Jean-Louis Fuchs
9d897a6092
Rank a word-boundary after a non-word-boundary
2019-09-07 22:53:29 +02:00
Maxime Coste
08df409a53
RankedMatch: Do not compare word boundary match count on single word matches
...
As the computation of word boundary matches is separate from the
actual subsequence matching, we sometimes have candidate that match
as a single word while still having multiple word boundary matches.
For example, with query "expresins", candidate "expressionism's"
will match as single word ("expressins" is a subsequence of
"expressionism"), and will have two word boundaries match (it does
match the last "s", which is considered as a separate word).
This should not be taken into account when compared against
candidate "expresions", which should be considered a better
match.
Fixes #1925
2018-03-15 23:20:03 +11:00
Maxime Coste
86fcc55e53
RankedMatch: Make punctuation ordered *before* alphanumeric characters
2017-12-06 13:58:34 +08:00
Maxime Coste
6ada6e6d77
Move all non-core string code to string_utils.{hh,cc}
2017-10-10 10:52:32 +08:00
Maxime Coste
ec1824d3c3
Make non smart case full match better than smart case full match
2017-07-19 20:18:47 +02:00
Maxime Coste
016a50f213
Use smart case matching for contiguous/prefix/fullmatch detection
...
Fixes #1498
2017-07-19 20:16:28 +02:00
Maxime Coste
6101138606
Change RankedMatch ordering to favor / characters
...
This will improve matching of filenames, as 'foo/' will be sorted
before 'foo-bar' due to `/` coming before `-` in the new ordering
(it comes after in ascii/unicode order).
Fixes #1395
2017-06-04 08:27:53 +01:00
Maxime Coste
db9b863222
Migrate WithBitOps template specialization to with_bit_ops function
...
This way we dont depend on knowing the base template to enable bit ops
on an enum type.
2017-03-15 18:00:58 +00:00
Maxime Coste
6a0354a597
Tweak ranked match ordering
2017-01-31 22:30:23 +00:00
Maxime Coste
9d09d14d99
Warning fix in ranked_match.cc
2017-01-30 11:22:14 +00:00
Maxime Coste
632e05d830
Fix infinite loop when comparing RankedMatches containing invalid utf8
...
If we had a word containing some invalid utf8, like a wrong sequence
of continuation bytes, we would infinitely loop back to the previous
valid character start.
Fixes #1157
2017-01-29 23:50:33 +00:00
Maxime Coste
cb395d39f8
Use iswlower instead of islower
...
islower can crash with big codepoints, and is incorrect anyway.
2017-01-29 23:37:10 +00:00
Maxime Coste
12856066b1
Cleanup include dependencies a bit
2016-11-28 23:58:08 +00:00
Maxime Coste
3f3ed0b333
Fix literal type that must be 64 bits
2016-11-22 22:20:30 +00:00
Maxime Coste
6d79ade019
Tweak RankedMatch behaviour and fix bug in its comparison function
...
casting TestableFlag<T> to UnderlyingType<T> was going through bool
conversion... Not sure how things worked earlier.
2016-11-14 19:39:35 +00:00
Maxime Coste
dc735450a8
Tweak ranked match comparison, give contiguous matches an edge
2016-09-30 22:13:01 +01:00
Maxime Coste
b5450aa375
Tweak RankedMatch logic, prioritize matches that are in a single word
2016-09-26 21:54:39 +01:00
Maxime Coste
f52f5f7c53
Only decode utf8 when strictly necessary in RankedMatch::operator<
2016-08-30 00:51:21 +01:00
Maxime Coste
2b631837d8
Tweak RankedMatch, compare max match index instead of match indices sum
2016-08-30 00:30:15 +01:00
Maxime Coste
cbe38b2f96
Use flags and bit operations instead of bools in RankedMatch
...
full match is now the most important flag for comparison.
2016-08-29 23:56:22 +01:00
Maxime Coste
fa35330b42
Fix RankedMatch ordering where 'a < a' was true
...
Fixes #679
2016-05-20 18:42:01 +01:00
Maxime Coste
1834a67b87
Go back to libc locale and use c_regex_traits
...
Unfortunately, cygwin does not support c++ locales.
2016-05-19 21:45:23 +01:00
Maxime Coste
49e674fe55
Fix RankedMatch::operator< with matching prefix candidates
2016-05-17 19:40:36 +01:00
Maxime Coste
84f62e6ff2
Use C++ locale based functions instead of the libc ones
2016-05-11 09:49:45 +01:00
Maxime Coste
989d64abd4
Tweak RankedMatch behaviour
2016-04-04 13:42:58 +01:00
Maxime Coste
0478a8f11a
Give priority to lower case in RankedMatch
2016-04-04 13:42:58 +01:00
Maxime Coste
94f4531bc8
Tweak subsequence_match_smart_case
2016-03-28 15:18:15 +01:00
Maxime Coste
2a9875bd58
Use manual lexicographic comparison in RankedMatch::operator<
2016-03-28 14:44:49 +01:00
Maxime Coste
c756f8f124
Move UsedLetters with RankedMatch
2016-03-25 20:38:27 +00:00
Maxime Coste
87704227ad
Tweak implementation of subsequence_match_smart_case
...
Remove use of utf8 iterators and use directly utf8 functions
2016-03-24 23:45:56 +00:00
Maxime Coste
ccb53eca42
Fix uninitialized value in RankedMatch
2016-03-24 22:04:56 +00:00
Maxime Coste
761d316af6
Tweak RankedMatch ordering, give priority to prefix matches
2016-03-02 13:30:54 +00:00
Maxime Coste
9b70346a00
Take subsequence matches index when sorting RankedMatch
2016-02-28 23:05:51 +00:00
Maxime Coste
53821be7cc
Tweak RankedMatch::operator<
2016-02-22 23:07:29 +00:00
Maxime Coste
85a8a0c26d
Fix count_word_boundaries_match
2016-02-17 23:05:08 +00:00
Maxime Coste
892c3647e4
Fix to_lower/to_upper handling to correctly support non unicode chars
...
require a proper unicode locale setup on the system
Fixes #94
2015-11-11 00:21:20 +00:00
Maxime Coste
2bf44b6b49
Make word insert completion work better with unicode char
2015-10-30 13:57:46 +00:00
Maxime Coste
24043bbffe
Use an heuristic based match ranking algorithm inspired by what YouCompleteMe does
2015-10-29 13:36:30 +00:00
Maxime Coste
89d22f3335
Move more logic into RankedMatch
2015-10-27 21:25:18 +00:00
Maxime Coste
c77cb7c777
Extract WordDB::RankedWord as RankedMatch in its own file
2015-10-22 19:49:08 +01:00
