnexistingIndex(uid))
+    }
+
+    async fn handle_list(&self) -> Result<Vec<(String, Uuid)>> {
+        let result = self.store.list().await?;
+        Ok(result)
+    }
+
+    async fn handle_snapshot(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        self.store.snapshot(path).await
+    }
+
+    async fn handle_dump(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        self.store.dump(path).await
+    }
+
+    async fn handle_insert(&self, uid: String, uuid: Uuid) -> Result<()> {
+        if !is_index_uid_valid(&uid) {
+            return Err(UuidResolverError::BadlyFormatted(uid));
+        }
+        self.store.insert(uid, uuid).await?;
+        Ok(())
+    }
+
+    async fn handle_get_size(&self) -> Result<u64> {
+        self.store.get_size().await
+    }
+}
+
+fn is_index_uid_valid(uid: &str) -> bool {
+    uid.chars()
+        .all(|x| x.is_ascii_alphanumeric() || x == '-' || x == '_')
+}

meilisearch-http/src/index_controller/uuid_resolver/error.rs

@@ -0,0 +1,34 @@
+use meilisearch_error::{Code, ErrorCode};
+
+pub type Result<T> = std::result::Result<T, UuidResolverError>;
+
+#[derive(Debug, thiserror::Error)]
+pub enum UuidResolverError {
+    #[error("Index already exists.")]
+    NameAlreadyExist,
+    #[error("Index \"{0}\" not found.")]
+    UnexistingIndex(String),
+    #[error("Index must have a valid uid; Index uid can be of type integer or string only composed of alphanumeric characters, hyphens (-) and underscores (_).")]
+    BadlyFormatted(String),
+    #[error("Internal error: {0}")]
+    Internal(Box<dyn std::error::Error + Sync + Send + 'static>),
+}
+
+internal_error!(
+    UuidResolverError: heed::Error,
+    uuid::Error,
+    std::io::Error,
+    tokio::task::JoinError,
+    serde_json::Error
+);
+
+impl ErrorCode for UuidResolverError {
+    fn error_code(&self) -> Code {
+        match self {
+            UuidResolverError::NameAlreadyExist => Code::IndexAlreadyExists,
+            UuidResolverError::UnexistingIndex(_) => Code::IndexNotFound,
+            UuidResolverError::BadlyFormatted(_) => Code::InvalidIndexUid,
+            UuidResolverError::Internal(_) => Code::Internal,
+        }
+    }
+}

meilisearch-http/src/index_controller/uuid_resolver/handle_impl.rs

@@ -0,0 +1,87 @@
+use std::collections::HashSet;
+use std::path::{Path, PathBuf};
+
+use tokio::sync::{mpsc, oneshot};
+use uuid::Uuid;
+
+use super::{HeedUuidStore, Result, UuidResolveMsg, UuidResolverActor, UuidResolverHandle};
+
+#[derive(Clone)]
+pub struct UuidResolverHandleImpl {
+    sender: mpsc::Sender<UuidResolveMsg>,
+}
+
+impl UuidResolverHandleImpl {
+    pub fn new(path: impl AsRef<Path>) -> Result<Self> {
+        let (sender, reveiver) = mpsc::channel(100);
+        let store = HeedUuidStore::new(path)?;
+        let actor = UuidResolverActor::new(reveiver, store);
+        tokio::spawn(actor.run());
+        Ok(Self { sender })
+    }
+}
+
+#[async_trait::async_trait]
+impl UuidResolverHandle for UuidResolverHandleImpl {
+    async fn get(&self, name: String) -> Result<Uuid> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::Get { uid: name, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+
+    async fn delete(&self, name: String) -> Result<Uuid> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::Delete { uid: name, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+
+    async fn list(&self) -> Result<Vec<(String, Uuid)>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::List { ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+
+    async fn insert(&self, name: String, uuid: Uuid) -> Result<()> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::Insert { ret, name, uuid };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+
+    async fn snapshot(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::SnapshotRequest { path, ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+
+    async fn get_size(&self) -> Result<u64> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::GetSize { ret };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+    async fn dump(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        let (ret, receiver) = oneshot::channel();
+        let msg = UuidResolveMsg::DumpRequest { ret, path };
+        let _ = self.sender.send(msg).await;
+        Ok(receiver
+            .await
+            .expect("Uuid resolver actor has been killed")?)
+    }
+}

meilisearch-http/src/index_controller/uuid_resolver/message.rs

@@ -0,0 +1,37 @@
+use std::collections::HashSet;
+use std::path::PathBuf;
+
+use tokio::sync::oneshot;
+use uuid::Uuid;
+
+use super::Result;
+
+pub enum UuidResolveMsg {
+    Get {
+        uid: String,
+        ret: oneshot::Sender<Result<Uuid>>,
+    },
+    Delete {
+        uid: String,
+        ret: oneshot::Sender<Result<Uuid>>,
+    },
+    List {
+        ret: oneshot::Sender<Result<Vec<(String, Uuid)>>>,
+    },
+    Insert {
+        uuid: Uuid,
+        name: String,
+        ret: oneshot::Sender<Result<()>>,
+    },
+    SnapshotRequest {
+        path: PathBuf,
+        ret: oneshot::Sender<Result<HashSet<Uuid>>>,
+    },
+    GetSize {
+        ret: oneshot::Sender<Result<u64>>,
+    },
+    DumpRequest {
+        path: PathBuf,
+        ret: oneshot::Sender<Result<HashSet<Uuid>>>,
+    },
+}

meilisearch-http/src/index_controller/uuid_resolver/mod.rs

@@ -0,0 +1,35 @@
+mod actor;
+pub mod error;
+mod handle_impl;
+mod message;
+pub mod store;
+
+use std::collections::HashSet;
+use std::path::PathBuf;
+
+use uuid::Uuid;
+
+use actor::UuidResolverActor;
+use error::Result;
+use message::UuidResolveMsg;
+use store::UuidStore;
+
+#[cfg(test)]
+use mockall::automock;
+
+pub use handle_impl::UuidResolverHandleImpl;
+pub use store::HeedUuidStore;
+
+const UUID_STORE_SIZE: usize = 1_073_741_824; //1GiB
+
+#[async_trait::async_trait]
+#[cfg_attr(test, automock)]
+pub trait UuidResolverHandle {
+    async fn get(&self, name: String) -> Result<Uuid>;
+    async fn insert(&self, name: String, uuid: Uuid) -> Result<()>;
+    async fn delete(&self, name: String) -> Result<Uuid>;
+    async fn list(&self) -> Result<Vec<(String, Uuid)>>;
+    async fn snapshot(&self, path: PathBuf) -> Result<HashSet<Uuid>>;
+    async fn get_size(&self) -> Result<u64>;
+    async fn dump(&self, path: PathBuf) -> Result<HashSet<Uuid>>;
+}

meilisearch-http/src/index_controller/uuid_resolver/store.rs

@@ -0,0 +1,224 @@
+use std::collections::HashSet;
+use std::fs::{create_dir_all, File};
+use std::io::{BufRead, BufReader, Write};
+use std::path::{Path, PathBuf};
+
+use heed::types::{ByteSlice, Str};
+use heed::{CompactionOption, Database, Env, EnvOpenOptions};
+use serde::{Deserialize, Serialize};
+use uuid::Uuid;
+
+use super::{error::UuidResolverError, Result, UUID_STORE_SIZE};
+use crate::helpers::EnvSizer;
+
+#[derive(Serialize, Deserialize)]
+struct DumpEntry {
+    uuid: Uuid,
+    uid: String,
+}
+
+const UUIDS_DB_PATH: &str = "index_uuids";
+
+#[async_trait::async_trait]
+pub trait UuidStore: Sized {
+    // Create a new entry for `name`. Return an error if `err` and the entry already exists, return
+    // the uuid otherwise.
+    async fn get_uuid(&self, uid: String) -> Result<Option<Uuid>>;
+    async fn delete(&self, uid: String) -> Result<Option<Uuid>>;
+    async fn list(&self) -> Result<Vec<(String, Uuid)>>;
+    async fn insert(&self, name: String, uuid: Uuid) -> Result<()>;
+    async fn snapshot(&self, path: PathBuf) -> Result<HashSet<Uuid>>;
+    async fn get_size(&self) -> Result<u64>;
+    async fn dump(&self, path: PathBuf) -> Result<HashSet<Uuid>>;
+}
+
+#[derive(Clone)]
+pub struct HeedUuidStore {
+    env: Env,
+    db: Database<Str, ByteSlice>,
+}
+
+impl HeedUuidStore {
+    pub fn new(path: impl AsRef<Path>) -> Result<Self> {
+        let path = path.as_ref().join(UUIDS_DB_PATH);
+        create_dir_all(&path)?;
+        let mut options = EnvOpenOptions::new();
+        options.map_size(UUID_STORE_SIZE); // 1GB
+        let env = options.open(path)?;
+        let db = env.create_database(None)?;
+        Ok(Self { env, db })
+    }
+
+    pub fn get_uuid(&self, name: String) -> Result<Option<Uuid>> {
+        let env = self.env.clone();
+        let db = self.db;
+        let txn = env.read_txn()?;
+        match db.get(&txn, &name)? {
+            Some(uuid) => {
+                let uuid = Uuid::from_slice(uuid)?;
+                Ok(Some(uuid))
+            }
+            None => Ok(None),
+        }
+    }
+
+    pub fn delete(&self, uid: String) -> Result<Option<Uuid>> {
+        let env = self.env.clone();
+        let db = self.db;
+        let mut txn = env.write_txn()?;
+        match db.get(&txn, &uid)? {
+            Some(uuid) => {
+                let uuid = Uuid::from_slice(uuid)?;
+                db.delete(&mut txn, &uid)?;
+                txn.commit()?;
+                Ok(Some(uuid))
+            }
+            None => Ok(None),
+        }
+    }
+
+    pub fn list(&self) -> Result<Vec<(String, Uuid)>> {
+        let env = self.env.clone();
+        let db = self.db;
+        let txn = env.read_txn()?;
+        let mut entries = Vec::new();
+        for entry in db.iter(&txn)? {
+            let (name, uuid) = entry?;
+            let uuid = Uuid::from_slice(uuid)?;
+            entries.push((name.to_owned(), uuid))
+        }
+        Ok(entries)
+    }
+
+    pub fn insert(&self, name: String, uuid: Uuid) -> Result<()> {
+        let env = self.env.clone();
+        let db = self.db;
+        let mut txn = env.write_txn()?;
+
+        if db.get(&txn, &name)?.is_some() {
+            return Err(UuidResolverError::NameAlreadyExist);
+        }
+
+        db.put(&mut txn, &name, uuid.as_bytes())?;
+        txn.commit()?;
+        Ok(())
+    }
+
+    pub fn snapshot(&self, mut path: PathBuf) -> Result<HashSet<Uuid>> {
+        let env = self.env.clone();
+        let db = self.db;
+        // Write transaction to acquire a lock on the database.
+        let txn = env.write_txn()?;
+        let mut entries = HashSet::new();
+        for entry in db.iter(&txn)? {
+            let (_, uuid) = entry?;
+            let uuid = Uuid::from_slice(uuid)?;
+            entries.insert(uuid);
+        }
+
+        // only perform snapshot if there are indexes
+        if !entries.is_empty() {
+            path.push(UUIDS_DB_PATH);
+            create_dir_all(&path).unwrap();
+            path.push("data.mdb");
+            env.copy_to_path(path, CompactionOption::Enabled)?;
+        }
+        Ok(entries)
+    }
+
+    pub fn get_size(&self) -> Result<u64> {
+        Ok(self.env.size())
+    }
+
+    pub fn dump(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        let dump_path = path.join(UUIDS_DB_PATH);
+        create_dir_all(&dump_path)?;
+        let dump_file_path = dump_path.join("data.jsonl");
+        let mut dump_file = File::create(&dump_file_path)?;
+        let mut uuids = HashSet::new();
+
+        let txn = self.env.read_txn()?;
+        for entry in self.db.iter(&txn)? {
+            let (uid, uuid) = entry?;
+            let uid = uid.to_string();
+            let uuid = Uuid::from_slice(uuid)?;
+
+            let entry = DumpEntry { uuid, uid };
+            serde_json::to_writer(&mut dump_file, &entry)?;
+            dump_file.write_all(b"\n").unwrap();
+
+            uuids.insert(uuid);
+        }
+
+        Ok(uuids)
+    }
+
+    pub fn load_dump(src: impl AsRef<Path>, dst: impl AsRef<Path>) -> Result<()> {
+        let uuid_resolver_path = dst.as_ref().join(UUIDS_DB_PATH);
+        std::fs::create_dir_all(&uuid_resolver_path)?;
+
+        let src_indexes = src.as_ref().join(UUIDS_DB_PATH).join("data.jsonl");
+        let indexes = File::open(&src_indexes)?;
+        let mut indexes = BufReader::new(indexes);
+        let mut line = String::new();
+
+        let db = Self::new(dst)?;
+        let mut txn = db.env.write_txn()?;
+
+        loop {
+            match indexes.read_line(&mut line) {
+                Ok(0) => break,
+                Ok(_) => {
+                    let DumpEntry { uuid, uid } = serde_json::from_str(&line)?;
+                    println!("importing {} {}", uid, uuid);
+                    db.db.put(&mut txn, &uid, uuid.as_bytes())?;
+                }
+                Err(e) => return Err(e.into()),
+            }
+
+            line.clear();
+        }
+        txn.commit()?;
+
+        db.env.prepare_for_closing().wait();
+
+        Ok(())
+    }
+}
+
+#[async_trait::async_trait]
+impl UuidStore for HeedUuidStore {
+    async fn get_uuid(&self, name: String) -> Result<Option<Uuid>> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.get_uuid(name)).await?
+    }
+
+    async fn delete(&self, uid: String) -> Result<Option<Uuid>> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.delete(uid)).await?
+    }
+
+    async fn list(&self) -> Result<Vec<(String, Uuid)>> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.list()).await?
+    }
+
+    async fn insert(&self, name: String, uuid: Uuid) -> Result<()> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.insert(name, uuid)).await?
+    }
+
+    async fn snapshot(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.snapshot(path)).await?
+    }
+
+    async fn get_size(&self) -> Result<u64> {
+        self.get_size()
+    }
+
+    async fn dump(&self, path: PathBuf) -> Result<HashSet<Uuid>> {
+        let this = self.clone();
+        tokio::task::spawn_blocking(move || this.dump(path)).await?
+    }
+}

meilisearch-http/src/lib.rs

@@ -0,0 +1,171 @@
+//! # MeiliSearch
+//! Hello there, future contributors. If you are here and see this code, it's probably because you want to add a super new fancy feature in MeiliSearch or fix a bug and first of all, thank you for that!
+//!
+//! To help you in this task, we'll try to do a little overview of the project.
+//! ## Milli
+//! [Milli](https://github.com/meilisearch/milli) is the core library of MeiliSearch. It's where we actually index documents and perform searches. Its purpose is to do these two tasks as fast as possible. You can give an update to milli, and it'll uses as many cores as provided to perform it as fast as possible. Nothing more. You can perform searches at the same time (search only uses one core).
+//! As you can see, we're missing quite a lot of features here; milli does not handle multiples indexes, it can't queue updates, it doesn't provide any web / API frontend, it doesn't implement dumps or snapshots, etc...
+//!
+//! ## `Index` module
+//! The [index] module is what encapsulates one milli index. It abstracts over its transaction and isolates a task that can be run into a thread. This is the unit of interaction with milli.
+//! If you add a feature to milli, you'll probably need to add it in this module too before exposing it to the rest of meilisearch.
+//!
+//! ## `IndexController` module
+//! To handle multiple indexes, we created an [index_controller]. It's in charge of creating new indexes, keeping references to all its indexes, forward asynchronous updates to its indexes, and provide an API to search in its indexes synchronously.
+//! To achieves this goal, we use an [actor model](https://en.wikipedia.org/wiki/Actor_model).
+//!
+//! ### The actor model
+//! Every actor is composed of at least three files:
+//! - `mod.rs` declare and import all the files used by the actor. We also describe the interface (= all the methods) used to interact with the actor. If you are not modifying anything inside of an actor, this is usually all you need to see.
+//! - `handle_impl.rs` implements the interface described in the `mod.rs`; in reality, there is no code logic in this file. Every method is only wrapping its parameters in a structure that is sent to the actor. This is useful for test and futureproofing.
+//! - `message.rs` contains an enum that describes all the interactions you can have with the actor.
+//! - `actor.rs` is used to create and execute the actor. It's where we'll write the loop looking for new messages and actually perform the tasks.
+//!
+//! MeiliSearch currently uses four actors:
+//! - [`uuid_resolver`](index_controller/uuid_resolver/index.html) hold the association between the user-provided indexes name and the internal [`uuid`](https://en.wikipedia.org/wiki/Universally_unique_identifier) representation we use.
+//! - [`index_actor`](index_controller::index_actor) is our representation of multiples indexes. Any request made to MeiliSearch that needs to talk to milli will pass through this actor.
+//! - [`update_actor`](index_controller/update_actor/index.html) is in charge of indexes updates. Since updates can take a long time to receive and  process, we need to:
+//!   1. Store them as fast as possible so we can continue to receive other updates even if nothing has been processed
+//!   2. Feed the `index_actor` with a new update every time it finished its current job.
+//! - [`dump_actor`](index_controller/dump_actor/index.html) this actor handle the  [dumps](https://docs.meilisearch.com/reference/api/dump.html). It needs to contact all the others actors and create a dump of everything that was currently happening.
+//!
+//! ## Data module
+//! The [data] module provide a unified interface to communicate with the index controller and other services (snapshot, dumps, ...), initialize the MeiliSearch instance
+//!
+//! ## HTTP server
+//! To handle the web and API part, we are using [actix-web](https://docs.rs/actix-web/); you can find all routes in the [routes] module.
+//! Currently, the configuration of actix-web is made in the [lib.rs](crate).
+//! Most of the routes use [extractors] to handle the authentication.
+
+#![allow(rustdoc::private_intra_doc_links)]
+pub mod data;
+#[macro_use]
+pub mod error;
+#[macro_use]
+pub mod extractors;
+pub mod helpers;
+mod index;
+mod index_controller;
+pub mod option;
+pub mod routes;
+
+#[cfg(all(not(debug_assertions), feature = "analytics"))]
+pub mod analytics;
+
+use crate::extractors::authentication::AuthConfig;
+
+pub use self::data::Data;
+pub use option::Opt;
+
+use actix_web::web;
+
+use extractors::authentication::policies::*;
+use extractors::payload::PayloadConfig;
+
+pub fn configure_data(config: &mut web::ServiceConfig, data: Data) {
+    let http_payload_size_limit = data.http_payload_size_limit();
+    config
+        .data(data.clone())
+        .app_data(data)
+        .app_data(
+            web::JsonConfig::default()
+                .limit(http_payload_size_limit)
+                .content_type(|_mime| true) // Accept all mime types
+                .error_handler(|err, _req| error::payload_error_handler(err).into()),
+        )
+        .app_data(PayloadConfig::new(http_payload_size_limit))
+        .app_data(
+            web::QueryConfig::default()
+                .error_handler(|err, _req| error::payload_error_handler(err).into()),
+        );
+}
+
+pub fn configure_auth(config: &mut web::ServiceConfig, data: &Data) {
+    let keys = data.api_keys();
+    let auth_config = if let Some(ref master_key) = keys.master {
+        let private_key = keys.private.as_ref().unwrap();
+        let public_key = keys.public.as_ref().unwrap();
+        let mut policies = init_policies!(Public, Private, Admin);
+        create_users!(
+            policies,
+            master_key.as_bytes() => { Admin, Private, Public },
+            private_key.as_bytes() => { Private, Public },
+            public_key.as_bytes() => { Public }
+        );
+        AuthConfig::Auth(policies)
+    } else {
+        AuthConfig::NoAuth
+    };
+
+    config.app_data(auth_config);
+}
+
+#[cfg(feature = "mini-dashboard")]
+pub fn dashboard(config: &mut web::ServiceConfig, enable_frontend: bool) {
+    use actix_web::HttpResponse;
+    use actix_web_static_files::Resource;
+
+    mod generated {
+        include!(concat!(env!("OUT_DIR"), "/generated.rs"));
+    }
+
+    if enable_frontend {
+        let generated = generated::generate();
+        let mut scope = web::scope("/");
+        // Generate routes for mini-dashboard assets
+        for (path, resource) in generated.into_iter() {
+            let Resource {
+                mime_type, data, ..
+            } = resource;
+            // Redirect index.html to /
+            if path == "index.html" {
+                config.service(web::resource("/").route(
+                    web::get().to(move || HttpResponse::Ok().content_type(mime_type).body(data)),
+                ));
+            } else {
+                scope = scope.service(web::resource(path).route(
+                    web::get().to(move || HttpResponse::Ok().content_type(mime_type).body(data)),
+                ));
+            }
+        }
+        config.service(scope);
+    } else {
+        config.service(web::resource("/").route(web::get().to(routes::running)));
+    }
+}
+
+#[cfg(not(feature = "mini-dashboard"))]
+pub fn dashboard(config: &mut web::ServiceConfig, _enable_frontend: bool) {
+    config.service(web::resource("/").route(web::get().to(routes::running)));
+}
+
+#[macro_export]
+macro_rules! create_app {
+    ($data:expr, $enable_frontend:expr) => {{
+        use actix_cors::Cors;
+        use actix_web::middleware::TrailingSlash;
+        use actix_web::App;
+        use actix_web::{middleware, web};
+        use meilisearch_http::routes;
+        use meilisearch_http::{configure_auth, configure_data, dashboard};
+
+        App::new()
+            .configure(|s| configure_data(s, $data.clone()))
+            .configure(|s| configure_auth(s, &$data))
+            .configure(routes::configure)
+            .configure(|s| dashboard(s, $enable_frontend))
+            .wrap(
+                Cors::default()
+                    .send_wildcard()
+                    .allowed_headers(vec!["content-type", "x-meili-api-key"])
+                    .allow_any_origin()
+                    .allow_any_method()
+                    .max_age(86_400), // 24h
+            )
+            .wrap(middleware::Logger::default())
+            .wrap(middleware::Compress::default())
+            .wrap(middleware::NormalizePath::new(
+                middleware::TrailingSlash::Trim,
+            ))
+    }};
+}

meilisearch-http/src/main.rs

@@ -0,0 +1,131 @@
+use std::env;
+
+use actix_web::HttpServer;
+use main_error::MainError;
+use meilisearch_http::{create_app, Data, Opt};
+use structopt::StructOpt;
+
+#[cfg(all(not(debug_assertions), feature = "analytics"))]
+use meilisearch_http::analytics;
+
+#[cfg(target_os = "linux")]
+#[global_allocator]
+static ALLOC: jemallocator::Jemalloc = jemallocator::Jemalloc;
+
+#[actix_web::main]
+async fn main() -> Result<(), MainError> {
+    let opt = Opt::from_args();
+
+    let mut log_builder = env_logger::Builder::new();
+    log_builder.parse_filters(&opt.log_level);
+    if opt.log_level == "info" {
+        // if we are in info we only allow the warn log_level for milli
+        log_builder.filter_module("milli", log::LevelFilter::Warn);
+    }
+
+    log_builder.init();
+
+    match opt.env.as_ref() {
+        "production" => {
+            if opt.master_key.is_none() {
+                return Err(
+                    "In production mode, the environment variable MEILI_MASTER_KEY is mandatory"
+                        .into(),
+                );
+            }
+        }
+        "development" => (),
+        _ => unreachable!(),
+    }
+
+    let data = Data::new(opt.clone())?;
+
+    #[cfg(all(not(debug_assertions), feature = "analytics"))]
+    if !opt.no_analytics {
+        let analytics_data = data.clone();
+        let analytics_opt = opt.clone();
+        tokio::task::spawn(analytics::analytics_sender(analytics_data, analytics_opt));
+    }
+
+    print_launch_resume(&opt, &data);
+
+    run_http(data, opt).await?;
+
+    Ok(())
+}
+
+async fn run_http(data: Data, opt: Opt) -> Result<(), Box<dyn std::error::Error>> {
+    let _enable_dashboard = &opt.env == "development";
+    let http_server = HttpServer::new(move || create_app!(data, _enable_dashboard))
+        // Disable signals allows the server to terminate immediately when a user enter CTRL-C
+        .disable_signals();
+
+    if let Some(config) = opt.get_ssl_config()? {
+        http_server
+            .bind_rustls(opt.http_addr, config)?
+            .run()
+            .await?;
+    } else {
+        http_server.bind(opt.http_addr)?.run().await?;
+    }
+    Ok(())
+}
+
+pub fn print_launch_resume(opt: &Opt, data: &Data) {
+    let commit_sha = option_env!("VERGEN_GIT_SHA").unwrap_or("unknown");
+    let commit_date = option_env!("VERGEN_GIT_COMMIT_TIMESTAMP").unwrap_or("unknown");
+
+    let ascii_name = r#"
+888b     d888          d8b 888 d8b  .d8888b.                                    888
+8888b   d8888          Y8P 888 Y8P d88P  Y88b                                   888
+88888b.d88888              888     Y88b.                                        888
+888Y88888P888  .d88b.  888 888 888  "Y888b.    .d88b.   8888b.  888d888 .d8888b 88888b.
+888 Y888P 888 d8P  Y8b 888 888 888     "Y88b. d8P  Y8b     "88b 888P"  d88P"    888 "88b
+888  Y8P  888 88888888 888 888 888       "888 88888888 .d888888 888    888      888  888
+888   "   888 Y8b.     888 888 888 Y88b  d88P Y8b.     888  888 888    Y88b.    888  888
+888       888  "Y8888  888 888 888  "Y8888P"   "Y8888  "Y888888 888     "Y8888P 888  888
+"#;
+
+    eprintln!("{}", ascii_name);
+
+    eprintln!("Database path:\t\t{:?}", opt.db_path);
+    eprintln!("Server listening on:\t\"http://{}\"", opt.http_addr);
+    eprintln!("Environment:\t\t{:?}", opt.env);
+    eprintln!("Commit SHA:\t\t{:?}", commit_sha.to_string());
+    eprintln!("Commit date:\t\t{:?}", commit_date.to_string());
+    eprintln!(
+        "Package version:\t{:?}",
+        env!("CARGO_PKG_VERSION").to_string()
+    );
+
+    #[cfg(all(not(debug_assertions), feature = "analytics"))]
+    {
+        if opt.no_analytics {
+            eprintln!("Anonymous telemetry:\t\"Disabled\"");
+        } else {
+            eprintln!(
+                "
+Thank you for using MeiliSearch!
+
+We collect anonymized analytics to improve our product and your experience. To learn more, including how to turn off analytics, visit our dedicated documentation page: https://docs.meilisearch.com/learn/what_is_meilisearch/telemetry.html
+
+Anonymous telemetry:   \"Enabled\""
+            );
+        }
+    }
+
+    eprintln!();
+
+    if data.api_keys().master.is_some() {
+        eprintln!("A Master Key has been set. Requests to MeiliSearch won't be authorized unless you provide an authentication key.");
+    } else {
+        eprintln!("No master key found; The server will accept unidentified requests. \
+            If you need some protection in development mode, please export a key: export MEILI_MASTER_KEY=xxx");
+    }
+
+    eprintln!();
+    eprintln!("Documentation:\t\thttps://docs.meilisearch.com");
+    eprintln!("Source code:\t\thttps://github.com/meilisearch/meilisearch");
+    eprintln!("Contact:\t\thttps://docs.meilisearch.com/resources/contact.html or bonjour@meilisearch.com");
+    eprintln!();
+}

meilisearch-http/src/option.rs

@@ -0,0 +1,328 @@
+use byte_unit::ByteError;
+use std::fmt;
+use std::io::{BufReader, Read};
+use std::ops::Deref;
+use std::path::PathBuf;
+use std::str::FromStr;
+use std::sync::Arc;
+use std::{error, fs};
+
+use byte_unit::Byte;
+use milli::CompressionType;
+use rustls::internal::pemfile::{certs, pkcs8_private_keys, rsa_private_keys};
+use rustls::{
+    AllowAnyAnonymousOrAuthenticatedClient, AllowAnyAuthenticatedClient, NoClientAuth,
+    RootCertStore,
+};
+use structopt::StructOpt;
+use sysinfo::{RefreshKind, System, SystemExt};
+
+#[derive(Debug, Clone, StructOpt)]
+pub struct IndexerOpts {
+    /// The amount of documents to skip before printing
+    /// a log regarding the indexing advancement.
+    #[structopt(long, default_value = "100000")] // 100k
+    pub log_every_n: usize,
+
+    /// Grenad max number of chunks in bytes.
+    #[structopt(long)]
+    pub max_nb_chunks: Option<usize>,
+
+    /// The maximum amount of memory the indexer will use. It defaults to 2/3
+    /// of the available memory. It is recommended to use something like 80%-90%
+    /// of the available memory, no more.
+    ///
+    /// In case the engine is unable to retrieve the available memory the engine will
+    /// try to use the memory it needs but without real limit, this can lead to
+    /// Out-Of-Memory issues and it is recommended to specify the amount of memory to use.
+    #[structopt(long, default_value)]
+    pub max_memory: MaxMemory,
+
+    /// The name of the compression algorithm to use when compressing intermediate
+    /// Grenad chunks while indexing documents.
+    ///
+    /// Choosing a fast algorithm will make the indexing faster but may consume more memory.
+    #[structopt(long, default_value = "snappy", possible_values = &["snappy", "zlib", "lz4", "lz4hc", "zstd"])]
+    pub chunk_compression_type: CompressionType,
+
+    /// The level of compression of the chosen algorithm.
+    #[structopt(long, requires = "chunk-compression-type")]
+    pub chunk_compression_level: Option<u32>,
+
+    /// Number of parallel jobs for indexing, defaults to # of CPUs.
+    #[structopt(long)]
+    pub indexing_jobs: Option<usize>,
+}
+
+impl Default for IndexerOpts {
+    fn default() -> Self {
+        Self {
+            log_every_n: 100_000,
+            max_nb_chunks: None,
+            max_memory: MaxMemory::default(),
+            chunk_compression_type: CompressionType::None,
+            chunk_compression_level: None,
+            indexing_jobs: None,
+        }
+    }
+}
+
+const POSSIBLE_ENV: [&str; 2] = ["development", "production"];
+
+#[derive(Debug, Clone, StructOpt)]
+pub struct Opt {
+    /// The destination where the database must be created.
+    #[structopt(long, env = "MEILI_DB_PATH", default_value = "./data.ms")]
+    pub db_path: PathBuf,
+
+    /// The address on which the http server will listen.
+    #[structopt(long, env = "MEILI_HTTP_ADDR", default_value = "127.0.0.1:7700")]
+    pub http_addr: String,
+
+    /// The master key allowing you to do everything on the server.
+    #[structopt(long, env = "MEILI_MASTER_KEY")]
+    pub master_key: Option<String>,
+
+    /// This environment variable must be set to `production` if you are running in production.
+    /// If the server is running in development mode more logs will be displayed,
+    /// and the master key can be avoided which implies that there is no security on the updates routes.
+    /// This is useful to debug when integrating the engine with another service.
+    #[structopt(long, env = "MEILI_ENV", default_value = "development", possible_values = &POSSIBLE_ENV)]
+    pub env: String,
+
+    /// Do not send analytics to Meili.
+    #[cfg(all(not(debug_assertions), feature = "analytics"))]
+    #[structopt(long, env = "MEILI_NO_ANALYTICS")]
+    pub no_analytics: bool,
+
+    /// The maximum size, in bytes, of the main lmdb database directory
+    #[structopt(long, env = "MEILI_MAX_INDEX_SIZE", default_value = "100 GiB")]
+    pub max_index_size: Byte,
+
+    /// The maximum size, in bytes, of the update lmdb database directory
+    #[structopt(long, env = "MEILI_MAX_UDB_SIZE", default_value = "100 GiB")]
+    pub max_udb_size: Byte,
+
+    /// The maximum size, in bytes, of accepted JSON payloads
+    #[structopt(long, env = "MEILI_HTTP_PAYLOAD_SIZE_LIMIT", default_value = "100 MB")]
+    pub http_payload_size_limit: Byte,
+
+    /// Read server certificates from CERTFILE.
+    /// This should contain PEM-format certificates
+    /// in the right order (the first certificate should
+    /// certify KEYFILE, the last should be a root CA).
+    #[structopt(long, env = "MEILI_SSL_CERT_PATH", parse(from_os_str))]
+    pub ssl_cert_path: Option<PathBuf>,
+
+    /// Read private key from KEYFILE.  This should be a RSA
+    /// private key or PKCS8-encoded private key, in PEM format.
+    #[structopt(long, env = "MEILI_SSL_KEY_PATH", parse(from_os_str))]
+    pub ssl_key_path: Option<PathBuf>,
+
+    /// Enable client authentication, and accept certificates
+    /// signed by those roots provided in CERTFILE.
+    #[structopt(long, env = "MEILI_SSL_AUTH_PATH", parse(from_os_str))]
+    pub ssl_auth_path: Option<PathBuf>,
+
+    /// Read DER-encoded OCSP response from OCSPFILE and staple to certificate.
+    /// Optional
+    #[structopt(long, env = "MEILI_SSL_OCSP_PATH", parse(from_os_str))]
+    pub ssl_ocsp_path: Option<PathBuf>,
+
+    /// Send a fatal alert if the client does not complete client authentication.
+    #[structopt(long, env = "MEILI_SSL_REQUIRE_AUTH")]
+    pub ssl_require_auth: bool,
+
+    /// SSL support session resumption
+    #[structopt(long, env = "MEILI_SSL_RESUMPTION")]
+    pub ssl_resumption: bool,
+
+    /// SSL support tickets.
+    #[structopt(long, env = "MEILI_SSL_TICKETS")]
+    pub ssl_tickets: bool,
+
+    /// Defines the path of the snapshot file to import.
+    /// This option will, by default, stop the process if a database already exist or if no snapshot exists at
+    /// the given path. If this option is not specified no snapshot is imported.
+    #[structopt(long)]
+    pub import_snapshot: Option<PathBuf>,
+
+    /// The engine will ignore a missing snapshot and not return an error in such case.
+    #[structopt(long, requires = "import-snapshot")]
+    pub ignore_missing_snapshot: bool,
+
+    /// The engine will skip snapshot importation and not return an error in such case.
+    #[structopt(long, requires = "import-snapshot")]
+    pub ignore_snapshot_if_db_exists: bool,
+
+    /// Defines the directory path where meilisearch will create snapshot each snapshot_time_gap.
+    #[structopt(long, env = "MEILI_SNAPSHOT_DIR", default_value = "snapshots/")]
+    pub snapshot_dir: PathBuf,
+
+    /// Activate snapshot scheduling.
+    #[structopt(long, env = "MEILI_SCHEDULE_SNAPSHOT")]
+    pub schedule_snapshot: bool,
+
+    /// Defines time interval, in seconds, between each snapshot creation.
+    #[structopt(long, env = "MEILI_SNAPSHOT_INTERVAL_SEC", default_value = "86400")] // 24h
+    pub snapshot_interval_sec: u64,
+
+    /// Folder where dumps are created when the dump route is called.
+    #[structopt(long, env = "MEILI_DUMPS_DIR", default_value = "dumps/")]
+    pub dumps_dir: PathBuf,
+
+    /// Import a dump from the specified path, must be a `.dump` file.
+    #[structopt(long, conflicts_with = "import-snapshot")]
+    pub import_dump: Option<PathBuf>,
+
+    /// Set the log level
+    #[structopt(long, env = "MEILI_LOG_LEVEL", default_value = "info")]
+    pub log_level: String,
+
+    #[structopt(skip)]
+    pub indexer_options: IndexerOpts,
+}
+
+impl Opt {
+    pub fn get_ssl_config(&self) -> Result<Option<rustls::ServerConfig>, Box<dyn error::Error>> {
+        if let (Some(cert_path), Some(key_path)) = (&self.ssl_cert_path, &self.ssl_key_path) {
+            let client_auth = match &self.ssl_auth_path {
+                Some(auth_path) => {
+                    let roots = load_certs(auth_path.to_path_buf())?;
+                    let mut client_auth_roots = RootCertStore::empty();
+                    for root in roots {
+                        client_auth_roots.add(&root).unwrap();
+                    }
+                    if self.ssl_require_auth {
+                        AllowAnyAuthenticatedClient::new(client_auth_roots)
+                    } else {
+                        AllowAnyAnonymousOrAuthenticatedClient::new(client_auth_roots)
+                    }
+                }
+                None => NoClientAuth::new(),
+            };
+
+            let mut config = rustls::ServerConfig::new(client_auth);
+            config.key_log = Arc::new(rustls::KeyLogFile::new());
+
+            let certs = load_certs(cert_path.to_path_buf())?;
+            let privkey = load_private_key(key_path.to_path_buf())?;
+            let ocsp = load_ocsp(&self.ssl_ocsp_path)?;
+            config
+                .set_single_cert_with_ocsp_and_sct(certs, privkey, ocsp, vec![])
+                .map_err(|_| "bad certificates/private key")?;
+
+            if self.ssl_resumption {
+                config.set_persistence(rustls::ServerSessionMemoryCache::new(256));
+            }
+
+            if self.ssl_tickets {
+                config.ticketer = rustls::Ticketer::new();
+            }
+
+            Ok(Some(config))
+        } else {
+            Ok(None)
+        }
+    }
+}
+
+/// A type used to detect the max memory available and use 2/3 of it.
+#[derive(Debug, Clone, Copy)]
+pub struct MaxMemory(Option<Byte>);
+
+impl FromStr for MaxMemory {
+    type Err = ByteError;
+
+    fn from_str(s: &str) -> Result<MaxMemory, ByteError> {
+        Byte::from_str(s).map(Some).map(MaxMemory)
+    }
+}
+
+impl Default for MaxMemory {
+    fn default() -> MaxMemory {
+        MaxMemory(
+            total_memory_bytes()
+                .map(|bytes| bytes * 2 / 3)
+                .map(Byte::from_bytes),
+        )
+    }
+}
+
+impl fmt::Display for MaxMemory {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        match self.0 {
+            Some(memory) => write!(f, "{}", memory.get_appropriate_unit(true)),
+            None => f.write_str("unknown"),
+        }
+    }
+}
+
+impl Deref for MaxMemory {
+    type Target = Option<Byte>;
+
+    fn deref(&self) -> &Self::Target {
+        &self.0
+    }
+}
+
+impl MaxMemory {
+    pub fn unlimited() -> Self {
+        Self(None)
+    }
+}
+
+/// Returns the total amount of bytes available or `None` if this system isn't supported.
+fn total_memory_bytes() -> Option<u64> {
+    if System::IS_SUPPORTED {
+        let memory_kind = RefreshKind::new().with_memory();
+        let mut system = System::new_with_specifics(memory_kind);
+        system.refresh_memory();
+        Some(system.total_memory() * 1024) // KiB into bytes
+    } else {
+        None
+    }
+}
+
+fn load_certs(filename: PathBuf) -> Result<Vec<rustls::Certificate>, Box<dyn error::Error>> {
+    let certfile = fs::File::open(filename).map_err(|_| "cannot open certificate file")?;
+    let mut reader = BufReader::new(certfile);
+    Ok(certs(&mut reader).map_err(|_| "cannot read certificate file")?)
+}
+
+fn load_private_key(filename: PathBuf) -> Result<rustls::PrivateKey, Box<dyn error::Error>> {
+    let rsa_keys = {
+        let keyfile =
+            fs::File::open(filename.clone()).map_err(|_| "cannot open private key file")?;
+        let mut reader = BufReader::new(keyfile);
+        rsa_private_keys(&mut reader).map_err(|_| "file contains invalid rsa private key")?
+    };
+
+    let pkcs8_keys = {
+        let keyfile = fs::File::open(filename).map_err(|_| "cannot open private key file")?;
+        let mut reader = BufReader::new(keyfile);
+        pkcs8_private_keys(&mut reader)
+            .map_err(|_| "file contains invalid pkcs8 private key (encrypted keys not supported)")?
+    };
+
+    // prefer to load pkcs8 keys
+    if !pkcs8_keys.is_empty() {
+        Ok(pkcs8_keys[0].clone())
+    } else {
+        assert!(!rsa_keys.is_empty());
+        Ok(rsa_keys[0].clone())
+    }
+}
+
+fn load_ocsp(filename: &Option<PathBuf>) -> Result<Vec<u8>, Box<dyn error::Error>> {
+    let mut ret = Vec::new();
+
+    if let Some(ref name) = filename {
+        fs::File::open(name)
+            .map_err(|_| "cannot open ocsp file")?
+            .read_to_end(&mut ret)
+            .map_err(|_| "cannot read oscp file")?;
+    }
+
+    Ok(ret)
+}

meilisearch-http/src/routes/dump.rs

@@ -0,0 +1,40 @@
+use actix_web::{web, HttpResponse};
+use log::debug;
+use serde::{Deserialize, Serialize};
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::Data;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(web::resource("").route(web::post().to(create_dump)))
+        .service(web::resource("/{dump_uid}/status").route(web::get().to(get_dump_status)));
+}
+
+pub async fn create_dump(data: GuardedData<Private, Data>) -> Result<HttpResponse, ResponseError> {
+    let res = data.create_dump().await?;
+
+    debug!("returns: {:?}", res);
+    Ok(HttpResponse::Accepted().json(res))
+}
+
+#[derive(Debug, Serialize)]
+#[serde(rename_all = "camelCase")]
+struct DumpStatusResponse {
+    status: String,
+}
+
+#[derive(Deserialize)]
+struct DumpParam {
+    dump_uid: String,
+}
+
+async fn get_dump_status(
+    data: GuardedData<Private, Data>,
+    path: web::Path<DumpParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let res = data.dump_status(path.dump_uid.clone()).await?;
+
+    debug!("returns: {:?}", res);
+    Ok(HttpResponse::Ok().json(res))
+}

meilisearch-http/src/routes/indexes/documents.rs

@@ -0,0 +1,214 @@
+use actix_web::{web, HttpResponse};
+use log::debug;
+use milli::update::{IndexDocumentsMethod, UpdateFormat};
+use serde::Deserialize;
+use serde_json::Value;
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::extractors::payload::Payload;
+use crate::routes::IndexParam;
+use crate::Data;
+
+const DEFAULT_RETRIEVE_DOCUMENTS_OFFSET: usize = 0;
+const DEFAULT_RETRIEVE_DOCUMENTS_LIMIT: usize = 20;
+
+/*
+macro_rules! guard_content_type {
+    ($fn_name:ident, $guard_value:literal) => {
+        fn $fn_name(head: &actix_web::dev::RequestHead) -> bool {
+            if let Some(content_type) = head.headers.get("Content-Type") {
+                content_type
+                    .to_str()
+                    .map(|v| v.contains($guard_value))
+                    .unwrap_or(false)
+            } else {
+                false
+            }
+        }
+    };
+}
+
+guard_content_type!(guard_json, "application/json");
+*/
+
+fn guard_json(head: &actix_web::dev::RequestHead) -> bool {
+    if let Some(_content_type) = head.headers.get("Content-Type") {
+        // CURRENTLY AND FOR THIS RELEASE ONLY WE DECIDED TO INTERPRET ALL CONTENT-TYPES AS JSON
+        true
+        /*
+        content_type
+            .to_str()
+            .map(|v| v.contains("application/json"))
+            .unwrap_or(false)
+        */
+    } else {
+        // if no content-type is specified we still accept the data as json!
+        true
+    }
+}
+
+#[derive(Deserialize)]
+pub struct DocumentParam {
+    index_uid: String,
+    document_id: String,
+}
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(
+        web::resource("")
+            .route(web::get().to(get_all_documents))
+            .route(web::post().guard(guard_json).to(add_documents))
+            .route(web::put().guard(guard_json).to(update_documents))
+            .route(web::delete().to(clear_all_documents)),
+    )
+    // this route needs to be before the /documents/{document_id} to match properly
+    .service(web::resource("/delete-batch").route(web::post().to(delete_documents)))
+    .service(
+        web::resource("/{document_id}")
+            .route(web::get().to(get_document))
+            .route(web::delete().to(delete_document)),
+    );
+}
+
+pub async fn get_document(
+    data: GuardedData<Public, Data>,
+    path: web::Path<DocumentParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let index = path.index_uid.clone();
+    let id = path.document_id.clone();
+    let document = data
+        .retrieve_document(index, id, None as Option<Vec<String>>)
+        .await?;
+    debug!("returns: {:?}", document);
+    Ok(HttpResponse::Ok().json(document))
+}
+
+pub async fn delete_document(
+    data: GuardedData<Private, Data>,
+    path: web::Path<DocumentParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let update_status = data
+        .delete_documents(path.index_uid.clone(), vec![path.document_id.clone()])
+        .await?;
+    debug!("returns: {:?}", update_status);
+    Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+}
+
+#[derive(Deserialize, Debug)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+pub struct BrowseQuery {
+    offset: Option<usize>,
+    limit: Option<usize>,
+    attributes_to_retrieve: Option<String>,
+}
+
+pub async fn get_all_documents(
+    data: GuardedData<Public, Data>,
+    path: web::Path<IndexParam>,
+    params: web::Query<BrowseQuery>,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", params);
+    let attributes_to_retrieve = params.attributes_to_retrieve.as_ref().and_then(|attrs| {
+        let mut names = Vec::new();
+        for name in attrs.split(',').map(String::from) {
+            if name == "*" {
+                return None;
+            }
+            names.push(name);
+        }
+        Some(names)
+    });
+
+    let documents = data
+        .retrieve_documents(
+            path.index_uid.clone(),
+            params.offset.unwrap_or(DEFAULT_RETRIEVE_DOCUMENTS_OFFSET),
+            params.limit.unwrap_or(DEFAULT_RETRIEVE_DOCUMENTS_LIMIT),
+            attributes_to_retrieve,
+        )
+        .await?;
+    debug!("returns: {:?}", documents);
+    Ok(HttpResponse::Ok().json(documents))
+}
+
+#[derive(Deserialize, Debug)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+pub struct UpdateDocumentsQuery {
+    primary_key: Option<String>,
+}
+
+/// Route used when the payload type is "application/json"
+/// Used to add or replace documents
+pub async fn add_documents(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+    params: web::Query<UpdateDocumentsQuery>,
+    body: Payload,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", params);
+    let update_status = data
+        .add_documents(
+            path.into_inner().index_uid,
+            IndexDocumentsMethod::ReplaceDocuments,
+            UpdateFormat::Json,
+            body,
+            params.primary_key.clone(),
+        )
+        .await?;
+
+    debug!("returns: {:?}", update_status);
+    Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+}
+
+/// Route used when the payload type is "application/json"
+/// Used to add or replace documents
+pub async fn update_documents(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+    params: web::Query<UpdateDocumentsQuery>,
+    body: Payload,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", params);
+    let update = data
+        .add_documents(
+            path.into_inner().index_uid,
+            IndexDocumentsMethod::UpdateDocuments,
+            UpdateFormat::Json,
+            body,
+            params.primary_key.clone(),
+        )
+        .await?;
+
+    debug!("returns: {:?}", update);
+    Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update.id() })))
+}
+
+pub async fn delete_documents(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+    body: web::Json<Vec<Value>>,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", body);
+    let ids = body
+        .iter()
+        .map(|v| {
+            v.as_str()
+                .map(String::from)
+                .unwrap_or_else(|| v.to_string())
+        })
+        .collect();
+
+    let update_status = data.delete_documents(path.index_uid.clone(), ids).await?;
+    debug!("returns: {:?}", update_status);
+    Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+}
+
+pub async fn clear_all_documents(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let update_status = data.clear_documents(path.index_uid.clone()).await?;
+    debug!("returns: {:?}", update_status);
+    Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+}

meilisearch-http/src/routes/indexes/mod.rs

@@ -0,0 +1,116 @@
+use actix_web::{web, HttpResponse};
+use chrono::{DateTime, Utc};
+use log::debug;
+use serde::{Deserialize, Serialize};
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::routes::IndexParam;
+use crate::Data;
+
+pub mod documents;
+pub mod search;
+pub mod settings;
+pub mod updates;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(
+        web::resource("")
+            .route(web::get().to(list_indexes))
+            .route(web::post().to(create_index)),
+    )
+    .service(
+        web::scope("/{index_uid}")
+            .service(
+                web::resource("")
+                    .route(web::get().to(get_index))
+                    .route(web::put().to(update_index))
+                    .route(web::delete().to(delete_index)),
+            )
+            .service(web::resource("/stats").route(web::get().to(get_index_stats)))
+            .service(web::scope("/documents").configure(documents::configure))
+            .service(web::scope("/search").configure(search::configure))
+            .service(web::scope("/updates").configure(updates::configure))
+            .service(web::scope("/settings").configure(settings::configure)),
+    );
+}
+
+pub async fn list_indexes(data: GuardedData<Private, Data>) -> Result<HttpResponse, ResponseError> {
+    let indexes = data.list_indexes().await?;
+    debug!("returns: {:?}", indexes);
+    Ok(HttpResponse::Ok().json(indexes))
+}
+
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+pub struct IndexCreateRequest {
+    uid: String,
+    primary_key: Option<String>,
+}
+
+pub async fn create_index(
+    data: GuardedData<Private, Data>,
+    body: web::Json<IndexCreateRequest>,
+) -> Result<HttpResponse, ResponseError> {
+    let body = body.into_inner();
+    let meta = data.create_index(body.uid, body.primary_key).await?;
+    Ok(HttpResponse::Created().json(meta))
+}
+
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+pub struct UpdateIndexRequest {
+    uid: Option<String>,
+    primary_key: Option<String>,
+}
+
+#[derive(Debug, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct UpdateIndexResponse {
+    name: String,
+    uid: String,
+    created_at: DateTime<Utc>,
+    updated_at: DateTime<Utc>,
+    primary_key: Option<String>,
+}
+
+pub async fn get_index(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let meta = data.index(path.index_uid.clone()).await?;
+    debug!("returns: {:?}", meta);
+    Ok(HttpResponse::Ok().json(meta))
+}
+
+pub async fn update_index(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+    body: web::Json<UpdateIndexRequest>,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", body);
+    let body = body.into_inner();
+    let meta = data
+        .update_index(path.into_inner().index_uid, body.primary_key, body.uid)
+        .await?;
+    debug!("returns: {:?}", meta);
+    Ok(HttpResponse::Ok().json(meta))
+}
+
+pub async fn delete_index(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+) -> Result<HttpResponse, ResponseError> {
+    data.delete_index(path.index_uid.clone()).await?;
+    Ok(HttpResponse::NoContent().finish())
+}
+
+pub async fn get_index_stats(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let response = data.get_index_stats(path.index_uid.clone()).await?;
+
+    debug!("returns: {:?}", response);
+    Ok(HttpResponse::Ok().json(response))
+}

meilisearch-http/src/routes/indexes/search.rs

@@ -0,0 +1,117 @@
+use actix_web::{web, HttpResponse};
+use log::debug;
+use serde::Deserialize;
+use serde_json::Value;
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::index::{default_crop_length, SearchQuery, DEFAULT_SEARCH_LIMIT};
+use crate::routes::IndexParam;
+use crate::Data;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(
+        web::resource("")
+            .route(web::get().to(search_with_url_query))
+            .route(web::post().to(search_with_post)),
+    );
+}
+
+#[derive(Deserialize, Debug)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+pub struct SearchQueryGet {
+    q: Option<String>,
+    offset: Option<usize>,
+    limit: Option<usize>,
+    attributes_to_retrieve: Option<String>,
+    attributes_to_crop: Option<String>,
+    #[serde(default = "default_crop_length")]
+    crop_length: usize,
+    attributes_to_highlight: Option<String>,
+    filter: Option<String>,
+    sort: Option<String>,
+    #[serde(default = "Default::default")]
+    matches: bool,
+    facets_distribution: Option<String>,
+}
+
+impl From<SearchQueryGet> for SearchQuery {
+    fn from(other: SearchQueryGet) -> Self {
+        let attributes_to_retrieve = other
+            .attributes_to_retrieve
+            .map(|attrs| attrs.split(',').map(String::from).collect());
+
+        let attributes_to_crop = other
+            .attributes_to_crop
+            .map(|attrs| attrs.split(',').map(String::from).collect());
+
+        let attributes_to_highlight = other
+            .attributes_to_highlight
+            .map(|attrs| attrs.split(',').map(String::from).collect());
+
+        let facets_distribution = other
+            .facets_distribution
+            .map(|attrs| attrs.split(',').map(String::from).collect());
+
+        let filter = match other.filter {
+            Some(f) => match serde_json::from_str(&f) {
+                Ok(v) => Some(v),
+                _ => Some(Value::String(f)),
+            },
+            None => None,
+        };
+
+        let sort = other
+            .sort
+            .map(|attrs| attrs.split(',').map(String::from).collect());
+
+        Self {
+            q: other.q,
+            offset: other.offset,
+            limit: other.limit.unwrap_or(DEFAULT_SEARCH_LIMIT),
+            attributes_to_retrieve,
+            attributes_to_crop,
+            crop_length: other.crop_length,
+            attributes_to_highlight,
+            filter,
+            sort,
+            matches: other.matches,
+            facets_distribution,
+        }
+    }
+}
+
+pub async fn search_with_url_query(
+    data: GuardedData<Public, Data>,
+    path: web::Path<IndexParam>,
+    params: web::Query<SearchQueryGet>,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("called with params: {:?}", params);
+    let query = params.into_inner().into();
+    let search_result = data.search(path.into_inner().index_uid, query).await?;
+
+    // Tests that the nb_hits is always set to false
+    #[cfg(test)]
+    assert!(!search_result.exhaustive_nb_hits);
+
+    debug!("returns: {:?}", search_result);
+    Ok(HttpResponse::Ok().json(search_result))
+}
+
+pub async fn search_with_post(
+    data: GuardedData<Public, Data>,
+    path: web::Path<IndexParam>,
+    params: web::Json<SearchQuery>,
+) -> Result<HttpResponse, ResponseError> {
+    debug!("search called with params: {:?}", params);
+    let search_result = data
+        .search(path.into_inner().index_uid, params.into_inner())
+        .await?;
+
+    // Tests that the nb_hits is always set to false
+    #[cfg(test)]
+    assert!(!search_result.exhaustive_nb_hits);
+
+    debug!("returns: {:?}", search_result);
+    Ok(HttpResponse::Ok().json(search_result))
+}

meilisearch-http/src/routes/indexes/settings.rs

@@ -0,0 +1,185 @@
+use actix_web::{web, HttpResponse};
+use log::debug;
+
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::index::Settings;
+use crate::Data;
+use crate::{error::ResponseError, index::Unchecked};
+
+#[macro_export]
+macro_rules! make_setting_route {
+    ($route:literal, $type:ty, $attr:ident, $camelcase_attr:literal) => {
+        pub mod $attr {
+            use log::debug;
+            use actix_web::{web, HttpResponse, Resource};
+
+            use milli::update::Setting;
+
+            use crate::data;
+            use crate::error::ResponseError;
+            use crate::index::Settings;
+            use crate::extractors::authentication::{GuardedData, policies::*};
+
+            pub async fn delete(
+                data: GuardedData<Private, data::Data>,
+                index_uid: web::Path<String>,
+            ) -> Result<HttpResponse, ResponseError> {
+                use crate::index::Settings;
+                let settings = Settings {
+                    $attr: Setting::Reset,
+                    ..Default::default()
+                };
+                let update_status = data.update_settings(index_uid.into_inner(), settings, false).await?;
+                debug!("returns: {:?}", update_status);
+                Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+            }
+
+            pub async fn update(
+                data: GuardedData<Private, data::Data>,
+                index_uid: actix_web::web::Path<String>,
+                body: actix_web::web::Json<Option<$type>>,
+            ) -> std::result::Result<HttpResponse, ResponseError> {
+                let settings = Settings {
+                    $attr: match body.into_inner() {
+                        Some(inner_body) => Setting::Set(inner_body),
+                        None => Setting::Reset
+                    },
+                    ..Default::default()
+                };
+
+                let update_status = data.update_settings(index_uid.into_inner(), settings, true).await?;
+                debug!("returns: {:?}", update_status);
+                Ok(HttpResponse::Accepted().json(serde_json::json!({ "updateId": update_status.id() })))
+            }
+
+            pub async fn get(
+                data: GuardedData<Private, data::Data>,
+                index_uid: actix_web::web::Path<String>,
+            ) -> std::result::Result<HttpResponse, ResponseError> {
+                let settings = data.settings(index_uid.into_inner()).await?;
+                debug!("returns: {:?}", settings);
+                let mut json = serde_json::json!(&settings);
+                let val = json[$camelcase_attr].take();
+                Ok(HttpResponse::Ok().json(val))
+            }
+
+            pub fn resources() -> Resource {
+                Resource::new($route)
+                    .route(web::get().to(get))
+                    .route(web::post().to(update))
+                    .route(web::delete().to(delete))
+            }
+        }
+    };
+}
+
+make_setting_route!(
+    "/filterable-attributes",
+    std::collections::BTreeSet<String>,
+    filterable_attributes,
+    "filterableAttributes"
+);
+
+make_setting_route!(
+    "/sortable-attributes",
+    std::collections::BTreeSet<String>,
+    sortable_attributes,
+    "sortableAttributes"
+);
+
+make_setting_route!(
+    "/displayed-attributes",
+    Vec<String>,
+    displayed_attributes,
+    "displayedAttributes"
+);
+
+make_setting_route!(
+    "/searchable-attributes",
+    Vec<String>,
+    searchable_attributes,
+    "searchableAttributes"
+);
+
+make_setting_route!(
+    "/stop-words",
+    std::collections::BTreeSet<String>,
+    stop_words,
+    "stopWords"
+);
+
+make_setting_route!(
+    "/synonyms",
+    std::collections::BTreeMap<String, Vec<String>>,
+    synonyms,
+    "synonyms"
+);
+
+make_setting_route!(
+    "/distinct-attribute",
+    String,
+    distinct_attribute,
+    "distinctAttribute"
+);
+
+make_setting_route!("/ranking-rules", Vec<String>, ranking_rules, "rankingRules");
+
+macro_rules! generate_configure {
+    ($($mod:ident),*) => {
+        pub fn configure(cfg: &mut web::ServiceConfig) {
+            cfg.service(
+                web::resource("")
+                .route(web::post().to(update_all))
+                .route(web::get().to(get_all))
+                .route(web::delete().to(delete_all)))
+                $(.service($mod::resources()))*;
+        }
+    };
+}
+
+generate_configure!(
+    filterable_attributes,
+    sortable_attributes,
+    displayed_attributes,
+    searchable_attributes,
+    distinct_attribute,
+    stop_words,
+    synonyms,
+    ranking_rules
+);
+
+pub async fn update_all(
+    data: GuardedData<Private, Data>,
+    index_uid: web::Path<String>,
+    body: web::Json<Settings<Unchecked>>,
+) -> Result<HttpResponse, ResponseError> {
+    let settings = body.into_inner().check();
+    let update_result = data
+        .update_settings(index_uid.into_inner(), settings, true)
+        .await?;
+    let json = serde_json::json!({ "updateId": update_result.id() });
+    debug!("returns: {:?}", json);
+    Ok(HttpResponse::Accepted().json(json))
+}
+
+pub async fn get_all(
+    data: GuardedData<Private, Data>,
+    index_uid: web::Path<String>,
+) -> Result<HttpResponse, ResponseError> {
+    let settings = data.settings(index_uid.into_inner()).await?;
+    debug!("returns: {:?}", settings);
+    Ok(HttpResponse::Ok().json(settings))
+}
+
+pub async fn delete_all(
+    data: GuardedData<Private, Data>,
+    index_uid: web::Path<String>,
+) -> Result<HttpResponse, ResponseError> {
+    let settings = Settings::cleared();
+    let update_result = data
+        .update_settings(index_uid.into_inner(), settings, false)
+        .await?;
+    let json = serde_json::json!({ "updateId": update_result.id() });
+    debug!("returns: {:?}", json);
+    Ok(HttpResponse::Accepted().json(json))
+}

meilisearch-http/src/routes/indexes/updates.rs

@@ -0,0 +1,64 @@
+use actix_web::{web, HttpResponse};
+use chrono::{DateTime, Utc};
+use log::debug;
+use serde::{Deserialize, Serialize};
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::routes::{IndexParam, UpdateStatusResponse};
+use crate::Data;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(web::resource("").route(web::get().to(get_all_updates_status)))
+        .service(web::resource("{update_id}").route(web::get().to(get_update_status)));
+}
+
+#[derive(Debug, Deserialize)]
+#[serde(rename_all = "camelCase", deny_unknown_fields)]
+struct UpdateIndexRequest {
+    uid: Option<String>,
+    primary_key: Option<String>,
+}
+
+#[derive(Debug, Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct UpdateIndexResponse {
+    name: String,
+    uid: String,
+    created_at: DateTime<Utc>,
+    updated_at: DateTime<Utc>,
+    primary_key: Option<String>,
+}
+
+#[derive(Deserialize)]
+pub struct UpdateParam {
+    index_uid: String,
+    update_id: u64,
+}
+
+pub async fn get_update_status(
+    data: GuardedData<Private, Data>,
+    path: web::Path<UpdateParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let params = path.into_inner();
+    let meta = data
+        .get_update_status(params.index_uid, params.update_id)
+        .await?;
+    let meta = UpdateStatusResponse::from(meta);
+    debug!("returns: {:?}", meta);
+    Ok(HttpResponse::Ok().json(meta))
+}
+
+pub async fn get_all_updates_status(
+    data: GuardedData<Private, Data>,
+    path: web::Path<IndexParam>,
+) -> Result<HttpResponse, ResponseError> {
+    let metas = data.get_updates_status(path.into_inner().index_uid).await?;
+    let metas = metas
+        .into_iter()
+        .map(UpdateStatusResponse::from)
+        .collect::<Vec<_>>();
+
+    debug!("returns: {:?}", metas);
+    Ok(HttpResponse::Ok().json(metas))
+}

meilisearch-http/src/routes/mod.rs

@@ -0,0 +1,368 @@
+use std::time::Duration;
+
+use actix_web::{web, HttpResponse};
+use chrono::{DateTime, Utc};
+use log::debug;
+use serde::{Deserialize, Serialize};
+
+use crate::error::ResponseError;
+use crate::extractors::authentication::{policies::*, GuardedData};
+use crate::index::{Settings, Unchecked};
+use crate::index_controller::{UpdateMeta, UpdateResult, UpdateStatus};
+use crate::Data;
+
+mod dump;
+mod indexes;
+
+pub fn configure(cfg: &mut web::ServiceConfig) {
+    cfg.service(web::resource("/health").route(web::get().to(get_health)))
+        .service(web::scope("/dumps").configure(dump::configure))
+        .service(web::resource("/keys").route(web::get().to(list_keys)))
+        .service(web::resource("/stats").route(web::get().to(get_stats)))
+        .service(web::resource("/version").route(web::get().to(get_version)))
+        .service(web::scope("/indexes").configure(indexes::configure));
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[allow(clippy::large_enum_variant)]
+#[serde(tag = "name")]
+pub enum UpdateType {
+    ClearAll,
+    Customs,
+    DocumentsAddition {
+        #[serde(skip_serializing_if = "Option::is_none")]
+        number: Option<usize>,
+    },
+    DocumentsPartial {
+        #[serde(skip_serializing_if = "Option::is_none")]
+        number: Option<usize>,
+    },
+    DocumentsDeletion {
+        #[serde(skip_serializing_if = "Option::is_none")]
+        number: Option<usize>,
+    },
+    Settings {
+        settings: Settings<Unchecked>,
+    },
+}
+
+impl From<&UpdateStatus> for UpdateType {
+    fn from(other: &UpdateStatus) -> Self {
+        use milli::update::IndexDocumentsMethod::*;
+        match other.meta() {
+            UpdateMeta::DocumentsAddition { method, .. } => {
+                let number = match other {
+                    UpdateStatus::Processed(processed) => match processed.success {
+                        UpdateResult::DocumentsAddition(ref addition) => {
+                            Some(addition.nb_documents)
+                        }
+                        _ => None,
+                    },
+                    _ => None,
+                };
+
+                match method {
+                    ReplaceDocuments => UpdateType::DocumentsAddition { number },
+                    UpdateDocuments => UpdateType::DocumentsPartial { number },
+                    _ => unreachable!(),
+                }
+            }
+            UpdateMeta::ClearDocuments => UpdateType::ClearAll,
+            UpdateMeta::DeleteDocuments { ids } => UpdateType::DocumentsDeletion {
+                number: Some(ids.len()),
+            },
+            UpdateMeta::Settings(settings) => UpdateType::Settings {
+                settings: settings.clone(),
+            },
+        }
+    }
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct ProcessedUpdateResult {
+    pub update_id: u64,
+    #[serde(rename = "type")]
+    pub update_type: UpdateType,
+    pub duration: f64, // in seconds
+    pub enqueued_at: DateTime<Utc>,
+    pub processed_at: DateTime<Utc>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct FailedUpdateResult {
+    pub update_id: u64,
+    #[serde(rename = "type")]
+    pub update_type: UpdateType,
+    #[serde(flatten)]
+    pub response: ResponseError,
+    pub duration: f64, // in seconds
+    pub enqueued_at: DateTime<Utc>,
+    pub processed_at: DateTime<Utc>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase")]
+pub struct EnqueuedUpdateResult {
+    pub update_id: u64,
+    #[serde(rename = "type")]
+    pub update_type: UpdateType,
+    pub enqueued_at: DateTime<Utc>,
+    #[serde(skip_serializing_if = "Option::is_none")]
+    pub started_processing_at: Option<DateTime<Utc>>,
+}
+
+#[derive(Debug, Clone, Serialize, Deserialize)]
+#[serde(rename_all = "camelCase", tag = "status")]
+pub enum UpdateStatusResponse {
+    Enqueued {
+        #[serde(flatten)]
+        content: EnqueuedUpdateResult,
+    },
+    Processing {
+        #[serde(flatten)]
+        content: EnqueuedUpdateResult,
+    },
+    Failed {
+        #[serde(flatten)]
+        content: FailedUpdateResult,
+    },
+    Processed {
+        #[serde(flatten)]
+        content: ProcessedUpdateResult,
+    },
+}
+
+impl From<UpdateStatus> for UpdateStatusResponse {
+    fn from(other: UpdateStatus) -> Self {
+        let update_type = UpdateType::from(&other);
+
+        match other {
+            UpdateStatus::Processing(processing) => {
+                let content = EnqueuedUpdateResult {
+                    update_id: processing.id(),
+                    update_type,
+                    enqueued_at: processing.from.enqueued_at,
+                    started_processing_at: Some(processing.started_processing_at),
+                };
+                UpdateStatusResponse::Processing { content }
+            }
+            UpdateStatus::Enqueued(enqueued) => {
+                let content = EnqueuedUpdateResult {
+                    update_id: enqueued.id(),
+                    update_type,
+                    enqueued_at: enqueued.enqueued_at,
+                    started_processing_at: None,
+                };
+                UpdateStatusResponse::Enqueued { content }
+            }
+            UpdateStatus::Processed(processed) => {
+                let duration = processed
+                    .processed_at
+                    .signed_duration_since(processed.from.started_processing_at)
+                    .num_milliseconds();
+
+                // necessary since chrono::duration don't expose a f64 secs method.
+                let duration = Duration::from_millis(duration as u64).as_secs_f64();
+
+                let content = ProcessedUpdateResult {
+                    update_id: processed.id(),
+                    update_type,
+                    duration,
+                    enqueued_at: processed.from.from.enqueued_at,
+                    processed_at: processed.processed_at,
+                };
+                UpdateStatusResponse::Processed { content }
+            }
+            UpdateStatus::Aborted(_) => unreachable!(),
+            UpdateStatus::Failed(failed) => {
+                let duration = failed
+                    .failed_at
+                    .signed_duration_since(failed.from.started_processing_at)
+                    .num_milliseconds();
+
+                // necessary since chrono::duration don't expose a f64 secs method.
+                let duration = Duration::from_millis(duration as u64).as_secs_f64();
+
+                let update_id = failed.id();
+                let response = failed.error;
+
+                let content = FailedUpdateResult {
+                    update_id,
+                    update_type,
+                    response,
+                    duration,
+                    enqueued_at: failed.from.from.enqueued_at,
+                    processed_at: failed.failed_at,
+                };
+                UpdateStatusResponse::Failed { content }
+            }
+        }
+    }
+}
+
+#[derive(Deserialize)]
+pub struct IndexParam {
+    index_uid: String,
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+pub struct IndexUpdateResponse {
+    pub update_id: u64,
+}
+
+impl IndexUpdateResponse {
+    pub fn with_id(update_id: u64) -> Self {
+        Self { update_id }
+    }
+}
+
+/// Always return a 200 with:
+/// ```json
+/// {
+///     "status": "Meilisearch is running"
+/// }
+/// ```
+pub async fn running() -> HttpResponse {
+    HttpResponse::Ok().json(serde_json::json!({ "status": "MeiliSearch is running" }))
+}
+
+async fn get_stats(data: GuardedData<Private, Data>) -> Result<HttpResponse, ResponseError> {
+    let response = data.get_all_stats().await?;
+
+    debug!("returns: {:?}", response);
+    Ok(HttpResponse::Ok().json(response))
+}
+
+#[derive(Serialize)]
+#[serde(rename_all = "camelCase")]
+struct VersionResponse {
+    commit_sha: String,
+    commit_date: String,
+    pkg_version: String,
+}
+
+async fn get_version(_data: GuardedData<Private, Data>) -> HttpResponse {
+    let commit_sha = option_env!("VERGEN_GIT_SHA").unwrap_or("unknown");
+    let commit_date = option_env!("VERGEN_GIT_COMMIT_TIMESTAMP").unwrap_or("unknown");
+
+    HttpResponse::Ok().json(VersionResponse {
+        commit_sha: commit_sha.to_string(),
+        commit_date: commit_date.to_string(),
+        pkg_version: env!("CARGO_PKG_VERSION").to_string(),
+    })
+}
+
+#[derive(Serialize)]
+struct KeysResponse {
+    private: Option<String>,
+    public: Option<String>,
+}
+
+pub async fn list_keys(data: GuardedData<Admin, Data>) -> HttpResponse {
+    let api_keys = data.api_keys.clone();
+    HttpResponse::Ok().json(&KeysResponse {
+        private: api_keys.private,
+        public: api_keys.public,
+    })
+}
+
+pub async fn get_health() -> Result<HttpResponse, ResponseError> {
+    Ok(HttpResponse::Ok().json(serde_json::json!({ "status": "available" })))
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+    use crate::data::Data;
+    use crate::extractors::authentication::GuardedData;
+
+    /// A type implemented for a route that uses a authentication policy `Policy`.
+    ///
+    /// This trait is used for regression testing of route authenticaton policies.
+    trait Is<Policy, T> {}
+
+    macro_rules! impl_is_policy {
+        ($($param:ident)*) => {
+            impl<Policy, Func, $($param,)* Res> Is<Policy, (($($param,)*), Res)> for Func
+                where Func: Fn(GuardedData<Policy, Data>, $($param,)*) -> Res {}
+
+        };
+    }
+
+    impl_is_policy! {}
+    impl_is_policy! {A}
+    impl_is_policy! {A B}
+    impl_is_policy! {A B C}
+    impl_is_policy! {A B C D}
+
+    /// Emits a compile error if a route doesn't have the correct authentication policy.
+    ///
+    /// This works by trying to cast the route function into a Is<Policy, _> type, where Policy it
+    /// the authentication policy defined for the route.
+    macro_rules! test_auth_routes {
+        ($($policy:ident => { $($route:expr,)*})*) => {
+            #[test]
+            fn test_auth() {
+                $($(let _: &dyn Is<$policy, _> = &$route;)*)*
+            }
+        };
+    }
+
+    test_auth_routes! {
+        Public => {
+            indexes::search::search_with_url_query,
+            indexes::search::search_with_post,
+
+            indexes::documents::get_document,
+            indexes::documents::get_all_documents,
+        }
+        Private => {
+            get_stats,
+            get_version,
+
+            indexes::create_index,
+            indexes::list_indexes,
+            indexes::get_index_stats,
+            indexes::delete_index,
+            indexes::update_index,
+            indexes::get_index,
+
+            dump::create_dump,
+
+            indexes::settings::filterable_attributes::get,
+            indexes::settings::displayed_attributes::get,
+            indexes::settings::searchable_attributes::get,
+            indexes::settings::stop_words::get,
+            indexes::settings::synonyms::get,
+            indexes::settings::distinct_attribute::get,
+            indexes::settings::filterable_attributes::update,
+            indexes::settings::displayed_attributes::update,
+            indexes::settings::searchable_attributes::update,
+            indexes::settings::stop_words::update,
+            indexes::settings::synonyms::update,
+            indexes::settings::distinct_attribute::update,
+            indexes::settings::filterable_attributes::delete,
+            indexes::settings::displayed_attributes::delete,
+            indexes::settings::searchable_attributes::delete,
+            indexes::settings::stop_words::delete,
+            indexes::settings::synonyms::delete,
+            indexes::settings::distinct_attribute::delete,
+            indexes::settings::delete_all,
+            indexes::settings::get_all,
+            indexes::settings::update_all,
+
+            indexes::documents::clear_all_documents,
+            indexes::documents::delete_documents,
+            indexes::documents::update_documents,
+            indexes::documents::add_documents,
+            indexes::documents::delete_document,
+
+            indexes::updates::get_all_updates_status,
+            indexes::updates::get_update_status,
+        }
+        Admin => { list_keys, }
+    }
+}