MeiliSearch/milli/tests/search/query_criteria.rs

471 lines
20 KiB
Rust
Raw Normal View History

use std::cmp::Reverse;
use std::io::Cursor;
2021-06-03 14:44:53 +02:00
use big_s::S;
use heed::EnvOpenOptions;
use itertools::Itertools;
use maplit::hashset;
use milli::documents::{DocumentsBatchBuilder, DocumentsBatchReader};
use milli::update::{IndexDocuments, IndexDocumentsConfig, IndexerConfig, Settings};
2022-08-18 17:36:08 +02:00
use milli::{AscDesc, Criterion, Index, Member, Search, SearchResult, TermsMatchingStrategy};
use rand::Rng;
2021-06-16 18:33:33 +02:00
use Criterion::*;
2021-06-03 14:44:53 +02:00
use crate::search::{self, EXTERNAL_DOCUMENTS_IDS};
2021-06-08 12:33:02 +02:00
const ALLOW_TYPOS: bool = true;
const DISALLOW_TYPOS: bool = false;
2022-08-18 17:36:08 +02:00
const ALLOW_OPTIONAL_WORDS: TermsMatchingStrategy = TermsMatchingStrategy::Last;
const DISALLOW_OPTIONAL_WORDS: TermsMatchingStrategy = TermsMatchingStrategy::All;
const ASC_DESC_CANDIDATES_THRESHOLD: usize = 1000;
2021-06-08 12:33:02 +02:00
macro_rules! test_criterion {
2021-08-23 11:37:18 +02:00
($func:ident, $optional_word:ident, $authorize_typos:ident, $criteria:expr, $sort_criteria:expr) => {
2021-06-08 12:33:02 +02:00
#[test]
fn $func() {
2021-06-17 14:24:59 +02:00
let criteria = $criteria;
2021-06-08 12:33:02 +02:00
let index = search::setup_search_index_with_criteria(&criteria);
2021-06-17 15:19:03 +02:00
let rtxn = index.read_txn().unwrap();
2021-06-08 12:33:02 +02:00
2021-06-17 15:19:03 +02:00
let mut search = Search::new(&rtxn, &index);
2021-06-08 12:33:02 +02:00
search.query(search::TEST_QUERY);
search.limit(EXTERNAL_DOCUMENTS_IDS.len());
search.authorize_typos($authorize_typos);
search.terms_matching_strategy($optional_word);
2021-08-23 11:37:18 +02:00
search.sort_criteria($sort_criteria);
2021-06-08 12:33:02 +02:00
2021-06-08 14:11:00 +02:00
let SearchResult { documents_ids, .. } = search.execute().unwrap();
2021-06-08 12:33:02 +02:00
2021-08-23 11:37:18 +02:00
let expected_external_ids: Vec<_> = search::expected_order(
&criteria,
$authorize_typos,
$optional_word,
&$sort_criteria[..],
)
.into_iter()
.map(|d| d.id)
.collect();
2021-06-08 12:33:02 +02:00
let documents_ids = search::internal_to_external_ids(&index, &documents_ids);
assert_eq!(documents_ids, expected_external_ids);
}
2021-06-17 13:56:09 +02:00
};
2021-06-08 12:33:02 +02:00
}
test_criterion!(none_allow_typo, DISALLOW_OPTIONAL_WORDS, ALLOW_TYPOS, vec![], vec![]);
2021-08-23 11:37:18 +02:00
test_criterion!(none_disallow_typo, DISALLOW_OPTIONAL_WORDS, DISALLOW_TYPOS, vec![], vec![]);
test_criterion!(words_allow_typo, ALLOW_OPTIONAL_WORDS, ALLOW_TYPOS, vec![Words], vec![]);
test_criterion!(
attribute_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
vec![Attribute],
vec![]
);
2022-02-02 18:45:11 +01:00
test_criterion!(typo, DISALLOW_OPTIONAL_WORDS, ALLOW_TYPOS, vec![Typo], vec![]);
2021-08-23 11:37:18 +02:00
test_criterion!(
attribute_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Attribute],
vec![]
);
test_criterion!(
exactness_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
vec![Exactness],
vec![]
);
test_criterion!(
exactness_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Exactness],
vec![]
);
test_criterion!(
proximity_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
vec![Proximity],
vec![]
);
test_criterion!(
proximity_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Proximity],
vec![]
);
2021-06-17 15:19:03 +02:00
test_criterion!(
asc_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Asc(S("asc_desc_rank"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
asc_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Asc(S("asc_desc_rank"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
desc_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Desc(S("asc_desc_rank"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
desc_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Desc(S("asc_desc_rank"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
asc_unexisting_field_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Asc(S("unexisting_field"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
asc_unexisting_field_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Asc(S("unexisting_field"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
desc_unexisting_field_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Desc(S("unexisting_field"))],
vec![]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
desc_unexisting_field_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Desc(S("unexisting_field"))],
vec![]
);
test_criterion!(empty_sort_by_allow_typo, DISALLOW_OPTIONAL_WORDS, ALLOW_TYPOS, vec![Sort], vec![]);
test_criterion!(
empty_sort_by_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Sort],
vec![]
);
test_criterion!(
sort_by_asc_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
vec![Sort],
2021-09-01 17:43:18 +02:00
vec![AscDesc::Asc(Member::Field(S("tag")))]
2021-08-23 11:37:18 +02:00
);
test_criterion!(
sort_by_asc_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Sort],
2021-09-01 17:43:18 +02:00
vec![AscDesc::Asc(Member::Field(S("tag")))]
2021-08-23 11:37:18 +02:00
);
test_criterion!(
sort_by_desc_allow_typo,
DISALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
vec![Sort],
2021-09-01 17:43:18 +02:00
vec![AscDesc::Desc(Member::Field(S("tag")))]
2021-08-23 11:37:18 +02:00
);
test_criterion!(
sort_by_desc_disallow_typo,
DISALLOW_OPTIONAL_WORDS,
DISALLOW_TYPOS,
vec![Sort],
2021-09-01 17:43:18 +02:00
vec![AscDesc::Desc(Member::Field(S("tag")))]
2021-06-17 15:19:03 +02:00
);
test_criterion!(
default_criteria_order,
ALLOW_OPTIONAL_WORDS,
ALLOW_TYPOS,
2021-08-23 11:37:18 +02:00
vec![Words, Typo, Proximity, Attribute, Exactness],
vec![]
2021-06-17 15:19:03 +02:00
);
2021-06-03 14:44:53 +02:00
#[test]
fn criteria_mixup() {
use Criterion::*;
let index = search::setup_search_index_with_criteria(&[
Words,
2021-06-16 18:33:33 +02:00
Attribute,
Desc(S("asc_desc_rank")),
Exactness,
Proximity,
Typo,
]);
2021-06-03 14:44:53 +02:00
2021-06-16 18:33:33 +02:00
#[rustfmt::skip]
2021-06-03 14:44:53 +02:00
let criteria_mix = {
2021-06-08 14:11:00 +02:00
// Criterion doesn't implement Copy, we create a new Criterion using a closure
2021-06-03 14:44:53 +02:00
let desc = || Desc(S("asc_desc_rank"));
// all possible criteria order
vec![
vec![Words, Attribute, desc(), Exactness, Proximity, Typo],
vec![Words, Attribute, desc(), Exactness, Typo, Proximity],
vec![Words, Attribute, desc(), Proximity, Exactness, Typo],
vec![Words, Attribute, desc(), Proximity, Typo, Exactness],
vec![Words, Attribute, desc(), Typo, Exactness, Proximity],
vec![Words, Attribute, desc(), Typo, Proximity, Exactness],
vec![Words, Attribute, Exactness, desc(), Proximity, Typo],
vec![Words, Attribute, Exactness, desc(), Typo, Proximity],
vec![Words, Attribute, Exactness, Proximity, desc(), Typo],
vec![Words, Attribute, Exactness, Proximity, Typo, desc()],
vec![Words, Attribute, Exactness, Typo, desc(), Proximity],
vec![Words, Attribute, Exactness, Typo, Proximity, desc()],
vec![Words, Attribute, Proximity, desc(), Exactness, Typo],
vec![Words, Attribute, Proximity, desc(), Typo, Exactness],
vec![Words, Attribute, Proximity, Exactness, desc(), Typo],
vec![Words, Attribute, Proximity, Exactness, Typo, desc()],
vec![Words, Attribute, Proximity, Typo, desc(), Exactness],
vec![Words, Attribute, Proximity, Typo, Exactness, desc()],
vec![Words, Attribute, Typo, desc(), Exactness, Proximity],
vec![Words, Attribute, Typo, desc(), Proximity, Exactness],
vec![Words, Attribute, Typo, Exactness, desc(), Proximity],
vec![Words, Attribute, Typo, Exactness, Proximity, desc()],
vec![Words, Attribute, Typo, Proximity, desc(), Exactness],
vec![Words, Attribute, Typo, Proximity, Exactness, desc()],
vec![Words, desc(), Attribute, Exactness, Proximity, Typo],
vec![Words, desc(), Attribute, Exactness, Typo, Proximity],
vec![Words, desc(), Attribute, Proximity, Exactness, Typo],
vec![Words, desc(), Attribute, Proximity, Typo, Exactness],
vec![Words, desc(), Attribute, Typo, Exactness, Proximity],
vec![Words, desc(), Attribute, Typo, Proximity, Exactness],
vec![Words, desc(), Exactness, Attribute, Proximity, Typo],
vec![Words, desc(), Exactness, Attribute, Typo, Proximity],
vec![Words, desc(), Exactness, Proximity, Attribute, Typo],
vec![Words, desc(), Exactness, Proximity, Typo, Attribute],
vec![Words, desc(), Exactness, Typo, Attribute, Proximity],
vec![Words, desc(), Exactness, Typo, Proximity, Attribute],
vec![Words, desc(), Proximity, Attribute, Exactness, Typo],
vec![Words, desc(), Proximity, Attribute, Typo, Exactness],
vec![Words, desc(), Proximity, Exactness, Attribute, Typo],
vec![Words, desc(), Proximity, Exactness, Typo, Attribute],
vec![Words, desc(), Proximity, Typo, Attribute, Exactness],
vec![Words, desc(), Proximity, Typo, Exactness, Attribute],
vec![Words, desc(), Typo, Attribute, Exactness, Proximity],
vec![Words, desc(), Typo, Attribute, Proximity, Exactness],
vec![Words, desc(), Typo, Exactness, Attribute, Proximity],
vec![Words, desc(), Typo, Exactness, Proximity, Attribute],
vec![Words, desc(), Typo, Proximity, Attribute, Exactness],
vec![Words, desc(), Typo, Proximity, Exactness, Attribute],
vec![Words, Exactness, Attribute, desc(), Proximity, Typo],
vec![Words, Exactness, Attribute, desc(), Typo, Proximity],
vec![Words, Exactness, Attribute, Proximity, desc(), Typo],
vec![Words, Exactness, Attribute, Proximity, Typo, desc()],
vec![Words, Exactness, Attribute, Typo, desc(), Proximity],
vec![Words, Exactness, Attribute, Typo, Proximity, desc()],
vec![Words, Exactness, desc(), Attribute, Proximity, Typo],
vec![Words, Exactness, desc(), Attribute, Typo, Proximity],
vec![Words, Exactness, desc(), Proximity, Attribute, Typo],
vec![Words, Exactness, desc(), Proximity, Typo, Attribute],
vec![Words, Exactness, desc(), Typo, Attribute, Proximity],
vec![Words, Exactness, desc(), Typo, Proximity, Attribute],
vec![Words, Exactness, Proximity, Attribute, desc(), Typo],
vec![Words, Exactness, Proximity, Attribute, Typo, desc()],
vec![Words, Exactness, Proximity, desc(), Attribute, Typo],
vec![Words, Exactness, Proximity, desc(), Typo, Attribute],
vec![Words, Exactness, Proximity, Typo, Attribute, desc()],
vec![Words, Exactness, Proximity, Typo, desc(), Attribute],
vec![Words, Exactness, Typo, Attribute, desc(), Proximity],
vec![Words, Exactness, Typo, Attribute, Proximity, desc()],
vec![Words, Exactness, Typo, desc(), Attribute, Proximity],
vec![Words, Exactness, Typo, desc(), Proximity, Attribute],
vec![Words, Exactness, Typo, Proximity, Attribute, desc()],
vec![Words, Exactness, Typo, Proximity, desc(), Attribute],
vec![Words, Proximity, Attribute, desc(), Exactness, Typo],
vec![Words, Proximity, Attribute, desc(), Typo, Exactness],
vec![Words, Proximity, Attribute, Exactness, desc(), Typo],
vec![Words, Proximity, Attribute, Exactness, Typo, desc()],
vec![Words, Proximity, Attribute, Typo, desc(), Exactness],
vec![Words, Proximity, Attribute, Typo, Exactness, desc()],
vec![Words, Proximity, desc(), Attribute, Exactness, Typo],
vec![Words, Proximity, desc(), Attribute, Typo, Exactness],
vec![Words, Proximity, desc(), Exactness, Attribute, Typo],
vec![Words, Proximity, desc(), Exactness, Typo, Attribute],
vec![Words, Proximity, desc(), Typo, Attribute, Exactness],
vec![Words, Proximity, desc(), Typo, Exactness, Attribute],
vec![Words, Proximity, Exactness, Attribute, desc(), Typo],
vec![Words, Proximity, Exactness, Attribute, Typo, desc()],
vec![Words, Proximity, Exactness, desc(), Attribute, Typo],
vec![Words, Proximity, Exactness, desc(), Typo, Attribute],
vec![Words, Proximity, Exactness, Typo, Attribute, desc()],
vec![Words, Proximity, Exactness, Typo, desc(), Attribute],
vec![Words, Proximity, Typo, Attribute, desc(), Exactness],
vec![Words, Proximity, Typo, Attribute, Exactness, desc()],
vec![Words, Proximity, Typo, desc(), Attribute, Exactness],
vec![Words, Proximity, Typo, desc(), Exactness, Attribute],
vec![Words, Proximity, Typo, Exactness, Attribute, desc()],
vec![Words, Proximity, Typo, Exactness, desc(), Attribute],
vec![Words, Typo, Attribute, desc(), Exactness, Proximity],
vec![Words, Typo, Attribute, desc(), Proximity, Exactness],
vec![Words, Typo, Attribute, Exactness, desc(), Proximity],
vec![Words, Typo, Attribute, Exactness, Proximity, desc()],
vec![Words, Typo, Attribute, Proximity, desc(), Exactness],
vec![Words, Typo, Attribute, Proximity, Exactness, desc()],
vec![Words, Typo, desc(), Attribute, Proximity, Exactness],
vec![Words, Typo, desc(), Exactness, Attribute, Proximity],
vec![Words, Typo, desc(), Exactness, Attribute, Proximity],
vec![Words, Typo, desc(), Exactness, Proximity, Attribute],
vec![Words, Typo, desc(), Proximity, Attribute, Exactness],
vec![Words, Typo, desc(), Proximity, Exactness, Attribute],
vec![Words, Typo, Exactness, Attribute, desc(), Proximity],
vec![Words, Typo, Exactness, Attribute, Proximity, desc()],
vec![Words, Typo, Exactness, desc(), Attribute, Proximity],
vec![Words, Typo, Exactness, desc(), Proximity, Attribute],
vec![Words, Typo, Exactness, Proximity, Attribute, desc()],
vec![Words, Typo, Exactness, Proximity, desc(), Attribute],
vec![Words, Typo, Proximity, Attribute, desc(), Exactness],
vec![Words, Typo, Proximity, Attribute, Exactness, desc()],
vec![Words, Typo, Proximity, desc(), Attribute, Exactness],
vec![Words, Typo, Proximity, desc(), Exactness, Attribute],
vec![Words, Typo, Proximity, Exactness, Attribute, desc()],
vec![Words, Typo, Proximity, Exactness, desc(), Attribute],
]
};
let config = IndexerConfig::default();
2021-06-03 14:44:53 +02:00
for criteria in criteria_mix {
eprintln!("Testing with criteria order: {:?}", &criteria);
//update criteria
let mut wtxn = index.write_txn().unwrap();
let mut builder = Settings::new(&mut wtxn, &index, &config);
builder.set_criteria(criteria.iter().map(ToString::to_string).collect());
builder.execute(|_| (), || false).unwrap();
2021-06-03 14:44:53 +02:00
wtxn.commit().unwrap();
let mut rtxn = index.read_txn().unwrap();
let mut search = Search::new(&mut rtxn, &index);
search.query(search::TEST_QUERY);
search.limit(EXTERNAL_DOCUMENTS_IDS.len());
search.terms_matching_strategy(ALLOW_OPTIONAL_WORDS);
2021-06-08 14:11:00 +02:00
search.authorize_typos(ALLOW_TYPOS);
2021-06-03 14:44:53 +02:00
2021-06-08 14:11:00 +02:00
let SearchResult { documents_ids, .. } = search.execute().unwrap();
2021-06-03 14:44:53 +02:00
2021-06-16 18:33:33 +02:00
let expected_external_ids: Vec<_> =
2022-08-18 17:36:08 +02:00
search::expected_order(&criteria, ALLOW_TYPOS, ALLOW_OPTIONAL_WORDS, &[])
2021-06-16 18:33:33 +02:00
.into_iter()
.map(|d| d.id)
.collect();
2021-06-03 14:44:53 +02:00
let documents_ids = search::internal_to_external_ids(&index, &documents_ids);
assert_eq!(documents_ids, expected_external_ids);
}
}
#[test]
fn criteria_ascdesc() {
let path = tempfile::tempdir().unwrap();
let mut options = EnvOpenOptions::new();
options.map_size(12 * 1024 * 1024); // 10 MB
let index = Index::new(options, &path).unwrap();
let mut wtxn = index.write_txn().unwrap();
let config = IndexerConfig::default();
let mut builder = Settings::new(&mut wtxn, &index, &config);
builder.set_sortable_fields(hashset! {
S("name"),
S("age"),
});
builder.execute(|_| (), || false).unwrap();
// index documents
let config = IndexerConfig { max_memory: Some(10 * 1024 * 1024), ..Default::default() };
let indexing_config = IndexDocumentsConfig { autogenerate_docids: true, ..Default::default() };
let builder =
IndexDocuments::new(&mut wtxn, &index, &config, indexing_config, |_| (), || false).unwrap();
let mut batch_builder = DocumentsBatchBuilder::new(Vec::new());
(0..ASC_DESC_CANDIDATES_THRESHOLD + 1).for_each(|_| {
let mut rng = rand::thread_rng();
let age = rng.gen::<u32>().to_string();
let name = rng
.sample_iter(&rand::distributions::Alphanumeric)
.map(char::from)
.filter(|c| *c >= 'a' && *c <= 'z')
.take(10)
.collect::<String>();
let json = serde_json::json!({
"name": name,
"age": age,
});
let object = match json {
serde_json::Value::Object(object) => object,
_ => panic!(),
};
batch_builder.append_json_object(&object).unwrap();
});
let vector = batch_builder.into_inner().unwrap();
let reader = DocumentsBatchReader::from_reader(Cursor::new(vector)).unwrap();
2022-06-15 14:35:19 +02:00
let (builder, user_error) = builder.add_documents(reader).unwrap();
user_error.unwrap();
builder.execute().unwrap();
wtxn.commit().unwrap();
let rtxn = index.read_txn().unwrap();
let documents = index.all_documents(&rtxn).unwrap().map(|doc| doc.unwrap()).collect::<Vec<_>>();
for criterion in [Asc(S("name")), Desc(S("name")), Asc(S("age")), Desc(S("age"))] {
eprintln!("Testing with criterion: {:?}", &criterion);
let mut wtxn = index.write_txn().unwrap();
let mut builder = Settings::new(&mut wtxn, &index, &config);
builder.set_criteria(vec![criterion.to_string()]);
builder.execute(|_| (), || false).unwrap();
wtxn.commit().unwrap();
let mut rtxn = index.read_txn().unwrap();
let mut search = Search::new(&mut rtxn, &index);
search.limit(ASC_DESC_CANDIDATES_THRESHOLD + 1);
let SearchResult { documents_ids, .. } = search.execute().unwrap();
let expected_document_ids = match criterion {
Asc(field_name) if field_name == "name" => {
documents.iter().sorted_by_key(|(_, obkv)| obkv.get(0).unwrap())
}
Desc(field_name) if field_name == "name" => {
documents.iter().sorted_by_key(|(_, obkv)| Reverse(obkv.get(0).unwrap()))
}
Asc(field_name) if field_name == "name" => {
documents.iter().sorted_by_key(|(_, obkv)| obkv.get(1).unwrap())
}
Desc(field_name) if field_name == "name" => {
documents.iter().sorted_by_key(|(_, obkv)| Reverse(obkv.get(1).unwrap()))
}
_ => continue,
}
.map(|(id, _)| *id)
.collect::<Vec<_>>();
assert_eq!(documents_ids, expected_document_ids);
}
}