Merge pull request #232 from meilisearch/support-splitted-words

Support splitted words

meilidb-core/src/automaton/mod.rs

@@ -2,7 +2,7 @@ mod dfa;
 mod query_enhancer;
 
 use std::cmp::Reverse;
-use std::vec;
+use std::{cmp, vec};
 
 use fst::{IntoStreamer, Streamer};
 use levenshtein_automata::DFA;
@@ -18,7 +18,7 @@ use self::query_enhancer::QueryEnhancerBuilder;
 const NGRAMS: usize = 3;
 
 pub struct AutomatonProducer {
-    automatons: Vec<Vec<Automaton>>,
+    automatons: Vec<AutomatonGroup>,
 }
 
 impl AutomatonProducer {
@@ -26,19 +26,47 @@ impl AutomatonProducer {
         reader: &heed::RoTxn,
         query: &str,
         main_store: store::Main,
+        postings_list_store: store::PostingsLists,
         synonyms_store: store::Synonyms,
     ) -> MResult<(AutomatonProducer, QueryEnhancer)> {
-        let (automatons, query_enhancer) =
+        let (automatons, query_enhancer) = generate_automatons(
-            generate_automatons(reader, query, main_store, synonyms_store)?;
+            reader,
+            query,
+            main_store,
+            postings_list_store,
+            synonyms_store,
+        )?;
 
         Ok((AutomatonProducer { automatons }, query_enhancer))
     }
 
-    pub fn into_iter(self) -> vec::IntoIter<Vec<Automaton>> {
+    pub fn into_iter(self) -> vec::IntoIter<AutomatonGroup> {
         self.automatons.into_iter()
     }
 }
 
+#[derive(Debug)]
+pub struct AutomatonGroup {
+    pub is_phrase_query: bool,
+    pub automatons: Vec<Automaton>,
+}
+
+impl AutomatonGroup {
+    fn normal(automatons: Vec<Automaton>) -> AutomatonGroup {
+        AutomatonGroup {
+            is_phrase_query: false,
+            automatons,
+        }
+    }
+
+    fn phrase_query(automatons: Vec<Automaton>) -> AutomatonGroup {
+        AutomatonGroup {
+            is_phrase_query: true,
+            automatons,
+        }
+    }
+}
+
 #[derive(Debug)]
 pub struct Automaton {
     pub index: usize,
@@ -102,12 +130,41 @@ pub fn normalize_str(string: &str) -> String {
     string
 }
 
+fn split_best_frequency<'a>(
+    reader: &heed::RoTxn,
+    word: &'a str,
+    postings_lists_store: store::PostingsLists,
+) -> MResult<Option<(&'a str, &'a str)>> {
+    let chars = word.char_indices().skip(1);
+    let mut best = None;
+
+    for (i, _) in chars {
+        let (left, right) = word.split_at(i);
+
+        let left_freq = postings_lists_store
+            .postings_list(reader, left.as_ref())?
+            .map_or(0, |i| i.len());
+
+        let right_freq = postings_lists_store
+            .postings_list(reader, right.as_ref())?
+            .map_or(0, |i| i.len());
+
+        let min_freq = cmp::min(left_freq, right_freq);
+        if min_freq != 0 && best.map_or(true, |(old, _, _)| min_freq > old) {
+            best = Some((min_freq, left, right));
+        }
+    }
+
+    Ok(best.map(|(_, l, r)| (l, r)))
+}
+
 fn generate_automatons(
     reader: &heed::RoTxn,
     query: &str,
     main_store: store::Main,
+    postings_lists_store: store::PostingsLists,
     synonym_store: store::Synonyms,
-) -> MResult<(Vec<Vec<Automaton>>, QueryEnhancer)> {
+) -> MResult<(Vec<AutomatonGroup>, QueryEnhancer)> {
     let has_end_whitespace = query.chars().last().map_or(false, char::is_whitespace);
     let query_words: Vec<_> = split_query_string(query).map(str::to_lowercase).collect();
     let synonyms = match main_store.synonyms_fst(reader)? {
@@ -136,7 +193,7 @@ fn generate_automatons(
         original_automatons.push(automaton);
     }
 
-    automatons.push(original_automatons);
+    automatons.push(AutomatonGroup::normal(original_automatons));
 
     for n in 1..=NGRAMS {
         let mut ngrams = query_words.windows(n).enumerate().peekable();
@@ -188,13 +245,27 @@ fn generate_automatons(
                                 Automaton::non_exact(automaton_index, n, synonym)
                             };
                             automaton_index += 1;
-                            automatons.push(vec![automaton]);
+                            automatons.push(AutomatonGroup::normal(vec![automaton]));
                         }
                     }
                 }
             }
 
-            if n != 1 {
+            if n == 1 {
+                if let Some((left, right)) =
+                    split_best_frequency(reader, &normalized, postings_lists_store)?
+                {
+                    let a = Automaton::exact(automaton_index, 1, left);
+                    enhancer_builder.declare(query_range.clone(), automaton_index, &[left]);
+                    automaton_index += 1;
+
+                    let b = Automaton::exact(automaton_index, 1, right);
+                    enhancer_builder.declare(query_range.clone(), automaton_index, &[left]);
+                    automaton_index += 1;
+
+                    automatons.push(AutomatonGroup::phrase_query(vec![a, b]));
+                }
+            } else {
                 // automaton of concatenation of query words
                 let concat = ngram_slice.concat();
                 let normalized = normalize_str(&concat);
@@ -204,15 +275,15 @@ fn generate_automatons(
 
                 let automaton = Automaton::exact(automaton_index, n, &normalized);
                 automaton_index += 1;
-                automatons.push(vec![automaton]);
+                automatons.push(AutomatonGroup::normal(vec![automaton]));
             }
         }
     }
 
     // order automatons, the most important first,
     // we keep the original automatons at the front.
-    automatons[1..].sort_by_key(|a| {
+    automatons[1..].sort_by_key(|group| {
-        let a = a.first().unwrap();
+        let a = group.automatons.first().unwrap();
         (Reverse(a.is_exact), a.ngram)
     });
 

meilidb-core/src/query_builder.rs

@@ -8,7 +8,7 @@ use fst::{IntoStreamer, Streamer};
 use sdset::SetBuf;
 use slice_group_by::{GroupBy, GroupByMut};
 
-use crate::automaton::{Automaton, AutomatonProducer, QueryEnhancer};
+use crate::automaton::{Automaton, AutomatonGroup, AutomatonProducer, QueryEnhancer};
 use crate::distinct_map::{BufferedDistinctMap, DistinctMap};
 use crate::raw_document::{raw_documents_from, RawDocument};
 use crate::{criterion::Criteria, Document, DocumentId, Highlight, TmpMatch};
@@ -138,7 +138,7 @@ fn multiword_rewrite_matches(
 
 fn fetch_raw_documents(
     reader: &heed::RoTxn,
-    automatons: &[Automaton],
+    automatons_groups: &[AutomatonGroup],
     query_enhancer: &QueryEnhancer,
     searchables: Option<&ReorderedAttrs>,
     main_store: store::Main,
@@ -148,7 +148,15 @@ fn fetch_raw_documents(
     let mut matches = Vec::new();
     let mut highlights = Vec::new();
 
-    for automaton in automatons {
+    for group in automatons_groups {
+        let AutomatonGroup {
+            is_phrase_query,
+            automatons,
+        } = group;
+        let phrase_query_len = automatons.len();
+
+        let mut tmp_matches = Vec::new();
+        for (id, automaton) in automatons.into_iter().enumerate() {
             let Automaton {
                 index,
                 is_exact,
@@ -172,8 +180,7 @@ fn fetch_raw_documents(
                     None => continue,
                 };
 
-            matches.reserve(doc_indexes.len());
+                tmp_matches.reserve(doc_indexes.len());
-            highlights.reserve(doc_indexes.len());
 
                 for di in doc_indexes.as_ref() {
                     let attribute = searchables.map_or(Some(di.attribute), |r| r.get(di.attribute));
@@ -192,13 +199,45 @@ fn fetch_raw_documents(
                             char_length: di.char_length,
                         };
 
-                    matches.push((di.document_id, match_));
+                        tmp_matches.push((di.document_id, id, match_, highlight));
-                    highlights.push((di.document_id, highlight));
                     }
                 }
             }
         }
 
+        if *is_phrase_query {
+            tmp_matches.sort_unstable_by_key(|(id, _, m, _)| (*id, m.attribute, m.word_index));
+            for group in tmp_matches.linear_group_by_key(|(id, _, m, _)| (*id, m.attribute)) {
+                for window in group.windows(2) {
+                    let (ida, ia, ma, ha) = window[0];
+                    let (idb, ib, mb, hb) = window[1];
+
+                    debug_assert_eq!(ida, idb);
+
+                    // if matches must follow and actually follows themselves
+                    if ia + 1 == ib && ma.word_index + 1 == mb.word_index {
+                        // TODO we must make it work for phrase query longer than 2
+                        // if the second match is the last phrase query word
+                        if ib + 1 == phrase_query_len {
+                            // insert first match
+                            matches.push((ida, ma));
+                            highlights.push((ida, ha));
+
+                            // insert second match
+                            matches.push((idb, mb));
+                            highlights.push((idb, hb));
+                        }
+                    }
+                }
+            }
+        } else {
+            for (id, _, match_, highlight) in tmp_matches {
+                matches.push((id, match_));
+                highlights.push((id, highlight));
+            }
+        }
+    }
+
     let matches = multiword_rewrite_matches(matches, &query_enhancer);
     let highlights = {
         highlights.sort_unstable_by_key(|(id, _)| *id);
@@ -367,15 +406,20 @@ where
     let start_processing = Instant::now();
     let mut raw_documents_processed = Vec::with_capacity(range.len());
 
-    let (automaton_producer, query_enhancer) =
+    let (automaton_producer, query_enhancer) = AutomatonProducer::new(
-        AutomatonProducer::new(reader, query, main_store, synonyms_store)?;
+        reader,
+        query,
+        main_store,
+        postings_lists_store,
+        synonyms_store,
+    )?;
 
     let automaton_producer = automaton_producer.into_iter();
     let mut automatons = Vec::new();
 
     // aggregate automatons groups by groups after time
     for auts in automaton_producer {
-        automatons.extend(auts);
+        automatons.push(auts);
 
         // we must retrieve the documents associated
         // with the current automatons
@@ -480,15 +524,20 @@ where
     let start_processing = Instant::now();
     let mut raw_documents_processed = Vec::new();
 
-    let (automaton_producer, query_enhancer) =
+    let (automaton_producer, query_enhancer) = AutomatonProducer::new(
-        AutomatonProducer::new(reader, query, main_store, synonyms_store)?;
+        reader,
+        query,
+        main_store,
+        postings_lists_store,
+        synonyms_store,
+    )?;
 
     let automaton_producer = automaton_producer.into_iter();
     let mut automatons = Vec::new();
 
     // aggregate automatons groups by groups after time
     for auts in automaton_producer {
-        automatons.extend(auts);
+        automatons.push(auts);
 
         // we must retrieve the documents associated
         // with the current automatons
@@ -1697,4 +1746,68 @@ mod tests {
         });
         assert_matches!(iter.next(), None);
     }
+
+    #[test]
+    fn simple_phrase_query_splitting() {
+        let store = TempDatabase::from_iter(vec![
+            ("search", &[doc_index(0, 0)][..]),
+            ("engine", &[doc_index(0, 1)][..]),
+            ("search", &[doc_index(1, 0)][..]),
+            ("slow", &[doc_index(1, 1)][..]),
+            ("engine", &[doc_index(1, 2)][..]),
+        ]);
+
+        let env = &store.database.env;
+        let reader = env.read_txn().unwrap();
+
+        let builder = store.query_builder();
+        let results = builder.query(&reader, "searchengine", 0..20).unwrap();
+        let mut iter = results.into_iter();
+
+        assert_matches!(iter.next(), Some(Document { id: DocumentId(0), matches, .. }) => {
+            let mut iter = matches.into_iter();
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 0, distance: 0, .. })); // search
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 1, distance: 0, .. })); // engine
+            assert_matches!(iter.next(), None);
+        });
+        assert_matches!(iter.next(), None);
+    }
+
+    #[test]
+    fn harder_phrase_query_splitting() {
+        let store = TempDatabase::from_iter(vec![
+            ("search", &[doc_index(0, 0)][..]),
+            ("search", &[doc_index(0, 1)][..]),
+            ("engine", &[doc_index(0, 2)][..]),
+            ("search", &[doc_index(1, 0)][..]),
+            ("slow", &[doc_index(1, 1)][..]),
+            ("search", &[doc_index(1, 2)][..]),
+            ("engine", &[doc_index(1, 3)][..]),
+            ("search", &[doc_index(1, 0)][..]),
+            ("search", &[doc_index(1, 1)][..]),
+            ("slow", &[doc_index(1, 2)][..]),
+            ("engine", &[doc_index(1, 3)][..]),
+        ]);
+
+        let env = &store.database.env;
+        let reader = env.read_txn().unwrap();
+
+        let builder = store.query_builder();
+        let results = builder.query(&reader, "searchengine", 0..20).unwrap();
+        let mut iter = results.into_iter();
+
+        assert_matches!(iter.next(), Some(Document { id: DocumentId(0), matches, .. }) => {
+            let mut iter = matches.into_iter();
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 1, distance: 0, .. })); // search
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 2, distance: 0, .. })); // engine
+            assert_matches!(iter.next(), None);
+        });
+        assert_matches!(iter.next(), Some(Document { id: DocumentId(1), matches, .. }) => {
+            let mut iter = matches.into_iter();
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 2, distance: 0, .. })); // search
+            assert_matches!(iter.next(), Some(TmpMatch { query_index: 0, word_index: 3, distance: 0, .. })); // engine
+            assert_matches!(iter.next(), None);
+        });
+        assert_matches!(iter.next(), None);
+    }
 }