Compare commits

...

6 Commits

Author SHA1 Message Date
Halvor Bø
d09720c989 Fixed bench 2021-07-07 00:32:01 +01:00
Halvor Bø
d9bcf6e26b Cleaned up variable names 2021-07-07 00:25:42 +01:00
Halvor Bø
7ba0d0df1c Made the implementation respect ordering 2021-07-06 23:58:15 +01:00
Halvor Bø
f1f922ea1b MProof of concept implementation of slop 2021-07-06 20:50:41 +01:00
Halvor Bø
c35343d361 Added todo 2021-07-05 23:30:34 +01:00
Halvor Bø
7fe421ae2a Initial 2021-07-05 23:04:09 +01:00
4 changed files with 290 additions and 32 deletions

View File

@@ -179,6 +179,87 @@ pub mod tests {
assert_nearly_equals!(scores[1], 0.46844664);
}
#[ignore]
#[test]
pub fn test_phrase_score_with_slop() {
let index = create_index(&["a c b", "a b c a b"]);
let schema = index.schema();
let text_field = schema.get_field("text").unwrap();
let searcher = index.reader().unwrap().searcher();
let test_query = |texts: Vec<&str>| {
let terms: Vec<Term> = texts
.iter()
.map(|text| Term::from_field_text(text_field, text))
.collect();
let mut phrase_query = PhraseQuery::new(terms);
phrase_query.slop(1);
searcher
.search(&phrase_query, &TEST_COLLECTOR_WITH_SCORE)
.expect("search should succeed")
.scores()
.to_vec()
};
let scores = test_query(vec!["a", "b"]);
assert_nearly_equals!(scores[0], 0.40618482);
assert_nearly_equals!(scores[1], 0.46844664);
}
#[test]
pub fn test_phrase_score_with_slop_size() {
let index = create_index(&["a b e c", "a e e e c", "a e e e e c"]);
let schema = index.schema();
let text_field = schema.get_field("text").unwrap();
let searcher = index.reader().unwrap().searcher();
let test_query = |texts: Vec<&str>| {
let terms: Vec<Term> = texts
.iter()
.map(|text| Term::from_field_text(text_field, text))
.collect();
let mut phrase_query = PhraseQuery::new(terms);
phrase_query.slop(3);
searcher
.search(&phrase_query, &TEST_COLLECTOR_WITH_SCORE)
.expect("search should succeed")
.scores()
.to_vec()
};
let scores = test_query(vec!["a", "c"]);
assert_nearly_equals!(scores[0], 0.29086056);
assert_nearly_equals!(scores[1], 0.26706287);
}
#[test]
pub fn test_phrase_score_with_slop_ordering() {
let index = create_index(&[
"a e b e c",
"a e e e e e b e e e e c",
"a c b",
"a c e b e",
"a e c b",
"a e b c",
]);
let schema = index.schema();
let text_field = schema.get_field("text").unwrap();
let searcher = index.reader().unwrap().searcher();
let test_query = |texts: Vec<&str>| {
let terms: Vec<Term> = texts
.iter()
.map(|text| Term::from_field_text(text_field, text))
.collect();
let mut phrase_query = PhraseQuery::new(terms);
phrase_query.slop(3);
searcher
.search(&phrase_query, &TEST_COLLECTOR_WITH_SCORE)
.expect("search should succeed")
.scores()
.to_vec()
};
let scores = test_query(vec!["a", "b", "c"]);
// The first and last matches.
assert_nearly_equals!(scores[0], 0.23091172);
assert_nearly_equals!(scores[1], 0.25024384);
}
#[test] // motivated by #234
pub fn test_phrase_query_docfreq_order() {
let mut schema_builder = Schema::builder();

View File

@@ -26,6 +26,7 @@ use crate::schema::{Field, Term};
pub struct PhraseQuery {
field: Field,
phrase_terms: Vec<(usize, Term)>,
slop: u32,
}
impl PhraseQuery {
@@ -56,9 +57,15 @@ impl PhraseQuery {
PhraseQuery {
field,
phrase_terms: terms,
slop: 0,
}
}
/// Slop allowed for the phrase.
pub fn slop(&mut self, value: u32) {
self.slop = value;
}
/// The `Field` this `PhraseQuery` is targeting.
pub fn field(&self) -> Field {
self.field
@@ -97,11 +104,11 @@ impl PhraseQuery {
}
let terms = self.phrase_terms();
let bm25_weight = Bm25Weight::for_terms(searcher, &terms)?;
Ok(PhraseWeight::new(
self.phrase_terms.clone(),
bm25_weight,
scoring_enabled,
))
let mut weight = PhraseWeight::new(self.phrase_terms.clone(), bm25_weight, scoring_enabled);
if self.slop > 0 {
weight.slop(self.slop);
}
Ok(weight)
}
}

View File

@@ -51,6 +51,7 @@ pub struct PhraseScorer<TPostings: Postings> {
fieldnorm_reader: FieldNormReader,
similarity_weight: Bm25Weight,
scoring_enabled: bool,
slop: u32,
}
/// Returns true iff the two sorted array contain a common element
@@ -130,12 +131,82 @@ fn intersection(left: &mut [u32], right: &[u32]) -> usize {
count
}
/// Intersect twos sorted arrays `left` and `right` and outputs the
/// resulting array in left.
///
/// Condition for match is that the value stored in left is less than the value in right,
/// and that the difference between the value in left_begin is within
///
/// Returns the length of the intersection
fn intersection_with_distance(
left: &mut [u32],
left_begin: &mut [u32],
right: &mut [u32],
max_distance_to_begin: u32,
) -> usize {
// TODO: Improve variable names?
let mut left_i = 0;
let mut right_i = 0;
let mut count = 0;
let left_len = left.len();
let right_len = right.len();
// Is the current last value guaranteed to be the final value.
let mut is_temporary = false;
while left_i < left_len && right_i < right_len {
let left_val = left[left_i];
let right_val = right[right_i];
match left_val.cmp(&right_val) {
Ordering::Less => {
if right_val - left_begin[left_i] <= max_distance_to_begin {
if is_temporary {
// If the value was temporary we have found a closer match.
count -= 1;
};
left[count] = left_val;
left_begin[count] = left_begin[left_i];
right[count] = right_val;
count += 1;
left_i += 1;
// Still possible to find a closer match.
is_temporary = true;
} else {
left_i += 1;
}
}
Ordering::Equal => {
if is_temporary {
// If the value was temporary we have found an.
count -= 1;
is_temporary = false;
}
left[count] = left_val;
left_begin[count] = left_begin[left_i];
right[count] = right_val;
count += 1;
left_i += 1;
right_i += 1;
}
Ordering::Greater => {
right_i += 1;
// Given the constraint that left cannot be greater than right we know that the value in left is
// final.
is_temporary = false;
}
}
}
for i in 0..count {
left[i] = right[i];
}
count
}
impl<TPostings: Postings> PhraseScorer<TPostings> {
pub fn new(
term_postings: Vec<(usize, TPostings)>,
similarity_weight: Bm25Weight,
fieldnorm_reader: FieldNormReader,
scoring_enabled: bool,
slop: u32,
) -> PhraseScorer<TPostings> {
let max_offset = term_postings
.iter()
@@ -158,6 +229,7 @@ impl<TPostings: Postings> PhraseScorer<TPostings> {
similarity_weight,
fieldnorm_reader,
scoring_enabled,
slop,
};
if scorer.doc() != TERMINATED && !scorer.phrase_match() {
scorer.advance();
@@ -170,6 +242,7 @@ impl<TPostings: Postings> PhraseScorer<TPostings> {
}
fn phrase_match(&mut self) -> bool {
// Need to add support for slop in phrase_count and phrase_exists.
if self.scoring_enabled {
let count = self.compute_phrase_count();
self.phrase_count = count;
@@ -180,29 +253,25 @@ impl<TPostings: Postings> PhraseScorer<TPostings> {
}
fn phrase_exists(&mut self) -> bool {
self.intersection_docset
.docset_mut_specialized(0)
.positions(&mut self.left);
let mut intersection_len = self.left.len();
for i in 1..self.num_terms - 1 {
{
self.intersection_docset
.docset_mut_specialized(i)
.positions(&mut self.right);
}
intersection_len = intersection(&mut self.left[..intersection_len], &self.right[..]);
if intersection_len == 0 {
return false;
}
}
self.intersection_docset
.docset_mut_specialized(self.num_terms - 1)
.positions(&mut self.right);
let intersection_len = if self.has_slop() {
self.compute_match_with_slop()
} else {
self.compute_match()
};
intersection_exists(&self.left[..intersection_len], &self.right[..])
}
fn compute_phrase_count(&mut self) -> u32 {
let intersection_len = if self.has_slop() {
self.compute_match_with_slop()
} else {
self.compute_match()
};
intersection_count(&self.left[..intersection_len], &self.right[..]) as u32
}
/// Computes match without slop.
fn compute_match(&mut self) -> usize {
{
self.intersection_docset
.docset_mut_specialized(0)
@@ -217,14 +286,52 @@ impl<TPostings: Postings> PhraseScorer<TPostings> {
}
intersection_len = intersection(&mut self.left[..intersection_len], &self.right[..]);
if intersection_len == 0 {
return 0u32;
return 0;
}
}
self.intersection_docset
.docset_mut_specialized(self.num_terms - 1)
.positions(&mut self.right);
intersection_count(&self.left[..intersection_len], &self.right[..]) as u32
intersection_len
}
// Computes match with slop.
fn compute_match_with_slop(&mut self) -> usize {
{
self.intersection_docset
.docset_mut_specialized(0)
.positions(&mut self.left);
}
let mut intersection_len = self.left.len();
// We'll increment the values to be equal to the next match in the right array to achieve ordered slop.
let mut left_begin_vec = self.left.clone();
let left_begin = &mut left_begin_vec[..];
for i in 1..self.num_terms {
{
self.intersection_docset
.docset_mut_specialized(i)
.positions(&mut self.right);
}
intersection_len = intersection_with_distance(
&mut self.left[..intersection_len],
&mut left_begin[..intersection_len],
&mut self.right[..],
self.slop,
);
// Update the left to be equal to the right. Merge the initial left.
if intersection_len == 0 {
return 0;
}
}
self.intersection_docset
.docset_mut_specialized(self.num_terms - 1)
.positions(&mut self.right);
intersection_len
}
fn has_slop(&self) -> bool {
self.slop > 0
}
}
@@ -267,18 +374,28 @@ impl<TPostings: Postings> Scorer for PhraseScorer<TPostings> {
#[cfg(test)]
mod tests {
use super::{intersection, intersection_count};
use super::{intersection, intersection_count, intersection_with_distance};
fn test_intersection_sym(left: &[u32], right: &[u32], expected: &[u32]) {
test_intersection_aux(left, right, expected);
test_intersection_aux(right, left, expected);
test_intersection_aux(left, right, expected, 0);
test_intersection_aux(right, left, expected, 0);
}
fn test_intersection_aux(left: &[u32], right: &[u32], expected: &[u32]) {
fn test_intersection_aux(left: &[u32], right: &[u32], expected: &[u32], slop: u32) {
let mut left_vec = Vec::from(left);
let left_mut = &mut left_vec[..];
assert_eq!(intersection_count(left_mut, right), expected.len());
let count = intersection(left_mut, right);
if slop == 0 {
let left_mut = &mut left_vec[..];
assert_eq!(intersection_count(left_mut, right), expected.len());
let count = intersection(left_mut, right);
assert_eq!(&left_mut[..count], expected);
return;
}
let mut right_vec = Vec::from(right);
let right_mut = &mut right_vec[..];
let mut left_begin_vec = Vec::from(left);
let left_begin_mut = &mut left_begin_vec[..];
let count = intersection_with_distance(left_mut, left_begin_mut, right_mut, slop);
assert_eq!(&left_mut[..count], expected);
}
@@ -290,6 +407,51 @@ mod tests {
test_intersection_sym(&[5, 7], &[1, 5, 10, 12], &[5]);
test_intersection_sym(&[1, 5, 6, 9, 10, 12], &[6, 8, 9, 12], &[6, 9, 12]);
}
#[test]
fn test_slop() {
// The slop is not symetric. It does not allow for the phrase to be out of order.
test_intersection_aux(&[1], &[2], &[2], 1);
test_intersection_aux(&[1], &[3], &[], 1);
test_intersection_aux(&[1], &[3], &[3], 2);
test_intersection_aux(&[], &[2], &[], 100000);
test_intersection_aux(&[5, 7, 11], &[1, 5, 10, 12], &[5, 12], 1);
test_intersection_aux(&[1, 5, 6, 9, 10, 12], &[6, 8, 9, 12], &[6, 9, 12], 1);
test_intersection_aux(&[1, 5, 6, 9, 10, 12], &[6, 8, 9, 12], &[6, 9, 12], 10);
}
fn test_merge(
left: &[u32],
right: &[u32],
left_begin: &[u32],
expected_left: &[u32],
expected_left_begin: &[u32],
slop: u32,
) {
let mut left_vec = Vec::from(left);
let left_mut = &mut left_vec[..];
let mut right_vec = Vec::from(right);
let right_mut = &mut right_vec[..];
let mut left_begin_vec = Vec::from(left_begin);
let left_begin_mut = &mut left_begin_vec[..];
let count = intersection_with_distance(left_mut, left_begin_mut, right_mut, slop);
assert_eq!(&left_mut[..count], expected_left);
assert_eq!(&left_begin_mut[..count], expected_left_begin);
}
#[test]
fn test_merge_slop() {
test_merge(&[1, 2], &[1], &[0, 1], &[1], &[0], 1);
test_merge(&[3], &[4], &[2], &[4], &[2], 2);
test_merge(&[3], &[4], &[2], &[4], &[2], 2);
test_merge(
&[1, 5, 6, 9, 10, 12],
&[6, 8, 9, 12],
&[0, 1, 2, 3, 4, 5],
&[6, 9, 12],
&[2, 3, 5],
10,
);
}
}
#[cfg(all(test, feature = "unstable"))]

View File

@@ -16,6 +16,7 @@ pub struct PhraseWeight {
phrase_terms: Vec<(usize, Term)>,
similarity_weight: Bm25Weight,
scoring_enabled: bool,
slop: u32,
}
impl PhraseWeight {
@@ -25,10 +26,12 @@ impl PhraseWeight {
similarity_weight: Bm25Weight,
scoring_enabled: bool,
) -> PhraseWeight {
let slop = 0;
PhraseWeight {
phrase_terms,
similarity_weight,
scoring_enabled,
slop,
}
}
@@ -77,8 +80,13 @@ impl PhraseWeight {
similarity_weight,
fieldnorm_reader,
self.scoring_enabled,
self.slop,
)))
}
pub fn slop(&mut self, slop: u32) {
self.slop = slop;
}
}
impl Weight for PhraseWeight {