snowykami/meilisearch
1
0
Fork 0
mirror of https://github.com/meilisearch/meilisearch.git synced 2024-11-25 11:35:05 +08:00
Code Issues Actions 3 Packages Projects Releases Wiki Activity

Move MatchPosition impl to Match, adjust counting score for phrases

Browse Source
This commit is contained in:
F. Levi 2024-09-13 21:20:06 +03:00
parent cab63abc84
commit a2a16bf846
1 changed files with 43 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

66
milli/src/search/new/matches/mod.rs
View File

@ -117,30 +117,30 @@ pub struct Match {
position: MatchPosition, position: MatchPosition,
} }
impl MatchPosition { impl Match {
fn get_first_word(m: &Match) -> usize { fn get_first_word_pos(&self) -> usize {
match m.position { match self.position {
MatchPosition::Word { word_position, .. } => word_position, MatchPosition::Word { word_position, .. } => word_position,
MatchPosition::Phrase { word_positions: (fwp, _), .. } => fwp, MatchPosition::Phrase { word_positions: (fwp, _), .. } => fwp,
} }
} }
fn get_last_word(m: &Match) -> usize { fn get_last_word_pos(&self) -> usize {
match m.position { match self.position {
MatchPosition::Word { word_position, .. } => word_position, MatchPosition::Word { word_position, .. } => word_position,
MatchPosition::Phrase { word_positions: (_, lwp), .. } => lwp, MatchPosition::Phrase { word_positions: (_, lwp), .. } => lwp,
} }
} }
fn get_first_token(m: &Match) -> usize { fn get_first_token_pos(&self) -> usize {
match m.position { match self.position {
MatchPosition::Word { token_position, .. } => token_position, MatchPosition::Word { token_position, .. } => token_position,
MatchPosition::Phrase { token_positions: (ftp, _), .. } => ftp, MatchPosition::Phrase { token_positions: (ftp, _), .. } => ftp,
} }
} }
fn get_last_token(m: &Match) -> usize { fn get_last_token_pos(&self) -> usize {
match m.position { match self.position {
MatchPosition::Word { token_position, .. } => token_position, MatchPosition::Word { token_position, .. } => token_position,
MatchPosition::Phrase { token_positions: (_, ltp), .. } => ltp, MatchPosition::Phrase { token_positions: (_, ltp), .. } => ltp,
} }
@ -272,7 +272,7 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
Some((tokens, matches)) => matches Some((tokens, matches)) => matches
.iter() .iter()
.map(|m| MatchBounds { .map(|m| MatchBounds {
start: tokens[MatchPosition::get_first_token(m)].byte_start, start: tokens[m.get_first_token_pos()].byte_start,
length: m.match_len, length: m.match_len,
}) })
.collect(), .collect(),
@ -288,13 +288,11 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
) -> (usize, usize) { ) -> (usize, usize) {
// if there is no match, we start from the beginning of the string by default. // if there is no match, we start from the beginning of the string by default.
let first_match_word_position = let first_match_word_position =
matches.first().map(|m| MatchPosition::get_first_word(m)).unwrap_or(0); matches.first().map(|m| m.get_first_word_pos()).unwrap_or(0);
let first_match_token_position = let first_match_token_position =
matches.first().map(|m| MatchPosition::get_first_token(m)).unwrap_or(0); matches.first().map(|m| m.get_first_token_pos()).unwrap_or(0);
let last_match_word_position = let last_match_word_position = matches.last().map(|m| m.get_last_word_pos()).unwrap_or(0);
matches.last().map(|m| MatchPosition::get_last_word(m)).unwrap_or(0); let last_match_token_position = matches.last().map(|m| m.get_last_token_pos()).unwrap_or(0);
let last_match_token_position =
matches.last().map(|m| MatchPosition::get_last_token(m)).unwrap_or(0);
// matches needs to be counted in the crop len. // matches needs to be counted in the crop len.
let mut remaining_words = crop_size + first_match_word_position - last_match_word_position; let mut remaining_words = crop_size + first_match_word_position - last_match_word_position;
@ -389,6 +387,16 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
let mut order_score = 0; let mut order_score = 0;
let mut distance_score = 0; let mut distance_score = 0;
// Count score for phrases
let tally_phrase_scores =
|fwp: &usize, lwp: &usize, order_score: &mut i16, distance_score: &mut i16| {
let words_in_phrase_minus_one = (lwp - fwp) as i16;
// will always be ordered, so +1 for each space between words
*order_score += words_in_phrase_minus_one;
// distance will always be 1, so -1 for each space between words
*distance_score -= words_in_phrase_minus_one;
};
let mut iter = matches.iter().peekable(); let mut iter = matches.iter().peekable();
while let Some(m) = iter.next() { while let Some(m) = iter.next() {
if let Some(next_match) = iter.peek() { if let Some(next_match) = iter.peek() {
@ -397,12 +405,24 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
order_score += 1; order_score += 1;
} }
let next_match_first_word_pos = MatchPosition::get_first_word(next_match); let m_last_word_pos = match m.position {
let current_match_first_word_pos = MatchPosition::get_first_word(m); MatchPosition::Word { word_position, .. } => word_position,
MatchPosition::Phrase { word_positions: (fwp, lwp), .. } => {
tally_phrase_scores(&fwp, &lwp, &mut order_score, &mut distance_score);
lwp
}
};
let next_match_first_word_pos = match next_match.position {
MatchPosition::Word { word_position, .. } => word_position,
MatchPosition::Phrase { word_positions: (fwp, _), .. } => fwp,
};
// compute distance between matches // compute distance between matches
distance_score -= distance_score -= (next_match_first_word_pos - m_last_word_pos).min(7) as i16;
(next_match_first_word_pos - current_match_first_word_pos).min(7) as i16; } else if let MatchPosition::Phrase { word_positions: (fwp, lwp), .. } = m.position {
// in case last match is a phrase, count score for its words
tally_phrase_scores(&fwp, &lwp, &mut order_score, &mut distance_score);
} }
ids.extend(m.ids.iter()); ids.extend(m.ids.iter());
@ -430,9 +450,9 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
// if next match would make interval gross more than crop_size, // if next match would make interval gross more than crop_size,
// we compare the current interval with the best one, // we compare the current interval with the best one,
// then we increase `interval_first` until next match can be added. // then we increase `interval_first` until next match can be added.
let next_match_word_pos = MatchPosition::get_first_word(next_match); let next_match_word_pos = next_match.get_last_word_pos();
let mut interval_first_match_word_pos = let mut interval_first_match_word_pos =
MatchPosition::get_last_word(&matches[interval_first]); matches[interval_first].get_first_word_pos();
if next_match_word_pos - interval_first_match_word_pos >= crop_size { if next_match_word_pos - interval_first_match_word_pos >= crop_size {
let interval_score = let interval_score =
@ -448,7 +468,7 @@ impl<'t, 'tokenizer> Matcher<'t, 'tokenizer, '_, '_> {
loop { loop {
interval_first += 1; interval_first += 1;
interval_first_match_word_pos = interval_first_match_word_pos =
MatchPosition::get_last_word(&matches[interval_first]); matches[interval_first].get_first_word_pos();
if next_match_word_pos - interval_first_match_word_pos < crop_size { if next_match_word_pos - interval_first_match_word_pos < crop_size {
break; break;