mirror of
https://github.com/meilisearch/MeiliSearch
synced 2024-12-23 21:20:24 +01:00
test(update, settings): set & reset synonyms
fixes after review more fixes after review
This commit is contained in:
parent
e39aabbfe6
commit
33860bc3b7
@ -1021,11 +1021,11 @@ mod tests {
|
||||
faceted_attributes: Setting::Set(hashmap! { "age".into() => "integer".into() }),
|
||||
criteria: Setting::Set(vec!["asc(age)".to_string()]),
|
||||
stop_words: Setting::Set(btreeset! { "and".to_string() }),
|
||||
synonyms: Setting::NotSet
|
||||
synonyms: Setting::Set(hashmap! { "alex".to_string() => vec!["alexey".to_string()] })
|
||||
};
|
||||
|
||||
assert_tokens(&settings, &[
|
||||
Token::Struct { name: "Settings", len: 5 },
|
||||
Token::Struct { name: "Settings", len: 6 },
|
||||
Token::Str("displayedAttributes"),
|
||||
Token::Some,
|
||||
Token::Seq { len: Some(1) },
|
||||
@ -1052,6 +1052,14 @@ mod tests {
|
||||
Token::Seq { len: Some(1) },
|
||||
Token::Str("and"),
|
||||
Token::SeqEnd,
|
||||
Token::Str("synonyms"),
|
||||
Token::Some,
|
||||
Token::Map { len: Some(1) },
|
||||
Token::Str("alex"),
|
||||
Token::Seq {len: Some(1) },
|
||||
Token::Str("alexey"),
|
||||
Token::SeqEnd,
|
||||
Token::MapEnd,
|
||||
Token::StructEnd,
|
||||
]);
|
||||
}
|
||||
@ -1064,11 +1072,11 @@ mod tests {
|
||||
faceted_attributes: Setting::Reset,
|
||||
criteria: Setting::Reset,
|
||||
stop_words: Setting::Reset,
|
||||
synonyms: Setting::NotSet
|
||||
synonyms: Setting::Reset,
|
||||
};
|
||||
|
||||
assert_tokens(&settings, &[
|
||||
Token::Struct { name: "Settings", len: 5 },
|
||||
Token::Struct { name: "Settings", len: 6 },
|
||||
Token::Str("displayedAttributes"),
|
||||
Token::None,
|
||||
Token::Str("searchableAttributes"),
|
||||
@ -1079,6 +1087,8 @@ mod tests {
|
||||
Token::None,
|
||||
Token::Str("stopWords"),
|
||||
Token::None,
|
||||
Token::Str("synonyms"),
|
||||
Token::None,
|
||||
Token::StructEnd,
|
||||
]);
|
||||
}
|
||||
@ -1091,6 +1101,7 @@ mod tests {
|
||||
faceted_attributes: Setting::NotSet,
|
||||
criteria: Setting::NotSet,
|
||||
stop_words: Setting::NotSet,
|
||||
synonyms: Setting::NotSet,
|
||||
};
|
||||
|
||||
assert_tokens(&settings, &[
|
||||
|
@ -417,22 +417,13 @@ impl Index {
|
||||
self.main.delete::<_, Str>(wtxn, SYNONYMS_KEY)
|
||||
}
|
||||
|
||||
pub fn synonyms(&self, rtxn: &RoTxn) -> anyhow::Result<Option<HashMap<Vec<String>, Vec<Vec<String>>>>> {
|
||||
match self.main.get::<_, Str, SerdeBincode<HashMap<Vec<String>, Vec<Vec<String>>>>>(rtxn, SYNONYMS_KEY)? {
|
||||
Some(synonyms) => Ok(Some(synonyms)),
|
||||
None => Ok(None),
|
||||
}
|
||||
pub fn synonyms(&self, rtxn: &RoTxn) -> heed::Result<HashMap<Vec<String>, Vec<Vec<String>>>> {
|
||||
Ok(self.main.get::<_, Str, SerdeBincode<_>>(rtxn, SYNONYMS_KEY)?.unwrap_or_default())
|
||||
}
|
||||
|
||||
pub fn words_synonyms<S: AsRef<str>>(&self, rtxn: &RoTxn, words: &[S]) -> anyhow::Result<Option<Vec<Vec<String>>>> {
|
||||
let words: Vec<_> = words.iter().map(|s| s.as_ref().to_string()).collect();
|
||||
|
||||
match self.synonyms(rtxn)? {
|
||||
Some(synonyms) => Ok(Some(
|
||||
synonyms.get(&words).cloned().unwrap_or(Vec::default())
|
||||
)),
|
||||
None => Ok(None)
|
||||
}
|
||||
pub fn words_synonyms<S: AsRef<str>>(&self, rtxn: &RoTxn, words: &[S]) -> heed::Result<Option<Vec<Vec<String>>>> {
|
||||
let words: Vec<_> = words.iter().map(|s| s.as_ref().to_owned()).collect();
|
||||
Ok(self.synonyms(rtxn)?.remove(&words))
|
||||
}
|
||||
|
||||
/* words prefixes fst */
|
||||
|
@ -155,7 +155,7 @@ impl fmt::Debug for Query {
|
||||
|
||||
trait Context {
|
||||
fn word_docids(&self, word: &str) -> heed::Result<Option<RoaringBitmap>>;
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> anyhow::Result<Option<Vec<Vec<String>>>>;
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> heed::Result<Option<Vec<Vec<String>>>>;
|
||||
fn word_documents_count(&self, word: &str) -> heed::Result<Option<u64>> {
|
||||
match self.word_docids(word)? {
|
||||
Some(rb) => Ok(Some(rb.len())),
|
||||
@ -177,7 +177,7 @@ impl<'a> Context for QueryTreeBuilder<'a> {
|
||||
self.index.word_docids.get(self.rtxn, word)
|
||||
}
|
||||
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> anyhow::Result<Option<Vec<Vec<String>>>> {
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> heed::Result<Option<Vec<Vec<String>>>> {
|
||||
self.index.words_synonyms(self.rtxn, words)
|
||||
}
|
||||
|
||||
@ -270,10 +270,10 @@ fn typos(word: String, authorize_typos: bool) -> QueryKind {
|
||||
}
|
||||
}
|
||||
|
||||
/// Fetch synonyms from the `Context` for the provided words
|
||||
/// Fetch synonyms from the `Context` for the provided word
|
||||
/// and create the list of operations for the query tree
|
||||
fn synonyms(ctx: &impl Context, words: &[&str]) -> anyhow::Result<Option<Vec<Operation>>> {
|
||||
let synonyms = ctx.synonyms(words)?;
|
||||
fn synonyms(ctx: &impl Context, word: &[&str]) -> heed::Result<Option<Vec<Operation>>> {
|
||||
let synonyms = ctx.synonyms(word)?;
|
||||
|
||||
Ok(synonyms.map(|synonyms| {
|
||||
synonyms.into_iter().map(|synonym| {
|
||||
@ -581,8 +581,8 @@ mod test {
|
||||
Ok(self.postings.get(word).cloned())
|
||||
}
|
||||
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> anyhow::Result<Option<Vec<Vec<String>>>> {
|
||||
let words: Vec<_> = words.iter().map(|s| s.as_ref().to_string()).collect();
|
||||
fn synonyms<S: AsRef<str>>(&self, words: &[S]) -> heed::Result<Option<Vec<Vec<String>>>> {
|
||||
let words: Vec<_> = words.iter().map(|s| s.as_ref().to_owned()).collect();
|
||||
Ok(self.synonyms.get(&words).cloned())
|
||||
}
|
||||
}
|
||||
|
@ -5,6 +5,7 @@ use anyhow::Context;
|
||||
use chrono::Utc;
|
||||
use grenad::CompressionType;
|
||||
use itertools::Itertools;
|
||||
use meilisearch_tokenizer::{Analyzer, AnalyzerConfig};
|
||||
use rayon::ThreadPool;
|
||||
use serde::{Deserialize, Deserializer, Serialize, Serializer};
|
||||
|
||||
@ -13,7 +14,6 @@ use crate::criterion::Criterion;
|
||||
use crate::facet::FacetType;
|
||||
use crate::update::{ClearDocuments, IndexDocuments, UpdateIndexingStep};
|
||||
use crate::update::index_documents::{IndexDocumentsMethod, Transform};
|
||||
use meilisearch_tokenizer::{AnalyzerConfig, Analyzer};
|
||||
|
||||
#[derive(Debug, Clone, PartialEq)]
|
||||
pub enum Setting<T> {
|
||||
@ -328,18 +328,7 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
|
||||
fn update_synonyms(&mut self) -> anyhow::Result<bool> {
|
||||
match self.synonyms {
|
||||
Setting::Set(ref synonyms) => {
|
||||
let old_synonyms = self.index.synonyms(self.wtxn)?.unwrap_or_default();
|
||||
|
||||
let mut config = AnalyzerConfig::default();
|
||||
|
||||
let stop_words = self.index.stop_words(self.wtxn)?;
|
||||
if let Some(stop_words) = &stop_words {
|
||||
config.stop_words(stop_words);
|
||||
}
|
||||
|
||||
let analyzer = Analyzer::new(config);
|
||||
|
||||
let normalize = |text: &String| {
|
||||
fn normalize(analyzer: &Analyzer<&[u8]>, text: &str) -> Vec<String> {
|
||||
analyzer
|
||||
.analyze(text)
|
||||
.tokens()
|
||||
@ -347,20 +336,40 @@ impl<'a, 't, 'u, 'i> Settings<'a, 't, 'u, 'i> {
|
||||
if token.is_word() { Some(token.text().to_string()) } else { None }
|
||||
)
|
||||
.collect::<Vec<_>>()
|
||||
};
|
||||
}
|
||||
|
||||
let new_synonyms = synonyms
|
||||
.iter()
|
||||
.map(|(word, synonyms)| {
|
||||
let normalized_word = normalize(word);
|
||||
let normalized_synonyms = synonyms.iter()
|
||||
.map(normalize)
|
||||
.unique()
|
||||
.collect::<Vec<_>>();
|
||||
let mut config = AnalyzerConfig::default();
|
||||
let stop_words = self.index.stop_words(self.wtxn)?;
|
||||
if let Some(stop_words) = &stop_words {
|
||||
config.stop_words(stop_words);
|
||||
}
|
||||
let analyzer = Analyzer::new(config);
|
||||
|
||||
(normalized_word, normalized_synonyms)
|
||||
})
|
||||
.collect();
|
||||
let mut new_synonyms = HashMap::new();
|
||||
for (word, synonyms) in synonyms {
|
||||
// Normalize both the word and associated synonyms.
|
||||
let normalized_word = normalize(&analyzer, word);
|
||||
let normalized_synonyms = synonyms
|
||||
.iter()
|
||||
.map(|synonym| normalize(&analyzer, synonym));
|
||||
|
||||
// Store the normalized synonyms under the normalized word,
|
||||
// merging the possible duplicate words.
|
||||
let entry = new_synonyms
|
||||
.entry(normalized_word)
|
||||
.or_insert_with(Vec::new);
|
||||
entry.extend(normalized_synonyms);
|
||||
}
|
||||
|
||||
// Make sure that we don't have duplicate synonyms.
|
||||
new_synonyms
|
||||
.iter_mut()
|
||||
.for_each(|(_, synonyms)| {
|
||||
synonyms.sort_unstable();
|
||||
synonyms.dedup();
|
||||
});
|
||||
|
||||
let old_synonyms = self.index.synonyms(self.wtxn)?;
|
||||
|
||||
if new_synonyms != old_synonyms {
|
||||
self.index.put_synonyms(self.wtxn, &new_synonyms)?;
|
||||
@ -734,6 +743,64 @@ mod tests {
|
||||
assert_eq!(result.documents_ids.len(), 1); // there is one benoit in our data
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn set_and_reset_synonyms() {
|
||||
let path = tempfile::tempdir().unwrap();
|
||||
let mut options = EnvOpenOptions::new();
|
||||
options.map_size(10 * 1024 * 1024); // 10 MB
|
||||
let index = Index::new(options, &path).unwrap();
|
||||
|
||||
// Send 3 documents with ids from 1 to 3.
|
||||
let mut wtxn = index.write_txn().unwrap();
|
||||
let content = &b"name,age,maxim\nkevin,23,I love dogs\nkevina,21,Doggos are the best\nbenoit,34,The crepes are really good\n"[..];
|
||||
let mut builder = IndexDocuments::new(&mut wtxn, &index, 0);
|
||||
builder.update_format(UpdateFormat::Csv);
|
||||
builder.execute(content, |_, _| ()).unwrap();
|
||||
|
||||
// In the same transaction provide some synonyms
|
||||
let mut builder = Settings::new(&mut wtxn, &index, 0);
|
||||
builder.set_synonyms(hashmap! {
|
||||
"blini".to_string() => vec!["crepes".to_string()],
|
||||
"super like".to_string() => vec!["love".to_string()],
|
||||
"puppies".to_string() => vec!["dogs".to_string(), "doggos".to_string()]
|
||||
});
|
||||
builder.execute(|_, _| ()).unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
|
||||
// Ensure synonyms are effectively stored
|
||||
let rtxn = index.read_txn().unwrap();
|
||||
let synonyms = index.synonyms(&rtxn).unwrap();
|
||||
assert!(!synonyms.is_empty()); // at this point the index should return something
|
||||
|
||||
// Check that we can use synonyms
|
||||
let result = index.search(&rtxn).query("blini").execute().unwrap();
|
||||
assert_eq!(result.documents_ids.len(), 1);
|
||||
let result = index.search(&rtxn).query("super like").execute().unwrap();
|
||||
assert_eq!(result.documents_ids.len(), 1);
|
||||
let result = index.search(&rtxn).query("puppies").execute().unwrap();
|
||||
assert_eq!(result.documents_ids.len(), 2);
|
||||
|
||||
// Reset the synonyms
|
||||
let mut wtxn = index.write_txn().unwrap();
|
||||
let mut builder = Settings::new(&mut wtxn, &index, 0);
|
||||
builder.reset_synonyms();
|
||||
builder.execute(|_, _| ()).unwrap();
|
||||
wtxn.commit().unwrap();
|
||||
|
||||
// Ensure synonyms are reset
|
||||
let rtxn = index.read_txn().unwrap();
|
||||
let synonyms = index.synonyms(&rtxn).unwrap();
|
||||
assert!(synonyms.is_empty());
|
||||
|
||||
// Check that synonyms are no longer work
|
||||
let result = index.search(&rtxn).query("blini").execute().unwrap();
|
||||
assert!(result.documents_ids.is_empty());
|
||||
let result = index.search(&rtxn).query("super like").execute().unwrap();
|
||||
assert!(result.documents_ids.is_empty());
|
||||
let result = index.search(&rtxn).query("puppies").execute().unwrap();
|
||||
assert!(result.documents_ids.is_empty());
|
||||
}
|
||||
|
||||
#[test]
|
||||
fn setting_searchable_recomputes_other_settings() {
|
||||
let path = tempfile::tempdir().unwrap();
|
||||
|
Loading…
x
Reference in New Issue
Block a user