mirror of
https://github.com/meilisearch/meilisearch.git
synced 2025-11-22 04:36:32 +00:00
Compare commits
63 Commits
v1.21.0
...
chat-route
| Author | SHA1 | Date | |
|---|---|---|---|
|
|
d63cf4a5a8 | ||
|
|
5ccb24377a | ||
|
|
d361cbf04e | ||
|
|
bd5268b4b2 | ||
|
|
a7ee848df3 | ||
|
|
a3600afa20 | ||
|
|
a4967dfaf7 | ||
|
|
ac33f10f81 | ||
|
|
7fd34b5f19 | ||
|
|
ef0f877bb9 | ||
|
|
85c20447e0 | ||
|
|
0f75ae9f25 | ||
|
|
11ace7f209 | ||
|
|
dd7155abb3 | ||
|
|
6959b01aa5 | ||
|
|
358abf662e | ||
|
|
5bb0714c5f | ||
|
|
7e42293a23 | ||
|
|
8c66e63760 | ||
|
|
5d8cdb075b | ||
|
|
fa139ee601 | ||
|
|
f635a8f9f3 | ||
|
|
807157b8cd | ||
|
|
ca5a87a606 | ||
|
|
bf3286ba41 | ||
|
|
1e67a47a8f | ||
|
|
7bfb7b7d58 | ||
|
|
1eb8249a51 | ||
|
|
f9ecb0ff31 | ||
|
|
723f1b13a0 | ||
|
|
97d74bd2b9 | ||
|
|
39ecea5e9e | ||
|
|
d389e49147 | ||
|
|
971249c59c | ||
|
|
d6f2bd9b57 | ||
|
|
14c8e5cb56 | ||
|
|
1159af1219 | ||
|
|
0fcc7e1377 | ||
|
|
a074384d69 | ||
|
|
bc7e4598db | ||
|
|
487002f2f0 | ||
|
|
a51579e923 | ||
|
|
afadfa420a | ||
|
|
116fbb4c24 | ||
|
|
64fe283abc | ||
|
|
9af1a29949 | ||
|
|
d5d30ed4f2 | ||
|
|
81ea8933c1 | ||
|
|
9875f55384 | ||
|
|
eb4445070f | ||
|
|
e603e221d5 | ||
|
|
d94f16b1d2 | ||
|
|
59873fdc6c | ||
|
|
0f2ddf560b | ||
|
|
fb08d62af9 | ||
|
|
f8f2f8df97 | ||
|
|
5b642b558a | ||
|
|
ae89825b37 | ||
|
|
fbd3558f04 | ||
|
|
223f5b7283 | ||
|
|
3bca6ff771 | ||
|
|
5d426bd95e | ||
|
|
f69e39171f |
2467
Cargo.lock
generated
2467
Cargo.lock
generated
File diff suppressed because it is too large
Load Diff
@@ -31,7 +31,7 @@ anyhow = "1.0.95"
|
||||
bytes = "1.9.0"
|
||||
convert_case = "0.6.0"
|
||||
flate2 = "1.0.35"
|
||||
reqwest = { version = "0.12.12", features = ["blocking", "rustls-tls"], default-features = false }
|
||||
reqwest = { version = "0.12.15", features = ["blocking", "rustls-tls"], default-features = false }
|
||||
|
||||
[features]
|
||||
default = ["milli/all-tokenizations"]
|
||||
|
||||
@@ -2,7 +2,8 @@ mod datasets_paths;
|
||||
mod utils;
|
||||
|
||||
use criterion::{criterion_group, criterion_main};
|
||||
use milli::{update::Settings, FilterableAttributesRule};
|
||||
use milli::update::Settings;
|
||||
use milli::FilterableAttributesRule;
|
||||
use utils::Conf;
|
||||
|
||||
#[cfg(not(windows))]
|
||||
|
||||
@@ -2,7 +2,8 @@ mod datasets_paths;
|
||||
mod utils;
|
||||
|
||||
use criterion::{criterion_group, criterion_main};
|
||||
use milli::{update::Settings, FilterableAttributesRule};
|
||||
use milli::update::Settings;
|
||||
use milli::FilterableAttributesRule;
|
||||
use utils::Conf;
|
||||
|
||||
#[cfg(not(windows))]
|
||||
|
||||
@@ -305,6 +305,7 @@ pub(crate) mod test {
|
||||
localized_attributes: Setting::NotSet,
|
||||
facet_search: Setting::NotSet,
|
||||
prefix_search: Setting::NotSet,
|
||||
chat: Setting::NotSet,
|
||||
_kind: std::marker::PhantomData,
|
||||
};
|
||||
settings.check()
|
||||
@@ -418,6 +419,7 @@ pub(crate) mod test {
|
||||
uid: Uuid::from_str("9f8a34da-b6b2-42f0-939b-dbd4c3448655").unwrap(),
|
||||
actions: vec![Action::DocumentsAll],
|
||||
indexes: vec![IndexUidPattern::from_str("doggos").unwrap()],
|
||||
chats: vec![IndexUidPattern::from_str("OpenAI").unwrap()],
|
||||
expires_at: Some(datetime!(4130-03-14 12:21 UTC)),
|
||||
created_at: datetime!(1960-11-15 0:00 UTC),
|
||||
updated_at: datetime!(2022-11-10 0:00 UTC),
|
||||
@@ -428,6 +430,7 @@ pub(crate) mod test {
|
||||
uid: Uuid::from_str("4622f717-1c00-47bb-a494-39d76a49b591").unwrap(),
|
||||
actions: vec![Action::All],
|
||||
indexes: vec![IndexUidPattern::all()],
|
||||
chats: vec![IndexUidPattern::all()],
|
||||
expires_at: None,
|
||||
created_at: datetime!(0000-01-01 00:01 UTC),
|
||||
updated_at: datetime!(1964-05-04 17:25 UTC),
|
||||
@@ -438,6 +441,7 @@ pub(crate) mod test {
|
||||
uid: Uuid::from_str("fb80b58b-0a34-412f-8ba7-1ce868f8ac5c").unwrap(),
|
||||
actions: vec![],
|
||||
indexes: vec![],
|
||||
chats: vec![],
|
||||
expires_at: None,
|
||||
created_at: datetime!(400-02-29 0:00 UTC),
|
||||
updated_at: datetime!(1024-02-29 0:00 UTC),
|
||||
|
||||
@@ -187,6 +187,7 @@ impl CompatV5ToV6 {
|
||||
v5::StarOr::Other(uid) => v6::IndexUidPattern::new_unchecked(uid.as_str()),
|
||||
})
|
||||
.collect(),
|
||||
chats: vec![],
|
||||
expires_at: key.expires_at,
|
||||
created_at: key.created_at,
|
||||
updated_at: key.updated_at,
|
||||
@@ -405,6 +406,7 @@ impl<T> From<v5::Settings<T>> for v6::Settings<v6::Unchecked> {
|
||||
search_cutoff_ms: v6::Setting::NotSet,
|
||||
facet_search: v6::Setting::NotSet,
|
||||
prefix_search: v6::Setting::NotSet,
|
||||
chat: v6::Setting::NotSet,
|
||||
_kind: std::marker::PhantomData,
|
||||
}
|
||||
}
|
||||
|
||||
@@ -131,6 +131,19 @@ impl RoFeatures {
|
||||
.into())
|
||||
}
|
||||
}
|
||||
|
||||
pub fn check_chat_completions(&self, disabled_action: &'static str) -> Result<()> {
|
||||
if self.runtime.chat_completions {
|
||||
Ok(())
|
||||
} else {
|
||||
Err(FeatureNotEnabledError {
|
||||
disabled_action,
|
||||
feature: "chat completions",
|
||||
issue_link: "https://github.com/orgs/meilisearch/discussions/835",
|
||||
}
|
||||
.into())
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl FeatureData {
|
||||
|
||||
@@ -34,6 +34,7 @@ pub fn snapshot_index_scheduler(scheduler: &IndexScheduler) -> String {
|
||||
planned_failures: _,
|
||||
run_loop_iteration: _,
|
||||
embedders: _,
|
||||
chat_settings: _,
|
||||
} = scheduler;
|
||||
|
||||
let rtxn = env.read_txn().unwrap();
|
||||
|
||||
@@ -51,10 +51,12 @@ pub use features::RoFeatures;
|
||||
use flate2::bufread::GzEncoder;
|
||||
use flate2::Compression;
|
||||
use meilisearch_types::batches::Batch;
|
||||
use meilisearch_types::features::{InstanceTogglableFeatures, Network, RuntimeTogglableFeatures};
|
||||
use meilisearch_types::features::{
|
||||
ChatCompletionSettings, InstanceTogglableFeatures, Network, RuntimeTogglableFeatures,
|
||||
};
|
||||
use meilisearch_types::heed::byteorder::BE;
|
||||
use meilisearch_types::heed::types::I128;
|
||||
use meilisearch_types::heed::{self, Env, RoTxn, WithoutTls};
|
||||
use meilisearch_types::heed::types::{SerdeJson, Str, I128};
|
||||
use meilisearch_types::heed::{self, Database, Env, RoTxn, RwTxn, WithoutTls};
|
||||
use meilisearch_types::milli::index::IndexEmbeddingConfig;
|
||||
use meilisearch_types::milli::update::IndexerConfig;
|
||||
use meilisearch_types::milli::vector::{Embedder, EmbedderOptions, EmbeddingConfigs};
|
||||
@@ -153,6 +155,9 @@ pub struct IndexScheduler {
|
||||
/// In charge of fetching and setting the status of experimental features.
|
||||
features: features::FeatureData,
|
||||
|
||||
/// Stores the custom chat prompts and other settings of the indexes.
|
||||
pub(crate) chat_settings: Database<Str, SerdeJson<ChatCompletionSettings>>,
|
||||
|
||||
/// Everything related to the processing of the tasks
|
||||
pub scheduler: scheduler::Scheduler,
|
||||
|
||||
@@ -211,11 +216,16 @@ impl IndexScheduler {
|
||||
#[cfg(test)]
|
||||
run_loop_iteration: self.run_loop_iteration.clone(),
|
||||
features: self.features.clone(),
|
||||
chat_settings: self.chat_settings,
|
||||
}
|
||||
}
|
||||
|
||||
pub(crate) const fn nb_db() -> u32 {
|
||||
Versioning::nb_db() + Queue::nb_db() + IndexMapper::nb_db() + features::FeatureData::nb_db()
|
||||
Versioning::nb_db()
|
||||
+ Queue::nb_db()
|
||||
+ IndexMapper::nb_db()
|
||||
+ features::FeatureData::nb_db()
|
||||
+ 1 // chat-prompts
|
||||
}
|
||||
|
||||
/// Create an index scheduler and start its run loop.
|
||||
@@ -269,6 +279,7 @@ impl IndexScheduler {
|
||||
let features = features::FeatureData::new(&env, &mut wtxn, options.instance_features)?;
|
||||
let queue = Queue::new(&env, &mut wtxn, &options)?;
|
||||
let index_mapper = IndexMapper::new(&env, &mut wtxn, &options, budget)?;
|
||||
let chat_settings = env.create_database(&mut wtxn, Some("chat-settings"))?;
|
||||
wtxn.commit()?;
|
||||
|
||||
// allow unreachable_code to get rids of the warning in the case of a test build.
|
||||
@@ -292,12 +303,21 @@ impl IndexScheduler {
|
||||
#[cfg(test)]
|
||||
run_loop_iteration: Arc::new(RwLock::new(0)),
|
||||
features,
|
||||
chat_settings,
|
||||
};
|
||||
|
||||
this.run();
|
||||
Ok(this)
|
||||
}
|
||||
|
||||
pub fn write_txn(&self) -> Result<RwTxn> {
|
||||
self.env.write_txn().map_err(|e| e.into())
|
||||
}
|
||||
|
||||
pub fn read_txn(&self) -> Result<RoTxn<WithoutTls>> {
|
||||
self.env.read_txn().map_err(|e| e.into())
|
||||
}
|
||||
|
||||
/// Return `Ok(())` if the index scheduler is able to access one of its database.
|
||||
pub fn health(&self) -> Result<()> {
|
||||
let rtxn = self.env.read_txn()?;
|
||||
@@ -374,10 +394,6 @@ impl IndexScheduler {
|
||||
}
|
||||
}
|
||||
|
||||
pub fn read_txn(&self) -> Result<RoTxn<WithoutTls>> {
|
||||
self.env.read_txn().map_err(|e| e.into())
|
||||
}
|
||||
|
||||
/// Start the run loop for the given index scheduler.
|
||||
///
|
||||
/// This function will execute in a different thread and must be called
|
||||
@@ -495,7 +511,7 @@ impl IndexScheduler {
|
||||
|
||||
/// Returns the total number of indexes available for the specified filter.
|
||||
/// And a `Vec` of the index_uid + its stats
|
||||
pub fn get_paginated_indexes_stats(
|
||||
pub fn paginated_indexes_stats(
|
||||
&self,
|
||||
filters: &meilisearch_auth::AuthFilter,
|
||||
from: usize,
|
||||
@@ -536,6 +552,25 @@ impl IndexScheduler {
|
||||
ret.map(|ret| (total, ret))
|
||||
}
|
||||
|
||||
/// Returns the total number of chat workspaces available ~~for the specified filter~~.
|
||||
/// And a `Vec` of the workspace_uids
|
||||
pub fn paginated_chat_workspace_uids(
|
||||
&self,
|
||||
_filters: &meilisearch_auth::AuthFilter,
|
||||
from: usize,
|
||||
limit: usize,
|
||||
) -> Result<(usize, Vec<String>)> {
|
||||
let rtxn = self.read_txn()?;
|
||||
let total = self.chat_settings.len(&rtxn)?;
|
||||
let mut iter = self.chat_settings.iter(&rtxn)?.skip(from);
|
||||
iter.by_ref()
|
||||
.take(limit)
|
||||
.map(|ret| ret.map_err(Error::from))
|
||||
.map(|ret| ret.map(|(uid, _)| uid.to_string()))
|
||||
.collect::<Result<Vec<_>, Error>>()
|
||||
.map(|ret| (total as usize, ret))
|
||||
}
|
||||
|
||||
/// The returned structure contains:
|
||||
/// 1. The name of the property being observed can be `statuses`, `types`, or `indexes`.
|
||||
/// 2. The name of the specific data related to the property can be `enqueued` for the `statuses`, `settingsUpdate` for the `types`, or the name of the index for the `indexes`, for example.
|
||||
@@ -864,6 +899,23 @@ impl IndexScheduler {
|
||||
.collect();
|
||||
res.map(EmbeddingConfigs::new)
|
||||
}
|
||||
|
||||
pub fn chat_settings(&self, rtxn: &RoTxn, uid: &str) -> Result<Option<ChatCompletionSettings>> {
|
||||
self.chat_settings.get(rtxn, uid).map_err(Into::into)
|
||||
}
|
||||
|
||||
pub fn put_chat_settings(
|
||||
&self,
|
||||
wtxn: &mut RwTxn,
|
||||
uid: &str,
|
||||
settings: &ChatCompletionSettings,
|
||||
) -> Result<()> {
|
||||
self.chat_settings.put(wtxn, uid, settings).map_err(Into::into)
|
||||
}
|
||||
|
||||
pub fn delete_chat_settings(&self, wtxn: &mut RwTxn, uid: &str) -> Result<bool> {
|
||||
self.chat_settings.delete(wtxn, uid).map_err(Into::into)
|
||||
}
|
||||
}
|
||||
|
||||
/// The outcome of calling the [`IndexScheduler::tick`] function.
|
||||
|
||||
@@ -375,9 +375,11 @@ impl IndexScheduler {
|
||||
post_commit_dabases_sizes
|
||||
.get(dbname)
|
||||
.map(|post_size| {
|
||||
use byte_unit::{Byte, UnitType::Binary};
|
||||
use std::cmp::Ordering::{Equal, Greater, Less};
|
||||
|
||||
use byte_unit::Byte;
|
||||
use byte_unit::UnitType::Binary;
|
||||
|
||||
let post = Byte::from_u64(*post_size as u64).get_appropriate_unit(Binary);
|
||||
let diff_size = post_size.abs_diff(*pre_size) as u64;
|
||||
let diff = Byte::from_u64(diff_size).get_appropriate_unit(Binary);
|
||||
|
||||
@@ -894,7 +894,7 @@ fn create_and_list_index() {
|
||||
|
||||
let err = index_scheduler.index("kefir").map(|_| ()).unwrap_err();
|
||||
snapshot!(err, @"Index `kefir` not found.");
|
||||
let empty = index_scheduler.get_paginated_indexes_stats(&AuthFilter::default(), 0, 20).unwrap();
|
||||
let empty = index_scheduler.paginated_indexes_stats(&AuthFilter::default(), 0, 20).unwrap();
|
||||
snapshot!(format!("{empty:?}"), @"(0, [])");
|
||||
|
||||
// After advancing just once the index should've been created, the wtxn has been released and commited
|
||||
@@ -902,7 +902,7 @@ fn create_and_list_index() {
|
||||
handle.advance_till([InsideProcessBatch]);
|
||||
|
||||
index_scheduler.index("kefir").unwrap();
|
||||
let list = index_scheduler.get_paginated_indexes_stats(&AuthFilter::default(), 0, 20).unwrap();
|
||||
let list = index_scheduler.paginated_indexes_stats(&AuthFilter::default(), 0, 20).unwrap();
|
||||
snapshot!(json_string!(list, { "[1][0][1].created_at" => "[date]", "[1][0][1].updated_at" => "[date]", "[1][0][1].used_database_size" => "[bytes]", "[1][0][1].database_size" => "[bytes]" }), @r###"
|
||||
[
|
||||
1,
|
||||
|
||||
@@ -2,10 +2,9 @@ use std::time::Instant;
|
||||
|
||||
use big_s::S;
|
||||
use meili_snap::snapshot;
|
||||
use meilisearch_types::milli::obkv_to_json;
|
||||
use meilisearch_types::milli::update::IndexDocumentsMethod::*;
|
||||
use meilisearch_types::milli::update::Setting;
|
||||
use meilisearch_types::milli::FilterableAttributesRule;
|
||||
use meilisearch_types::milli::{obkv_to_json, FilterableAttributesRule};
|
||||
use meilisearch_types::tasks::{Kind, KindWithContent};
|
||||
|
||||
use crate::insta_snapshot::snapshot_index_scheduler;
|
||||
|
||||
@@ -165,6 +165,7 @@ impl AuthController {
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct AuthFilter {
|
||||
search_rules: Option<SearchRules>,
|
||||
key_authorized_indexes: SearchRules,
|
||||
@@ -351,6 +352,7 @@ pub struct IndexSearchRules {
|
||||
fn generate_default_keys(store: &HeedAuthStore) -> Result<()> {
|
||||
store.put_api_key(Key::default_admin())?;
|
||||
store.put_api_key(Key::default_search())?;
|
||||
store.put_api_key(Key::default_chat())?;
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
@@ -125,6 +125,13 @@ impl HeedAuthStore {
|
||||
Action::MetricsAll => {
|
||||
actions.insert(Action::MetricsGet);
|
||||
}
|
||||
Action::ChatsSettingsAll => {
|
||||
actions.extend([
|
||||
Action::ChatsSettingsGet,
|
||||
Action::ChatsSettingsUpdate,
|
||||
Action::ChatsSettingsDelete,
|
||||
]);
|
||||
}
|
||||
other => {
|
||||
actions.insert(*other);
|
||||
}
|
||||
|
||||
@@ -4,9 +4,12 @@ use std::marker::PhantomData;
|
||||
use std::ops::ControlFlow;
|
||||
|
||||
use deserr::errors::{JsonError, QueryParamError};
|
||||
use deserr::{take_cf_content, DeserializeError, IntoValue, MergeWithError, ValuePointerRef};
|
||||
use deserr::{
|
||||
take_cf_content, DeserializeError, Deserr, IntoValue, MergeWithError, ValuePointerRef,
|
||||
};
|
||||
use milli::update::ChatSettings;
|
||||
|
||||
use crate::error::deserr_codes::*;
|
||||
use crate::error::deserr_codes::{self, *};
|
||||
use crate::error::{
|
||||
Code, DeserrParseBoolError, DeserrParseIntError, ErrorCode, InvalidTaskDateError,
|
||||
ParseOffsetDateTimeError,
|
||||
@@ -33,6 +36,7 @@ pub struct DeserrError<Format, C: Default + ErrorCode> {
|
||||
pub code: Code,
|
||||
_phantom: PhantomData<(Format, C)>,
|
||||
}
|
||||
|
||||
impl<Format, C: Default + ErrorCode> DeserrError<Format, C> {
|
||||
pub fn new(msg: String, code: Code) -> Self {
|
||||
Self { msg, code, _phantom: PhantomData }
|
||||
@@ -117,6 +121,16 @@ impl<C: Default + ErrorCode> DeserializeError for DeserrQueryParamError<C> {
|
||||
}
|
||||
}
|
||||
|
||||
impl Deserr<DeserrError<DeserrJson, deserr_codes::InvalidSettingsIndexChat>> for ChatSettings {
|
||||
fn deserialize_from_value<V: IntoValue>(
|
||||
value: deserr::Value<V>,
|
||||
location: ValuePointerRef,
|
||||
) -> Result<Self, DeserrError<DeserrJson, deserr_codes::InvalidSettingsIndexChat>> {
|
||||
Deserr::<JsonError>::deserialize_from_value(value, location)
|
||||
.map_err(|e| DeserrError::new(e.to_string(), InvalidSettingsIndexChat.error_code()))
|
||||
}
|
||||
}
|
||||
|
||||
pub fn immutable_field_error(field: &str, accepted: &[&str], code: Code) -> DeserrJsonError {
|
||||
let msg = format!(
|
||||
"Immutable field `{field}`: expected one of {}",
|
||||
|
||||
@@ -194,200 +194,213 @@ macro_rules! make_error_codes {
|
||||
|
||||
// An exhaustive list of all the error codes used by meilisearch.
|
||||
make_error_codes! {
|
||||
ApiKeyAlreadyExists , InvalidRequest , CONFLICT ;
|
||||
ApiKeyNotFound , InvalidRequest , NOT_FOUND ;
|
||||
BadParameter , InvalidRequest , BAD_REQUEST;
|
||||
BadRequest , InvalidRequest , BAD_REQUEST;
|
||||
DatabaseSizeLimitReached , Internal , INTERNAL_SERVER_ERROR;
|
||||
DocumentNotFound , InvalidRequest , NOT_FOUND;
|
||||
DumpAlreadyProcessing , InvalidRequest , CONFLICT;
|
||||
DumpNotFound , InvalidRequest , NOT_FOUND;
|
||||
DumpProcessFailed , Internal , INTERNAL_SERVER_ERROR;
|
||||
DuplicateIndexFound , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyActions , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyCreatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyExpiresAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyIndexes , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyKey , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyUid , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyUpdatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexCreatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexUid , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexUpdatedAt , InvalidRequest , BAD_REQUEST;
|
||||
IndexAlreadyExists , InvalidRequest , CONFLICT ;
|
||||
IndexCreationFailed , Internal , INTERNAL_SERVER_ERROR;
|
||||
IndexNotFound , InvalidRequest , NOT_FOUND;
|
||||
IndexPrimaryKeyAlreadyExists , InvalidRequest , BAD_REQUEST ;
|
||||
IndexPrimaryKeyMultipleCandidatesFound, InvalidRequest , BAD_REQUEST;
|
||||
IndexPrimaryKeyNoCandidateFound , InvalidRequest , BAD_REQUEST ;
|
||||
Internal , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidApiKey , Auth , FORBIDDEN ;
|
||||
InvalidApiKeyActions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyDescription , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyExpiresAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyName , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyUid , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidContentType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
InvalidDocumentCsvDelimiter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentFields , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
MissingDocumentFilter , InvalidRequest , BAD_REQUEST ;
|
||||
MissingDocumentEditionFunction , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentGeoField , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidVectorDimensions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidVectorsType , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentIds , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchEmbedder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarEmbedder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHybridQuery , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexPrimaryKey , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexUid , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacetsByIndex , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacetOrder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFederated , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFederationOptions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchMaxValuesPerFacet , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchMergeFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryPagination , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryRankingRules , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryPosition , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchRemote , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchWeight , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkRemotes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkSelf , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkSearchApiKey , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkUrl , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToSearchOn , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToCrop , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToHighlight , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarAttributesToRetrieve , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToRetrieve , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchRankingScoreThreshold , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarRankingScoreThreshold , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchCropLength , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchCropMarker , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchSemanticRatio , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchLocales , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchExhaustiveFacetCount, InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchFacetName , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHighlightPostTag , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHighlightPreTag , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHitsPerPage , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchMatchingStrategy , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchPage , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchQ , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchQuery , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchName , InvalidRequest , BAD_REQUEST ;
|
||||
FacetSearchDisabled , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchVector , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowMatchesPosition , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowRankingScore , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarShowRankingScore , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowRankingScoreDetails , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarShowRankingScoreDetails , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchSort , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchDistinct , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDisplayedAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDistinctAttribute , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsProximityPrecision , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFacetSearch , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsPrefixSearch , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFaceting , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFilterableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsPagination , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSearchCutoffMs , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsEmbedders , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsRankingRules , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSearchableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSortableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsStopWords , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsNonSeparatorTokens , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSeparatorTokens , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDictionary , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSynonyms , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsTypoTolerance , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsLocalizedAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidState , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidStoreFile , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidSwapDuplicateIndexFound , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSwapIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterEnqueuedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterFinishedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterStartedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeEnqueuedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeFinishedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeStartedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskCanceledBy , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskFrom , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskReverse , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskStatuses , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskTypes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskUids , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidBatchUids , InvalidRequest , BAD_REQUEST ;
|
||||
IoError , System , UNPROCESSABLE_ENTITY;
|
||||
FeatureNotEnabled , InvalidRequest , BAD_REQUEST ;
|
||||
MalformedPayload , InvalidRequest , BAD_REQUEST ;
|
||||
MaxFieldsLimitExceeded , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyActions , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyExpiresAt , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
MissingAuthorizationHeader , Auth , UNAUTHORIZED ;
|
||||
MissingContentType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
MissingDocumentId , InvalidRequest , BAD_REQUEST ;
|
||||
MissingFacetSearchFacetName , InvalidRequest , BAD_REQUEST ;
|
||||
MissingIndexUid , InvalidRequest , BAD_REQUEST ;
|
||||
MissingMasterKey , Auth , UNAUTHORIZED ;
|
||||
MissingNetworkUrl , InvalidRequest , BAD_REQUEST ;
|
||||
MissingPayload , InvalidRequest , BAD_REQUEST ;
|
||||
MissingSearchHybrid , InvalidRequest , BAD_REQUEST ;
|
||||
MissingSwapIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
MissingTaskFilters , InvalidRequest , BAD_REQUEST ;
|
||||
NoSpaceLeftOnDevice , System , UNPROCESSABLE_ENTITY;
|
||||
PayloadTooLarge , InvalidRequest , PAYLOAD_TOO_LARGE ;
|
||||
RemoteBadResponse , System , BAD_GATEWAY ;
|
||||
RemoteBadRequest , InvalidRequest , BAD_REQUEST ;
|
||||
RemoteCouldNotSendRequest , System , BAD_GATEWAY ;
|
||||
RemoteInvalidApiKey , Auth , FORBIDDEN ;
|
||||
RemoteRemoteError , System , BAD_GATEWAY ;
|
||||
RemoteTimeout , System , BAD_GATEWAY ;
|
||||
TooManySearchRequests , System , SERVICE_UNAVAILABLE ;
|
||||
TaskNotFound , InvalidRequest , NOT_FOUND ;
|
||||
TaskFileNotFound , InvalidRequest , NOT_FOUND ;
|
||||
BatchNotFound , InvalidRequest , NOT_FOUND ;
|
||||
TooManyOpenFiles , System , UNPROCESSABLE_ENTITY ;
|
||||
TooManyVectors , InvalidRequest , BAD_REQUEST ;
|
||||
UnretrievableDocument , Internal , BAD_REQUEST ;
|
||||
UnretrievableErrorCode , InvalidRequest , BAD_REQUEST ;
|
||||
UnsupportedMediaType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
ApiKeyAlreadyExists , InvalidRequest , CONFLICT ;
|
||||
ApiKeyNotFound , InvalidRequest , NOT_FOUND ;
|
||||
BadParameter , InvalidRequest , BAD_REQUEST;
|
||||
BadRequest , InvalidRequest , BAD_REQUEST;
|
||||
DatabaseSizeLimitReached , Internal , INTERNAL_SERVER_ERROR;
|
||||
DocumentNotFound , InvalidRequest , NOT_FOUND;
|
||||
DumpAlreadyProcessing , InvalidRequest , CONFLICT;
|
||||
DumpNotFound , InvalidRequest , NOT_FOUND;
|
||||
DumpProcessFailed , Internal , INTERNAL_SERVER_ERROR;
|
||||
DuplicateIndexFound , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyActions , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyCreatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyExpiresAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyIndexes , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyKey , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyUid , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableApiKeyUpdatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexCreatedAt , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexUid , InvalidRequest , BAD_REQUEST;
|
||||
ImmutableIndexUpdatedAt , InvalidRequest , BAD_REQUEST;
|
||||
IndexAlreadyExists , InvalidRequest , CONFLICT ;
|
||||
IndexCreationFailed , Internal , INTERNAL_SERVER_ERROR;
|
||||
IndexNotFound , InvalidRequest , NOT_FOUND;
|
||||
IndexPrimaryKeyAlreadyExists , InvalidRequest , BAD_REQUEST ;
|
||||
IndexPrimaryKeyMultipleCandidatesFound , InvalidRequest , BAD_REQUEST;
|
||||
IndexPrimaryKeyNoCandidateFound , InvalidRequest , BAD_REQUEST ;
|
||||
Internal , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidApiKey , Auth , FORBIDDEN ;
|
||||
InvalidApiKeyActions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyDescription , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyExpiresAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyName , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidApiKeyUid , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidContentType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
InvalidDocumentCsvDelimiter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentFields , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
MissingDocumentFilter , InvalidRequest , BAD_REQUEST ;
|
||||
MissingDocumentEditionFunction , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentGeoField , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidVectorDimensions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidVectorsType , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentIds , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchEmbedder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarEmbedder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHybridQuery , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexPrimaryKey , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidIndexUid , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacetsByIndex , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFacetOrder , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFederated , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchFederationOptions , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchMaxValuesPerFacet , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchMergeFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryPagination , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryRankingRules , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchQueryPosition , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchRemote , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidMultiSearchWeight , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkRemotes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkSelf , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkSearchApiKey , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidNetworkUrl , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToSearchOn , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToCrop , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToHighlight , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarAttributesToRetrieve , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchAttributesToRetrieve , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchRankingScoreThreshold , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarRankingScoreThreshold , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchRetrieveVectors , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchCropLength , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchCropMarker , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchFacets , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchSemanticRatio , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchLocales , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchExhaustiveFacetCount , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchFacetName , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarFilter , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHighlightPostTag , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHighlightPreTag , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchHitsPerPage , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchMatchingStrategy , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchOffset , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchPage , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchQ , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchQuery , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidFacetSearchName , InvalidRequest , BAD_REQUEST ;
|
||||
FacetSearchDisabled , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchVector , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowMatchesPosition , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowRankingScore , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarShowRankingScore , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchShowRankingScoreDetails , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSimilarShowRankingScoreDetails , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchSort , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSearchDistinct , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDisplayedAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDistinctAttribute , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsProximityPrecision , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFacetSearch , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsPrefixSearch , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFaceting , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsFilterableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsPagination , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSearchCutoffMs , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsEmbedders , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsRankingRules , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSearchableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSortableAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsStopWords , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsNonSeparatorTokens , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSeparatorTokens , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsDictionary , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsSynonyms , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsTypoTolerance , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsLocalizedAttributes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidState , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidStoreFile , Internal , INTERNAL_SERVER_ERROR ;
|
||||
InvalidSwapDuplicateIndexFound , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSwapIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterEnqueuedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterFinishedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskAfterStartedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeEnqueuedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeFinishedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskBeforeStartedAt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskCanceledBy , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskFrom , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskLimit , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskReverse , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskStatuses , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskTypes , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidTaskUids , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidBatchUids , InvalidRequest , BAD_REQUEST ;
|
||||
IoError , System , UNPROCESSABLE_ENTITY;
|
||||
FeatureNotEnabled , InvalidRequest , BAD_REQUEST ;
|
||||
MalformedPayload , InvalidRequest , BAD_REQUEST ;
|
||||
MaxFieldsLimitExceeded , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyActions , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyExpiresAt , InvalidRequest , BAD_REQUEST ;
|
||||
MissingApiKeyIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
MissingAuthorizationHeader , Auth , UNAUTHORIZED ;
|
||||
MissingContentType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
MissingDocumentId , InvalidRequest , BAD_REQUEST ;
|
||||
MissingFacetSearchFacetName , InvalidRequest , BAD_REQUEST ;
|
||||
MissingIndexUid , InvalidRequest , BAD_REQUEST ;
|
||||
MissingMasterKey , Auth , UNAUTHORIZED ;
|
||||
MissingNetworkUrl , InvalidRequest , BAD_REQUEST ;
|
||||
MissingPayload , InvalidRequest , BAD_REQUEST ;
|
||||
MissingSearchHybrid , InvalidRequest , BAD_REQUEST ;
|
||||
MissingSwapIndexes , InvalidRequest , BAD_REQUEST ;
|
||||
MissingTaskFilters , InvalidRequest , BAD_REQUEST ;
|
||||
NoSpaceLeftOnDevice , System , UNPROCESSABLE_ENTITY;
|
||||
PayloadTooLarge , InvalidRequest , PAYLOAD_TOO_LARGE ;
|
||||
RemoteBadResponse , System , BAD_GATEWAY ;
|
||||
RemoteBadRequest , InvalidRequest , BAD_REQUEST ;
|
||||
RemoteCouldNotSendRequest , System , BAD_GATEWAY ;
|
||||
RemoteInvalidApiKey , Auth , FORBIDDEN ;
|
||||
RemoteRemoteError , System , BAD_GATEWAY ;
|
||||
RemoteTimeout , System , BAD_GATEWAY ;
|
||||
TooManySearchRequests , System , SERVICE_UNAVAILABLE ;
|
||||
TaskNotFound , InvalidRequest , NOT_FOUND ;
|
||||
TaskFileNotFound , InvalidRequest , NOT_FOUND ;
|
||||
BatchNotFound , InvalidRequest , NOT_FOUND ;
|
||||
TooManyOpenFiles , System , UNPROCESSABLE_ENTITY ;
|
||||
TooManyVectors , InvalidRequest , BAD_REQUEST ;
|
||||
UnretrievableDocument , Internal , BAD_REQUEST ;
|
||||
UnretrievableErrorCode , InvalidRequest , BAD_REQUEST ;
|
||||
UnsupportedMediaType , InvalidRequest , UNSUPPORTED_MEDIA_TYPE ;
|
||||
|
||||
// Experimental features
|
||||
VectorEmbeddingError , InvalidRequest , BAD_REQUEST ;
|
||||
NotFoundSimilarId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentEditionContext , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentEditionFunctionFilter , InvalidRequest , BAD_REQUEST ;
|
||||
EditDocumentsByFunctionError , InvalidRequest , BAD_REQUEST
|
||||
VectorEmbeddingError , InvalidRequest , BAD_REQUEST ;
|
||||
NotFoundSimilarId , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentEditionContext , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidDocumentEditionFunctionFilter , InvalidRequest , BAD_REQUEST ;
|
||||
EditDocumentsByFunctionError , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidSettingsIndexChat , InvalidRequest , BAD_REQUEST ;
|
||||
// Experimental features - Chat Completions
|
||||
InvalidApiKeyChats , InvalidRequest , BAD_REQUEST ;
|
||||
ChatWorkspaceNotFound , InvalidRequest , NOT_FOUND ;
|
||||
InvalidChatCompletionSource , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionBaseApi , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionApiKey , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionPrompts , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionSystemPrompt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionSearchDescriptionPrompt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionSearchQueryParamPrompt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionSearchIndexUidParamPrompt , InvalidRequest , BAD_REQUEST ;
|
||||
InvalidChatCompletionPreQueryPrompt , InvalidRequest , BAD_REQUEST
|
||||
}
|
||||
|
||||
impl ErrorCode for JoinError {
|
||||
|
||||
@@ -2,6 +2,13 @@ use std::collections::BTreeMap;
|
||||
|
||||
use serde::{Deserialize, Serialize};
|
||||
|
||||
pub const DEFAULT_CHAT_SYSTEM_PROMPT: &str = "You are a highly capable research assistant with access to powerful search tools. IMPORTANT INSTRUCTIONS:1. When answering questions, you MUST make multiple tool calls (at least 2-3) to gather comprehensive information.2. Use different search queries for each tool call - vary keywords, rephrase questions, and explore different semantic angles to ensure broad coverage.3. Always explicitly announce BEFORE making each tool call by saying: \"I'll search for [specific information] now.\"4. Combine information from ALL tool calls to provide complete, nuanced answers rather than relying on a single source.5. For complex topics, break down your research into multiple targeted queries rather than using a single generic search.";
|
||||
pub const DEFAULT_CHAT_SEARCH_DESCRIPTION_PROMPT: &str =
|
||||
"Search the database for relevant JSON documents using an optional query.";
|
||||
pub const DEFAULT_CHAT_SEARCH_Q_PARAM_PROMPT: &str = "The search query string used to find relevant documents in the index. This should contain keywords or phrases that best represent what the user is looking for. More specific queries will yield more precise results.";
|
||||
pub const DEFAULT_CHAT_SEARCH_INDEX_UID_PARAM_PROMPT: &str = "The name of the index to search within. An index is a collection of documents organized for search. Selecting the right index ensures the most relevant results for the user query. You have access to two indexes: movies, steam. The movies index contains movies with overviews. The steam index contains steam games from the Steam platform with their prices";
|
||||
pub const DEFAULT_CHAT_PRE_QUERY_PROMPT: &str = "";
|
||||
|
||||
#[derive(Serialize, Deserialize, Debug, Clone, Copy, Default, PartialEq, Eq)]
|
||||
#[serde(rename_all = "camelCase", default)]
|
||||
pub struct RuntimeTogglableFeatures {
|
||||
@@ -12,6 +19,7 @@ pub struct RuntimeTogglableFeatures {
|
||||
pub network: bool,
|
||||
pub get_task_documents_route: bool,
|
||||
pub composite_embedders: bool,
|
||||
pub chat_completions: bool,
|
||||
}
|
||||
|
||||
#[derive(Default, Debug, Clone, Copy)]
|
||||
@@ -37,3 +45,69 @@ pub struct Network {
|
||||
#[serde(default)]
|
||||
pub remotes: BTreeMap<String, Remote>,
|
||||
}
|
||||
|
||||
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Default)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct ChatCompletionSettings {
|
||||
pub source: ChatCompletionSource,
|
||||
#[serde(default)]
|
||||
pub base_api: Option<String>,
|
||||
#[serde(default)]
|
||||
pub api_key: Option<String>,
|
||||
#[serde(default)]
|
||||
pub prompts: ChatCompletionPrompts,
|
||||
}
|
||||
|
||||
impl ChatCompletionSettings {
|
||||
pub fn hide_secrets(&mut self) {
|
||||
if let Some(api_key) = &mut self.api_key {
|
||||
Self::hide_secret(api_key);
|
||||
}
|
||||
}
|
||||
|
||||
fn hide_secret(secret: &mut String) {
|
||||
match secret.len() {
|
||||
x if x < 10 => {
|
||||
secret.replace_range(.., "XXX...");
|
||||
}
|
||||
x if x < 20 => {
|
||||
secret.replace_range(2.., "XXXX...");
|
||||
}
|
||||
x if x < 30 => {
|
||||
secret.replace_range(3.., "XXXXX...");
|
||||
}
|
||||
_x => {
|
||||
secret.replace_range(5.., "XXXXXX...");
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq, Default)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub enum ChatCompletionSource {
|
||||
#[default]
|
||||
OpenAi,
|
||||
}
|
||||
|
||||
#[derive(Serialize, Deserialize, Debug, Clone, PartialEq, Eq)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct ChatCompletionPrompts {
|
||||
pub system: String,
|
||||
pub search_description: String,
|
||||
pub search_q_param: String,
|
||||
pub search_index_uid_param: String,
|
||||
pub pre_query: String,
|
||||
}
|
||||
|
||||
impl Default for ChatCompletionPrompts {
|
||||
fn default() -> Self {
|
||||
Self {
|
||||
system: DEFAULT_CHAT_SYSTEM_PROMPT.to_string(),
|
||||
search_description: DEFAULT_CHAT_SEARCH_DESCRIPTION_PROMPT.to_string(),
|
||||
search_q_param: DEFAULT_CHAT_SEARCH_Q_PARAM_PROMPT.to_string(),
|
||||
search_index_uid_param: DEFAULT_CHAT_SEARCH_INDEX_UID_PARAM_PROMPT.to_string(),
|
||||
pre_query: DEFAULT_CHAT_PRE_QUERY_PROMPT.to_string(),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@@ -53,10 +53,14 @@ pub struct CreateApiKey {
|
||||
#[schema(example = json!(["documents.add"]))]
|
||||
#[deserr(error = DeserrJsonError<InvalidApiKeyActions>, missing_field_error = DeserrJsonError::missing_api_key_actions)]
|
||||
pub actions: Vec<Action>,
|
||||
/// A list of accesible indexes permitted for the key. `["*"]` for all indexes. The `*` character can be used as a wildcard when located at the last position. e.g. `products_*` to allow access to all indexes whose names start with `products_`.
|
||||
/// A list of accessible indexes permitted for the key. `["*"]` for all indexes. The `*` character can be used as a wildcard when located at the last position. e.g. `products_*` to allow access to all indexes whose names start with `products_`.
|
||||
#[deserr(error = DeserrJsonError<InvalidApiKeyIndexes>, missing_field_error = DeserrJsonError::missing_api_key_indexes)]
|
||||
#[schema(value_type = Vec<String>, example = json!(["products"]))]
|
||||
pub indexes: Vec<IndexUidPattern>,
|
||||
/// A list of accessible chats permitted for the key. `["*"]` for all chats. The `*` character can be used as a wildcard when located at the last position. e.g. `chats_*` to allow access to all indexes whose names start with `chats_`.
|
||||
#[deserr(default, error = DeserrJsonError<InvalidApiKeyChats>)]
|
||||
#[schema(value_type = Vec<String>, example = json!(["OpenAI", "ChatGPT"]))]
|
||||
pub chats: Vec<IndexUidPattern>,
|
||||
/// Represent the expiration date and time as RFC 3339 format. `null` equals to no expiration time.
|
||||
#[deserr(error = DeserrJsonError<InvalidApiKeyExpiresAt>, try_from(Option<String>) = parse_expiration_date -> ParseOffsetDateTimeError, missing_field_error = DeserrJsonError::missing_api_key_expires_at)]
|
||||
pub expires_at: Option<OffsetDateTime>,
|
||||
@@ -64,7 +68,7 @@ pub struct CreateApiKey {
|
||||
|
||||
impl CreateApiKey {
|
||||
pub fn to_key(self) -> Key {
|
||||
let CreateApiKey { description, name, uid, actions, indexes, expires_at } = self;
|
||||
let CreateApiKey { description, name, uid, actions, indexes, chats, expires_at } = self;
|
||||
let now = OffsetDateTime::now_utc();
|
||||
Key {
|
||||
description,
|
||||
@@ -72,6 +76,7 @@ impl CreateApiKey {
|
||||
uid,
|
||||
actions,
|
||||
indexes,
|
||||
chats,
|
||||
expires_at,
|
||||
created_at: now,
|
||||
updated_at: now,
|
||||
@@ -120,6 +125,7 @@ pub struct Key {
|
||||
pub uid: KeyId,
|
||||
pub actions: Vec<Action>,
|
||||
pub indexes: Vec<IndexUidPattern>,
|
||||
pub chats: Vec<IndexUidPattern>,
|
||||
#[serde(with = "time::serde::rfc3339::option")]
|
||||
pub expires_at: Option<OffsetDateTime>,
|
||||
#[serde(with = "time::serde::rfc3339")]
|
||||
@@ -138,6 +144,7 @@ impl Key {
|
||||
uid,
|
||||
actions: vec![Action::All],
|
||||
indexes: vec![IndexUidPattern::all()],
|
||||
chats: vec![IndexUidPattern::all()],
|
||||
expires_at: None,
|
||||
created_at: now,
|
||||
updated_at: now,
|
||||
@@ -153,6 +160,23 @@ impl Key {
|
||||
uid,
|
||||
actions: vec![Action::Search],
|
||||
indexes: vec![IndexUidPattern::all()],
|
||||
chats: vec![],
|
||||
expires_at: None,
|
||||
created_at: now,
|
||||
updated_at: now,
|
||||
}
|
||||
}
|
||||
|
||||
pub fn default_chat() -> Self {
|
||||
let now = OffsetDateTime::now_utc();
|
||||
let uid = Uuid::new_v4();
|
||||
Self {
|
||||
name: Some("Default Chat API Key".to_string()),
|
||||
description: Some("Use it to chat and search from the frontend".to_string()),
|
||||
uid,
|
||||
actions: vec![Action::ChatCompletions, Action::Search],
|
||||
indexes: vec![IndexUidPattern::all()],
|
||||
chats: vec![IndexUidPattern::all()],
|
||||
expires_at: None,
|
||||
created_at: now,
|
||||
updated_at: now,
|
||||
@@ -308,6 +332,25 @@ pub enum Action {
|
||||
#[serde(rename = "network.update")]
|
||||
#[deserr(rename = "network.update")]
|
||||
NetworkUpdate,
|
||||
// TODO should we rename it chatCompletions.get ?
|
||||
#[serde(rename = "chat")]
|
||||
#[deserr(rename = "chat")]
|
||||
ChatCompletions,
|
||||
#[serde(rename = "chats.get")]
|
||||
#[deserr(rename = "chats.get")]
|
||||
ChatsGet,
|
||||
#[serde(rename = "chatsSettings.*")]
|
||||
#[deserr(rename = "chatsSettings.*")]
|
||||
ChatsSettingsAll,
|
||||
#[serde(rename = "chatsSettings.get")]
|
||||
#[deserr(rename = "chatsSettings.get")]
|
||||
ChatsSettingsGet,
|
||||
#[serde(rename = "chatsSettings.update")]
|
||||
#[deserr(rename = "chatsSettings.update")]
|
||||
ChatsSettingsUpdate,
|
||||
#[serde(rename = "chatsSettings.delete")]
|
||||
#[deserr(rename = "chatsSettings.delete")]
|
||||
ChatsSettingsDelete,
|
||||
}
|
||||
|
||||
impl Action {
|
||||
@@ -333,6 +376,12 @@ impl Action {
|
||||
SETTINGS_ALL => Some(Self::SettingsAll),
|
||||
SETTINGS_GET => Some(Self::SettingsGet),
|
||||
SETTINGS_UPDATE => Some(Self::SettingsUpdate),
|
||||
CHAT_COMPLETIONS => Some(Self::ChatCompletions),
|
||||
CHATS_GET => Some(Self::ChatsGet),
|
||||
CHATS_SETTINGS_ALL => Some(Self::ChatsSettingsAll),
|
||||
CHATS_SETTINGS_GET => Some(Self::ChatsSettingsGet),
|
||||
CHATS_SETTINGS_UPDATE => Some(Self::ChatsSettingsUpdate),
|
||||
CHATS_SETTINGS_DELETE => Some(Self::ChatsSettingsDelete),
|
||||
STATS_ALL => Some(Self::StatsAll),
|
||||
STATS_GET => Some(Self::StatsGet),
|
||||
METRICS_ALL => Some(Self::MetricsAll),
|
||||
@@ -397,4 +446,11 @@ pub mod actions {
|
||||
|
||||
pub const NETWORK_GET: u8 = NetworkGet.repr();
|
||||
pub const NETWORK_UPDATE: u8 = NetworkUpdate.repr();
|
||||
|
||||
pub const CHAT_COMPLETIONS: u8 = ChatCompletions.repr();
|
||||
pub const CHATS_GET: u8 = ChatsGet.repr();
|
||||
pub const CHATS_SETTINGS_ALL: u8 = ChatsSettingsAll.repr();
|
||||
pub const CHATS_SETTINGS_GET: u8 = ChatsSettingsGet.repr();
|
||||
pub const CHATS_SETTINGS_UPDATE: u8 = ChatsSettingsUpdate.repr();
|
||||
pub const CHATS_SETTINGS_DELETE: u8 = ChatsSettingsDelete.repr();
|
||||
}
|
||||
|
||||
@@ -11,6 +11,7 @@ use fst::IntoStreamer;
|
||||
use milli::disabled_typos_terms::DisabledTyposTerms;
|
||||
use milli::index::{IndexEmbeddingConfig, PrefixSearch};
|
||||
use milli::proximity::ProximityPrecision;
|
||||
pub use milli::update::ChatSettings;
|
||||
use milli::update::Setting;
|
||||
use milli::{Criterion, CriterionError, FilterableAttributesRule, Index, DEFAULT_VALUES_PER_FACET};
|
||||
use serde::{Deserialize, Serialize, Serializer};
|
||||
@@ -185,7 +186,7 @@ impl<E: DeserializeError> Deserr<E> for SettingEmbeddingSettings {
|
||||
/// Holds all the settings for an index. `T` can either be `Checked` if they represents settings
|
||||
/// whose validity is guaranteed, or `Unchecked` if they need to be validated. In the later case, a
|
||||
/// call to `check` will return a `Settings<Checked>` from a `Settings<Unchecked>`.
|
||||
#[derive(Debug, Clone, Default, Serialize, Deserialize, PartialEq, Eq, Deserr, ToSchema)]
|
||||
#[derive(Debug, Clone, Default, Serialize, Deserialize, PartialEq, Deserr, ToSchema)]
|
||||
#[serde(
|
||||
deny_unknown_fields,
|
||||
rename_all = "camelCase",
|
||||
@@ -199,72 +200,86 @@ pub struct Settings<T> {
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsDisplayedAttributes>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["id", "title", "description", "url"]))]
|
||||
pub displayed_attributes: WildcardSetting,
|
||||
|
||||
/// Fields in which to search for matching query words sorted by order of importance.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsSearchableAttributes>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["title", "description"]))]
|
||||
pub searchable_attributes: WildcardSetting,
|
||||
|
||||
/// Attributes to use for faceting and filtering. See [Filtering and Faceted Search](https://www.meilisearch.com/docs/learn/filtering_and_sorting/search_with_facet_filters).
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsFilterableAttributes>)]
|
||||
#[schema(value_type = Option<Vec<FilterableAttributesRule>>, example = json!(["release_date", "genre"]))]
|
||||
pub filterable_attributes: Setting<Vec<FilterableAttributesRule>>,
|
||||
|
||||
/// Attributes to use when sorting search results.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsSortableAttributes>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["release_date"]))]
|
||||
pub sortable_attributes: Setting<BTreeSet<String>>,
|
||||
|
||||
/// List of ranking rules sorted by order of importance. The order is customizable.
|
||||
/// [A list of ordered built-in ranking rules](https://www.meilisearch.com/docs/learn/relevancy/relevancy).
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsRankingRules>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!([RankingRuleView::Words, RankingRuleView::Typo, RankingRuleView::Proximity, RankingRuleView::Attribute, RankingRuleView::Exactness]))]
|
||||
pub ranking_rules: Setting<Vec<RankingRuleView>>,
|
||||
|
||||
/// List of words ignored when present in search queries.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsStopWords>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["the", "a", "them", "their"]))]
|
||||
pub stop_words: Setting<BTreeSet<String>>,
|
||||
|
||||
/// List of characters not delimiting where one term begins and ends.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsNonSeparatorTokens>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!([" ", "\n"]))]
|
||||
pub non_separator_tokens: Setting<BTreeSet<String>>,
|
||||
|
||||
/// List of characters delimiting where one term begins and ends.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsSeparatorTokens>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["S"]))]
|
||||
pub separator_tokens: Setting<BTreeSet<String>>,
|
||||
|
||||
/// List of strings Meilisearch should parse as a single term.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsDictionary>)]
|
||||
#[schema(value_type = Option<Vec<String>>, example = json!(["iPhone pro"]))]
|
||||
pub dictionary: Setting<BTreeSet<String>>,
|
||||
|
||||
/// List of associated words treated similarly. A word associated to an array of word as synonyms.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsSynonyms>)]
|
||||
#[schema(value_type = Option<BTreeMap<String, Vec<String>>>, example = json!({ "he": ["she", "they", "them"], "phone": ["iPhone", "android"]}))]
|
||||
pub synonyms: Setting<BTreeMap<String, Vec<String>>>,
|
||||
|
||||
/// Search returns documents with distinct (different) values of the given field.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsDistinctAttribute>)]
|
||||
#[schema(value_type = Option<String>, example = json!("sku"))]
|
||||
pub distinct_attribute: Setting<String>,
|
||||
|
||||
/// Precision level when calculating the proximity ranking rule.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsProximityPrecision>)]
|
||||
#[schema(value_type = Option<String>, example = json!(ProximityPrecisionView::ByAttribute))]
|
||||
pub proximity_precision: Setting<ProximityPrecisionView>,
|
||||
|
||||
/// Customize typo tolerance feature.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsTypoTolerance>)]
|
||||
#[schema(value_type = Option<TypoSettings>, example = json!({ "enabled": true, "disableOnAttributes": ["title"]}))]
|
||||
pub typo_tolerance: Setting<TypoSettings>,
|
||||
|
||||
/// Faceting settings.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsFaceting>)]
|
||||
#[schema(value_type = Option<FacetingSettings>, example = json!({ "maxValuesPerFacet": 10, "sortFacetValuesBy": { "genre": FacetValuesSort::Count }}))]
|
||||
pub faceting: Setting<FacetingSettings>,
|
||||
|
||||
/// Pagination settings.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsPagination>)]
|
||||
@@ -276,24 +291,34 @@ pub struct Settings<T> {
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsEmbedders>)]
|
||||
#[schema(value_type = Option<BTreeMap<String, SettingEmbeddingSettings>>)]
|
||||
pub embedders: Setting<BTreeMap<String, SettingEmbeddingSettings>>,
|
||||
|
||||
/// Maximum duration of a search query.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsSearchCutoffMs>)]
|
||||
#[schema(value_type = Option<u64>, example = json!(50))]
|
||||
pub search_cutoff_ms: Setting<u64>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsLocalizedAttributes>)]
|
||||
#[schema(value_type = Option<Vec<LocalizedAttributesRuleView>>, example = json!(50))]
|
||||
pub localized_attributes: Setting<Vec<LocalizedAttributesRuleView>>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsFacetSearch>)]
|
||||
#[schema(value_type = Option<bool>, example = json!(true))]
|
||||
pub facet_search: Setting<bool>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsPrefixSearch>)]
|
||||
#[schema(value_type = Option<PrefixSearchSettings>, example = json!("Hemlo"))]
|
||||
pub prefix_search: Setting<PrefixSearchSettings>,
|
||||
|
||||
/// Customize the chat prompting.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSettingsIndexChat>)]
|
||||
#[schema(value_type = Option<ChatSettings>)]
|
||||
pub chat: Setting<ChatSettings>,
|
||||
|
||||
#[serde(skip)]
|
||||
#[deserr(skip)]
|
||||
pub _kind: PhantomData<T>,
|
||||
@@ -359,6 +384,7 @@ impl Settings<Checked> {
|
||||
localized_attributes: Setting::Reset,
|
||||
facet_search: Setting::Reset,
|
||||
prefix_search: Setting::Reset,
|
||||
chat: Setting::Reset,
|
||||
_kind: PhantomData,
|
||||
}
|
||||
}
|
||||
@@ -385,6 +411,7 @@ impl Settings<Checked> {
|
||||
localized_attributes: localized_attributes_rules,
|
||||
facet_search,
|
||||
prefix_search,
|
||||
chat,
|
||||
_kind,
|
||||
} = self;
|
||||
|
||||
@@ -409,6 +436,7 @@ impl Settings<Checked> {
|
||||
localized_attributes: localized_attributes_rules,
|
||||
facet_search,
|
||||
prefix_search,
|
||||
chat,
|
||||
_kind: PhantomData,
|
||||
}
|
||||
}
|
||||
@@ -459,6 +487,7 @@ impl Settings<Unchecked> {
|
||||
localized_attributes: self.localized_attributes,
|
||||
facet_search: self.facet_search,
|
||||
prefix_search: self.prefix_search,
|
||||
chat: self.chat,
|
||||
_kind: PhantomData,
|
||||
}
|
||||
}
|
||||
@@ -533,8 +562,9 @@ impl Settings<Unchecked> {
|
||||
Setting::Set(this)
|
||||
}
|
||||
},
|
||||
prefix_search: other.prefix_search.or(self.prefix_search),
|
||||
facet_search: other.facet_search.or(self.facet_search),
|
||||
prefix_search: other.prefix_search.or(self.prefix_search),
|
||||
chat: other.chat.clone().or(self.chat.clone()),
|
||||
_kind: PhantomData,
|
||||
}
|
||||
}
|
||||
@@ -573,6 +603,7 @@ pub fn apply_settings_to_builder(
|
||||
localized_attributes: localized_attributes_rules,
|
||||
facet_search,
|
||||
prefix_search,
|
||||
chat,
|
||||
_kind,
|
||||
} = settings;
|
||||
|
||||
@@ -783,6 +814,12 @@ pub fn apply_settings_to_builder(
|
||||
Setting::Reset => builder.reset_facet_search(),
|
||||
Setting::NotSet => (),
|
||||
}
|
||||
|
||||
match chat {
|
||||
Setting::Set(chat) => builder.set_chat(chat.clone()),
|
||||
Setting::Reset => builder.reset_chat(),
|
||||
Setting::NotSet => (),
|
||||
}
|
||||
}
|
||||
|
||||
pub enum SecretPolicy {
|
||||
@@ -880,14 +917,11 @@ pub fn settings(
|
||||
})
|
||||
.collect();
|
||||
let embedders = Setting::Set(embedders);
|
||||
|
||||
let search_cutoff_ms = index.search_cutoff(rtxn)?;
|
||||
|
||||
let localized_attributes_rules = index.localized_attributes_rules(rtxn)?;
|
||||
|
||||
let prefix_search = index.prefix_search(rtxn)?.map(PrefixSearchSettings::from);
|
||||
|
||||
let facet_search = index.facet_search(rtxn)?;
|
||||
let chat = index.chat_config(rtxn).map(ChatSettings::from)?;
|
||||
|
||||
let mut settings = Settings {
|
||||
displayed_attributes: match displayed_attributes {
|
||||
@@ -925,8 +959,9 @@ pub fn settings(
|
||||
Some(rules) => Setting::Set(rules.into_iter().map(|r| r.into()).collect()),
|
||||
None => Setting::Reset,
|
||||
},
|
||||
prefix_search: Setting::Set(prefix_search.unwrap_or_default()),
|
||||
facet_search: Setting::Set(facet_search),
|
||||
prefix_search: Setting::Set(prefix_search.unwrap_or_default()),
|
||||
chat: Setting::Set(chat),
|
||||
_kind: PhantomData,
|
||||
};
|
||||
|
||||
@@ -1154,6 +1189,7 @@ pub(crate) mod test {
|
||||
search_cutoff_ms: Setting::NotSet,
|
||||
facet_search: Setting::NotSet,
|
||||
prefix_search: Setting::NotSet,
|
||||
chat: Setting::NotSet,
|
||||
_kind: PhantomData::<Unchecked>,
|
||||
};
|
||||
|
||||
@@ -1185,6 +1221,8 @@ pub(crate) mod test {
|
||||
search_cutoff_ms: Setting::NotSet,
|
||||
facet_search: Setting::NotSet,
|
||||
prefix_search: Setting::NotSet,
|
||||
chat: Setting::NotSet,
|
||||
|
||||
_kind: PhantomData::<Unchecked>,
|
||||
};
|
||||
|
||||
|
||||
@@ -8,7 +8,7 @@ use crate::error::ResponseError;
|
||||
use crate::settings::{Settings, Unchecked};
|
||||
use crate::tasks::{serialize_duration, Details, IndexSwap, Kind, Status, Task, TaskId};
|
||||
|
||||
#[derive(Debug, Clone, PartialEq, Eq, Serialize, ToSchema)]
|
||||
#[derive(Debug, Clone, PartialEq, Serialize, ToSchema)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
#[schema(rename_all = "camelCase")]
|
||||
pub struct TaskView {
|
||||
@@ -67,7 +67,7 @@ impl TaskView {
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Default, Debug, PartialEq, Eq, Clone, Serialize, Deserialize, ToSchema)]
|
||||
#[derive(Default, Debug, PartialEq, Clone, Serialize, Deserialize, ToSchema)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
#[schema(rename_all = "camelCase")]
|
||||
pub struct DetailsView {
|
||||
|
||||
@@ -597,7 +597,7 @@ impl fmt::Display for ParseTaskKindError {
|
||||
}
|
||||
impl std::error::Error for ParseTaskKindError {}
|
||||
|
||||
#[derive(Debug, PartialEq, Eq, Clone, Serialize, Deserialize)]
|
||||
#[derive(Debug, PartialEq, Clone, Serialize, Deserialize)]
|
||||
pub enum Details {
|
||||
DocumentAdditionOrUpdate {
|
||||
received_documents: u64,
|
||||
|
||||
@@ -32,6 +32,7 @@ async-trait = "0.1.85"
|
||||
bstr = "1.11.3"
|
||||
byte-unit = { version = "5.1.6", features = ["serde"] }
|
||||
bytes = "1.9.0"
|
||||
bumpalo = "3.16.0"
|
||||
clap = { version = "4.5.24", features = ["derive", "env"] }
|
||||
crossbeam-channel = "0.5.15"
|
||||
deserr = { version = "0.6.3", features = ["actix-web"] }
|
||||
@@ -48,6 +49,7 @@ is-terminal = "0.4.13"
|
||||
itertools = "0.14.0"
|
||||
jsonwebtoken = "9.3.0"
|
||||
lazy_static = "1.5.0"
|
||||
liquid = "0.26.9"
|
||||
meilisearch-auth = { path = "../meilisearch-auth" }
|
||||
meilisearch-types = { path = "../meilisearch-types" }
|
||||
mimalloc = { version = "0.1.43", default-features = false }
|
||||
@@ -111,6 +113,8 @@ utoipa = { version = "5.3.1", features = [
|
||||
"openapi_extensions",
|
||||
] }
|
||||
utoipa-scalar = { version = "0.3.0", optional = true, features = ["actix-web"] }
|
||||
async-openai = { git = "https://github.com/meilisearch/async-openai", branch = "better-error-handling" }
|
||||
actix-web-lab = { version = "0.24.1", default-features = false }
|
||||
|
||||
[dev-dependencies]
|
||||
actix-rt = "2.10.0"
|
||||
|
||||
@@ -197,6 +197,7 @@ struct Infos {
|
||||
experimental_max_number_of_batched_tasks: usize,
|
||||
experimental_limit_batched_tasks_total_size: u64,
|
||||
experimental_network: bool,
|
||||
experimental_chat_completions: bool,
|
||||
experimental_get_task_documents_route: bool,
|
||||
experimental_composite_embedders: bool,
|
||||
experimental_embedding_cache_entries: usize,
|
||||
@@ -296,6 +297,7 @@ impl Infos {
|
||||
network,
|
||||
get_task_documents_route,
|
||||
composite_embedders,
|
||||
chat_completions,
|
||||
} = features;
|
||||
|
||||
// We're going to override every sensible information.
|
||||
@@ -314,6 +316,7 @@ impl Infos {
|
||||
experimental_enable_logs_route: experimental_enable_logs_route | logs_route,
|
||||
experimental_reduce_indexing_memory_usage,
|
||||
experimental_network: network,
|
||||
experimental_chat_completions: chat_completions,
|
||||
experimental_get_task_documents_route: get_task_documents_route,
|
||||
experimental_composite_embedders: composite_embedders,
|
||||
experimental_embedding_cache_entries,
|
||||
|
||||
@@ -4,6 +4,7 @@ use std::marker::PhantomData;
|
||||
use std::ops::Deref;
|
||||
use std::pin::Pin;
|
||||
|
||||
use actix_web::http::header::AUTHORIZATION;
|
||||
use actix_web::web::Data;
|
||||
use actix_web::FromRequest;
|
||||
pub use error::AuthenticationError;
|
||||
@@ -94,36 +95,44 @@ impl<P: Policy + 'static, D: 'static + Clone> FromRequest for GuardedData<P, D>
|
||||
_payload: &mut actix_web::dev::Payload,
|
||||
) -> Self::Future {
|
||||
match req.app_data::<Data<AuthController>>().cloned() {
|
||||
Some(auth) => match req
|
||||
.headers()
|
||||
.get("Authorization")
|
||||
.map(|type_token| type_token.to_str().unwrap_or_default().splitn(2, ' '))
|
||||
{
|
||||
Some(mut type_token) => match type_token.next() {
|
||||
Some("Bearer") => {
|
||||
// TODO: find a less hardcoded way?
|
||||
let index = req.match_info().get("index_uid");
|
||||
match type_token.next() {
|
||||
Some(token) => Box::pin(Self::auth_bearer(
|
||||
auth,
|
||||
token.to_string(),
|
||||
index.map(String::from),
|
||||
req.app_data::<D>().cloned(),
|
||||
)),
|
||||
None => Box::pin(err(AuthenticationError::InvalidToken.into())),
|
||||
}
|
||||
}
|
||||
_otherwise => {
|
||||
Box::pin(err(AuthenticationError::MissingAuthorizationHeader.into()))
|
||||
}
|
||||
},
|
||||
None => Box::pin(Self::auth_token(auth, req.app_data::<D>().cloned())),
|
||||
Some(auth) => match extract_token_from_request(req) {
|
||||
Ok(Some(token)) => {
|
||||
// TODO: find a less hardcoded way?
|
||||
let index = req.match_info().get("index_uid");
|
||||
Box::pin(Self::auth_bearer(
|
||||
auth,
|
||||
token.to_string(),
|
||||
index.map(String::from),
|
||||
req.app_data::<D>().cloned(),
|
||||
))
|
||||
}
|
||||
Ok(None) => Box::pin(Self::auth_token(auth, req.app_data::<D>().cloned())),
|
||||
Err(e) => Box::pin(err(e.into())),
|
||||
},
|
||||
None => Box::pin(err(AuthenticationError::IrretrievableState.into())),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
pub fn extract_token_from_request(
|
||||
req: &actix_web::HttpRequest,
|
||||
) -> Result<Option<&str>, AuthenticationError> {
|
||||
match req
|
||||
.headers()
|
||||
.get(AUTHORIZATION)
|
||||
.map(|type_token| type_token.to_str().unwrap_or_default().splitn(2, ' '))
|
||||
{
|
||||
Some(mut type_token) => match type_token.next() {
|
||||
Some("Bearer") => match type_token.next() {
|
||||
Some(token) => Ok(Some(token)),
|
||||
None => Err(AuthenticationError::InvalidToken),
|
||||
},
|
||||
_otherwise => Err(AuthenticationError::MissingAuthorizationHeader),
|
||||
},
|
||||
None => Ok(None),
|
||||
}
|
||||
}
|
||||
|
||||
pub trait Policy {
|
||||
fn authenticate(
|
||||
auth: Data<AuthController>,
|
||||
@@ -156,16 +165,22 @@ pub mod policies {
|
||||
ExpiredTenantToken { exp: i64, now: i64 },
|
||||
#[error("The provided API key is invalid.")]
|
||||
InvalidApiKey,
|
||||
#[error("The provided tenant token cannot acces the index `{index}`, allowed indexes are {allowed:?}.")]
|
||||
#[error("The provided tenant token cannot access the index `{index}`, allowed indexes are {allowed:?}.")]
|
||||
TenantTokenAccessingnUnauthorizedIndex { index: String, allowed: Vec<String> },
|
||||
#[error(
|
||||
"The API key used to generate this tenant token cannot acces the index `{index}`."
|
||||
"The API key used to generate this tenant token cannot access the index `{index}`."
|
||||
)]
|
||||
TenantTokenApiKeyAccessingnUnauthorizedIndex { index: String },
|
||||
#[error(
|
||||
"The API key cannot acces the index `{index}`, authorized indexes are {allowed:?}."
|
||||
"The API key cannot access the index `{index}`, authorized indexes are {allowed:?}."
|
||||
)]
|
||||
ApiKeyAccessingnUnauthorizedIndex { index: String, allowed: Vec<String> },
|
||||
#[error("The provided tenant token cannot access the chats `{chat}`, allowed chats are {allowed:?}.")]
|
||||
TenantTokenAccessingnUnauthorizedChat { chat: String, allowed: Vec<String> },
|
||||
#[error("The API key used to generate this tenant token cannot access the chat `{chat}`.")]
|
||||
TenantTokenApiKeyAccessingnUnauthorizedChat { chat: String },
|
||||
#[error("The API key cannot access the chat `{chat}`, authorized chats are {allowed:?}.")]
|
||||
ApiKeyAccessingnUnauthorizedChat { chat: String, allowed: Vec<String> },
|
||||
#[error("The provided tenant token is invalid.")]
|
||||
InvalidTenantToken,
|
||||
#[error("Could not decode tenant token, {0}.")]
|
||||
@@ -222,8 +237,8 @@ pub mod policies {
|
||||
pub struct ActionPolicy<const A: u8>;
|
||||
|
||||
impl<const A: u8> Policy for ActionPolicy<A> {
|
||||
/// Attempts to grant authentication from a bearer token (that can be a tenant token or an API key), the requested Action,
|
||||
/// and a list of requested indexes.
|
||||
/// Attempts to grant authentication from a bearer token (that can be a tenant token or an API key),
|
||||
/// the requested Action, and a list of requested indexes.
|
||||
///
|
||||
/// If the bearer token is not allowed for the specified indexes and action, returns `None`.
|
||||
/// Otherwise, returns an object containing the generated permissions: the search filters to add to a search, and the list of allowed indexes
|
||||
@@ -299,8 +314,8 @@ pub mod policies {
|
||||
auth: &AuthController,
|
||||
token: &str,
|
||||
) -> Result<TenantTokenOutcome, AuthError> {
|
||||
// Only search action can be accessed by a tenant token.
|
||||
if A != actions::SEARCH {
|
||||
// Only search and chat actions can be accessed by a tenant token.
|
||||
if A != actions::SEARCH && A != actions::CHAT_COMPLETIONS {
|
||||
return Ok(TenantTokenOutcome::NotATenantToken);
|
||||
}
|
||||
|
||||
|
||||
291
crates/meilisearch/src/routes/chat.rs
Normal file
291
crates/meilisearch/src/routes/chat.rs
Normal file
@@ -0,0 +1,291 @@
|
||||
use std::mem;
|
||||
|
||||
use actix_web::web::{self, Data};
|
||||
use actix_web::{Either, HttpResponse, Responder};
|
||||
use actix_web_lab::sse::{self, Event};
|
||||
use async_openai::config::OpenAIConfig;
|
||||
use async_openai::types::{
|
||||
ChatCompletionRequestAssistantMessageArgs, ChatCompletionRequestMessage,
|
||||
ChatCompletionRequestToolMessage, ChatCompletionRequestToolMessageContent,
|
||||
ChatCompletionToolArgs, ChatCompletionToolType, CreateChatCompletionRequest, FinishReason,
|
||||
FunctionObjectArgs,
|
||||
};
|
||||
use async_openai::Client;
|
||||
use futures::StreamExt;
|
||||
use index_scheduler::IndexScheduler;
|
||||
use meilisearch_types::error::ResponseError;
|
||||
use meilisearch_types::keys::actions;
|
||||
use meilisearch_types::milli::index::IndexEmbeddingConfig;
|
||||
use meilisearch_types::milli::prompt::PromptData;
|
||||
use meilisearch_types::milli::vector::EmbeddingConfig;
|
||||
use meilisearch_types::{Document, Index};
|
||||
use serde::{Deserialize, Serialize};
|
||||
use serde_json::json;
|
||||
|
||||
use crate::extractors::authentication::policies::ActionPolicy;
|
||||
use crate::extractors::authentication::GuardedData;
|
||||
use crate::metrics::MEILISEARCH_DEGRADED_SEARCH_REQUESTS;
|
||||
use crate::routes::indexes::search::search_kind;
|
||||
use crate::search::{
|
||||
add_search_rules, perform_search, HybridQuery, RetrieveVectors, SearchQuery, SemanticRatio,
|
||||
};
|
||||
use crate::search_queue::SearchQueue;
|
||||
|
||||
/// The default description of the searchInIndex tool provided to OpenAI.
|
||||
const DEFAULT_SEARCH_IN_INDEX_TOOL_DESCRIPTION: &str =
|
||||
"Search the database for relevant JSON documents using an optional query.";
|
||||
/// The default description of the searchInIndex `q` parameter tool provided to OpenAI.
|
||||
const DEFAULT_SEARCH_IN_INDEX_Q_PARAMETER_TOOL_DESCRIPTION: &str =
|
||||
"The search query string used to find relevant documents in the index. \
|
||||
This should contain keywords or phrases that best represent what the user is looking for. \
|
||||
More specific queries will yield more precise results.";
|
||||
/// The default description of the searchInIndex `index` parameter tool provided to OpenAI.
|
||||
const DEFAULT_SEARCH_IN_INDEX_INDEX_PARAMETER_TOOL_DESCRIPTION: &str =
|
||||
"The name of the index to search within. An index is a collection of documents organized for search. \
|
||||
Selecting the right index ensures the most relevant results for the user query";
|
||||
|
||||
const EMBEDDER_NAME: &str = "openai";
|
||||
|
||||
pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
cfg.service(web::resource("").route(web::post().to(chat)));
|
||||
}
|
||||
|
||||
/// Get a chat completion
|
||||
async fn chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_GET }>, Data<IndexScheduler>>,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
web::Json(mut chat_completion): web::Json<CreateChatCompletionRequest>,
|
||||
) -> impl Responder {
|
||||
// To enable later on, when the feature will be experimental
|
||||
// index_scheduler.features().check_chat("Using the /chat route")?;
|
||||
|
||||
if chat_completion.stream.unwrap_or(false) {
|
||||
Either::Right(streamed_chat(index_scheduler, search_queue, chat_completion).await)
|
||||
} else {
|
||||
Either::Left(non_streamed_chat(index_scheduler, search_queue, chat_completion).await)
|
||||
}
|
||||
}
|
||||
|
||||
async fn non_streamed_chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_GET }>, Data<IndexScheduler>>,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
mut chat_completion: CreateChatCompletionRequest,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
let api_key = std::env::var("MEILI_OPENAI_API_KEY")
|
||||
.expect("cannot find OpenAI API Key (MEILI_OPENAI_API_KEY)");
|
||||
let config = OpenAIConfig::default().with_api_key(&api_key); // we can also change the API base
|
||||
let client = Client::with_config(config);
|
||||
|
||||
assert_eq!(
|
||||
chat_completion.n.unwrap_or(1),
|
||||
1,
|
||||
"Meilisearch /chat only support one completion at a time (n = 1, n = null)"
|
||||
);
|
||||
|
||||
let rtxn = index_scheduler.read_txn().unwrap();
|
||||
let search_in_index_description = index_scheduler
|
||||
.chat_prompts(&rtxn, "searchInIndex-description")
|
||||
.unwrap()
|
||||
.unwrap_or(DEFAULT_SEARCH_IN_INDEX_TOOL_DESCRIPTION)
|
||||
.to_string();
|
||||
let search_in_index_q_param_description = index_scheduler
|
||||
.chat_prompts(&rtxn, "searchInIndex-q-param-description")
|
||||
.unwrap()
|
||||
.unwrap_or(DEFAULT_SEARCH_IN_INDEX_Q_PARAMETER_TOOL_DESCRIPTION)
|
||||
.to_string();
|
||||
let search_in_index_index_description = index_scheduler
|
||||
.chat_prompts(&rtxn, "searchInIndex-index-param-description")
|
||||
.unwrap()
|
||||
.unwrap_or(DEFAULT_SEARCH_IN_INDEX_INDEX_PARAMETER_TOOL_DESCRIPTION)
|
||||
.to_string();
|
||||
drop(rtxn);
|
||||
|
||||
let mut response;
|
||||
loop {
|
||||
let tools = chat_completion.tools.get_or_insert_default();
|
||||
tools.push(
|
||||
ChatCompletionToolArgs::default()
|
||||
.r#type(ChatCompletionToolType::Function)
|
||||
.function(
|
||||
FunctionObjectArgs::default()
|
||||
.name("searchInIndex")
|
||||
.description(&search_in_index_description)
|
||||
.parameters(json!({
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"index_uid": {
|
||||
"type": "string",
|
||||
"enum": ["main"],
|
||||
"description": search_in_index_index_description,
|
||||
},
|
||||
"q": {
|
||||
"type": ["string", "null"],
|
||||
"description": search_in_index_q_param_description,
|
||||
}
|
||||
},
|
||||
"required": ["index_uid", "q"],
|
||||
"additionalProperties": false,
|
||||
}))
|
||||
.strict(true)
|
||||
.build()
|
||||
.unwrap(),
|
||||
)
|
||||
.build()
|
||||
.unwrap(),
|
||||
);
|
||||
response = client.chat().create(chat_completion.clone()).await.unwrap();
|
||||
|
||||
let choice = &mut response.choices[0];
|
||||
match choice.finish_reason {
|
||||
Some(FinishReason::ToolCalls) => {
|
||||
let tool_calls = mem::take(&mut choice.message.tool_calls).unwrap_or_default();
|
||||
|
||||
let (meili_calls, other_calls): (Vec<_>, Vec<_>) =
|
||||
tool_calls.into_iter().partition(|call| call.function.name == "searchInIndex");
|
||||
|
||||
chat_completion.messages.push(
|
||||
ChatCompletionRequestAssistantMessageArgs::default()
|
||||
.tool_calls(meili_calls.clone())
|
||||
.build()
|
||||
.unwrap()
|
||||
.into(),
|
||||
);
|
||||
|
||||
for call in meili_calls {
|
||||
let SearchInIndexParameters { index_uid, q } =
|
||||
serde_json::from_str(&call.function.arguments).unwrap();
|
||||
|
||||
let mut query = SearchQuery {
|
||||
q,
|
||||
hybrid: Some(HybridQuery {
|
||||
semantic_ratio: SemanticRatio::default(),
|
||||
embedder: EMBEDDER_NAME.to_string(),
|
||||
}),
|
||||
limit: 20,
|
||||
..Default::default()
|
||||
};
|
||||
|
||||
// Tenant token search_rules.
|
||||
if let Some(search_rules) =
|
||||
index_scheduler.filters().get_index_search_rules(&index_uid)
|
||||
{
|
||||
add_search_rules(&mut query.filter, search_rules);
|
||||
}
|
||||
|
||||
// TBD
|
||||
// let mut aggregate = SearchAggregator::<SearchPOST>::from_query(&query);
|
||||
|
||||
let index = index_scheduler.index(&index_uid)?;
|
||||
let search_kind = search_kind(
|
||||
&query,
|
||||
index_scheduler.get_ref(),
|
||||
index_uid.to_string(),
|
||||
&index,
|
||||
)?;
|
||||
|
||||
let permit = search_queue.try_get_search_permit().await?;
|
||||
let features = index_scheduler.features();
|
||||
let index_cloned = index.clone();
|
||||
let search_result = tokio::task::spawn_blocking(move || {
|
||||
perform_search(
|
||||
index_uid.to_string(),
|
||||
&index_cloned,
|
||||
query,
|
||||
search_kind,
|
||||
RetrieveVectors::new(false),
|
||||
features,
|
||||
)
|
||||
})
|
||||
.await;
|
||||
permit.drop().await;
|
||||
|
||||
let search_result = search_result?;
|
||||
if let Ok(ref search_result) = search_result {
|
||||
// aggregate.succeed(search_result);
|
||||
if search_result.degraded {
|
||||
MEILISEARCH_DEGRADED_SEARCH_REQUESTS.inc();
|
||||
}
|
||||
}
|
||||
// analytics.publish(aggregate, &req);
|
||||
|
||||
let search_result = search_result?;
|
||||
let formatted = format_documents(
|
||||
&index,
|
||||
search_result.hits.into_iter().map(|doc| doc.document),
|
||||
);
|
||||
let text = formatted.join("\n");
|
||||
chat_completion.messages.push(ChatCompletionRequestMessage::Tool(
|
||||
ChatCompletionRequestToolMessage {
|
||||
tool_call_id: call.id,
|
||||
content: ChatCompletionRequestToolMessageContent::Text(text),
|
||||
},
|
||||
));
|
||||
}
|
||||
|
||||
// Let the client call other tools by themselves
|
||||
if !other_calls.is_empty() {
|
||||
response.choices[0].message.tool_calls = Some(other_calls);
|
||||
break;
|
||||
}
|
||||
}
|
||||
_ => break,
|
||||
}
|
||||
}
|
||||
|
||||
Ok(HttpResponse::Ok().json(response))
|
||||
}
|
||||
|
||||
async fn streamed_chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_GET }>, Data<IndexScheduler>>,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
mut chat_completion: CreateChatCompletionRequest,
|
||||
) -> impl Responder {
|
||||
assert!(chat_completion.stream.unwrap_or(false));
|
||||
|
||||
let api_key = std::env::var("MEILI_OPENAI_API_KEY")
|
||||
.expect("cannot find OpenAI API Key (MEILI_OPENAI_API_KEY)");
|
||||
let config = OpenAIConfig::default().with_api_key(&api_key); // we can also change the API base
|
||||
let client = Client::with_config(config);
|
||||
let response = client.chat().create_stream(chat_completion).await.unwrap();
|
||||
actix_web_lab::sse::Sse::from_stream(response.map(|response| {
|
||||
response
|
||||
.map(|mut r| Event::Data(sse::Data::new_json(r.choices.pop().unwrap().delta).unwrap()))
|
||||
}))
|
||||
}
|
||||
|
||||
#[derive(Deserialize)]
|
||||
struct SearchInIndexParameters {
|
||||
/// The index uid to search in.
|
||||
index_uid: String,
|
||||
/// The query parameter to use.
|
||||
q: Option<String>,
|
||||
}
|
||||
|
||||
fn format_documents(index: &Index, documents: impl Iterator<Item = Document>) -> Vec<String> {
|
||||
let rtxn = index.read_txn().unwrap();
|
||||
let IndexEmbeddingConfig { name: _, config, user_provided: _ } = index
|
||||
.embedding_configs(&rtxn)
|
||||
.unwrap()
|
||||
.into_iter()
|
||||
.find(|conf| conf.name == EMBEDDER_NAME)
|
||||
.unwrap();
|
||||
|
||||
let EmbeddingConfig {
|
||||
embedder_options: _,
|
||||
prompt: PromptData { template, max_bytes },
|
||||
quantized: _,
|
||||
} = config;
|
||||
|
||||
#[derive(Serialize)]
|
||||
struct Doc<T: Serialize> {
|
||||
doc: T,
|
||||
}
|
||||
|
||||
let template = liquid::ParserBuilder::with_stdlib().build().unwrap().parse(&template).unwrap();
|
||||
documents
|
||||
.map(|doc| {
|
||||
let object = liquid::to_object(&Doc { doc }).unwrap();
|
||||
template.render(&object).unwrap()
|
||||
})
|
||||
.collect()
|
||||
}
|
||||
710
crates/meilisearch/src/routes/chats/chat_completions.rs
Normal file
710
crates/meilisearch/src/routes/chats/chat_completions.rs
Normal file
@@ -0,0 +1,710 @@
|
||||
use std::collections::HashMap;
|
||||
use std::fmt::Write as _;
|
||||
use std::mem;
|
||||
use std::ops::ControlFlow;
|
||||
use std::time::Duration;
|
||||
|
||||
use actix_web::web::{self, Data};
|
||||
use actix_web::{Either, HttpRequest, HttpResponse, Responder};
|
||||
use actix_web_lab::sse::{Event, Sse};
|
||||
use async_openai::config::{Config, OpenAIConfig};
|
||||
use async_openai::types::{
|
||||
ChatCompletionMessageToolCall, ChatCompletionMessageToolCallChunk,
|
||||
ChatCompletionRequestAssistantMessageArgs, ChatCompletionRequestMessage,
|
||||
ChatCompletionRequestSystemMessage, ChatCompletionRequestSystemMessageContent,
|
||||
ChatCompletionRequestToolMessage, ChatCompletionRequestToolMessageContent,
|
||||
ChatCompletionStreamResponseDelta, ChatCompletionToolArgs, ChatCompletionToolType,
|
||||
CreateChatCompletionRequest, CreateChatCompletionStreamResponse, FinishReason, FunctionCall,
|
||||
FunctionCallStream, FunctionObjectArgs,
|
||||
};
|
||||
use async_openai::Client;
|
||||
use bumpalo::Bump;
|
||||
use futures::StreamExt;
|
||||
use index_scheduler::IndexScheduler;
|
||||
use meilisearch_auth::AuthController;
|
||||
use meilisearch_types::error::{Code, ResponseError};
|
||||
use meilisearch_types::features::{
|
||||
ChatCompletionPrompts as DbChatCompletionPrompts, ChatCompletionSettings as DbChatSettings,
|
||||
};
|
||||
use meilisearch_types::keys::actions;
|
||||
use meilisearch_types::milli::index::ChatConfig;
|
||||
use meilisearch_types::milli::{all_obkv_to_json, obkv_to_json, TimeBudget};
|
||||
use meilisearch_types::{Document, Index};
|
||||
use serde::Deserialize;
|
||||
use serde_json::json;
|
||||
use tokio::runtime::Handle;
|
||||
use tokio::sync::mpsc::error::SendError;
|
||||
|
||||
use super::errors::StreamErrorEvent;
|
||||
use super::utils::format_documents;
|
||||
use super::{
|
||||
ChatsParam, MEILI_APPEND_CONVERSATION_MESSAGE_NAME, MEILI_SEARCH_IN_INDEX_FUNCTION_NAME,
|
||||
MEILI_SEARCH_PROGRESS_NAME, MEILI_SEARCH_SOURCES_NAME,
|
||||
};
|
||||
use crate::error::MeilisearchHttpError;
|
||||
use crate::extractors::authentication::policies::ActionPolicy;
|
||||
use crate::extractors::authentication::{extract_token_from_request, GuardedData, Policy as _};
|
||||
use crate::metrics::MEILISEARCH_DEGRADED_SEARCH_REQUESTS;
|
||||
use crate::routes::chats::utils::SseEventSender;
|
||||
use crate::routes::indexes::search::search_kind;
|
||||
use crate::search::{add_search_rules, prepare_search, search_from_kind, SearchQuery};
|
||||
use crate::search_queue::SearchQueue;
|
||||
|
||||
pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
cfg.service(web::resource("").route(web::post().to(chat)));
|
||||
}
|
||||
|
||||
/// Get a chat completion
|
||||
async fn chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_COMPLETIONS }>, Data<IndexScheduler>>,
|
||||
auth_ctrl: web::Data<AuthController>,
|
||||
chats_param: web::Path<ChatsParam>,
|
||||
req: HttpRequest,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
web::Json(chat_completion): web::Json<CreateChatCompletionRequest>,
|
||||
) -> impl Responder {
|
||||
let ChatsParam { workspace_uid } = chats_param.into_inner();
|
||||
|
||||
assert_eq!(
|
||||
chat_completion.n.unwrap_or(1),
|
||||
1,
|
||||
"Meilisearch /chat only support one completion at a time (n = 1, n = null)"
|
||||
);
|
||||
|
||||
if chat_completion.stream.unwrap_or(false) {
|
||||
Either::Right(
|
||||
streamed_chat(
|
||||
index_scheduler,
|
||||
auth_ctrl,
|
||||
search_queue,
|
||||
&workspace_uid,
|
||||
req,
|
||||
chat_completion,
|
||||
)
|
||||
.await,
|
||||
)
|
||||
} else {
|
||||
Either::Left(
|
||||
non_streamed_chat(
|
||||
index_scheduler,
|
||||
auth_ctrl,
|
||||
search_queue,
|
||||
&workspace_uid,
|
||||
req,
|
||||
chat_completion,
|
||||
)
|
||||
.await,
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Default, Debug, Clone, Copy)]
|
||||
pub struct FunctionSupport {
|
||||
/// Defines if we can call the _meiliSearchProgress function
|
||||
/// to inform the front-end about what we are searching for.
|
||||
report_progress: bool,
|
||||
/// Defines if we can call the _meiliSearchSources function
|
||||
/// to inform the front-end about the sources of the search.
|
||||
report_sources: bool,
|
||||
/// Defines if we can call the _meiliAppendConversationMessage
|
||||
/// function to provide the messages to append into the conversation.
|
||||
append_to_conversation: bool,
|
||||
}
|
||||
|
||||
/// Setup search tool in chat completion request
|
||||
fn setup_search_tool(
|
||||
index_scheduler: &Data<IndexScheduler>,
|
||||
filters: &meilisearch_auth::AuthFilter,
|
||||
chat_completion: &mut CreateChatCompletionRequest,
|
||||
prompts: &DbChatCompletionPrompts,
|
||||
) -> Result<FunctionSupport, ResponseError> {
|
||||
let tools = chat_completion.tools.get_or_insert_default();
|
||||
if tools.iter().any(|t| t.function.name == MEILI_SEARCH_IN_INDEX_FUNCTION_NAME) {
|
||||
panic!("{MEILI_SEARCH_IN_INDEX_FUNCTION_NAME} function already set");
|
||||
}
|
||||
|
||||
// Remove internal tools used for front-end notifications as they should be hidden from the LLM.
|
||||
let mut report_progress = false;
|
||||
let mut report_sources = false;
|
||||
let mut append_to_conversation = false;
|
||||
tools.retain(|tool| {
|
||||
match tool.function.name.as_str() {
|
||||
MEILI_SEARCH_PROGRESS_NAME => {
|
||||
report_progress = true;
|
||||
false
|
||||
}
|
||||
MEILI_SEARCH_SOURCES_NAME => {
|
||||
report_sources = true;
|
||||
false
|
||||
}
|
||||
MEILI_APPEND_CONVERSATION_MESSAGE_NAME => {
|
||||
append_to_conversation = true;
|
||||
false
|
||||
}
|
||||
_ => true, // keep other tools
|
||||
}
|
||||
});
|
||||
|
||||
let mut index_uids = Vec::new();
|
||||
let mut function_description = prompts.search_description.clone();
|
||||
index_scheduler.try_for_each_index::<_, ()>(|name, index| {
|
||||
// Make sure to skip unauthorized indexes
|
||||
if !filters.is_index_authorized(name) {
|
||||
return Ok(());
|
||||
}
|
||||
|
||||
let rtxn = index.read_txn()?;
|
||||
let chat_config = index.chat_config(&rtxn)?;
|
||||
let index_description = chat_config.description;
|
||||
let _ = writeln!(&mut function_description, "\n\n - {name}: {index_description}\n");
|
||||
index_uids.push(name.to_string());
|
||||
|
||||
Ok(())
|
||||
})?;
|
||||
|
||||
let tool = ChatCompletionToolArgs::default()
|
||||
.r#type(ChatCompletionToolType::Function)
|
||||
.function(
|
||||
FunctionObjectArgs::default()
|
||||
.name(MEILI_SEARCH_IN_INDEX_FUNCTION_NAME)
|
||||
.description(&function_description)
|
||||
.parameters(json!({
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"index_uid": {
|
||||
"type": "string",
|
||||
"enum": index_uids,
|
||||
"description": prompts.search_index_uid_param,
|
||||
},
|
||||
"q": {
|
||||
// Unfortunately, Mistral does not support an array of types, here.
|
||||
// "type": ["string", "null"],
|
||||
"type": "string",
|
||||
"description": prompts.search_q_param,
|
||||
}
|
||||
},
|
||||
"required": ["index_uid", "q"],
|
||||
"additionalProperties": false,
|
||||
}))
|
||||
.strict(true)
|
||||
.build()
|
||||
.unwrap(),
|
||||
)
|
||||
.build()
|
||||
.unwrap();
|
||||
|
||||
tools.push(tool);
|
||||
|
||||
chat_completion.messages.insert(
|
||||
0,
|
||||
ChatCompletionRequestMessage::System(ChatCompletionRequestSystemMessage {
|
||||
content: ChatCompletionRequestSystemMessageContent::Text(prompts.system.clone()),
|
||||
name: None,
|
||||
}),
|
||||
);
|
||||
|
||||
Ok(FunctionSupport { report_progress, report_sources, append_to_conversation })
|
||||
}
|
||||
|
||||
/// Process search request and return formatted results
|
||||
async fn process_search_request(
|
||||
index_scheduler: &GuardedData<
|
||||
ActionPolicy<{ actions::CHAT_COMPLETIONS }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
auth_ctrl: web::Data<AuthController>,
|
||||
search_queue: &web::Data<SearchQueue>,
|
||||
auth_token: &str,
|
||||
index_uid: String,
|
||||
q: Option<String>,
|
||||
) -> Result<(Index, Vec<Document>, String), ResponseError> {
|
||||
// TBD
|
||||
// let mut aggregate = SearchAggregator::<SearchPOST>::from_query(&query);
|
||||
|
||||
let index = index_scheduler.index(&index_uid)?;
|
||||
let rtxn = index.static_read_txn()?;
|
||||
let ChatConfig { description: _, prompt: _, search_parameters } = index.chat_config(&rtxn)?;
|
||||
let mut query = SearchQuery { q, ..SearchQuery::from(search_parameters) };
|
||||
let auth_filter = ActionPolicy::<{ actions::SEARCH }>::authenticate(
|
||||
auth_ctrl,
|
||||
auth_token,
|
||||
Some(index_uid.as_str()),
|
||||
)?;
|
||||
|
||||
// Tenant token search_rules.
|
||||
if let Some(search_rules) = auth_filter.get_index_search_rules(&index_uid) {
|
||||
add_search_rules(&mut query.filter, search_rules);
|
||||
}
|
||||
let search_kind =
|
||||
search_kind(&query, index_scheduler.get_ref(), index_uid.to_string(), &index)?;
|
||||
|
||||
let permit = search_queue.try_get_search_permit().await?;
|
||||
let features = index_scheduler.features();
|
||||
let index_cloned = index.clone();
|
||||
let output = tokio::task::spawn_blocking(move || -> Result<_, ResponseError> {
|
||||
let time_budget = match index_cloned
|
||||
.search_cutoff(&rtxn)
|
||||
.map_err(|e| MeilisearchHttpError::from_milli(e, Some(index_uid.clone())))?
|
||||
{
|
||||
Some(cutoff) => TimeBudget::new(Duration::from_millis(cutoff)),
|
||||
None => TimeBudget::default(),
|
||||
};
|
||||
|
||||
let (search, _is_finite_pagination, _max_total_hits, _offset) =
|
||||
prepare_search(&index_cloned, &rtxn, &query, &search_kind, time_budget, features)?;
|
||||
|
||||
search_from_kind(index_uid, search_kind, search)
|
||||
.map(|(search_results, _)| (rtxn, search_results))
|
||||
.map_err(ResponseError::from)
|
||||
})
|
||||
.await;
|
||||
permit.drop().await;
|
||||
|
||||
let output = output?;
|
||||
let mut documents = Vec::new();
|
||||
if let Ok((ref rtxn, ref search_result)) = output {
|
||||
// aggregate.succeed(search_result);
|
||||
if search_result.degraded {
|
||||
MEILISEARCH_DEGRADED_SEARCH_REQUESTS.inc();
|
||||
}
|
||||
|
||||
let fields_ids_map = index.fields_ids_map(rtxn)?;
|
||||
let displayed_fields = index.displayed_fields_ids(rtxn)?;
|
||||
for &document_id in &search_result.documents_ids {
|
||||
let obkv = index.document(rtxn, document_id)?;
|
||||
let document = match displayed_fields {
|
||||
Some(ref fields) => obkv_to_json(fields, &fields_ids_map, obkv)?,
|
||||
None => all_obkv_to_json(obkv, &fields_ids_map)?,
|
||||
};
|
||||
documents.push(document);
|
||||
}
|
||||
}
|
||||
// analytics.publish(aggregate, &req);
|
||||
|
||||
let (rtxn, search_result) = output?;
|
||||
let render_alloc = Bump::new();
|
||||
let formatted = format_documents(&rtxn, &index, &render_alloc, search_result.documents_ids)?;
|
||||
let text = formatted.join("\n");
|
||||
drop(rtxn);
|
||||
|
||||
Ok((index, documents, text))
|
||||
}
|
||||
|
||||
async fn non_streamed_chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_COMPLETIONS }>, Data<IndexScheduler>>,
|
||||
auth_ctrl: web::Data<AuthController>,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
workspace_uid: &str,
|
||||
req: HttpRequest,
|
||||
mut chat_completion: CreateChatCompletionRequest,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats chat completions route")?;
|
||||
let filters = index_scheduler.filters();
|
||||
|
||||
let rtxn = index_scheduler.read_txn()?;
|
||||
let chat_settings = match index_scheduler.chat_settings(&rtxn, workspace_uid).unwrap() {
|
||||
Some(settings) => settings,
|
||||
None => {
|
||||
return Err(ResponseError::from_msg(
|
||||
format!("Chat `{workspace_uid}` not found"),
|
||||
Code::ChatWorkspaceNotFound,
|
||||
))
|
||||
}
|
||||
};
|
||||
|
||||
let mut config = OpenAIConfig::default();
|
||||
if let Some(api_key) = chat_settings.api_key.as_ref() {
|
||||
config = config.with_api_key(api_key);
|
||||
}
|
||||
if let Some(base_api) = chat_settings.base_api.as_ref() {
|
||||
config = config.with_api_base(base_api);
|
||||
}
|
||||
let client = Client::with_config(config);
|
||||
|
||||
let auth_token = extract_token_from_request(&req)?.unwrap();
|
||||
// TODO do function support later
|
||||
let _function_support =
|
||||
setup_search_tool(&index_scheduler, filters, &mut chat_completion, &chat_settings.prompts)?;
|
||||
|
||||
let mut response;
|
||||
loop {
|
||||
response = client.chat().create(chat_completion.clone()).await.unwrap();
|
||||
|
||||
let choice = &mut response.choices[0];
|
||||
match choice.finish_reason {
|
||||
Some(FinishReason::ToolCalls) => {
|
||||
let tool_calls = mem::take(&mut choice.message.tool_calls).unwrap_or_default();
|
||||
|
||||
let (meili_calls, other_calls): (Vec<_>, Vec<_>) = tool_calls
|
||||
.into_iter()
|
||||
.partition(|call| call.function.name == MEILI_SEARCH_IN_INDEX_FUNCTION_NAME);
|
||||
|
||||
chat_completion.messages.push(
|
||||
ChatCompletionRequestAssistantMessageArgs::default()
|
||||
.tool_calls(meili_calls.clone())
|
||||
.build()
|
||||
.unwrap()
|
||||
.into(),
|
||||
);
|
||||
|
||||
for call in meili_calls {
|
||||
let result = match serde_json::from_str(&call.function.arguments) {
|
||||
Ok(SearchInIndexParameters { index_uid, q }) => process_search_request(
|
||||
&index_scheduler,
|
||||
auth_ctrl.clone(),
|
||||
&search_queue,
|
||||
auth_token,
|
||||
index_uid,
|
||||
q,
|
||||
)
|
||||
.await
|
||||
.map_err(|e| e.to_string()),
|
||||
Err(err) => Err(err.to_string()),
|
||||
};
|
||||
|
||||
// TODO report documents sources later
|
||||
let text = match result {
|
||||
Ok((_, _documents, text)) => text,
|
||||
Err(err) => err,
|
||||
};
|
||||
|
||||
let answer = format!("{}\n\n{text}", chat_settings.prompts.pre_query);
|
||||
chat_completion.messages.push(ChatCompletionRequestMessage::Tool(
|
||||
ChatCompletionRequestToolMessage {
|
||||
tool_call_id: call.id.clone(),
|
||||
content: ChatCompletionRequestToolMessageContent::Text(answer),
|
||||
},
|
||||
));
|
||||
}
|
||||
|
||||
// Let the client call other tools by themselves
|
||||
if !other_calls.is_empty() {
|
||||
response.choices[0].message.tool_calls = Some(other_calls);
|
||||
break;
|
||||
}
|
||||
}
|
||||
_ => break,
|
||||
}
|
||||
}
|
||||
|
||||
Ok(HttpResponse::Ok().json(response))
|
||||
}
|
||||
|
||||
async fn streamed_chat(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHAT_COMPLETIONS }>, Data<IndexScheduler>>,
|
||||
auth_ctrl: web::Data<AuthController>,
|
||||
search_queue: web::Data<SearchQueue>,
|
||||
workspace_uid: &str,
|
||||
req: HttpRequest,
|
||||
mut chat_completion: CreateChatCompletionRequest,
|
||||
) -> Result<impl Responder, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats chat completions route")?;
|
||||
let filters = index_scheduler.filters();
|
||||
|
||||
let rtxn = index_scheduler.read_txn()?;
|
||||
let chat_settings = match index_scheduler.chat_settings(&rtxn, workspace_uid)? {
|
||||
Some(settings) => settings,
|
||||
None => {
|
||||
return Err(ResponseError::from_msg(
|
||||
format!("Chat `{workspace_uid}` not found"),
|
||||
Code::ChatWorkspaceNotFound,
|
||||
))
|
||||
}
|
||||
};
|
||||
drop(rtxn);
|
||||
|
||||
let mut config = OpenAIConfig::default();
|
||||
if let Some(api_key) = chat_settings.api_key.as_ref() {
|
||||
config = config.with_api_key(api_key);
|
||||
}
|
||||
if let Some(base_api) = chat_settings.base_api.as_ref() {
|
||||
config = config.with_api_base(base_api);
|
||||
}
|
||||
|
||||
let auth_token = extract_token_from_request(&req)?.unwrap().to_string();
|
||||
let function_support =
|
||||
setup_search_tool(&index_scheduler, filters, &mut chat_completion, &chat_settings.prompts)?;
|
||||
|
||||
tracing::debug!("Conversation function support: {function_support:?}");
|
||||
|
||||
let (tx, rx) = tokio::sync::mpsc::channel(10);
|
||||
let tx = SseEventSender::new(tx);
|
||||
let _join_handle = Handle::current().spawn(async move {
|
||||
let client = Client::with_config(config.clone());
|
||||
let mut global_tool_calls = HashMap::<u32, Call>::new();
|
||||
|
||||
// Limit the number of internal calls to satisfy the search requests of the LLM
|
||||
for _ in 0..20 {
|
||||
let output = run_conversation(
|
||||
&index_scheduler,
|
||||
&auth_ctrl,
|
||||
&search_queue,
|
||||
&auth_token,
|
||||
&client,
|
||||
&chat_settings,
|
||||
&mut chat_completion,
|
||||
&tx,
|
||||
&mut global_tool_calls,
|
||||
function_support,
|
||||
);
|
||||
|
||||
match output.await {
|
||||
Ok(ControlFlow::Continue(())) => (),
|
||||
Ok(ControlFlow::Break(_finish_reason)) => break,
|
||||
// If the connection is closed we must stop
|
||||
Err(SendError(_)) => return,
|
||||
}
|
||||
}
|
||||
|
||||
let _ = tx.stop().await;
|
||||
});
|
||||
|
||||
Ok(Sse::from_infallible_receiver(rx).with_retry_duration(Duration::from_secs(10)))
|
||||
}
|
||||
|
||||
/// Updates the chat completion with the new messages, streams the LLM tokens,
|
||||
/// and report progress and errors.
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
async fn run_conversation<C: Config>(
|
||||
index_scheduler: &GuardedData<
|
||||
ActionPolicy<{ actions::CHAT_COMPLETIONS }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
auth_ctrl: &web::Data<AuthController>,
|
||||
search_queue: &web::Data<SearchQueue>,
|
||||
auth_token: &str,
|
||||
client: &Client<C>,
|
||||
chat_settings: &DbChatSettings,
|
||||
chat_completion: &mut CreateChatCompletionRequest,
|
||||
tx: &SseEventSender,
|
||||
global_tool_calls: &mut HashMap<u32, Call>,
|
||||
function_support: FunctionSupport,
|
||||
) -> Result<ControlFlow<Option<FinishReason>, ()>, SendError<Event>> {
|
||||
let mut finish_reason = None;
|
||||
// safety: The unwrap can only happen if the stream is not correctly configured.
|
||||
let mut response = client.chat().create_stream(chat_completion.clone()).await.unwrap();
|
||||
while let Some(result) = response.next().await {
|
||||
match result {
|
||||
Ok(resp) => {
|
||||
let choice = &resp.choices[0];
|
||||
finish_reason = choice.finish_reason;
|
||||
|
||||
let ChatCompletionStreamResponseDelta { ref tool_calls, .. } = &choice.delta;
|
||||
|
||||
match tool_calls {
|
||||
Some(tool_calls) => {
|
||||
for chunk in tool_calls {
|
||||
let ChatCompletionMessageToolCallChunk {
|
||||
index,
|
||||
id,
|
||||
r#type: _,
|
||||
function,
|
||||
} = chunk;
|
||||
let FunctionCallStream { name, arguments } = function.as_ref().unwrap();
|
||||
|
||||
global_tool_calls
|
||||
.entry(*index)
|
||||
.and_modify(|call| {
|
||||
if call.is_internal() {
|
||||
call.append(arguments.as_ref().unwrap())
|
||||
}
|
||||
})
|
||||
.or_insert_with(|| {
|
||||
if name.as_deref() == Some(MEILI_SEARCH_IN_INDEX_FUNCTION_NAME)
|
||||
{
|
||||
Call::Internal {
|
||||
id: id.as_ref().unwrap().clone(),
|
||||
function_name: name.as_ref().unwrap().clone(),
|
||||
arguments: arguments.as_ref().unwrap().clone(),
|
||||
}
|
||||
} else {
|
||||
Call::External
|
||||
}
|
||||
});
|
||||
|
||||
if global_tool_calls.get(index).is_some_and(Call::is_external) {
|
||||
todo!("Support forwarding external tool calls");
|
||||
}
|
||||
}
|
||||
}
|
||||
None => {
|
||||
if !global_tool_calls.is_empty() {
|
||||
let (meili_calls, other_calls): (Vec<_>, Vec<_>) =
|
||||
mem::take(global_tool_calls)
|
||||
.into_values()
|
||||
.flat_map(|call| match call {
|
||||
Call::Internal { id, function_name: name, arguments } => {
|
||||
Some(ChatCompletionMessageToolCall {
|
||||
id,
|
||||
r#type: Some(ChatCompletionToolType::Function),
|
||||
function: FunctionCall { name, arguments },
|
||||
})
|
||||
}
|
||||
Call::External => None,
|
||||
})
|
||||
.partition(|call| {
|
||||
call.function.name == MEILI_SEARCH_IN_INDEX_FUNCTION_NAME
|
||||
});
|
||||
|
||||
chat_completion.messages.push(
|
||||
ChatCompletionRequestAssistantMessageArgs::default()
|
||||
.tool_calls(meili_calls.clone())
|
||||
.build()
|
||||
.unwrap()
|
||||
.into(),
|
||||
);
|
||||
|
||||
assert!(
|
||||
other_calls.is_empty(),
|
||||
"We do not support external tool forwarding for now"
|
||||
);
|
||||
|
||||
handle_meili_tools(
|
||||
index_scheduler,
|
||||
auth_ctrl,
|
||||
search_queue,
|
||||
auth_token,
|
||||
chat_settings,
|
||||
tx,
|
||||
meili_calls,
|
||||
chat_completion,
|
||||
&resp,
|
||||
function_support,
|
||||
)
|
||||
.await?;
|
||||
} else {
|
||||
tx.forward_response(&resp).await?;
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
Err(error) => {
|
||||
let error = StreamErrorEvent::from_openai_error(error).await.unwrap();
|
||||
tx.send_error(&error).await?;
|
||||
return Ok(ControlFlow::Break(None));
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// We must stop if the finish reason is not something we can solve with Meilisearch
|
||||
match finish_reason {
|
||||
Some(FinishReason::ToolCalls) => Ok(ControlFlow::Continue(())),
|
||||
otherwise => Ok(ControlFlow::Break(otherwise)),
|
||||
}
|
||||
}
|
||||
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
async fn handle_meili_tools(
|
||||
index_scheduler: &GuardedData<
|
||||
ActionPolicy<{ actions::CHAT_COMPLETIONS }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
auth_ctrl: &web::Data<AuthController>,
|
||||
search_queue: &web::Data<SearchQueue>,
|
||||
auth_token: &str,
|
||||
chat_settings: &DbChatSettings,
|
||||
tx: &SseEventSender,
|
||||
meili_calls: Vec<ChatCompletionMessageToolCall>,
|
||||
chat_completion: &mut CreateChatCompletionRequest,
|
||||
resp: &CreateChatCompletionStreamResponse,
|
||||
FunctionSupport { report_progress, report_sources, append_to_conversation, .. }: FunctionSupport,
|
||||
) -> Result<(), SendError<Event>> {
|
||||
for call in meili_calls {
|
||||
if report_progress {
|
||||
tx.report_search_progress(
|
||||
resp.clone(),
|
||||
&call.id,
|
||||
&call.function.name,
|
||||
&call.function.arguments,
|
||||
)
|
||||
.await?;
|
||||
}
|
||||
|
||||
if append_to_conversation {
|
||||
tx.append_tool_call_conversation_message(
|
||||
resp.clone(),
|
||||
call.id.clone(),
|
||||
call.function.name.clone(),
|
||||
call.function.arguments.clone(),
|
||||
)
|
||||
.await?;
|
||||
}
|
||||
|
||||
let result = match serde_json::from_str(&call.function.arguments) {
|
||||
Ok(SearchInIndexParameters { index_uid, q }) => process_search_request(
|
||||
index_scheduler,
|
||||
auth_ctrl.clone(),
|
||||
search_queue,
|
||||
auth_token,
|
||||
index_uid,
|
||||
q,
|
||||
)
|
||||
.await
|
||||
.map_err(|e| e.to_string()),
|
||||
Err(err) => Err(err.to_string()),
|
||||
};
|
||||
|
||||
let text = match result {
|
||||
Ok((_index, documents, text)) => {
|
||||
if report_sources {
|
||||
tx.report_sources(resp.clone(), &call.id, &documents).await?;
|
||||
}
|
||||
|
||||
text
|
||||
}
|
||||
Err(err) => err,
|
||||
};
|
||||
|
||||
let answer = format!("{}\n\n{text}", chat_settings.prompts.pre_query);
|
||||
let tool = ChatCompletionRequestMessage::Tool(ChatCompletionRequestToolMessage {
|
||||
tool_call_id: call.id.clone(),
|
||||
content: ChatCompletionRequestToolMessageContent::Text(answer),
|
||||
});
|
||||
|
||||
if append_to_conversation {
|
||||
tx.append_conversation_message(resp.clone(), &tool).await?;
|
||||
}
|
||||
|
||||
chat_completion.messages.push(tool);
|
||||
}
|
||||
|
||||
Ok(())
|
||||
}
|
||||
|
||||
/// The structure used to aggregate the function calls to make.
|
||||
#[derive(Debug)]
|
||||
enum Call {
|
||||
/// Tool calls to tools that must be managed by Meilisearch internally.
|
||||
/// Typically the search functions.
|
||||
Internal { id: String, function_name: String, arguments: String },
|
||||
/// Tool calls that we track but only to know that its not our functions.
|
||||
/// We return the function calls as-is to the end-user.
|
||||
External,
|
||||
}
|
||||
|
||||
impl Call {
|
||||
fn is_internal(&self) -> bool {
|
||||
matches!(self, Call::Internal { .. })
|
||||
}
|
||||
|
||||
fn is_external(&self) -> bool {
|
||||
matches!(self, Call::External { .. })
|
||||
}
|
||||
|
||||
fn append(&mut self, more: &str) {
|
||||
match self {
|
||||
Call::Internal { arguments, .. } => arguments.push_str(more),
|
||||
Call::External { .. } => {
|
||||
panic!("Cannot append argument chunks to an external function")
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Deserialize)]
|
||||
struct SearchInIndexParameters {
|
||||
/// The index uid to search in.
|
||||
index_uid: String,
|
||||
/// The query parameter to use.
|
||||
q: Option<String>,
|
||||
}
|
||||
187
crates/meilisearch/src/routes/chats/errors.rs
Normal file
187
crates/meilisearch/src/routes/chats/errors.rs
Normal file
@@ -0,0 +1,187 @@
|
||||
use async_openai::error::{ApiError, OpenAIError};
|
||||
use async_openai::reqwest_eventsource::Error as EventSourceError;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use uuid::Uuid;
|
||||
|
||||
#[derive(Debug, Clone, Deserialize)]
|
||||
pub struct OpenAiOutsideError {
|
||||
/// Emitted when an error occurs.
|
||||
error: OpenAiInnerError,
|
||||
}
|
||||
|
||||
/// Emitted when an error occurs.
|
||||
#[derive(Debug, Clone, Deserialize)]
|
||||
pub struct OpenAiInnerError {
|
||||
/// The error code.
|
||||
code: Option<String>,
|
||||
/// The error message.
|
||||
message: String,
|
||||
/// The error parameter.
|
||||
param: Option<String>,
|
||||
/// The type of the event. Always `error`.
|
||||
r#type: String,
|
||||
}
|
||||
|
||||
/// An error that occurs during the streaming process.
|
||||
///
|
||||
/// It directly comes from the OpenAI API and you can
|
||||
/// read more about error events on their website:
|
||||
/// <https://platform.openai.com/docs/api-reference/realtime-server-events/error>
|
||||
#[derive(Debug, Serialize, Deserialize)]
|
||||
pub struct StreamErrorEvent {
|
||||
/// The unique ID of the server event.
|
||||
pub event_id: String,
|
||||
/// The event type, must be error.
|
||||
pub r#type: String,
|
||||
/// Details of the error.
|
||||
pub error: StreamError,
|
||||
}
|
||||
|
||||
/// Details of the error.
|
||||
#[derive(Debug, Serialize, Deserialize)]
|
||||
pub struct StreamError {
|
||||
/// The type of error (e.g., "invalid_request_error", "server_error").
|
||||
pub r#type: String,
|
||||
/// Error code, if any.
|
||||
pub code: Option<String>,
|
||||
/// A human-readable error message.
|
||||
pub message: String,
|
||||
/// Parameter related to the error, if any.
|
||||
pub param: Option<String>,
|
||||
/// The event_id of the client event that caused the error, if applicable.
|
||||
pub event_id: Option<String>,
|
||||
}
|
||||
|
||||
impl StreamErrorEvent {
|
||||
pub async fn from_openai_error(error: OpenAIError) -> Result<Self, reqwest::Error> {
|
||||
let error_type = "error".to_string();
|
||||
match error {
|
||||
OpenAIError::Reqwest(e) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "internal_reqwest_error".to_string(),
|
||||
code: Some("internal".to_string()),
|
||||
message: e.to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
OpenAIError::ApiError(ApiError { message, r#type, param, code }) => {
|
||||
Ok(StreamErrorEvent {
|
||||
r#type: error_type,
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
error: StreamError {
|
||||
r#type: r#type.unwrap_or_else(|| "unknown".to_string()),
|
||||
code,
|
||||
message,
|
||||
param,
|
||||
event_id: None,
|
||||
},
|
||||
})
|
||||
}
|
||||
OpenAIError::JSONDeserialize(error) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "json_deserialize_error".to_string(),
|
||||
code: Some("internal".to_string()),
|
||||
message: error.to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
OpenAIError::FileSaveError(_) | OpenAIError::FileReadError(_) => unreachable!(),
|
||||
OpenAIError::StreamError(error) => match error {
|
||||
EventSourceError::InvalidStatusCode(_status_code, response) => {
|
||||
let OpenAiOutsideError {
|
||||
error: OpenAiInnerError { code, message, param, r#type },
|
||||
} = response.json().await?;
|
||||
|
||||
Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError { r#type, code, message, param, event_id: None },
|
||||
})
|
||||
}
|
||||
EventSourceError::InvalidContentType(_header_value, response) => {
|
||||
let OpenAiOutsideError {
|
||||
error: OpenAiInnerError { code, message, param, r#type },
|
||||
} = response.json().await?;
|
||||
|
||||
Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError { r#type, code, message, param, event_id: None },
|
||||
})
|
||||
}
|
||||
EventSourceError::Utf8(error) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "invalid_utf8_error".to_string(),
|
||||
code: None,
|
||||
message: error.to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
EventSourceError::Parser(error) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "parser_error".to_string(),
|
||||
code: None,
|
||||
message: error.to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
EventSourceError::Transport(error) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "transport_error".to_string(),
|
||||
code: None,
|
||||
message: error.to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
EventSourceError::InvalidLastEventId(message) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "invalid_last_event_id".to_string(),
|
||||
code: None,
|
||||
message,
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
EventSourceError::StreamEnded => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "stream_ended".to_string(),
|
||||
code: None,
|
||||
message: "Stream ended".to_string(),
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
},
|
||||
OpenAIError::InvalidArgument(message) => Ok(StreamErrorEvent {
|
||||
event_id: Uuid::new_v4().to_string(),
|
||||
r#type: error_type,
|
||||
error: StreamError {
|
||||
r#type: "invalid_argument".to_string(),
|
||||
code: None,
|
||||
message,
|
||||
param: None,
|
||||
event_id: None,
|
||||
},
|
||||
}),
|
||||
}
|
||||
}
|
||||
}
|
||||
93
crates/meilisearch/src/routes/chats/mod.rs
Normal file
93
crates/meilisearch/src/routes/chats/mod.rs
Normal file
@@ -0,0 +1,93 @@
|
||||
use actix_web::web::{self, Data};
|
||||
use actix_web::HttpResponse;
|
||||
use deserr::actix_web::AwebQueryParameter;
|
||||
use deserr::Deserr;
|
||||
use index_scheduler::IndexScheduler;
|
||||
use meilisearch_types::deserr::query_params::Param;
|
||||
use meilisearch_types::deserr::DeserrQueryParamError;
|
||||
use meilisearch_types::error::deserr_codes::{InvalidIndexLimit, InvalidIndexOffset};
|
||||
use meilisearch_types::error::ResponseError;
|
||||
use meilisearch_types::keys::actions;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use tracing::debug;
|
||||
use utoipa::{IntoParams, ToSchema};
|
||||
|
||||
use super::Pagination;
|
||||
use crate::extractors::authentication::policies::ActionPolicy;
|
||||
use crate::extractors::authentication::GuardedData;
|
||||
use crate::routes::PAGINATION_DEFAULT_LIMIT;
|
||||
|
||||
pub mod chat_completions;
|
||||
mod errors;
|
||||
pub mod settings;
|
||||
mod utils;
|
||||
|
||||
/// The function name to report search progress.
|
||||
const MEILI_SEARCH_PROGRESS_NAME: &str = "_meiliSearchProgress";
|
||||
/// The function name to append a conversation message in the user conversation.
|
||||
const MEILI_APPEND_CONVERSATION_MESSAGE_NAME: &str = "_meiliAppendConversationMessage";
|
||||
/// The function name to report sources to the frontend.
|
||||
const MEILI_SEARCH_SOURCES_NAME: &str = "_meiliSearchSources";
|
||||
/// The *internal* function name to provide to the LLM to search in indexes.
|
||||
const MEILI_SEARCH_IN_INDEX_FUNCTION_NAME: &str = "_meiliSearchInIndex";
|
||||
|
||||
#[derive(Deserialize)]
|
||||
pub struct ChatsParam {
|
||||
workspace_uid: String,
|
||||
}
|
||||
|
||||
pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
cfg.service(web::resource("").route(web::get().to(list_workspaces))).service(
|
||||
web::scope("/{workspace_uid}")
|
||||
.service(web::scope("/chat/completions").configure(chat_completions::configure))
|
||||
.service(web::scope("/settings").configure(settings::configure)),
|
||||
);
|
||||
}
|
||||
|
||||
#[derive(Deserr, Debug, Clone, Copy, IntoParams)]
|
||||
#[deserr(error = DeserrQueryParamError, rename_all = camelCase, deny_unknown_fields)]
|
||||
#[into_params(rename_all = "camelCase", parameter_in = Query)]
|
||||
pub struct ListChats {
|
||||
/// The number of chat workspaces to skip before starting to retrieve anything
|
||||
#[param(value_type = Option<usize>, default, example = 100)]
|
||||
#[deserr(default, error = DeserrQueryParamError<InvalidIndexOffset>)]
|
||||
pub offset: Param<usize>,
|
||||
/// The number of chat workspaces to retrieve
|
||||
#[param(value_type = Option<usize>, default = 20, example = 1)]
|
||||
#[deserr(default = Param(PAGINATION_DEFAULT_LIMIT), error = DeserrQueryParamError<InvalidIndexLimit>)]
|
||||
pub limit: Param<usize>,
|
||||
}
|
||||
|
||||
impl ListChats {
|
||||
fn as_pagination(self) -> Pagination {
|
||||
Pagination { offset: self.offset.0, limit: self.limit.0 }
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Serialize, Clone, ToSchema)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct ChatWorkspaceView {
|
||||
/// Unique identifier for the index
|
||||
pub uid: String,
|
||||
}
|
||||
|
||||
pub async fn list_workspaces(
|
||||
index_scheduler: GuardedData<ActionPolicy<{ actions::CHATS_GET }>, Data<IndexScheduler>>,
|
||||
paginate: AwebQueryParameter<ListChats, DeserrQueryParamError>,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats settings route")?;
|
||||
|
||||
debug!(parameters = ?paginate, "List chat workspaces");
|
||||
let filters = index_scheduler.filters();
|
||||
let (total, workspaces) = index_scheduler.paginated_chat_workspace_uids(
|
||||
filters,
|
||||
*paginate.offset,
|
||||
*paginate.limit,
|
||||
)?;
|
||||
let workspaces =
|
||||
workspaces.into_iter().map(|uid| ChatWorkspaceView { uid }).collect::<Vec<_>>();
|
||||
let ret = paginate.as_pagination().format_with(total, workspaces);
|
||||
|
||||
debug!(returns = ?ret, "List chat workspaces");
|
||||
Ok(HttpResponse::Ok().json(ret))
|
||||
}
|
||||
228
crates/meilisearch/src/routes/chats/settings.rs
Normal file
228
crates/meilisearch/src/routes/chats/settings.rs
Normal file
@@ -0,0 +1,228 @@
|
||||
use actix_web::web::{self, Data};
|
||||
use actix_web::HttpResponse;
|
||||
use deserr::Deserr;
|
||||
use index_scheduler::IndexScheduler;
|
||||
use meilisearch_types::deserr::DeserrJsonError;
|
||||
use meilisearch_types::error::deserr_codes::*;
|
||||
use meilisearch_types::error::{Code, ResponseError};
|
||||
use meilisearch_types::features::{
|
||||
ChatCompletionPrompts as DbChatCompletionPrompts, ChatCompletionSettings,
|
||||
ChatCompletionSource as DbChatCompletionSource, DEFAULT_CHAT_PRE_QUERY_PROMPT,
|
||||
DEFAULT_CHAT_SEARCH_DESCRIPTION_PROMPT, DEFAULT_CHAT_SEARCH_INDEX_UID_PARAM_PROMPT,
|
||||
DEFAULT_CHAT_SEARCH_Q_PARAM_PROMPT, DEFAULT_CHAT_SYSTEM_PROMPT,
|
||||
};
|
||||
use meilisearch_types::keys::actions;
|
||||
use meilisearch_types::milli::update::Setting;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use utoipa::ToSchema;
|
||||
|
||||
use super::ChatsParam;
|
||||
use crate::extractors::authentication::policies::ActionPolicy;
|
||||
use crate::extractors::authentication::GuardedData;
|
||||
use crate::extractors::sequential_extractor::SeqHandler;
|
||||
|
||||
pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
cfg.service(
|
||||
web::resource("")
|
||||
.route(web::get().to(SeqHandler(get_settings)))
|
||||
.route(web::patch().to(SeqHandler(patch_settings)))
|
||||
.route(web::delete().to(SeqHandler(delete_settings))),
|
||||
);
|
||||
}
|
||||
|
||||
async fn get_settings(
|
||||
index_scheduler: GuardedData<
|
||||
ActionPolicy<{ actions::CHATS_SETTINGS_GET }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
chats_param: web::Path<ChatsParam>,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats settings route")?;
|
||||
|
||||
let ChatsParam { workspace_uid } = chats_param.into_inner();
|
||||
|
||||
// TODO do a spawn_blocking here ???
|
||||
let rtxn = index_scheduler.read_txn()?;
|
||||
let mut settings = match index_scheduler.chat_settings(&rtxn, &workspace_uid)? {
|
||||
Some(settings) => settings,
|
||||
None => {
|
||||
return Err(ResponseError::from_msg(
|
||||
format!("Chat `{workspace_uid}` not found"),
|
||||
Code::ChatWorkspaceNotFound,
|
||||
))
|
||||
}
|
||||
};
|
||||
settings.hide_secrets();
|
||||
Ok(HttpResponse::Ok().json(settings))
|
||||
}
|
||||
|
||||
async fn patch_settings(
|
||||
index_scheduler: GuardedData<
|
||||
ActionPolicy<{ actions::CHATS_SETTINGS_UPDATE }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
chats_param: web::Path<ChatsParam>,
|
||||
web::Json(new): web::Json<GlobalChatSettings>,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats settings route")?;
|
||||
let ChatsParam { workspace_uid } = chats_param.into_inner();
|
||||
|
||||
// TODO do a spawn_blocking here
|
||||
let mut wtxn = index_scheduler.write_txn()?;
|
||||
let old_settings = index_scheduler.chat_settings(&wtxn, &workspace_uid)?.unwrap_or_default();
|
||||
|
||||
let prompts = match new.prompts {
|
||||
Setting::Set(new_prompts) => DbChatCompletionPrompts {
|
||||
system: match new_prompts.system {
|
||||
Setting::Set(new_system) => new_system,
|
||||
Setting::Reset => DEFAULT_CHAT_SYSTEM_PROMPT.to_string(),
|
||||
Setting::NotSet => old_settings.prompts.system,
|
||||
},
|
||||
search_description: match new_prompts.search_description {
|
||||
Setting::Set(new_description) => new_description,
|
||||
Setting::Reset => DEFAULT_CHAT_SEARCH_DESCRIPTION_PROMPT.to_string(),
|
||||
Setting::NotSet => old_settings.prompts.search_description,
|
||||
},
|
||||
search_q_param: match new_prompts.search_q_param {
|
||||
Setting::Set(new_description) => new_description,
|
||||
Setting::Reset => DEFAULT_CHAT_SEARCH_Q_PARAM_PROMPT.to_string(),
|
||||
Setting::NotSet => old_settings.prompts.search_q_param,
|
||||
},
|
||||
search_index_uid_param: match new_prompts.search_index_uid_param {
|
||||
Setting::Set(new_description) => new_description,
|
||||
Setting::Reset => DEFAULT_CHAT_SEARCH_INDEX_UID_PARAM_PROMPT.to_string(),
|
||||
Setting::NotSet => old_settings.prompts.search_index_uid_param,
|
||||
},
|
||||
pre_query: match new_prompts.pre_query {
|
||||
Setting::Set(new_description) => new_description,
|
||||
Setting::Reset => DEFAULT_CHAT_PRE_QUERY_PROMPT.to_string(),
|
||||
Setting::NotSet => old_settings.prompts.pre_query,
|
||||
},
|
||||
},
|
||||
Setting::Reset => DbChatCompletionPrompts::default(),
|
||||
Setting::NotSet => old_settings.prompts,
|
||||
};
|
||||
|
||||
let settings = ChatCompletionSettings {
|
||||
source: match new.source {
|
||||
Setting::Set(new_source) => new_source.into(),
|
||||
Setting::Reset => DbChatCompletionSource::default(),
|
||||
Setting::NotSet => old_settings.source,
|
||||
},
|
||||
base_api: match new.base_api {
|
||||
Setting::Set(new_base_api) => Some(new_base_api),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => old_settings.base_api,
|
||||
},
|
||||
api_key: match new.api_key {
|
||||
Setting::Set(new_api_key) => Some(new_api_key),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => old_settings.api_key,
|
||||
},
|
||||
prompts,
|
||||
};
|
||||
|
||||
// TODO send analytics
|
||||
// analytics.publish(
|
||||
// PatchNetworkAnalytics {
|
||||
// network_size: merged_remotes.len(),
|
||||
// network_has_self: merged_self.is_some(),
|
||||
// },
|
||||
// &req,
|
||||
// );
|
||||
|
||||
index_scheduler.put_chat_settings(&mut wtxn, &workspace_uid, &settings)?;
|
||||
wtxn.commit()?;
|
||||
|
||||
Ok(HttpResponse::Ok().json(settings))
|
||||
}
|
||||
|
||||
async fn delete_settings(
|
||||
index_scheduler: GuardedData<
|
||||
ActionPolicy<{ actions::CHATS_SETTINGS_DELETE }>,
|
||||
Data<IndexScheduler>,
|
||||
>,
|
||||
chats_param: web::Path<ChatsParam>,
|
||||
) -> Result<HttpResponse, ResponseError> {
|
||||
index_scheduler.features().check_chat_completions("Using the /chats settings route")?;
|
||||
|
||||
let ChatsParam { workspace_uid } = chats_param.into_inner();
|
||||
|
||||
// TODO do a spawn_blocking here
|
||||
let mut wtxn = index_scheduler.write_txn()?;
|
||||
if index_scheduler.delete_chat_settings(&mut wtxn, &workspace_uid)? {
|
||||
wtxn.commit()?;
|
||||
Ok(HttpResponse::NoContent().finish())
|
||||
} else {
|
||||
Err(ResponseError::from_msg(
|
||||
format!("Chat `{workspace_uid}` not found"),
|
||||
Code::ChatWorkspaceNotFound,
|
||||
))
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Deserialize, Deserr, ToSchema)]
|
||||
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
||||
#[serde(deny_unknown_fields, rename_all = "camelCase")]
|
||||
#[schema(rename_all = "camelCase")]
|
||||
pub struct GlobalChatSettings {
|
||||
#[serde(default)]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<ChatCompletionSource>)]
|
||||
pub source: Setting<ChatCompletionSource>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionBaseApi>)]
|
||||
#[schema(value_type = Option<String>, example = json!("https://api.mistral.ai/v1"))]
|
||||
pub base_api: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionApiKey>)]
|
||||
#[schema(value_type = Option<String>, example = json!("abcd1234..."))]
|
||||
pub api_key: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default)]
|
||||
#[schema(inline, value_type = Option<ChatPrompts>)]
|
||||
pub prompts: Setting<ChatPrompts>,
|
||||
}
|
||||
|
||||
#[derive(Default, Debug, Clone, Copy, Serialize, Deserialize, Deserr, ToSchema)]
|
||||
#[serde(deny_unknown_fields, rename_all = "camelCase")]
|
||||
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
||||
pub enum ChatCompletionSource {
|
||||
#[default]
|
||||
OpenAi,
|
||||
}
|
||||
|
||||
impl From<ChatCompletionSource> for DbChatCompletionSource {
|
||||
fn from(source: ChatCompletionSource) -> Self {
|
||||
match source {
|
||||
ChatCompletionSource::OpenAi => DbChatCompletionSource::OpenAi,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Deserialize, Deserr, ToSchema)]
|
||||
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
||||
#[serde(deny_unknown_fields, rename_all = "camelCase")]
|
||||
#[schema(rename_all = "camelCase")]
|
||||
pub struct ChatPrompts {
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionSystemPrompt>)]
|
||||
#[schema(value_type = Option<String>, example = json!("You are a helpful assistant..."))]
|
||||
pub system: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionSearchDescriptionPrompt>)]
|
||||
#[schema(value_type = Option<String>, example = json!("This is the search function..."))]
|
||||
pub search_description: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionSearchQueryParamPrompt>)]
|
||||
#[schema(value_type = Option<String>, example = json!("This is query parameter..."))]
|
||||
pub search_q_param: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionSearchIndexUidParamPrompt>)]
|
||||
#[schema(value_type = Option<String>, example = json!("This is index you want to search in..."))]
|
||||
pub search_index_uid_param: Setting<String>,
|
||||
#[serde(default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidChatCompletionPreQueryPrompt>)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
pub pre_query: Setting<String>,
|
||||
}
|
||||
243
crates/meilisearch/src/routes/chats/utils.rs
Normal file
243
crates/meilisearch/src/routes/chats/utils.rs
Normal file
@@ -0,0 +1,243 @@
|
||||
use std::cell::RefCell;
|
||||
use std::sync::RwLock;
|
||||
|
||||
use actix_web_lab::sse::{self, Event};
|
||||
use async_openai::types::{
|
||||
ChatChoiceStream, ChatCompletionMessageToolCall, ChatCompletionMessageToolCallChunk,
|
||||
ChatCompletionRequestAssistantMessage, ChatCompletionRequestMessage,
|
||||
ChatCompletionStreamResponseDelta, ChatCompletionToolType, CreateChatCompletionStreamResponse,
|
||||
FunctionCall, FunctionCallStream, Role,
|
||||
};
|
||||
use bumpalo::Bump;
|
||||
use meilisearch_types::error::ResponseError;
|
||||
use meilisearch_types::heed::RoTxn;
|
||||
use meilisearch_types::milli::index::ChatConfig;
|
||||
use meilisearch_types::milli::prompt::{Prompt, PromptData};
|
||||
use meilisearch_types::milli::update::new::document::DocumentFromDb;
|
||||
use meilisearch_types::milli::{
|
||||
DocumentId, FieldIdMapWithMetadata, GlobalFieldsIdsMap, MetadataBuilder,
|
||||
};
|
||||
use meilisearch_types::{Document, Index};
|
||||
use serde::Serialize;
|
||||
use tokio::sync::mpsc::error::SendError;
|
||||
use tokio::sync::mpsc::Sender;
|
||||
|
||||
use super::errors::StreamErrorEvent;
|
||||
use super::MEILI_APPEND_CONVERSATION_MESSAGE_NAME;
|
||||
use crate::routes::chats::{MEILI_SEARCH_PROGRESS_NAME, MEILI_SEARCH_SOURCES_NAME};
|
||||
|
||||
pub struct SseEventSender(Sender<Event>);
|
||||
|
||||
impl SseEventSender {
|
||||
pub fn new(sender: Sender<Event>) -> Self {
|
||||
Self(sender)
|
||||
}
|
||||
|
||||
/// Ask the front-end user to append this tool *call* to the conversation
|
||||
pub async fn append_tool_call_conversation_message(
|
||||
&self,
|
||||
resp: CreateChatCompletionStreamResponse,
|
||||
call_id: String,
|
||||
function_name: String,
|
||||
function_arguments: String,
|
||||
) -> Result<(), SendError<Event>> {
|
||||
#[allow(deprecated)]
|
||||
let message =
|
||||
ChatCompletionRequestMessage::Assistant(ChatCompletionRequestAssistantMessage {
|
||||
content: None,
|
||||
refusal: None,
|
||||
name: None,
|
||||
audio: None,
|
||||
tool_calls: Some(vec![ChatCompletionMessageToolCall {
|
||||
id: call_id,
|
||||
r#type: Some(ChatCompletionToolType::Function),
|
||||
function: FunctionCall { name: function_name, arguments: function_arguments },
|
||||
}]),
|
||||
function_call: None,
|
||||
});
|
||||
|
||||
self.append_conversation_message(resp, &message).await
|
||||
}
|
||||
|
||||
/// Ask the front-end user to append this tool to the conversation
|
||||
pub async fn append_conversation_message(
|
||||
&self,
|
||||
mut resp: CreateChatCompletionStreamResponse,
|
||||
message: &ChatCompletionRequestMessage,
|
||||
) -> Result<(), SendError<Event>> {
|
||||
let call_text = serde_json::to_string(message).unwrap();
|
||||
let tool_call = ChatCompletionMessageToolCallChunk {
|
||||
index: 0,
|
||||
id: Some(uuid::Uuid::new_v4().to_string()),
|
||||
r#type: Some(ChatCompletionToolType::Function),
|
||||
function: Some(FunctionCallStream {
|
||||
name: Some(MEILI_APPEND_CONVERSATION_MESSAGE_NAME.to_string()),
|
||||
arguments: Some(call_text),
|
||||
}),
|
||||
};
|
||||
|
||||
resp.choices[0] = ChatChoiceStream {
|
||||
index: 0,
|
||||
#[allow(deprecated)]
|
||||
delta: ChatCompletionStreamResponseDelta {
|
||||
content: None,
|
||||
function_call: None,
|
||||
tool_calls: Some(vec![tool_call]),
|
||||
role: Some(Role::Assistant),
|
||||
refusal: None,
|
||||
},
|
||||
finish_reason: None,
|
||||
logprobs: None,
|
||||
};
|
||||
|
||||
self.send_json(&resp).await
|
||||
}
|
||||
|
||||
pub async fn report_search_progress(
|
||||
&self,
|
||||
mut resp: CreateChatCompletionStreamResponse,
|
||||
call_id: &str,
|
||||
function_name: &str,
|
||||
function_arguments: &str,
|
||||
) -> Result<(), SendError<Event>> {
|
||||
#[derive(Debug, Clone, Serialize)]
|
||||
/// Provides information about the current Meilisearch search operation.
|
||||
struct MeiliSearchProgress<'a> {
|
||||
/// The call ID to track the sources of the search.
|
||||
call_id: &'a str,
|
||||
/// The name of the function we are executing.
|
||||
function_name: &'a str,
|
||||
/// The arguments of the function we are executing, encoded in JSON.
|
||||
function_arguments: &'a str,
|
||||
}
|
||||
|
||||
let progress = MeiliSearchProgress { call_id, function_name, function_arguments };
|
||||
let call_text = serde_json::to_string(&progress).unwrap();
|
||||
let tool_call = ChatCompletionMessageToolCallChunk {
|
||||
index: 0,
|
||||
id: Some(uuid::Uuid::new_v4().to_string()),
|
||||
r#type: Some(ChatCompletionToolType::Function),
|
||||
function: Some(FunctionCallStream {
|
||||
name: Some(MEILI_SEARCH_PROGRESS_NAME.to_string()),
|
||||
arguments: Some(call_text),
|
||||
}),
|
||||
};
|
||||
|
||||
resp.choices[0] = ChatChoiceStream {
|
||||
index: 0,
|
||||
#[allow(deprecated)]
|
||||
delta: ChatCompletionStreamResponseDelta {
|
||||
content: None,
|
||||
function_call: None,
|
||||
tool_calls: Some(vec![tool_call]),
|
||||
role: Some(Role::Assistant),
|
||||
refusal: None,
|
||||
},
|
||||
finish_reason: None,
|
||||
logprobs: None,
|
||||
};
|
||||
|
||||
self.send_json(&resp).await
|
||||
}
|
||||
|
||||
pub async fn report_sources(
|
||||
&self,
|
||||
mut resp: CreateChatCompletionStreamResponse,
|
||||
call_id: &str,
|
||||
documents: &[Document],
|
||||
) -> Result<(), SendError<Event>> {
|
||||
#[derive(Debug, Clone, Serialize)]
|
||||
/// Provides sources of the search.
|
||||
struct MeiliSearchSources<'a> {
|
||||
/// The call ID to track the original search associated to those sources.
|
||||
call_id: &'a str,
|
||||
/// The documents associated with the search (call_id).
|
||||
/// Only the displayed attributes of the documents are returned.
|
||||
sources: &'a [Document],
|
||||
}
|
||||
|
||||
let sources = MeiliSearchSources { call_id, sources: documents };
|
||||
let call_text = serde_json::to_string(&sources).unwrap();
|
||||
let tool_call = ChatCompletionMessageToolCallChunk {
|
||||
index: 0,
|
||||
id: Some(uuid::Uuid::new_v4().to_string()),
|
||||
r#type: Some(ChatCompletionToolType::Function),
|
||||
function: Some(FunctionCallStream {
|
||||
name: Some(MEILI_SEARCH_SOURCES_NAME.to_string()),
|
||||
arguments: Some(call_text),
|
||||
}),
|
||||
};
|
||||
|
||||
resp.choices[0] = ChatChoiceStream {
|
||||
index: 0,
|
||||
#[allow(deprecated)]
|
||||
delta: ChatCompletionStreamResponseDelta {
|
||||
content: None,
|
||||
function_call: None,
|
||||
tool_calls: Some(vec![tool_call]),
|
||||
role: Some(Role::Assistant),
|
||||
refusal: None,
|
||||
},
|
||||
finish_reason: None,
|
||||
logprobs: None,
|
||||
};
|
||||
|
||||
self.send_json(&resp).await
|
||||
}
|
||||
|
||||
pub async fn forward_response(
|
||||
&self,
|
||||
resp: &CreateChatCompletionStreamResponse,
|
||||
) -> Result<(), SendError<Event>> {
|
||||
self.send_json(resp).await
|
||||
}
|
||||
|
||||
pub async fn send_error(&self, error: &StreamErrorEvent) -> Result<(), SendError<Event>> {
|
||||
self.send_json(error).await
|
||||
}
|
||||
|
||||
pub async fn stop(self) -> Result<(), SendError<Event>> {
|
||||
self.0.send(Event::Data(sse::Data::new("[DONE]"))).await
|
||||
}
|
||||
|
||||
async fn send_json<S: Serialize>(&self, data: &S) -> Result<(), SendError<Event>> {
|
||||
self.0.send(Event::Data(sse::Data::new_json(data).unwrap())).await
|
||||
}
|
||||
}
|
||||
|
||||
/// Format documents based on the provided template and maximum bytes.
|
||||
///
|
||||
/// This formatting function is usually used to generate a summary of the documents for LLMs.
|
||||
pub fn format_documents<'doc>(
|
||||
rtxn: &RoTxn<'_>,
|
||||
index: &Index,
|
||||
doc_alloc: &'doc Bump,
|
||||
internal_docids: Vec<DocumentId>,
|
||||
) -> Result<Vec<&'doc str>, ResponseError> {
|
||||
let ChatConfig { prompt: PromptData { template, max_bytes }, .. } = index.chat_config(rtxn)?;
|
||||
|
||||
let prompt = Prompt::new(template, max_bytes).unwrap();
|
||||
let fid_map = index.fields_ids_map(rtxn)?;
|
||||
let metadata_builder = MetadataBuilder::from_index(index, rtxn)?;
|
||||
let fid_map_with_meta = FieldIdMapWithMetadata::new(fid_map.clone(), metadata_builder);
|
||||
let global = RwLock::new(fid_map_with_meta);
|
||||
let gfid_map = RefCell::new(GlobalFieldsIdsMap::new(&global));
|
||||
|
||||
let external_ids: Vec<String> = index
|
||||
.external_id_of(rtxn, internal_docids.iter().copied())?
|
||||
.into_iter()
|
||||
.collect::<Result<_, _>>()?;
|
||||
|
||||
let mut renders = Vec::new();
|
||||
for (docid, external_docid) in internal_docids.into_iter().zip(external_ids) {
|
||||
let document = match DocumentFromDb::new(docid, rtxn, index, &fid_map)? {
|
||||
Some(doc) => doc,
|
||||
None => continue,
|
||||
};
|
||||
|
||||
let text = prompt.render_document(&external_docid, document, &gfid_map, doc_alloc).unwrap();
|
||||
renders.push(text);
|
||||
}
|
||||
|
||||
Ok(renders)
|
||||
}
|
||||
@@ -53,6 +53,7 @@ pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
network: Some(false),
|
||||
get_task_documents_route: Some(false),
|
||||
composite_embedders: Some(false),
|
||||
chat_completions: Some(false),
|
||||
})),
|
||||
(status = 401, description = "The authorization header is missing", body = ResponseError, content_type = "application/json", example = json!(
|
||||
{
|
||||
@@ -97,6 +98,8 @@ pub struct RuntimeTogglableFeatures {
|
||||
pub get_task_documents_route: Option<bool>,
|
||||
#[deserr(default)]
|
||||
pub composite_embedders: Option<bool>,
|
||||
#[deserr(default)]
|
||||
pub chat_completions: Option<bool>,
|
||||
}
|
||||
|
||||
impl From<meilisearch_types::features::RuntimeTogglableFeatures> for RuntimeTogglableFeatures {
|
||||
@@ -109,6 +112,7 @@ impl From<meilisearch_types::features::RuntimeTogglableFeatures> for RuntimeTogg
|
||||
network,
|
||||
get_task_documents_route,
|
||||
composite_embedders,
|
||||
chat_completions,
|
||||
} = value;
|
||||
|
||||
Self {
|
||||
@@ -119,6 +123,7 @@ impl From<meilisearch_types::features::RuntimeTogglableFeatures> for RuntimeTogg
|
||||
network: Some(network),
|
||||
get_task_documents_route: Some(get_task_documents_route),
|
||||
composite_embedders: Some(composite_embedders),
|
||||
chat_completions: Some(chat_completions),
|
||||
}
|
||||
}
|
||||
}
|
||||
@@ -132,6 +137,7 @@ pub struct PatchExperimentalFeatureAnalytics {
|
||||
network: bool,
|
||||
get_task_documents_route: bool,
|
||||
composite_embedders: bool,
|
||||
chat_completions: bool,
|
||||
}
|
||||
|
||||
impl Aggregate for PatchExperimentalFeatureAnalytics {
|
||||
@@ -148,6 +154,7 @@ impl Aggregate for PatchExperimentalFeatureAnalytics {
|
||||
network: new.network,
|
||||
get_task_documents_route: new.get_task_documents_route,
|
||||
composite_embedders: new.composite_embedders,
|
||||
chat_completions: new.chat_completions,
|
||||
})
|
||||
}
|
||||
|
||||
@@ -173,6 +180,7 @@ impl Aggregate for PatchExperimentalFeatureAnalytics {
|
||||
network: Some(false),
|
||||
get_task_documents_route: Some(false),
|
||||
composite_embedders: Some(false),
|
||||
chat_completions: Some(false),
|
||||
})),
|
||||
(status = 401, description = "The authorization header is missing", body = ResponseError, content_type = "application/json", example = json!(
|
||||
{
|
||||
@@ -214,6 +222,7 @@ async fn patch_features(
|
||||
.0
|
||||
.composite_embedders
|
||||
.unwrap_or(old_features.composite_embedders),
|
||||
chat_completions: new_features.0.chat_completions.unwrap_or(old_features.chat_completions),
|
||||
};
|
||||
|
||||
// explicitly destructure for analytics rather than using the `Serialize` implementation, because
|
||||
@@ -227,6 +236,7 @@ async fn patch_features(
|
||||
network,
|
||||
get_task_documents_route,
|
||||
composite_embedders,
|
||||
chat_completions,
|
||||
} = new_features;
|
||||
|
||||
analytics.publish(
|
||||
@@ -238,6 +248,7 @@ async fn patch_features(
|
||||
network,
|
||||
get_task_documents_route,
|
||||
composite_embedders,
|
||||
chat_completions,
|
||||
},
|
||||
&req,
|
||||
);
|
||||
|
||||
@@ -172,7 +172,7 @@ pub async fn list_indexes(
|
||||
debug!(parameters = ?paginate, "List indexes");
|
||||
let filters = index_scheduler.filters();
|
||||
let (total, indexes) =
|
||||
index_scheduler.get_paginated_indexes_stats(filters, *paginate.offset, *paginate.limit)?;
|
||||
index_scheduler.paginated_indexes_stats(filters, *paginate.offset, *paginate.limit)?;
|
||||
let indexes = indexes
|
||||
.into_iter()
|
||||
.map(|(name, stats)| IndexView {
|
||||
|
||||
@@ -6,7 +6,7 @@ use meilisearch_types::deserr::DeserrJsonError;
|
||||
use meilisearch_types::error::ResponseError;
|
||||
use meilisearch_types::index_uid::IndexUid;
|
||||
use meilisearch_types::settings::{
|
||||
settings, SecretPolicy, SettingEmbeddingSettings, Settings, Unchecked,
|
||||
settings, ChatSettings, SecretPolicy, SettingEmbeddingSettings, Settings, Unchecked,
|
||||
};
|
||||
use meilisearch_types::tasks::KindWithContent;
|
||||
use tracing::debug;
|
||||
@@ -508,6 +508,17 @@ make_setting_routes!(
|
||||
camelcase_attr: "prefixSearch",
|
||||
analytics: PrefixSearchAnalytics
|
||||
},
|
||||
{
|
||||
route: "/chat",
|
||||
update_verb: put,
|
||||
value_type: ChatSettings,
|
||||
err_type: meilisearch_types::deserr::DeserrJsonError<
|
||||
meilisearch_types::error::deserr_codes::InvalidSettingsIndexChat,
|
||||
>,
|
||||
attr: chat,
|
||||
camelcase_attr: "chat",
|
||||
analytics: ChatAnalytics
|
||||
},
|
||||
);
|
||||
|
||||
#[utoipa::path(
|
||||
@@ -597,6 +608,7 @@ pub async fn update_all(
|
||||
),
|
||||
facet_search: FacetSearchAnalytics::new(new_settings.facet_search.as_ref().set()),
|
||||
prefix_search: PrefixSearchAnalytics::new(new_settings.prefix_search.as_ref().set()),
|
||||
chat: ChatAnalytics::new(new_settings.chat.as_ref().set()),
|
||||
},
|
||||
&req,
|
||||
);
|
||||
|
||||
@@ -10,8 +10,8 @@ use meilisearch_types::locales::{Locale, LocalizedAttributesRuleView};
|
||||
use meilisearch_types::milli::update::Setting;
|
||||
use meilisearch_types::milli::FilterableAttributesRule;
|
||||
use meilisearch_types::settings::{
|
||||
FacetingSettings, PaginationSettings, PrefixSearchSettings, ProximityPrecisionView,
|
||||
RankingRuleView, SettingEmbeddingSettings, TypoSettings,
|
||||
ChatSettings, FacetingSettings, PaginationSettings, PrefixSearchSettings,
|
||||
ProximityPrecisionView, RankingRuleView, SettingEmbeddingSettings, TypoSettings,
|
||||
};
|
||||
use serde::Serialize;
|
||||
|
||||
@@ -39,6 +39,7 @@ pub struct SettingsAnalytics {
|
||||
pub non_separator_tokens: NonSeparatorTokensAnalytics,
|
||||
pub facet_search: FacetSearchAnalytics,
|
||||
pub prefix_search: PrefixSearchAnalytics,
|
||||
pub chat: ChatAnalytics,
|
||||
}
|
||||
|
||||
impl Aggregate for SettingsAnalytics {
|
||||
@@ -198,6 +199,7 @@ impl Aggregate for SettingsAnalytics {
|
||||
set: new.prefix_search.set | self.prefix_search.set,
|
||||
value: new.prefix_search.value.or(self.prefix_search.value),
|
||||
},
|
||||
chat: ChatAnalytics { set: new.chat.set | self.chat.set },
|
||||
})
|
||||
}
|
||||
|
||||
@@ -676,3 +678,18 @@ impl PrefixSearchAnalytics {
|
||||
SettingsAnalytics { prefix_search: self, ..Default::default() }
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Serialize, Default)]
|
||||
pub struct ChatAnalytics {
|
||||
pub set: bool,
|
||||
}
|
||||
|
||||
impl ChatAnalytics {
|
||||
pub fn new(settings: Option<&ChatSettings>) -> Self {
|
||||
Self { set: settings.is_some() }
|
||||
}
|
||||
|
||||
pub fn into_settings(self) -> SettingsAnalytics {
|
||||
SettingsAnalytics { chat: self, ..Default::default() }
|
||||
}
|
||||
}
|
||||
|
||||
@@ -52,6 +52,7 @@ const PAGINATION_DEFAULT_LIMIT_FN: fn() -> usize = || 20;
|
||||
|
||||
mod api_key;
|
||||
pub mod batches;
|
||||
pub mod chats;
|
||||
mod dump;
|
||||
pub mod features;
|
||||
pub mod indexes;
|
||||
@@ -113,7 +114,8 @@ pub fn configure(cfg: &mut web::ServiceConfig) {
|
||||
.service(web::scope("/swap-indexes").configure(swap_indexes::configure))
|
||||
.service(web::scope("/metrics").configure(metrics::configure))
|
||||
.service(web::scope("/experimental-features").configure(features::configure))
|
||||
.service(web::scope("/network").configure(network::configure));
|
||||
.service(web::scope("/network").configure(network::configure))
|
||||
.service(web::scope("/chats").configure(chats::configure));
|
||||
|
||||
#[cfg(feature = "swagger")]
|
||||
{
|
||||
|
||||
@@ -16,6 +16,7 @@ use meilisearch_types::error::{Code, ResponseError};
|
||||
use meilisearch_types::heed::RoTxn;
|
||||
use meilisearch_types::index_uid::IndexUid;
|
||||
use meilisearch_types::locales::Locale;
|
||||
use meilisearch_types::milli::index::{self, SearchParameters};
|
||||
use meilisearch_types::milli::score_details::{ScoreDetails, ScoringStrategy};
|
||||
use meilisearch_types::milli::vector::parsed_vectors::ExplicitVectors;
|
||||
use meilisearch_types::milli::vector::Embedder;
|
||||
@@ -56,7 +57,7 @@ pub const DEFAULT_HIGHLIGHT_PRE_TAG: fn() -> String = || "<em>".to_string();
|
||||
pub const DEFAULT_HIGHLIGHT_POST_TAG: fn() -> String = || "</em>".to_string();
|
||||
pub const DEFAULT_SEMANTIC_RATIO: fn() -> SemanticRatio = || SemanticRatio(0.5);
|
||||
|
||||
#[derive(Clone, Default, PartialEq, Deserr, ToSchema)]
|
||||
#[derive(Clone, PartialEq, Deserr, ToSchema)]
|
||||
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
|
||||
pub struct SearchQuery {
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSearchQ>)]
|
||||
@@ -119,9 +120,73 @@ pub struct SearchQuery {
|
||||
pub locales: Option<Vec<Locale>>,
|
||||
}
|
||||
|
||||
impl From<SearchParameters> for SearchQuery {
|
||||
fn from(parameters: SearchParameters) -> Self {
|
||||
let SearchParameters {
|
||||
hybrid,
|
||||
limit,
|
||||
sort,
|
||||
distinct,
|
||||
matching_strategy,
|
||||
attributes_to_search_on,
|
||||
ranking_score_threshold,
|
||||
} = parameters;
|
||||
|
||||
SearchQuery {
|
||||
hybrid: hybrid.map(|index::HybridQuery { semantic_ratio, embedder }| HybridQuery {
|
||||
semantic_ratio: SemanticRatio::try_from(semantic_ratio)
|
||||
.ok()
|
||||
.unwrap_or_else(DEFAULT_SEMANTIC_RATIO),
|
||||
embedder,
|
||||
}),
|
||||
limit: limit.unwrap_or_else(DEFAULT_SEARCH_LIMIT),
|
||||
sort,
|
||||
distinct,
|
||||
matching_strategy: matching_strategy.map(MatchingStrategy::from).unwrap_or_default(),
|
||||
attributes_to_search_on,
|
||||
ranking_score_threshold: ranking_score_threshold.map(RankingScoreThreshold::from),
|
||||
..Default::default()
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl Default for SearchQuery {
|
||||
fn default() -> Self {
|
||||
SearchQuery {
|
||||
q: None,
|
||||
vector: None,
|
||||
hybrid: None,
|
||||
offset: DEFAULT_SEARCH_OFFSET(),
|
||||
limit: DEFAULT_SEARCH_LIMIT(),
|
||||
page: None,
|
||||
hits_per_page: None,
|
||||
attributes_to_retrieve: None,
|
||||
retrieve_vectors: false,
|
||||
attributes_to_crop: None,
|
||||
crop_length: DEFAULT_CROP_LENGTH(),
|
||||
attributes_to_highlight: None,
|
||||
show_matches_position: false,
|
||||
show_ranking_score: false,
|
||||
show_ranking_score_details: false,
|
||||
filter: None,
|
||||
sort: None,
|
||||
distinct: None,
|
||||
facets: None,
|
||||
highlight_pre_tag: DEFAULT_HIGHLIGHT_PRE_TAG(),
|
||||
highlight_post_tag: DEFAULT_HIGHLIGHT_POST_TAG(),
|
||||
crop_marker: DEFAULT_CROP_MARKER(),
|
||||
matching_strategy: Default::default(),
|
||||
attributes_to_search_on: None,
|
||||
ranking_score_threshold: None,
|
||||
locales: None,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy, PartialEq, Deserr, ToSchema, Serialize)]
|
||||
#[deserr(try_from(f64) = TryFrom::try_from -> InvalidSearchRankingScoreThreshold)]
|
||||
pub struct RankingScoreThreshold(f64);
|
||||
|
||||
impl std::convert::TryFrom<f64> for RankingScoreThreshold {
|
||||
type Error = InvalidSearchRankingScoreThreshold;
|
||||
|
||||
@@ -136,6 +201,14 @@ impl std::convert::TryFrom<f64> for RankingScoreThreshold {
|
||||
}
|
||||
}
|
||||
|
||||
impl From<index::RankingScoreThreshold> for RankingScoreThreshold {
|
||||
fn from(threshold: index::RankingScoreThreshold) -> Self {
|
||||
let threshold = threshold.as_f64();
|
||||
assert!((0.0..=1.0).contains(&threshold));
|
||||
RankingScoreThreshold(threshold)
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy, PartialEq, Deserr)]
|
||||
#[deserr(try_from(f64) = TryFrom::try_from -> InvalidSimilarRankingScoreThreshold)]
|
||||
pub struct RankingScoreThresholdSimilar(f64);
|
||||
@@ -279,8 +352,8 @@ impl fmt::Debug for SearchQuery {
|
||||
#[deserr(error = DeserrJsonError<InvalidSearchHybridQuery>, rename_all = camelCase, deny_unknown_fields)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct HybridQuery {
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSearchSemanticRatio>, default)]
|
||||
#[schema(value_type = f32, default)]
|
||||
#[deserr(default, error = DeserrJsonError<InvalidSearchSemanticRatio>)]
|
||||
#[schema(default, value_type = f32)]
|
||||
#[serde(default)]
|
||||
pub semantic_ratio: SemanticRatio,
|
||||
#[deserr(error = DeserrJsonError<InvalidSearchEmbedder>)]
|
||||
@@ -717,6 +790,16 @@ impl From<MatchingStrategy> for TermsMatchingStrategy {
|
||||
}
|
||||
}
|
||||
|
||||
impl From<index::MatchingStrategy> for MatchingStrategy {
|
||||
fn from(other: index::MatchingStrategy) -> Self {
|
||||
match other {
|
||||
index::MatchingStrategy::Last => Self::Last,
|
||||
index::MatchingStrategy::All => Self::All,
|
||||
index::MatchingStrategy::Frequency => Self::Frequency,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Default, Clone, PartialEq, Eq, Deserr)]
|
||||
#[deserr(rename_all = camelCase)]
|
||||
pub enum FacetValuesSort {
|
||||
@@ -882,7 +965,7 @@ pub fn add_search_rules(filter: &mut Option<Value>, rules: IndexSearchRules) {
|
||||
}
|
||||
}
|
||||
|
||||
fn prepare_search<'t>(
|
||||
pub fn prepare_search<'t>(
|
||||
index: &'t Index,
|
||||
rtxn: &'t RoTxn,
|
||||
query: &'t SearchQuery,
|
||||
@@ -1260,7 +1343,7 @@ struct HitMaker<'a> {
|
||||
vectors_fid: Option<FieldId>,
|
||||
retrieve_vectors: RetrieveVectors,
|
||||
to_retrieve_ids: BTreeSet<FieldId>,
|
||||
embedding_configs: Vec<milli::index::IndexEmbeddingConfig>,
|
||||
embedding_configs: Vec<index::IndexEmbeddingConfig>,
|
||||
formatter_builder: MatcherBuilder<'a>,
|
||||
formatted_options: BTreeMap<FieldId, FormatOptions>,
|
||||
show_ranking_score: bool,
|
||||
|
||||
@@ -820,6 +820,22 @@ async fn list_api_keys() {
|
||||
"createdAt": "[ignored]",
|
||||
"updatedAt": "[ignored]"
|
||||
},
|
||||
{
|
||||
"name": "Default Chat API Key",
|
||||
"description": "Use it to chat and search from the frontend",
|
||||
"key": "[ignored]",
|
||||
"uid": "[ignored]",
|
||||
"actions": [
|
||||
"search",
|
||||
"chat.get"
|
||||
],
|
||||
"indexes": [
|
||||
"*"
|
||||
],
|
||||
"expiresAt": null,
|
||||
"createdAt": "[ignored]",
|
||||
"updatedAt": "[ignored]"
|
||||
},
|
||||
{
|
||||
"name": "Default Search API Key",
|
||||
"description": "Use it to search from the frontend",
|
||||
|
||||
@@ -1,8 +1,8 @@
|
||||
use crate::json;
|
||||
use meili_snap::{json_string, snapshot};
|
||||
use serde_json::Value;
|
||||
|
||||
use crate::common::{shared_does_not_exists_index, Server};
|
||||
use crate::json;
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn create_and_get_index() {
|
||||
|
||||
@@ -1,10 +1,9 @@
|
||||
use meili_snap::*;
|
||||
|
||||
use super::test_settings_documents_indexing_swapping_and_search;
|
||||
use crate::common::{shared_does_not_exists_index, Server, DOCUMENTS, NESTED_DOCUMENTS};
|
||||
use crate::json;
|
||||
|
||||
use super::test_settings_documents_indexing_swapping_and_search;
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn search_unexisting_index() {
|
||||
let index = shared_does_not_exists_index().await;
|
||||
|
||||
@@ -3,10 +3,10 @@ use meilisearch::Opt;
|
||||
use tempfile::TempDir;
|
||||
|
||||
use super::test_settings_documents_indexing_swapping_and_search;
|
||||
use crate::{
|
||||
common::{default_settings, shared_index_with_documents, Server, DOCUMENTS, NESTED_DOCUMENTS},
|
||||
json,
|
||||
use crate::common::{
|
||||
default_settings, shared_index_with_documents, Server, DOCUMENTS, NESTED_DOCUMENTS,
|
||||
};
|
||||
use crate::json;
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn search_with_filter_string_notation() {
|
||||
|
||||
@@ -2,11 +2,10 @@ use meili_snap::{json_string, snapshot};
|
||||
use meilisearch_types::milli::constants::RESERVED_GEO_FIELD_NAME;
|
||||
use once_cell::sync::Lazy;
|
||||
|
||||
use super::test_settings_documents_indexing_swapping_and_search;
|
||||
use crate::common::{Server, Value};
|
||||
use crate::json;
|
||||
|
||||
use super::test_settings_documents_indexing_swapping_and_search;
|
||||
|
||||
static DOCUMENTS: Lazy<Value> = Lazy::new(|| {
|
||||
json!([
|
||||
{
|
||||
|
||||
@@ -1,9 +1,6 @@
|
||||
use std::mem;
|
||||
|
||||
use heed::Database;
|
||||
use heed::DatabaseStat;
|
||||
use heed::RoTxn;
|
||||
use heed::Unspecified;
|
||||
use heed::{Database, DatabaseStat, RoTxn, Unspecified};
|
||||
use serde::{Deserialize, Serialize};
|
||||
|
||||
use crate::BEU32;
|
||||
|
||||
@@ -1,10 +1,9 @@
|
||||
use heed::{
|
||||
types::{SerdeJson, Str},
|
||||
RoTxn, RwTxn,
|
||||
};
|
||||
use heed::types::{SerdeJson, Str};
|
||||
use heed::{RoTxn, RwTxn};
|
||||
use serde::{Deserialize, Serialize};
|
||||
|
||||
use crate::{index::main_key, Index};
|
||||
use crate::index::main_key;
|
||||
use crate::Index;
|
||||
|
||||
#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, Default)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
|
||||
@@ -1,5 +1,4 @@
|
||||
use std::collections::BTreeSet;
|
||||
use std::collections::HashMap;
|
||||
use std::collections::{BTreeSet, HashMap};
|
||||
use std::convert::Infallible;
|
||||
use std::fmt::Write;
|
||||
use std::{io, str};
|
||||
|
||||
@@ -32,13 +32,13 @@ impl ExternalDocumentsIds {
|
||||
&self,
|
||||
rtxn: &RoTxn<'_>,
|
||||
external_id: A,
|
||||
) -> heed::Result<Option<u32>> {
|
||||
) -> heed::Result<Option<DocumentId>> {
|
||||
self.0.get(rtxn, external_id.as_ref())
|
||||
}
|
||||
|
||||
/// An helper function to debug this type, returns an `HashMap` of both,
|
||||
/// soft and hard fst maps, combined.
|
||||
pub fn to_hash_map(&self, rtxn: &RoTxn<'_>) -> heed::Result<HashMap<String, u32>> {
|
||||
pub fn to_hash_map(&self, rtxn: &RoTxn<'_>) -> heed::Result<HashMap<String, DocumentId>> {
|
||||
let mut map = HashMap::default();
|
||||
for result in self.0.iter(rtxn)? {
|
||||
let (external, internal) = result?;
|
||||
|
||||
@@ -7,6 +7,7 @@ use crate::FieldId;
|
||||
mod global;
|
||||
pub mod metadata;
|
||||
pub use global::GlobalFieldsIdsMap;
|
||||
pub use metadata::{FieldIdMapWithMetadata, MetadataBuilder};
|
||||
|
||||
#[derive(Debug, Clone, Serialize, Deserialize)]
|
||||
pub struct FieldsIdsMap {
|
||||
|
||||
@@ -1,13 +1,12 @@
|
||||
use std::collections::{BTreeSet, HashSet};
|
||||
|
||||
use deserr::{DeserializeError, Deserr, ValuePointerRef};
|
||||
use serde::{Deserialize, Serialize};
|
||||
use std::collections::{BTreeSet, HashSet};
|
||||
use utoipa::ToSchema;
|
||||
|
||||
use crate::{
|
||||
attribute_patterns::{match_distinct_field, match_field_legacy, PatternMatch},
|
||||
constants::RESERVED_GEO_FIELD_NAME,
|
||||
AttributePatterns,
|
||||
};
|
||||
use crate::attribute_patterns::{match_distinct_field, match_field_legacy, PatternMatch};
|
||||
use crate::constants::RESERVED_GEO_FIELD_NAME;
|
||||
use crate::AttributePatterns;
|
||||
|
||||
#[derive(Serialize, Deserialize, PartialEq, Eq, Clone, Debug, ToSchema)]
|
||||
#[serde(untagged)]
|
||||
|
||||
@@ -1,14 +1,18 @@
|
||||
use std::borrow::Cow;
|
||||
use std::collections::{BTreeMap, BTreeSet, HashMap, HashSet};
|
||||
use std::error::Error;
|
||||
use std::fmt;
|
||||
use std::fs::File;
|
||||
use std::path::Path;
|
||||
|
||||
use deserr::Deserr;
|
||||
use heed::types::*;
|
||||
use heed::{CompactionOption, Database, DatabaseStat, RoTxn, RwTxn, Unspecified, WithoutTls};
|
||||
use indexmap::IndexMap;
|
||||
use roaring::RoaringBitmap;
|
||||
use rstar::RTree;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use utoipa::ToSchema;
|
||||
|
||||
use crate::constants::{self, RESERVED_GEO_FIELD_NAME, RESERVED_VECTORS_FIELD_NAME};
|
||||
use crate::database_stats::DatabaseStats;
|
||||
@@ -23,7 +27,9 @@ use crate::heed_codec::facet::{
|
||||
use crate::heed_codec::version::VersionCodec;
|
||||
use crate::heed_codec::{BEU16StrCodec, FstSetCodec, StrBEU16Codec, StrRefCodec};
|
||||
use crate::order_by_map::OrderByMap;
|
||||
use crate::prompt::PromptData;
|
||||
use crate::proximity::ProximityPrecision;
|
||||
use crate::update::new::StdResult;
|
||||
use crate::vector::{ArroyStats, ArroyWrapper, Embedding, EmbeddingConfig};
|
||||
use crate::{
|
||||
default_criteria, CboRoaringBitmapCodec, Criterion, DocumentId, ExternalDocumentsIds,
|
||||
@@ -79,6 +85,7 @@ pub mod main_key {
|
||||
pub const PREFIX_SEARCH: &str = "prefix_search";
|
||||
pub const DOCUMENTS_STATS: &str = "documents_stats";
|
||||
pub const DISABLED_TYPOS_TERMS: &str = "disabled_typos_terms";
|
||||
pub const CHAT: &str = "chat";
|
||||
}
|
||||
|
||||
pub mod db_name {
|
||||
@@ -1691,6 +1698,25 @@ impl Index {
|
||||
self.main.remap_key_type::<Str>().delete(txn, main_key::FACET_SEARCH)
|
||||
}
|
||||
|
||||
pub fn chat_config(&self, txn: &RoTxn<'_>) -> heed::Result<ChatConfig> {
|
||||
self.main
|
||||
.remap_types::<Str, SerdeJson<_>>()
|
||||
.get(txn, main_key::CHAT)
|
||||
.map(|o| o.unwrap_or_default())
|
||||
}
|
||||
|
||||
pub(crate) fn put_chat_config(
|
||||
&self,
|
||||
txn: &mut RwTxn<'_>,
|
||||
val: &ChatConfig,
|
||||
) -> heed::Result<()> {
|
||||
self.main.remap_types::<Str, SerdeJson<_>>().put(txn, main_key::CHAT, &val)
|
||||
}
|
||||
|
||||
pub(crate) fn delete_chat_config(&self, txn: &mut RwTxn<'_>) -> heed::Result<bool> {
|
||||
self.main.remap_key_type::<Str>().delete(txn, main_key::CHAT)
|
||||
}
|
||||
|
||||
pub fn localized_attributes_rules(
|
||||
&self,
|
||||
rtxn: &RoTxn<'_>,
|
||||
@@ -1917,13 +1943,97 @@ pub struct IndexEmbeddingConfig {
|
||||
pub user_provided: RoaringBitmap,
|
||||
}
|
||||
|
||||
#[derive(Debug, Default, Deserialize, Serialize)]
|
||||
pub struct ChatConfig {
|
||||
pub description: String,
|
||||
/// Contains the document template and max template length.
|
||||
pub prompt: PromptData,
|
||||
pub search_parameters: SearchParameters,
|
||||
}
|
||||
|
||||
#[derive(Debug, Default, Deserialize, Serialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct SearchParameters {
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub hybrid: Option<HybridQuery>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub limit: Option<usize>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub sort: Option<Vec<String>>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub distinct: Option<String>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub matching_strategy: Option<MatchingStrategy>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub attributes_to_search_on: Option<Vec<String>>,
|
||||
#[serde(skip_serializing_if = "Option::is_none")]
|
||||
pub ranking_score_threshold: Option<RankingScoreThreshold>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy, Default, Deserialize, Serialize, PartialEq, Deserr, ToSchema)]
|
||||
#[deserr(try_from(f64) = TryFrom::try_from -> InvalidSearchRankingScoreThreshold)]
|
||||
pub struct RankingScoreThreshold(f64);
|
||||
|
||||
impl RankingScoreThreshold {
|
||||
pub fn as_f64(&self) -> f64 {
|
||||
self.0
|
||||
}
|
||||
}
|
||||
|
||||
impl TryFrom<f64> for RankingScoreThreshold {
|
||||
type Error = InvalidSearchRankingScoreThreshold;
|
||||
|
||||
fn try_from(value: f64) -> StdResult<Self, Self::Error> {
|
||||
if !(0.0..=1.0).contains(&value) {
|
||||
Err(InvalidSearchRankingScoreThreshold)
|
||||
} else {
|
||||
Ok(RankingScoreThreshold(value))
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct InvalidSearchRankingScoreThreshold;
|
||||
|
||||
impl Error for InvalidSearchRankingScoreThreshold {}
|
||||
|
||||
impl fmt::Display for InvalidSearchRankingScoreThreshold {
|
||||
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
||||
write!(
|
||||
f,
|
||||
"the value of `rankingScoreThreshold` is invalid, expected a float between `0.0` and `1.0`."
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Default, Deserialize, Serialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct HybridQuery {
|
||||
pub semantic_ratio: f32,
|
||||
pub embedder: String,
|
||||
}
|
||||
|
||||
#[derive(Debug, Deserialize, Serialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub struct PrefixSettings {
|
||||
pub prefix_count_threshold: usize,
|
||||
pub max_prefix_length: usize,
|
||||
pub compute_prefixes: PrefixSearch,
|
||||
}
|
||||
|
||||
#[derive(Default, Debug, Copy, Clone, PartialEq, Eq, Deserr, ToSchema, Serialize, Deserialize)]
|
||||
#[deserr(rename_all = camelCase)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub enum MatchingStrategy {
|
||||
/// Remove query words from last to first
|
||||
#[default]
|
||||
Last,
|
||||
/// All query words are mandatory
|
||||
All,
|
||||
/// Remove query words from the most frequent to the least
|
||||
Frequency,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy, Serialize, Deserialize, PartialEq, Eq, Default)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
pub enum PrefixSearch {
|
||||
|
||||
@@ -52,18 +52,19 @@ pub use search::new::{
|
||||
};
|
||||
use serde_json::Value;
|
||||
pub use thread_pool_no_abort::{PanicCatched, ThreadPoolNoAbort, ThreadPoolNoAbortBuilder};
|
||||
pub use {charabia as tokenizer, heed, rhai};
|
||||
pub use {arroy, charabia as tokenizer, heed, rhai};
|
||||
|
||||
pub use self::asc_desc::{AscDesc, AscDescError, Member, SortError};
|
||||
pub use self::attribute_patterns::AttributePatterns;
|
||||
pub use self::attribute_patterns::PatternMatch;
|
||||
pub use self::attribute_patterns::{AttributePatterns, PatternMatch};
|
||||
pub use self::criterion::{default_criteria, Criterion, CriterionError};
|
||||
pub use self::error::{
|
||||
Error, FieldIdMapMissingEntry, InternalError, SerializationError, UserError,
|
||||
};
|
||||
pub use self::external_documents_ids::ExternalDocumentsIds;
|
||||
pub use self::fieldids_weights_map::FieldidsWeightsMap;
|
||||
pub use self::fields_ids_map::{FieldsIdsMap, GlobalFieldsIdsMap};
|
||||
pub use self::fields_ids_map::{
|
||||
FieldIdMapWithMetadata, FieldsIdsMap, GlobalFieldsIdsMap, MetadataBuilder,
|
||||
};
|
||||
pub use self::filterable_attributes_rules::{
|
||||
FilterFeatures, FilterableAttributesFeatures, FilterableAttributesPatterns,
|
||||
FilterableAttributesRule,
|
||||
@@ -84,8 +85,6 @@ pub use self::search::{
|
||||
};
|
||||
pub use self::update::ChannelCongestion;
|
||||
|
||||
pub use arroy;
|
||||
|
||||
pub type Result<T> = std::result::Result<T, error::Error>;
|
||||
|
||||
pub type Attribute = u32;
|
||||
|
||||
@@ -1,4 +1,3 @@
|
||||
use enum_iterator::Sequence;
|
||||
use std::any::TypeId;
|
||||
use std::borrow::Cow;
|
||||
use std::marker::PhantomData;
|
||||
@@ -6,6 +5,7 @@ use std::sync::atomic::{AtomicU32, Ordering};
|
||||
use std::sync::{Arc, RwLock};
|
||||
use std::time::{Duration, Instant};
|
||||
|
||||
use enum_iterator::Sequence;
|
||||
use indexmap::IndexMap;
|
||||
use itertools::Itertools;
|
||||
use serde::Serialize;
|
||||
|
||||
@@ -64,7 +64,7 @@ fn default_template() -> liquid::Template {
|
||||
new_template(default_template_text()).unwrap()
|
||||
}
|
||||
|
||||
fn default_template_text() -> &'static str {
|
||||
pub fn default_template_text() -> &'static str {
|
||||
"{% for field in fields %}\
|
||||
{% if field.is_searchable and field.value != nil %}\
|
||||
{{ field.name }}: {{ field.value }}\n\
|
||||
|
||||
@@ -10,6 +10,7 @@ pub use self::facet::{FacetDistribution, Filter, OrderBy, DEFAULT_VALUES_PER_FAC
|
||||
pub use self::new::matches::{FormatOptions, MatchBounds, MatcherBuilder, MatchingWords};
|
||||
use self::new::{execute_vector_search, PartialSearchResult, VectorStoreStats};
|
||||
use crate::filterable_attributes_rules::{filtered_matching_patterns, matching_features};
|
||||
use crate::index::MatchingStrategy;
|
||||
use crate::score_details::{ScoreDetails, ScoringStrategy};
|
||||
use crate::vector::Embedder;
|
||||
use crate::{
|
||||
@@ -364,6 +365,16 @@ impl Default for TermsMatchingStrategy {
|
||||
}
|
||||
}
|
||||
|
||||
impl From<MatchingStrategy> for TermsMatchingStrategy {
|
||||
fn from(other: MatchingStrategy) -> Self {
|
||||
match other {
|
||||
MatchingStrategy::Last => Self::Last,
|
||||
MatchingStrategy::All => Self::All,
|
||||
MatchingStrategy::Frequency => Self::Frequency,
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fn get_first(s: &str) -> &str {
|
||||
match s.chars().next() {
|
||||
Some(c) => &s[..c.len_utf8()],
|
||||
|
||||
@@ -1,8 +1,9 @@
|
||||
use std::collections::VecDeque;
|
||||
|
||||
use heed::types::{Bytes, Unit};
|
||||
use heed::{RoPrefix, RoTxn};
|
||||
use roaring::RoaringBitmap;
|
||||
use rstar::RTree;
|
||||
use std::collections::VecDeque;
|
||||
|
||||
use super::facet_string_values;
|
||||
use super::ranking_rules::{RankingRule, RankingRuleOutput, RankingRuleQueryTrait};
|
||||
|
||||
@@ -47,8 +47,7 @@ use sort::Sort;
|
||||
|
||||
use self::distinct::facet_string_values;
|
||||
use self::geo_sort::GeoSort;
|
||||
pub use self::geo_sort::Parameter as GeoSortParameter;
|
||||
pub use self::geo_sort::Strategy as GeoSortStrategy;
|
||||
pub use self::geo_sort::{Parameter as GeoSortParameter, Strategy as GeoSortStrategy};
|
||||
use self::graph_based_ranking_rule::Words;
|
||||
use self::interner::Interned;
|
||||
use self::vector_sort::VectorSort;
|
||||
|
||||
190
crates/milli/src/update/chat.rs
Normal file
190
crates/milli/src/update/chat.rs
Normal file
@@ -0,0 +1,190 @@
|
||||
use std::error::Error;
|
||||
use std::fmt;
|
||||
|
||||
use deserr::errors::JsonError;
|
||||
use deserr::Deserr;
|
||||
use serde::{Deserialize, Serialize};
|
||||
use utoipa::ToSchema;
|
||||
|
||||
use crate::index::{self, ChatConfig, MatchingStrategy, RankingScoreThreshold, SearchParameters};
|
||||
use crate::prompt::{default_max_bytes, PromptData};
|
||||
use crate::update::Setting;
|
||||
|
||||
#[derive(Debug, Clone, Default, Serialize, Deserialize, PartialEq, Deserr, ToSchema)]
|
||||
#[serde(deny_unknown_fields, rename_all = "camelCase")]
|
||||
#[deserr(error = JsonError, deny_unknown_fields, rename_all = camelCase)]
|
||||
pub struct ChatSettings {
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
pub description: Setting<String>,
|
||||
|
||||
/// A liquid template used to render documents to a text that can be embedded.
|
||||
///
|
||||
/// Meillisearch interpolates the template for each document and sends the resulting text to the embedder.
|
||||
/// The embedder then generates document vectors based on this text.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
pub document_template: Setting<String>,
|
||||
|
||||
/// Rendered texts are truncated to this size. Defaults to 400.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<usize>)]
|
||||
pub document_template_max_bytes: Setting<usize>,
|
||||
|
||||
/// The search parameters to use for the LLM.
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<ChatSearchParams>)]
|
||||
pub search_parameters: Setting<ChatSearchParams>,
|
||||
}
|
||||
|
||||
impl From<ChatConfig> for ChatSettings {
|
||||
fn from(config: ChatConfig) -> Self {
|
||||
let ChatConfig {
|
||||
description,
|
||||
prompt: PromptData { template, max_bytes },
|
||||
search_parameters,
|
||||
} = config;
|
||||
ChatSettings {
|
||||
description: Setting::Set(description),
|
||||
document_template: Setting::Set(template),
|
||||
document_template_max_bytes: Setting::Set(
|
||||
max_bytes.unwrap_or(default_max_bytes()).get(),
|
||||
),
|
||||
search_parameters: Setting::Set({
|
||||
let SearchParameters {
|
||||
hybrid,
|
||||
limit,
|
||||
sort,
|
||||
distinct,
|
||||
matching_strategy,
|
||||
attributes_to_search_on,
|
||||
ranking_score_threshold,
|
||||
} = search_parameters;
|
||||
|
||||
let hybrid = hybrid.map(|index::HybridQuery { semantic_ratio, embedder }| {
|
||||
HybridQuery { semantic_ratio: SemanticRatio(semantic_ratio), embedder }
|
||||
});
|
||||
|
||||
ChatSearchParams {
|
||||
hybrid: Setting::some_or_not_set(hybrid),
|
||||
limit: Setting::some_or_not_set(limit),
|
||||
sort: Setting::some_or_not_set(sort),
|
||||
distinct: Setting::some_or_not_set(distinct),
|
||||
matching_strategy: Setting::some_or_not_set(matching_strategy),
|
||||
attributes_to_search_on: Setting::some_or_not_set(attributes_to_search_on),
|
||||
ranking_score_threshold: Setting::some_or_not_set(ranking_score_threshold),
|
||||
}
|
||||
}),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Default, Serialize, Deserialize, PartialEq, Deserr, ToSchema)]
|
||||
#[serde(deny_unknown_fields, rename_all = "camelCase")]
|
||||
#[deserr(error = JsonError, deny_unknown_fields, rename_all = camelCase)]
|
||||
pub struct ChatSearchParams {
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<HybridQuery>)]
|
||||
pub hybrid: Setting<HybridQuery>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default = Setting::Set(20))]
|
||||
#[schema(value_type = Option<usize>)]
|
||||
pub limit: Setting<usize>,
|
||||
|
||||
// #[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
// #[deserr(default)]
|
||||
// pub attributes_to_retrieve: Option<BTreeSet<String>>,
|
||||
|
||||
// #[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
// #[deserr(default)]
|
||||
// pub filter: Option<Value>,
|
||||
//
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<Vec<String>>)]
|
||||
pub sort: Setting<Vec<String>>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
pub distinct: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<MatchingStrategy>)]
|
||||
pub matching_strategy: Setting<MatchingStrategy>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<Vec<String>>)]
|
||||
pub attributes_to_search_on: Setting<Vec<String>>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<RankingScoreThreshold>)]
|
||||
pub ranking_score_threshold: Setting<RankingScoreThreshold>,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Default, Deserr, ToSchema, PartialEq, Serialize, Deserialize)]
|
||||
#[serde(rename_all = "camelCase")]
|
||||
#[deserr(error = JsonError, rename_all = camelCase, deny_unknown_fields)]
|
||||
pub struct HybridQuery {
|
||||
#[deserr(default)]
|
||||
#[serde(default)]
|
||||
#[schema(default, value_type = f32)]
|
||||
pub semantic_ratio: SemanticRatio,
|
||||
#[schema(value_type = String)]
|
||||
pub embedder: String,
|
||||
}
|
||||
|
||||
#[derive(Debug, Clone, Copy, Deserr, ToSchema, PartialEq, Serialize, Deserialize)]
|
||||
#[deserr(try_from(f32) = TryFrom::try_from -> InvalidSearchSemanticRatio)]
|
||||
pub struct SemanticRatio(f32);
|
||||
|
||||
impl Default for SemanticRatio {
|
||||
fn default() -> Self {
|
||||
SemanticRatio(0.5)
|
||||
}
|
||||
}
|
||||
|
||||
impl std::convert::TryFrom<f32> for SemanticRatio {
|
||||
type Error = InvalidSearchSemanticRatio;
|
||||
|
||||
fn try_from(f: f32) -> Result<Self, Self::Error> {
|
||||
// the suggested "fix" is: `!(0.0..=1.0).contains(&f)`` which is allegedly less readable
|
||||
#[allow(clippy::manual_range_contains)]
|
||||
if f > 1.0 || f < 0.0 {
|
||||
Err(InvalidSearchSemanticRatio)
|
||||
} else {
|
||||
Ok(SemanticRatio(f))
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
#[derive(Debug)]
|
||||
pub struct InvalidSearchSemanticRatio;
|
||||
|
||||
impl Error for InvalidSearchSemanticRatio {}
|
||||
|
||||
impl fmt::Display for InvalidSearchSemanticRatio {
|
||||
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
|
||||
write!(
|
||||
f,
|
||||
"the value of `semanticRatio` is invalid, expected a float between `0.0` and `1.0`."
|
||||
)
|
||||
}
|
||||
}
|
||||
|
||||
impl std::ops::Deref for SemanticRatio {
|
||||
type Target = f32;
|
||||
|
||||
fn deref(&self) -> &Self::Target {
|
||||
&self.0
|
||||
}
|
||||
}
|
||||
@@ -1,4 +1,5 @@
|
||||
pub use self::available_ids::AvailableIds;
|
||||
pub use self::chat::ChatSettings;
|
||||
pub use self::clear_documents::ClearDocuments;
|
||||
pub use self::concurrent_available_ids::ConcurrentAvailableIds;
|
||||
pub use self::facet::bulk::FacetsUpdateBulk;
|
||||
@@ -13,6 +14,7 @@ pub use self::words_prefix_integer_docids::WordPrefixIntegerDocids;
|
||||
pub use self::words_prefixes_fst::WordsPrefixesFst;
|
||||
|
||||
mod available_ids;
|
||||
mod chat;
|
||||
mod clear_documents;
|
||||
mod concurrent_available_ids;
|
||||
pub(crate) mod del_add;
|
||||
|
||||
@@ -4,6 +4,7 @@ use serde_json::Value;
|
||||
|
||||
use crate::attribute_patterns::PatternMatch;
|
||||
use crate::fields_ids_map::metadata::Metadata;
|
||||
use crate::filterable_attributes_rules::match_faceted_field;
|
||||
use crate::update::new::document::Document;
|
||||
use crate::update::new::extract::geo::extract_geo_coordinates;
|
||||
use crate::update::new::extract::perm_json_p;
|
||||
@@ -11,8 +12,6 @@ use crate::{
|
||||
FieldId, FilterableAttributesRule, GlobalFieldsIdsMap, InternalError, Result, UserError,
|
||||
};
|
||||
|
||||
use crate::filterable_attributes_rules::match_faceted_field;
|
||||
|
||||
#[allow(clippy::too_many_arguments)]
|
||||
pub fn extract_document_facets<'doc>(
|
||||
document: impl Document<'doc>,
|
||||
|
||||
@@ -18,7 +18,8 @@ pub use vectors::EmbeddingExtractor;
|
||||
pub mod perm_json_p {
|
||||
use serde_json::{Map, Value};
|
||||
|
||||
use crate::{attribute_patterns::PatternMatch, Result};
|
||||
use crate::attribute_patterns::PatternMatch;
|
||||
use crate::Result;
|
||||
const SPLIT_SYMBOL: char = '.';
|
||||
|
||||
/// Returns `true` if the `selector` match the `key`.
|
||||
|
||||
@@ -11,9 +11,10 @@ use roaring::RoaringBitmap;
|
||||
use serde::{Deserialize, Deserializer, Serialize, Serializer};
|
||||
use time::OffsetDateTime;
|
||||
|
||||
use super::chat::ChatSearchParams;
|
||||
use super::del_add::{DelAdd, DelAddOperation};
|
||||
use super::index_documents::{IndexDocumentsConfig, Transform};
|
||||
use super::IndexerConfig;
|
||||
use super::{ChatSettings, IndexerConfig};
|
||||
use crate::attribute_patterns::PatternMatch;
|
||||
use crate::constants::RESERVED_GEO_FIELD_NAME;
|
||||
use crate::criterion::Criterion;
|
||||
@@ -22,11 +23,11 @@ use crate::error::UserError;
|
||||
use crate::fields_ids_map::metadata::{FieldIdMapWithMetadata, MetadataBuilder};
|
||||
use crate::filterable_attributes_rules::match_faceted_field;
|
||||
use crate::index::{
|
||||
IndexEmbeddingConfig, PrefixSearch, DEFAULT_MIN_WORD_LEN_ONE_TYPO,
|
||||
DEFAULT_MIN_WORD_LEN_TWO_TYPOS,
|
||||
ChatConfig, IndexEmbeddingConfig, PrefixSearch, SearchParameters,
|
||||
DEFAULT_MIN_WORD_LEN_ONE_TYPO, DEFAULT_MIN_WORD_LEN_TWO_TYPOS,
|
||||
};
|
||||
use crate::order_by_map::OrderByMap;
|
||||
use crate::prompt::default_max_bytes;
|
||||
use crate::prompt::{default_max_bytes, default_template_text, PromptData};
|
||||
use crate::proximity::ProximityPrecision;
|
||||
use crate::update::index_documents::IndexDocumentsMethod;
|
||||
use crate::update::{IndexDocuments, UpdateIndexingStep};
|
||||
@@ -123,6 +124,15 @@ impl<T> Setting<T> {
|
||||
*self = new;
|
||||
true
|
||||
}
|
||||
|
||||
#[track_caller]
|
||||
pub fn unwrap(self) -> T {
|
||||
match self {
|
||||
Setting::Set(value) => value,
|
||||
Setting::Reset => panic!("Setting::Reset unwrapped"),
|
||||
Setting::NotSet => panic!("Setting::NotSet unwrapped"),
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
impl<T: Serialize> Serialize for Setting<T> {
|
||||
@@ -185,6 +195,7 @@ pub struct Settings<'a, 't, 'i> {
|
||||
localized_attributes_rules: Setting<Vec<LocalizedAttributesRule>>,
|
||||
prefix_search: Setting<PrefixSearch>,
|
||||
facet_search: Setting<bool>,
|
||||
chat: Setting<ChatSettings>,
|
||||
}
|
||||
|
||||
impl<'a, 't, 'i> Settings<'a, 't, 'i> {
|
||||
@@ -223,6 +234,7 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
|
||||
localized_attributes_rules: Setting::NotSet,
|
||||
prefix_search: Setting::NotSet,
|
||||
facet_search: Setting::NotSet,
|
||||
chat: Setting::NotSet,
|
||||
indexer_config,
|
||||
}
|
||||
}
|
||||
@@ -453,6 +465,14 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
|
||||
self.facet_search = Setting::Reset;
|
||||
}
|
||||
|
||||
pub fn set_chat(&mut self, value: ChatSettings) {
|
||||
self.chat = Setting::Set(value);
|
||||
}
|
||||
|
||||
pub fn reset_chat(&mut self) {
|
||||
self.chat = Setting::Reset;
|
||||
}
|
||||
|
||||
#[tracing::instrument(
|
||||
level = "trace"
|
||||
skip(self, progress_callback, should_abort, settings_diff),
|
||||
@@ -1238,6 +1258,109 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
|
||||
Ok(())
|
||||
}
|
||||
|
||||
fn update_chat_config(&mut self) -> heed::Result<bool> {
|
||||
match &mut self.chat {
|
||||
Setting::Set(ChatSettings {
|
||||
description: new_description,
|
||||
document_template: new_document_template,
|
||||
document_template_max_bytes: new_document_template_max_bytes,
|
||||
search_parameters: new_search_parameters,
|
||||
}) => {
|
||||
let ChatConfig { description, prompt, search_parameters } =
|
||||
self.index.chat_config(self.wtxn)?;
|
||||
|
||||
let description = match new_description {
|
||||
Setting::Set(new) => new.clone(),
|
||||
Setting::Reset => Default::default(),
|
||||
Setting::NotSet => description,
|
||||
};
|
||||
|
||||
let prompt = PromptData {
|
||||
template: match new_document_template {
|
||||
Setting::Set(new) => new.clone(),
|
||||
Setting::Reset => default_template_text().to_string(),
|
||||
Setting::NotSet => prompt.template.clone(),
|
||||
},
|
||||
max_bytes: match new_document_template_max_bytes {
|
||||
Setting::Set(m) => NonZeroUsize::new(*m),
|
||||
Setting::Reset => Some(default_max_bytes()),
|
||||
Setting::NotSet => prompt.max_bytes,
|
||||
},
|
||||
};
|
||||
|
||||
let search_parameters = match new_search_parameters {
|
||||
Setting::Set(sp) => {
|
||||
let ChatSearchParams {
|
||||
hybrid,
|
||||
limit,
|
||||
sort,
|
||||
distinct,
|
||||
matching_strategy,
|
||||
attributes_to_search_on,
|
||||
ranking_score_threshold,
|
||||
} = sp;
|
||||
|
||||
SearchParameters {
|
||||
hybrid: match hybrid {
|
||||
Setting::Set(hybrid) => Some(crate::index::HybridQuery {
|
||||
semantic_ratio: *hybrid.semantic_ratio,
|
||||
embedder: hybrid.embedder.clone(),
|
||||
}),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.hybrid.clone(),
|
||||
},
|
||||
limit: match limit {
|
||||
Setting::Set(limit) => Some(*limit),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.limit,
|
||||
},
|
||||
sort: match sort {
|
||||
Setting::Set(sort) => Some(sort.clone()),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.sort.clone(),
|
||||
},
|
||||
distinct: match distinct {
|
||||
Setting::Set(distinct) => Some(distinct.clone()),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.distinct.clone(),
|
||||
},
|
||||
matching_strategy: match matching_strategy {
|
||||
Setting::Set(matching_strategy) => Some(*matching_strategy),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.matching_strategy,
|
||||
},
|
||||
attributes_to_search_on: match attributes_to_search_on {
|
||||
Setting::Set(attributes_to_search_on) => {
|
||||
Some(attributes_to_search_on.clone())
|
||||
}
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => {
|
||||
search_parameters.attributes_to_search_on.clone()
|
||||
}
|
||||
},
|
||||
ranking_score_threshold: match ranking_score_threshold {
|
||||
Setting::Set(rst) => Some(*rst),
|
||||
Setting::Reset => None,
|
||||
Setting::NotSet => search_parameters.ranking_score_threshold,
|
||||
},
|
||||
}
|
||||
}
|
||||
Setting::Reset => Default::default(),
|
||||
Setting::NotSet => search_parameters,
|
||||
};
|
||||
|
||||
self.index.put_chat_config(
|
||||
self.wtxn,
|
||||
&ChatConfig { description, prompt, search_parameters },
|
||||
)?;
|
||||
|
||||
Ok(true)
|
||||
}
|
||||
Setting::Reset => self.index.delete_chat_config(self.wtxn),
|
||||
Setting::NotSet => Ok(false),
|
||||
}
|
||||
}
|
||||
|
||||
pub fn execute<FP, FA>(mut self, progress_callback: FP, should_abort: FA) -> Result<()>
|
||||
where
|
||||
FP: Fn(UpdateIndexingStep) + Sync,
|
||||
@@ -1275,6 +1398,7 @@ impl<'a, 't, 'i> Settings<'a, 't, 'i> {
|
||||
self.update_facet_search()?;
|
||||
self.update_localized_attributes_rules()?;
|
||||
self.update_disabled_typos_terms()?;
|
||||
self.update_chat_config()?;
|
||||
|
||||
let embedding_config_updates = self.update_embedding_configs()?;
|
||||
|
||||
|
||||
@@ -897,6 +897,7 @@ fn test_correct_settings_init() {
|
||||
prefix_search,
|
||||
facet_search,
|
||||
disable_on_numbers,
|
||||
chat,
|
||||
} = settings;
|
||||
assert!(matches!(searchable_fields, Setting::NotSet));
|
||||
assert!(matches!(displayed_fields, Setting::NotSet));
|
||||
@@ -925,6 +926,7 @@ fn test_correct_settings_init() {
|
||||
assert!(matches!(prefix_search, Setting::NotSet));
|
||||
assert!(matches!(facet_search, Setting::NotSet));
|
||||
assert!(matches!(disable_on_numbers, Setting::NotSet));
|
||||
assert!(matches!(chat, Setting::NotSet));
|
||||
})
|
||||
.unwrap();
|
||||
}
|
||||
|
||||
@@ -33,6 +33,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - Defaults to `openAi`
|
||||
pub source: Setting<EmbedderSource>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
@@ -55,6 +56,7 @@ pub struct EmbeddingSettings {
|
||||
/// - For source `openAi`, defaults to `text-embedding-3-small`
|
||||
/// - For source `huggingFace`, defaults to `BAAI/bge-base-en-v1.5`
|
||||
pub model: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
@@ -75,6 +77,7 @@ pub struct EmbeddingSettings {
|
||||
/// - When `model` is set to default, defaults to `617ca489d9e86b49b8167676d8220688b99db36e`
|
||||
/// - Otherwise, defaults to `null`
|
||||
pub revision: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<OverridePooling>)]
|
||||
@@ -96,6 +99,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - Embedders created before this parameter was available default to `forceMean` to preserve the existing behavior.
|
||||
pub pooling: Setting<OverridePooling>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
@@ -118,6 +122,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - This setting is partially hidden when returned by the settings
|
||||
pub api_key: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
@@ -141,6 +146,7 @@ pub struct EmbeddingSettings {
|
||||
/// - For source `openAi`, the dimensions is the maximum allowed by the model.
|
||||
/// - For sources `ollama` and `rest`, the dimensions are inferred by embedding a sample text.
|
||||
pub dimensions: Setting<usize>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<bool>)]
|
||||
@@ -167,6 +173,7 @@ pub struct EmbeddingSettings {
|
||||
/// first enabling it. If you are unsure of whether the performance-relevancy tradeoff is right for you,
|
||||
/// we recommend to use this parameter on a test index first.
|
||||
pub binary_quantized: Setting<bool>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<bool>)]
|
||||
@@ -183,6 +190,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - ๐๏ธ When modified, embeddings are regenerated for documents whose rendering through the template produces a different text.
|
||||
pub document_template: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<usize>)]
|
||||
@@ -201,6 +209,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - Defaults to 400
|
||||
pub document_template_max_bytes: Setting<usize>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<String>)]
|
||||
@@ -219,6 +228,7 @@ pub struct EmbeddingSettings {
|
||||
/// - ๐ฑ When modified for source `openAi`, embeddings are never regenerated
|
||||
/// - ๐๏ธ When modified for sources `ollama` and `rest`, embeddings are always regenerated
|
||||
pub url: Setting<String>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<serde_json::Value>)]
|
||||
@@ -236,6 +246,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - ๐๏ธ Changing the value of this parameter always regenerates embeddings
|
||||
pub request: Setting<serde_json::Value>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<serde_json::Value>)]
|
||||
@@ -253,6 +264,7 @@ pub struct EmbeddingSettings {
|
||||
///
|
||||
/// - ๐๏ธ Changing the value of this parameter always regenerates embeddings
|
||||
pub response: Setting<serde_json::Value>,
|
||||
|
||||
#[serde(default, skip_serializing_if = "Setting::is_not_set")]
|
||||
#[deserr(default)]
|
||||
#[schema(value_type = Option<BTreeMap<String, String>>)]
|
||||
|
||||
Reference in New Issue
Block a user