mirror of
https://github.com/meilisearch/meilisearch.git
synced 2024-11-23 02:27:40 +08:00
Merge #2281
2281: Hard limit the number of results returned by a search r=Kerollmops a=Kerollmops This PR fixes #2133 by hard-limiting the number of results that a search request can return at any time. I would like the guidance of `@MarinPostma` to test that, should I use a mocking test here? Or should I do anything else? I talked about touching the _nb_hits_ value with `@qdequele` and we concluded that it was not correct to do so. Could you please confirm that it is the right place to change that? Co-authored-by: Kerollmops <clement@meilisearch.com>
This commit is contained in:
commit
09a72cee03
@ -267,3 +267,79 @@ async fn displayed_attributes() {
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
assert!(response["hits"].get("title").is_none());
|
||||
}
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn placeholder_search_is_hard_limited() {
|
||||
let server = Server::new().await;
|
||||
let index = server.index("test");
|
||||
|
||||
let documents: Vec<_> = (0..1200)
|
||||
.map(|i| json!({ "id": i, "text": "I am unique!" }))
|
||||
.collect();
|
||||
index.add_documents(documents.into(), None).await;
|
||||
index.wait_task(0).await;
|
||||
|
||||
index
|
||||
.search(
|
||||
json!({
|
||||
"limit": 1500,
|
||||
}),
|
||||
|response, code| {
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
assert_eq!(response["hits"].as_array().unwrap().len(), 1000);
|
||||
},
|
||||
)
|
||||
.await;
|
||||
|
||||
index
|
||||
.search(
|
||||
json!({
|
||||
"offset": 800,
|
||||
"limit": 400,
|
||||
}),
|
||||
|response, code| {
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
assert_eq!(response["hits"].as_array().unwrap().len(), 200);
|
||||
},
|
||||
)
|
||||
.await;
|
||||
}
|
||||
|
||||
#[actix_rt::test]
|
||||
async fn search_is_hard_limited() {
|
||||
let server = Server::new().await;
|
||||
let index = server.index("test");
|
||||
|
||||
let documents: Vec<_> = (0..1200)
|
||||
.map(|i| json!({ "id": i, "text": "I am unique!" }))
|
||||
.collect();
|
||||
index.add_documents(documents.into(), None).await;
|
||||
index.wait_task(0).await;
|
||||
|
||||
index
|
||||
.search(
|
||||
json!({
|
||||
"q": "unique",
|
||||
"limit": 1500,
|
||||
}),
|
||||
|response, code| {
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
assert_eq!(response["hits"].as_array().unwrap().len(), 1000);
|
||||
},
|
||||
)
|
||||
.await;
|
||||
|
||||
index
|
||||
.search(
|
||||
json!({
|
||||
"q": "unique",
|
||||
"offset": 800,
|
||||
"limit": 400,
|
||||
}),
|
||||
|response, code| {
|
||||
assert_eq!(code, 200, "{}", response);
|
||||
assert_eq!(response["hits"].as_array().unwrap().len(), 200);
|
||||
},
|
||||
)
|
||||
.await;
|
||||
}
|
||||
|
@ -1,3 +1,4 @@
|
||||
use std::cmp::min;
|
||||
use std::collections::{BTreeMap, BTreeSet, HashSet};
|
||||
use std::str::FromStr;
|
||||
use std::time::Instant;
|
||||
@ -34,6 +35,10 @@ pub const fn default_crop_length() -> usize {
|
||||
DEFAULT_CROP_LENGTH
|
||||
}
|
||||
|
||||
/// The maximimum number of results that the engine
|
||||
/// will be able to return in one search call.
|
||||
pub const HARD_RESULT_LIMIT: usize = 1000;
|
||||
|
||||
#[derive(Deserialize, Debug, Clone, PartialEq)]
|
||||
#[serde(rename_all = "camelCase", deny_unknown_fields)]
|
||||
pub struct SearchQuery {
|
||||
@ -97,8 +102,13 @@ impl Index {
|
||||
search.query(query);
|
||||
}
|
||||
|
||||
search.limit(query.limit);
|
||||
search.offset(query.offset.unwrap_or_default());
|
||||
// Make sure that a user can't get more documents than the hard limit,
|
||||
// we align that on the offset too.
|
||||
let offset = min(query.offset.unwrap_or(0), HARD_RESULT_LIMIT);
|
||||
let limit = min(query.limit, HARD_RESULT_LIMIT.saturating_sub(offset));
|
||||
|
||||
search.offset(offset);
|
||||
search.limit(limit);
|
||||
|
||||
if let Some(ref filter) = query.filter {
|
||||
if let Some(facets) = parse_filter(filter)? {
|
||||
|
Loading…
Reference in New Issue
Block a user