From 039a9a4cc7463e6c18aed3c78c622b8bbb38b51d Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 11:11:55 +0100
Subject: [PATCH 1/7] feat: Pre-compute matches query index groups
---
src/rank/criterion/exact.rs | 7 +-
src/rank/criterion/number_of_words.rs | 8 +--
src/rank/criterion/sum_of_typos.rs | 11 ++--
src/rank/criterion/sum_of_words_attribute.rs | 10 ++-
src/rank/criterion/sum_of_words_position.rs | 12 ++--
src/rank/criterion/words_proximity.rs | 8 +--
src/rank/mod.rs | 69 ++++++++++++++++++--
src/rank/query_builder.rs | 2 +-
8 files changed, 86 insertions(+), 41 deletions(-)
diff --git a/src/rank/criterion/exact.rs b/src/rank/criterion/exact.rs
index 041a6ae67..6d33a6f38 100644
--- a/src/rank/criterion/exact.rs
+++ b/src/rank/criterion/exact.rs
@@ -2,9 +2,8 @@ use std::cmp::Ordering;
use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
-use crate::rank::{match_query_index, Document};
+use crate::rank::{Document, Matches};
use crate::rank::criterion::Criterion;
use crate::database::DatabaseView;
use crate::Match;
@@ -15,8 +14,8 @@ fn contains_exact(matches: &[Match]) -> bool {
}
#[inline]
-fn number_exact_matches(matches: &[Match]) -> usize {
- GroupBy::new(matches, match_query_index).map(contains_exact).count()
+fn number_exact_matches(matches: &Matches) -> usize {
+ matches.query_index_groups().map(contains_exact).count()
}
#[derive(Debug, Clone, Copy)]
diff --git a/src/rank/criterion/number_of_words.rs b/src/rank/criterion/number_of_words.rs
index 855d997ba..23cf36a2c 100644
--- a/src/rank/criterion/number_of_words.rs
+++ b/src/rank/criterion/number_of_words.rs
@@ -2,16 +2,14 @@ use std::cmp::Ordering;
use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
-use crate::rank::{match_query_index, Document};
+use crate::rank::{Document, Matches};
use crate::rank::criterion::Criterion;
use crate::database::DatabaseView;
-use crate::Match;
#[inline]
-fn number_of_query_words(matches: &[Match]) -> usize {
- GroupBy::new(matches, match_query_index).count()
+fn number_of_query_words(matches: &Matches) -> usize {
+ matches.query_index_groups().count()
}
#[derive(Debug, Clone, Copy)]
diff --git a/src/rank/criterion/sum_of_typos.rs b/src/rank/criterion/sum_of_typos.rs
index 409c37cb4..18a199113 100644
--- a/src/rank/criterion/sum_of_typos.rs
+++ b/src/rank/criterion/sum_of_typos.rs
@@ -3,21 +3,18 @@ use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
-
-use crate::rank::{match_query_index, Document};
+use crate::rank::{Document, Matches};
use crate::rank::criterion::Criterion;
use crate::database::DatabaseView;
-use crate::Match;
#[inline]
-fn sum_matches_typos(matches: &[Match]) -> i8 {
+fn sum_matches_typos(matches: &Matches) -> i8 {
let mut sum_typos = 0;
let mut number_words = 0;
// note that GroupBy will never return an empty group
// so we can do this assumption safely
- for group in GroupBy::new(matches, match_query_index) {
+ for group in matches.query_index_groups() {
sum_typos += unsafe { group.get_unchecked(0).distance } as i8;
number_words += 1;
}
@@ -44,7 +41,7 @@ where D: Deref<Target=DB>
mod tests {
use super::*;
- use crate::{DocumentId, Attribute, WordArea};
+ use crate::{Match, DocumentId, Attribute, WordArea};
// typing: "Geox CEO"
//
diff --git a/src/rank/criterion/sum_of_words_attribute.rs b/src/rank/criterion/sum_of_words_attribute.rs
index 718ae7447..cfcaaec68 100644
--- a/src/rank/criterion/sum_of_words_attribute.rs
+++ b/src/rank/criterion/sum_of_words_attribute.rs
@@ -2,19 +2,17 @@ use std::cmp::Ordering;
use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
+use crate::rank::{Document, Matches};
use crate::database::DatabaseView;
-use crate::rank::{match_query_index, Document};
use crate::rank::criterion::Criterion;
-use crate::Match;
#[inline]
-fn sum_matches_attributes(matches: &[Match]) -> u16 {
+fn sum_matches_attributes(matches: &Matches) -> u16 {
// note that GroupBy will never return an empty group
// so we can do this assumption safely
- GroupBy::new(matches, match_query_index).map(|group| unsafe {
- group.get_unchecked(0).attribute.attribute()
+ matches.query_index_groups().map(|group| {
+ unsafe { group.get_unchecked(0).attribute.attribute() }
}).sum()
}
diff --git a/src/rank/criterion/sum_of_words_position.rs b/src/rank/criterion/sum_of_words_position.rs
index d0ebaa74f..e1d650534 100644
--- a/src/rank/criterion/sum_of_words_position.rs
+++ b/src/rank/criterion/sum_of_words_position.rs
@@ -2,19 +2,17 @@ use std::cmp::Ordering;
use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
-use crate::database::DatabaseView;
-use crate::rank::{match_query_index, Document};
+use crate::rank::{Document, Matches};
use crate::rank::criterion::Criterion;
-use crate::Match;
+use crate::database::DatabaseView;
#[inline]
-fn sum_matches_attribute_index(matches: &[Match]) -> u32 {
+fn sum_matches_attribute_index(matches: &Matches) -> u32 {
// note that GroupBy will never return an empty group
// so we can do this assumption safely
- GroupBy::new(matches, match_query_index).map(|group| unsafe {
- group.get_unchecked(0).attribute.word_index()
+ matches.query_index_groups().map(|group| {
+ unsafe { group.get_unchecked(0).attribute.word_index() }
}).sum()
}
diff --git a/src/rank/criterion/words_proximity.rs b/src/rank/criterion/words_proximity.rs
index fc80dfaec..41f4b49b8 100644
--- a/src/rank/criterion/words_proximity.rs
+++ b/src/rank/criterion/words_proximity.rs
@@ -2,9 +2,8 @@ use std::cmp::{self, Ordering};
use std::ops::Deref;
use rocksdb::DB;
-use group_by::GroupBy;
-use crate::rank::{match_query_index, Document};
+use crate::rank::{Document, Matches};
use crate::rank::criterion::Criterion;
use crate::database::DatabaseView;
use crate::Match;
@@ -34,9 +33,9 @@ fn min_proximity(lhs: &[Match], rhs: &[Match]) -> u32 {
min_prox
}
-fn matches_proximity(matches: &[Match]) -> u32 {
+fn matches_proximity(matches: &Matches) -> u32 {
let mut proximity = 0;
- let mut iter = GroupBy::new(matches, match_query_index);
+ let mut iter = matches.query_index_groups();
// iterate over groups by windows of size 2
let mut last = iter.next();
@@ -91,6 +90,7 @@ mod tests {
// soup -> of = 8
// + of -> the = 1
// + the -> day = 8 (not 1)
+ let matches = Matches::from_unsorted_matches(matches.to_vec());
assert_eq!(matches_proximity(matches), 17);
}
diff --git a/src/rank/mod.rs b/src/rank/mod.rs
index 4d1b6b1ea..0d3f538d0 100644
--- a/src/rank/mod.rs
+++ b/src/rank/mod.rs
@@ -2,6 +2,11 @@ pub mod criterion;
mod query_builder;
mod distinct_map;
+use std::slice::Windows;
+
+use sdset::SetBuf;
+use group_by::GroupBy;
+
use crate::{Match, DocumentId};
pub use self::query_builder::{FilterFunc, QueryBuilder, DistinctQueryBuilder};
@@ -14,20 +19,70 @@ fn match_query_index(a: &Match, b: &Match) -> bool {
#[derive(Debug, Clone)]
pub struct Document {
pub id: DocumentId,
- pub matches: Vec<Match>,
+ pub matches: Matches,
}
impl Document {
pub fn new(doc: DocumentId, match_: Match) -> Self {
- unsafe { Self::from_sorted_matches(doc, vec![match_]) }
+ let matches = SetBuf::new_unchecked(vec![match_]);
+ Self::from_matches(doc, matches)
}
- pub fn from_matches(doc: DocumentId, mut matches: Vec<Match>) -> Self {
- matches.sort_unstable();
- unsafe { Self::from_sorted_matches(doc, matches) }
- }
+ pub fn from_matches(id: DocumentId, matches: SetBuf<Match>) -> Self {
+ let mut last = 0;
+ let mut slices = vec![0];
+ for group in GroupBy::new(&matches, match_query_index) {
+ let index = last + group.len();
+ slices.push(index);
+ last = index;
+ }
- pub unsafe fn from_sorted_matches(id: DocumentId, matches: Vec<Match>) -> Self {
+ let matches = Matches { matches, slices };
Self { id, matches }
}
+
+ pub fn from_unsorted_matches(doc: DocumentId, mut matches: Vec<Match>) -> Self {
+ matches.sort_unstable();
+ let matches = SetBuf::new_unchecked(matches);
+ Self::from_matches(doc, matches)
+ }
}
+
+#[derive(Debug, Clone)]
+pub struct Matches {
+ matches: SetBuf<Match>,
+ slices: Vec<usize>,
+}
+
+impl Matches {
+ pub fn query_index_groups(&self) -> QueryIndexGroups {
+ QueryIndexGroups {
+ matches: &self.matches,
+ windows: self.slices.windows(2),
+ }
+ }
+}
+
+pub struct QueryIndexGroups<'a, 'b> {
+ matches: &'a [Match],
+ windows: Windows<'b, usize>,
+}
+
+impl<'a, 'b> Iterator for QueryIndexGroups<'a, 'b> {
+ type Item = &'a [Match];
+
+ fn next(&mut self) -> Option<Self::Item> {
+ self.windows.next().map(|range| {
+ match *range {
+ [left, right] => &self.matches[left..right],
+ _ => unreachable!()
+ }
+ })
+ }
+}
+
+// impl ExactSizeIterator for QueryIndexGroups<'_, '_> {
+// fn len(&self) -> usize {
+// self.windows.len() // FIXME (+1) ?
+// }
+// }
diff --git a/src/rank/query_builder.rs b/src/rank/query_builder.rs
index 4df983f8a..9a17e0473 100644
--- a/src/rank/query_builder.rs
+++ b/src/rank/query_builder.rs
@@ -116,7 +116,7 @@ where D: Deref<Target=DB>,
}
}
- matches.into_iter().map(|(id, matches)| Document::from_matches(id, matches)).collect()
+ matches.into_iter().map(|(id, m)| Document::from_unsorted_matches(id, m)).collect()
}
}
From d21406a93908b2b3ed8cd57b2182996b43cf7806 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 11:23:21 +0100
Subject: [PATCH 2/7] feat: Allow Matches to be constructed
---
src/rank/criterion/sum_of_typos.rs | 30 ++++---------------
src/rank/criterion/words_proximity.rs | 9 ++++--
src/rank/mod.rs | 42 +++++++++++++++++----------
3 files changed, 38 insertions(+), 43 deletions(-)
diff --git a/src/rank/criterion/sum_of_typos.rs b/src/rank/criterion/sum_of_typos.rs
index 18a199113..400650ad2 100644
--- a/src/rank/criterion/sum_of_typos.rs
+++ b/src/rank/criterion/sum_of_typos.rs
@@ -66,10 +66,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(0),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(0), matches)
};
let doc1 = {
@@ -89,10 +86,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(1),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(1), matches)
};
let lhs = sum_matches_typos(&doc0.matches);
@@ -123,10 +117,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(0),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(0), matches)
};
let doc1 = {
@@ -139,10 +130,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(1),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(1), matches)
};
let lhs = sum_matches_typos(&doc0.matches);
@@ -173,10 +161,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(0),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(0), matches)
};
let doc1 = {
@@ -189,10 +174,7 @@ mod tests {
word_area: WordArea::new_faillible(0, 6)
},
];
- Document {
- id: DocumentId(1),
- matches: matches,
- }
+ Document::from_unsorted_matches(DocumentId(1), matches)
};
let lhs = sum_matches_typos(&doc0.matches);
diff --git a/src/rank/criterion/words_proximity.rs b/src/rank/criterion/words_proximity.rs
index 41f4b49b8..7fe3102d3 100644
--- a/src/rank/criterion/words_proximity.rs
+++ b/src/rank/criterion/words_proximity.rs
@@ -90,8 +90,8 @@ mod tests {
// soup -> of = 8
// + of -> the = 1
// + the -> day = 8 (not 1)
- let matches = Matches::from_unsorted_matches(matches.to_vec());
- assert_eq!(matches_proximity(matches), 17);
+ let matches = Matches::from_unsorted(matches.to_vec());
+ assert_eq!(matches_proximity(&matches), 17);
}
#[test]
@@ -118,7 +118,8 @@ mod tests {
// soup -> of = 1
// + of -> the = 1
// + the -> day = 1
- assert_eq!(matches_proximity(matches), 3);
+ let matches = Matches::from_unsorted(matches.to_vec());
+ assert_eq!(matches_proximity(&matches), 3);
}
}
@@ -152,6 +153,8 @@ mod bench {
matches.push(match_);
}
+ let matches = Matches::from_unsorted(matches.to_vec());
+
bench.iter(|| {
let proximity = matches_proximity(&matches);
test::black_box(move || proximity)
diff --git a/src/rank/mod.rs b/src/rank/mod.rs
index 0d3f538d0..ad91830bf 100644
--- a/src/rank/mod.rs
+++ b/src/rank/mod.rs
@@ -23,28 +23,19 @@ pub struct Document {
}
impl Document {
- pub fn new(doc: DocumentId, match_: Match) -> Self {
+ pub fn new(id: DocumentId, match_: Match) -> Self {
let matches = SetBuf::new_unchecked(vec![match_]);
- Self::from_matches(doc, matches)
+ Self::from_matches(id, matches)
}
pub fn from_matches(id: DocumentId, matches: SetBuf<Match>) -> Self {
- let mut last = 0;
- let mut slices = vec![0];
- for group in GroupBy::new(&matches, match_query_index) {
- let index = last + group.len();
- slices.push(index);
- last = index;
- }
-
- let matches = Matches { matches, slices };
+ let matches = Matches::new(matches);
Self { id, matches }
}
- pub fn from_unsorted_matches(doc: DocumentId, mut matches: Vec<Match>) -> Self {
- matches.sort_unstable();
- let matches = SetBuf::new_unchecked(matches);
- Self::from_matches(doc, matches)
+ pub fn from_unsorted_matches(id: DocumentId, matches: Vec<Match>) -> Self {
+ let matches = Matches::from_unsorted(matches);
+ Self { id, matches }
}
}
@@ -55,6 +46,25 @@ pub struct Matches {
}
impl Matches {
+ pub fn new(matches: SetBuf<Match>) -> Matches {
+ let mut last = 0;
+ let mut slices = vec![0];
+
+ for group in GroupBy::new(&matches, match_query_index) {
+ let index = last + group.len();
+ slices.push(index);
+ last = index;
+ }
+
+ Matches { matches, slices }
+ }
+
+ pub fn from_unsorted(mut matches: Vec<Match>) -> Matches {
+ matches.sort_unstable();
+ let matches = SetBuf::new_unchecked(matches);
+ Matches::new(matches)
+ }
+
pub fn query_index_groups(&self) -> QueryIndexGroups {
QueryIndexGroups {
matches: &self.matches,
@@ -75,7 +85,7 @@ impl<'a, 'b> Iterator for QueryIndexGroups<'a, 'b> {
self.windows.next().map(|range| {
match *range {
[left, right] => &self.matches[left..right],
- _ => unreachable!()
+ _ => unreachable!(),
}
})
}
From ef7ba96d4a3f2107c386aa33860e707b812ab732 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 11:23:42 +0100
Subject: [PATCH 3/7] feat: Introducing the Matches as_matches method
---
examples/query-database.rs | 4 ++--
src/rank/mod.rs | 4 ++++
2 files changed, 6 insertions(+), 2 deletions(-)
diff --git a/examples/query-database.rs b/examples/query-database.rs
index ce64870a9..edd311f16 100644
--- a/examples/query-database.rs
+++ b/examples/query-database.rs
@@ -115,7 +115,7 @@ fn main() -> Result<(), Box<Error>> {
};
print!("{}: ", name);
- let areas = create_highlight_areas(&text, &doc.matches, attr);
+ let areas = create_highlight_areas(&text, doc.matches.as_matches(), attr);
display_highlights(&text, &areas)?;
println!();
}
@@ -124,7 +124,7 @@ fn main() -> Result<(), Box<Error>> {
}
let mut matching_attributes = HashSet::new();
- for _match in doc.matches {
+ for _match in doc.matches.as_matches() {
let attr = SchemaAttr::new(_match.attribute.attribute());
let name = schema.attribute_name(attr);
matching_attributes.insert(name);
diff --git a/src/rank/mod.rs b/src/rank/mod.rs
index ad91830bf..5416ca882 100644
--- a/src/rank/mod.rs
+++ b/src/rank/mod.rs
@@ -71,6 +71,10 @@ impl Matches {
windows: self.slices.windows(2),
}
}
+
+ pub fn as_matches(&self) -> &[Match] {
+ &self.matches
+ }
}
pub struct QueryIndexGroups<'a, 'b> {
From c594597a01422087a66e78bdfddd32d5abc99ad2 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 11:29:43 +0100
Subject: [PATCH 4/7] feat: Introduce multiple Iterator impl for Matches
---
src/rank/mod.rs | 60 ++++++++++++++++++++++++++++++++++++++++++++-----
1 file changed, 54 insertions(+), 6 deletions(-)
diff --git a/src/rank/mod.rs b/src/rank/mod.rs
index 5416ca882..2023cb266 100644
--- a/src/rank/mod.rs
+++ b/src/rank/mod.rs
@@ -2,6 +2,7 @@ pub mod criterion;
mod query_builder;
mod distinct_map;
+use std::iter::FusedIterator;
use std::slice::Windows;
use sdset::SetBuf;
@@ -82,9 +83,10 @@ pub struct QueryIndexGroups<'a, 'b> {
windows: Windows<'b, usize>,
}
-impl<'a, 'b> Iterator for QueryIndexGroups<'a, 'b> {
+impl<'a> Iterator for QueryIndexGroups<'a, '_> {
type Item = &'a [Match];
+ #[inline]
fn next(&mut self) -> Option<Self::Item> {
self.windows.next().map(|range| {
match *range {
@@ -93,10 +95,56 @@ impl<'a, 'b> Iterator for QueryIndexGroups<'a, 'b> {
}
})
}
+
+ #[inline]
+ fn size_hint(&self) -> (usize, Option<usize>) {
+ self.windows.size_hint()
+ }
+
+ #[inline]
+ fn count(self) -> usize {
+ self.len()
+ }
+
+ #[inline]
+ fn nth(&mut self, n: usize) -> Option<Self::Item> {
+ self.windows.nth(n).map(|range| {
+ match *range {
+ [left, right] => &self.matches[left..right],
+ _ => unreachable!(),
+ }
+ })
+ }
+
+ #[inline]
+ fn last(self) -> Option<Self::Item> {
+ let (matches, windows) = (self.matches, self.windows);
+ windows.last().map(|range| {
+ match *range {
+ [left, right] => &matches[left..right],
+ _ => unreachable!(),
+ }
+ })
+ }
}
-// impl ExactSizeIterator for QueryIndexGroups<'_, '_> {
-// fn len(&self) -> usize {
-// self.windows.len() // FIXME (+1) ?
-// }
-// }
+impl ExactSizeIterator for QueryIndexGroups<'_, '_> {
+ #[inline]
+ fn len(&self) -> usize {
+ self.windows.len()
+ }
+}
+
+impl FusedIterator for QueryIndexGroups<'_, '_> { }
+
+impl DoubleEndedIterator for QueryIndexGroups<'_, '_> {
+ #[inline]
+ fn next_back(&mut self) -> Option<Self::Item> {
+ self.windows.next_back().map(|range| {
+ match *range {
+ [left, right] => &self.matches[left..right],
+ _ => unreachable!(),
+ }
+ })
+ }
+}
From 0d07af3caf432f6a4c62125f5765c191881f637a Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 11:41:47 +0100
Subject: [PATCH 5/7] fix: Filter and count the exact matching words
---
src/rank/criterion/exact.rs | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/rank/criterion/exact.rs b/src/rank/criterion/exact.rs
index 6d33a6f38..759bd951e 100644
--- a/src/rank/criterion/exact.rs
+++ b/src/rank/criterion/exact.rs
@@ -9,13 +9,13 @@ use crate::database::DatabaseView;
use crate::Match;
#[inline]
-fn contains_exact(matches: &[Match]) -> bool {
+fn contains_exact(matches: &&[Match]) -> bool {
matches.iter().any(|m| m.is_exact)
}
#[inline]
fn number_exact_matches(matches: &Matches) -> usize {
- matches.query_index_groups().map(contains_exact).count()
+ matches.query_index_groups().filter(contains_exact).count()
}
#[derive(Debug, Clone, Copy)]
From d899b8660382ea5a4c211761d28ea31e885a46ce Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 12:33:27 +0100
Subject: [PATCH 6/7] feat: Prefer using ranges and not using unreachable!
---
src/rank/mod.rs | 54 ++++++++++++++++-----------------------
src/rank/query_builder.rs | 2 +-
2 files changed, 23 insertions(+), 33 deletions(-)
diff --git a/src/rank/mod.rs b/src/rank/mod.rs
index 2023cb266..91392cbee 100644
--- a/src/rank/mod.rs
+++ b/src/rank/mod.rs
@@ -3,7 +3,8 @@ mod query_builder;
mod distinct_map;
use std::iter::FusedIterator;
-use std::slice::Windows;
+use std::slice::Iter;
+use std::ops::Range;
use sdset::SetBuf;
use group_by::GroupBy;
@@ -43,18 +44,19 @@ impl Document {
#[derive(Debug, Clone)]
pub struct Matches {
matches: SetBuf<Match>,
- slices: Vec<usize>,
+ slices: Vec<Range<usize>>,
}
impl Matches {
pub fn new(matches: SetBuf<Match>) -> Matches {
- let mut last = 0;
- let mut slices = vec![0];
+ let mut last_end = 0;
+ let mut slices = Vec::new();
for group in GroupBy::new(&matches, match_query_index) {
- let index = last + group.len();
- slices.push(index);
- last = index;
+ let start = last_end;
+ let end = last_end + group.len();
+ slices.push(Range { start, end });
+ last_end = end;
}
Matches { matches, slices }
@@ -69,7 +71,7 @@ impl Matches {
pub fn query_index_groups(&self) -> QueryIndexGroups {
QueryIndexGroups {
matches: &self.matches,
- windows: self.slices.windows(2),
+ slices: self.slices.iter(),
}
}
@@ -80,7 +82,7 @@ impl Matches {
pub struct QueryIndexGroups<'a, 'b> {
matches: &'a [Match],
- windows: Windows<'b, usize>,
+ slices: Iter<'b, Range<usize>>,
}
impl<'a> Iterator for QueryIndexGroups<'a, '_> {
@@ -88,17 +90,14 @@ impl<'a> Iterator for QueryIndexGroups<'a, '_> {
#[inline]
fn next(&mut self) -> Option<Self::Item> {
- self.windows.next().map(|range| {
- match *range {
- [left, right] => &self.matches[left..right],
- _ => unreachable!(),
- }
+ self.slices.next().cloned().map(|range| {
+ unsafe { self.matches.get_unchecked(range) }
})
}
#[inline]
fn size_hint(&self) -> (usize, Option<usize>) {
- self.windows.size_hint()
+ self.slices.size_hint()
}
#[inline]
@@ -108,22 +107,16 @@ impl<'a> Iterator for QueryIndexGroups<'a, '_> {
#[inline]
fn nth(&mut self, n: usize) -> Option<Self::Item> {
- self.windows.nth(n).map(|range| {
- match *range {
- [left, right] => &self.matches[left..right],
- _ => unreachable!(),
- }
+ self.slices.nth(n).cloned().map(|range| {
+ unsafe { self.matches.get_unchecked(range) }
})
}
#[inline]
fn last(self) -> Option<Self::Item> {
- let (matches, windows) = (self.matches, self.windows);
- windows.last().map(|range| {
- match *range {
- [left, right] => &matches[left..right],
- _ => unreachable!(),
- }
+ let (matches, slices) = (self.matches, self.slices);
+ slices.last().cloned().map(|range| {
+ unsafe { matches.get_unchecked(range) }
})
}
}
@@ -131,7 +124,7 @@ impl<'a> Iterator for QueryIndexGroups<'a, '_> {
impl ExactSizeIterator for QueryIndexGroups<'_, '_> {
#[inline]
fn len(&self) -> usize {
- self.windows.len()
+ self.slices.len()
}
}
@@ -140,11 +133,8 @@ impl FusedIterator for QueryIndexGroups<'_, '_> { }
impl DoubleEndedIterator for QueryIndexGroups<'_, '_> {
#[inline]
fn next_back(&mut self) -> Option<Self::Item> {
- self.windows.next_back().map(|range| {
- match *range {
- [left, right] => &self.matches[left..right],
- _ => unreachable!(),
- }
+ self.slices.next_back().cloned().map(|range| {
+ unsafe { self.matches.get_unchecked(range) }
})
}
}
diff --git a/src/rank/query_builder.rs b/src/rank/query_builder.rs
index 9a17e0473..3ec49d83d 100644
--- a/src/rank/query_builder.rs
+++ b/src/rank/query_builder.rs
@@ -116,7 +116,7 @@ where D: Deref<Target=DB>,
}
}
- matches.into_iter().map(|(id, m)| Document::from_unsorted_matches(id, m)).collect()
+ matches.into_iter().map(|(i, m)| Document::from_unsorted_matches(i, m)).collect()
}
}
From c74caa0f82ece8abb9179b3cf1416f6ac5e76bbb Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Cl=C3=A9ment=20Renault?= <renault.cle@gmail.com>
Date: Sun, 6 Jan 2019 13:18:04 +0100
Subject: [PATCH 7/7] feat: Sum usizes instead of little u16/u32
---
src/rank/criterion/sum_of_words_attribute.rs | 4 ++--
src/rank/criterion/sum_of_words_position.rs | 4 ++--
2 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/rank/criterion/sum_of_words_attribute.rs b/src/rank/criterion/sum_of_words_attribute.rs
index cfcaaec68..aea21c35f 100644
--- a/src/rank/criterion/sum_of_words_attribute.rs
+++ b/src/rank/criterion/sum_of_words_attribute.rs
@@ -8,11 +8,11 @@ use crate::database::DatabaseView;
use crate::rank::criterion::Criterion;
#[inline]
-fn sum_matches_attributes(matches: &Matches) -> u16 {
+fn sum_matches_attributes(matches: &Matches) -> usize {
// note that GroupBy will never return an empty group
// so we can do this assumption safely
matches.query_index_groups().map(|group| {
- unsafe { group.get_unchecked(0).attribute.attribute() }
+ unsafe { group.get_unchecked(0).attribute.attribute() as usize }
}).sum()
}
diff --git a/src/rank/criterion/sum_of_words_position.rs b/src/rank/criterion/sum_of_words_position.rs
index e1d650534..0b27184ba 100644
--- a/src/rank/criterion/sum_of_words_position.rs
+++ b/src/rank/criterion/sum_of_words_position.rs
@@ -8,11 +8,11 @@ use crate::rank::criterion::Criterion;
use crate::database::DatabaseView;
#[inline]
-fn sum_matches_attribute_index(matches: &Matches) -> u32 {
+fn sum_matches_attribute_index(matches: &Matches) -> usize {
// note that GroupBy will never return an empty group
// so we can do this assumption safely
matches.query_index_groups().map(|group| {
- unsafe { group.get_unchecked(0).attribute.word_index() }
+ unsafe { group.get_unchecked(0).attribute.word_index() as usize }
}).sum()
}