Move the documents MTBL database inside the Index

This commit is contained in:
Clément Renault
2020-08-10 13:47:19 +02:00
parent ecd2b2f217
commit 394844062f
5 changed files with 75 additions and 40 deletions

View File

@ -9,7 +9,6 @@ use std::time::Instant;
use askama_warp::Template;
use heed::EnvOpenOptions;
use oxidized_mtbl::Reader;
use serde::Deserialize;
use slice_group_by::StrGroupBy;
use structopt::StructOpt;
@ -99,22 +98,13 @@ async fn main() -> anyhow::Result<()> {
.open(&opt.database)?;
// Open the LMDB database.
let index = Index::new(&env)?;
// Open the documents MTBL database.
let path = opt.database.join("documents.mtbl");
let file = File::open(path)?;
let mmap = unsafe { memmap::Mmap::map(&file)? };
let mmap = TransitiveArc(Arc::new(mmap));
let documents = Reader::new(mmap)?;
let index = Index::new(&env, &opt.database)?;
// Retrieve the database the file stem (w/o the extension),
// the disk file size and the number of documents in the database.
let db_name = opt.database.file_stem().and_then(|s| s.to_str()).unwrap_or("").to_string();
let db_size = File::open(opt.database.join("data.mdb"))?.metadata()?.len() as usize;
// Retrieve the documents count.
let docs_count = documents.metadata().count_entries;
let docs_count = index.number_of_documents();
// We run and wait on the HTTP server
@ -198,7 +188,6 @@ async fn main() -> anyhow::Result<()> {
}
let env_cloned = env.clone();
let documents_cloned = documents.clone();
let disable_highlighting = opt.disable_highlighting;
let query_route = warp::filters::method::post()
.and(warp::path!("query"))
@ -213,13 +202,10 @@ async fn main() -> anyhow::Result<()> {
if let Some(headers) = index.headers(&rtxn).unwrap() {
// We write the headers
body.extend_from_slice(headers);
let documents = index.documents(documents_ids).unwrap();
for id in documents_ids {
let id_bytes = id.to_be_bytes();
let content = documents_cloned.clone().get(&id_bytes).unwrap();
let content = content.expect(&format!("could not find document {}", id));
for (_id, content) in documents {
let content = std::str::from_utf8(content.as_ref()).unwrap();
let content = if disable_highlighting {
Cow::from(content)
} else {