Changes to multi_search route

This commit is contained in:
Louis Dureuil 2024-07-11 16:37:01 +02:00
parent 5c323cecc7
commit 83d71662aa
No known key found for this signature in database

View File

@ -10,12 +10,14 @@ use serde::Serialize;
use tracing::debug; use tracing::debug;
use crate::analytics::{Analytics, MultiSearchAggregator}; use crate::analytics::{Analytics, MultiSearchAggregator};
use crate::error::MeilisearchHttpError;
use crate::extractors::authentication::policies::ActionPolicy; use crate::extractors::authentication::policies::ActionPolicy;
use crate::extractors::authentication::{AuthenticationError, GuardedData}; use crate::extractors::authentication::{AuthenticationError, GuardedData};
use crate::extractors::sequential_extractor::SeqHandler; use crate::extractors::sequential_extractor::SeqHandler;
use crate::routes::indexes::search::search_kind; use crate::routes::indexes::search::search_kind;
use crate::search::{ use crate::search::{
add_search_rules, perform_search, RetrieveVectors, SearchQueryWithIndex, SearchResultWithIndex, add_search_rules, perform_federated_search, perform_search, FederatedSearch, RetrieveVectors,
SearchQueryWithIndex, SearchResultWithIndex,
}; };
use crate::search_queue::SearchQueue; use crate::search_queue::SearchQueue;
@ -28,46 +30,86 @@ struct SearchResults {
results: Vec<SearchResultWithIndex>, results: Vec<SearchResultWithIndex>,
} }
#[derive(Debug, deserr::Deserr)]
#[deserr(error = DeserrJsonError, rename_all = camelCase, deny_unknown_fields)]
pub struct SearchQueries {
queries: Vec<SearchQueryWithIndex>,
}
pub async fn multi_search_with_post( pub async fn multi_search_with_post(
index_scheduler: GuardedData<ActionPolicy<{ actions::SEARCH }>, Data<IndexScheduler>>, index_scheduler: GuardedData<ActionPolicy<{ actions::SEARCH }>, Data<IndexScheduler>>,
search_queue: Data<SearchQueue>, search_queue: Data<SearchQueue>,
params: AwebJson<SearchQueries, DeserrJsonError>, params: AwebJson<FederatedSearch, DeserrJsonError>,
req: HttpRequest, req: HttpRequest,
analytics: web::Data<dyn Analytics>, analytics: web::Data<dyn Analytics>,
) -> Result<HttpResponse, ResponseError> { ) -> Result<HttpResponse, ResponseError> {
let queries = params.into_inner().queries;
let mut multi_aggregate = MultiSearchAggregator::from_queries(&queries, &req);
let features = index_scheduler.features();
// Since we don't want to process half of the search requests and then get a permit refused // Since we don't want to process half of the search requests and then get a permit refused
// we're going to get one permit for the whole duration of the multi-search request. // we're going to get one permit for the whole duration of the multi-search request.
let _permit = search_queue.try_get_search_permit().await?; let _permit = search_queue.try_get_search_permit().await?;
let federated_search = params.into_inner();
let mut multi_aggregate = MultiSearchAggregator::from_federated_search(&federated_search, &req);
let FederatedSearch { mut queries, federation } = federated_search;
let features = index_scheduler.features();
// regardless of federation, check authorization and apply search rules
let auth = 'check_authorization: {
for (query_index, federated_query) in queries.iter_mut().enumerate() {
let index_uid = federated_query.index_uid.as_str();
// Check index from API key
if !index_scheduler.filters().is_index_authorized(index_uid) {
break 'check_authorization Err(AuthenticationError::InvalidToken)
.with_index(query_index);
}
// Apply search rules from tenant token
if let Some(search_rules) = index_scheduler.filters().get_index_search_rules(index_uid)
{
add_search_rules(&mut federated_query.filter, search_rules);
}
}
Ok(())
};
auth.map_err(|(mut err, query_index)| {
// Add the query index that failed as context for the error message.
// We're doing it only here and not directly in the `WithIndex` trait so that the `with_index` function returns a different type
// of result and we can benefit from static typing.
err.message = format!("Inside `.queries[{query_index}]`: {}", err.message);
err
})?;
let response = match federation {
Some(federation) => {
let search_result = tokio::task::spawn_blocking(move || {
perform_federated_search(&index_scheduler, queries, federation, features)
})
.await;
if let Ok(Ok(_)) = search_result {
multi_aggregate.succeed();
}
analytics.post_multi_search(multi_aggregate);
HttpResponse::Ok().json(search_result??)
}
None => {
// Explicitly expect a `(ResponseError, usize)` for the error type rather than `ResponseError` only, // Explicitly expect a `(ResponseError, usize)` for the error type rather than `ResponseError` only,
// so that `?` doesn't work if it doesn't use `with_index`, ensuring that it is not forgotten in case of code // so that `?` doesn't work if it doesn't use `with_index`, ensuring that it is not forgotten in case of code
// changes. // changes.
let search_results: Result<_, (ResponseError, usize)> = async { let search_results: Result<_, (ResponseError, usize)> = async {
let mut search_results = Vec::with_capacity(queries.len()); let mut search_results = Vec::with_capacity(queries.len());
for (query_index, (index_uid, mut query)) in for (query_index, (index_uid, query, federation_options)) in queries
queries.into_iter().map(SearchQueryWithIndex::into_index_query).enumerate() .into_iter()
.map(SearchQueryWithIndex::into_index_query_federation)
.enumerate()
{ {
debug!(on_index = query_index, parameters = ?query, "Multi-search"); debug!(on_index = query_index, parameters = ?query, "Multi-search");
// Check index from API key if federation_options.is_some() {
if !index_scheduler.filters().is_index_authorized(&index_uid) { return Err((
return Err(AuthenticationError::InvalidToken).with_index(query_index); MeilisearchHttpError::FederationOptionsInNonFederatedRequest(
} query_index,
// Apply search rules from tenant token )
if let Some(search_rules) = index_scheduler.filters().get_index_search_rules(&index_uid) .into(),
{ query_index,
add_search_rules(&mut query.filter, search_rules); ));
} }
let index = index_scheduler let index = index_scheduler
@ -81,10 +123,11 @@ pub async fn multi_search_with_post(
}) })
.with_index(query_index)?; .with_index(query_index)?;
let search_kind = search_kind(&query, index_scheduler.get_ref(), &index, features) let search_kind =
search_kind(&query, index_scheduler.get_ref(), &index, features)
.with_index(query_index)?;
let retrieve_vector = RetrieveVectors::new(query.retrieve_vectors, features)
.with_index(query_index)?; .with_index(query_index)?;
let retrieve_vector =
RetrieveVectors::new(query.retrieve_vectors, features).with_index(query_index)?;
let search_result = tokio::task::spawn_blocking(move || { let search_result = tokio::task::spawn_blocking(move || {
perform_search(&index, query, search_kind, retrieve_vector) perform_search(&index, query, search_kind, retrieve_vector)
@ -116,7 +159,11 @@ pub async fn multi_search_with_post(
debug!(returns = ?search_results, "Multi-search"); debug!(returns = ?search_results, "Multi-search");
Ok(HttpResponse::Ok().json(SearchResults { results: search_results })) HttpResponse::Ok().json(SearchResults { results: search_results })
}
};
Ok(response)
} }
/// Local `Result` extension trait to avoid `map_err` boilerplate. /// Local `Result` extension trait to avoid `map_err` boilerplate.