patch 9.1.1046: fuzzymatching doesn't prefer matching camelcase
Problem: fuzzymatching doesn't prefer matching camelcase
(Tomasz N)
Solution: Add extra score when case matches (glepnir)
fixes: #16434
closes: #16439
Signed-off-by: glepnir <glephunter@gmail.com>
Signed-off-by: Christian Brabandt <cb@256bit.org>
diff --git a/src/search.c b/src/search.c
index 616331e..46fa7b9 100644
--- a/src/search.c
+++ b/src/search.c
@@ -42,7 +42,7 @@
static int is_zero_width(char_u *pattern, size_t patternlen, int move, pos_T *cur, int direction);
static void cmdline_search_stat(int dirc, pos_T *pos, pos_T *cursor_pos, int show_top_bot_msg, char_u *msgbuf, size_t msgbuflen, int recompute, int maxcount, long timeout);
static void update_search_stat(int dirc, pos_T *pos, pos_T *cursor_pos, searchstat_T *stat, int recompute, int maxcount, long timeout);
-static int fuzzy_match_compute_score(char_u *str, int strSz, int_u *matches, int numMatches);
+static int fuzzy_match_compute_score(char_u *fuzpat, char_u *str, int strSz, int_u *matches, int numMatches);
static int fuzzy_match_recursive(char_u *fuzpat, char_u *str, int_u strIdx, int *outScore, char_u *strBegin, int strLen, int_u *srcMatches, int_u *matches, int maxMatches, int nextMatch, int *recursionCount);
#if defined(FEAT_EVAL) || defined(FEAT_PROTO)
static int fuzzy_match_item_compare(const void *s1, const void *s2);
@@ -4355,6 +4355,10 @@
#define CAMEL_BONUS 30
// bonus if the first letter is matched
#define FIRST_LETTER_BONUS 15
+// bonus if exact match
+#define EXACT_MATCH_BONUS 100
+// bonus if case match when no ignorecase
+#define CASE_MATCH_BONUS 25
// penalty applied for every letter in str before the first match
#define LEADING_LETTER_PENALTY (-5)
// maximum penalty for leading letters
@@ -4374,6 +4378,7 @@
*/
static int
fuzzy_match_compute_score(
+ char_u *fuzpat,
char_u *str,
int strSz,
int_u *matches,
@@ -4386,6 +4391,11 @@
char_u *p = str;
int_u sidx = 0;
int is_exact_match = TRUE;
+ char_u *orig_fuzpat = fuzpat - numMatches;
+ char_u *curpat = orig_fuzpat;
+ int pat_idx = 0;
+ // Track consecutive camel case matches
+ int consecutive_camel = 0;
// Initialize score
score = 100;
@@ -4404,6 +4414,8 @@
for (i = 0; i < numMatches; ++i)
{
int_u currIdx = matches[i];
+ int curr;
+ int is_camel = FALSE;
if (i > 0)
{
@@ -4413,15 +4425,18 @@
if (currIdx == (prevIdx + 1))
score += SEQUENTIAL_BONUS;
else
+ {
score += GAP_PENALTY * (currIdx - prevIdx);
+ // Reset consecutive camel count on gap
+ consecutive_camel = 0;
+ }
}
// Check for bonuses based on neighbor character value
if (currIdx > 0)
{
// Camel case
- int neighbor = ' ';
- int curr;
+ int neighbor = ' ';
if (has_mbyte)
{
@@ -4439,8 +4454,18 @@
curr = str[currIdx];
}
+ // Enhanced camel case scoring
if (vim_islower(neighbor) && vim_isupper(curr))
- score += CAMEL_BONUS;
+ {
+ score += CAMEL_BONUS * 2; // Double the camel case bonus
+ is_camel = TRUE;
+ consecutive_camel++;
+ // Additional bonus for consecutive camel
+ if (consecutive_camel > 1)
+ score += CAMEL_BONUS;
+ }
+ else
+ consecutive_camel = 0;
// Bonus if the match follows a separator character
if (neighbor == '/' || neighbor == '\\')
@@ -4452,14 +4477,47 @@
{
// First letter
score += FIRST_LETTER_BONUS;
+ curr = has_mbyte ? (*mb_ptr2char)(p) : str[currIdx];
}
+
+ // Case matching bonus
+ if (vim_isalpha(curr))
+ {
+ while (pat_idx < i && *curpat)
+ {
+ if (has_mbyte)
+ MB_PTR_ADV(curpat);
+ else
+ curpat++;
+ pat_idx++;
+ }
+
+ if (has_mbyte)
+ {
+ if (curr == (*mb_ptr2char)(curpat))
+ {
+ score += CASE_MATCH_BONUS;
+ // Extra bonus for exact case match in camel
+ if (is_camel)
+ score += CASE_MATCH_BONUS / 2;
+ }
+ }
+ else if (curr == *curpat)
+ {
+ score += CASE_MATCH_BONUS;
+ if (is_camel)
+ score += CASE_MATCH_BONUS / 2;
+ }
+ }
+
// Check exact match condition
if (currIdx != (int_u)i)
is_exact_match = FALSE;
}
+
// Boost score for exact matches
if (is_exact_match && numMatches == strSz)
- score += 100;
+ score += EXACT_MATCH_BONUS;
return score;
}
@@ -4563,7 +4621,7 @@
// Calculate score
if (matched)
- *outScore = fuzzy_match_compute_score(strBegin, strLen, matches,
+ *outScore = fuzzy_match_compute_score(fuzpat, strBegin, strLen, matches,
nextMatch);
// Return best result
diff --git a/src/testdir/test_matchfuzzy.vim b/src/testdir/test_matchfuzzy.vim
index e880d73..cba0844 100644
--- a/src/testdir/test_matchfuzzy.vim
+++ b/src/testdir/test_matchfuzzy.vim
@@ -96,15 +96,15 @@
" Test for the matchfuzzypos() function
func Test_matchfuzzypos()
- call assert_equal([['curl', 'world'], [[2,3], [2,3]], [128, 127]], matchfuzzypos(['world', 'curl'], 'rl'))
- call assert_equal([['curl', 'world'], [[2,3], [2,3]], [128, 127]], matchfuzzypos(['world', 'one', 'curl'], 'rl'))
+ call assert_equal([['curl', 'world'], [[2,3], [2,3]], [178, 177]], matchfuzzypos(['world', 'curl'], 'rl'))
+ call assert_equal([['curl', 'world'], [[2,3], [2,3]], [178, 177]], matchfuzzypos(['world', 'one', 'curl'], 'rl'))
call assert_equal([['hello', 'hello world hello world'],
- \ [[0, 1, 2, 3, 4], [0, 1, 2, 3, 4]], [375, 257]],
+ \ [[0, 1, 2, 3, 4], [0, 1, 2, 3, 4]], [500, 382]],
\ matchfuzzypos(['hello world hello world', 'hello', 'world'], 'hello'))
- call assert_equal([['aaaaaaa'], [[0, 1, 2]], [191]], matchfuzzypos(['aaaaaaa'], 'aaa'))
- call assert_equal([['a b'], [[0, 3]], [219]], matchfuzzypos(['a b'], 'a b'))
- call assert_equal([['a b'], [[0, 3]], [219]], matchfuzzypos(['a b'], 'a b'))
- call assert_equal([['a b'], [[0]], [112]], matchfuzzypos(['a b'], ' a '))
+ call assert_equal([['aaaaaaa'], [[0, 1, 2]], [266]], matchfuzzypos(['aaaaaaa'], 'aaa'))
+ call assert_equal([['a b'], [[0, 3]], [269]], matchfuzzypos(['a b'], 'a b'))
+ call assert_equal([['a b'], [[0, 3]], [269]], matchfuzzypos(['a b'], 'a b'))
+ call assert_equal([['a b'], [[0]], [137]], matchfuzzypos(['a b'], ' a '))
call assert_equal([[], [], []], matchfuzzypos(['a b'], ' '))
call assert_equal([[], [], []], matchfuzzypos(['world', 'curl'], 'ab'))
let x = matchfuzzypos([repeat('a', 256)], repeat('a', 256))
@@ -113,33 +113,33 @@
call assert_equal([[], [], []], matchfuzzypos([], 'abc'))
" match in a long string
- call assert_equal([[repeat('x', 300) .. 'abc'], [[300, 301, 302]], [-135]],
+ call assert_equal([[repeat('x', 300) .. 'abc'], [[300, 301, 302]], [-60]],
\ matchfuzzypos([repeat('x', 300) .. 'abc'], 'abc'))
" preference for camel case match
- call assert_equal([['xabcxxaBc'], [[6, 7, 8]], [189]], matchfuzzypos(['xabcxxaBc'], 'abc'))
+ call assert_equal([['xabcxxaBc'], [[6, 7, 8]], [269]], matchfuzzypos(['xabcxxaBc'], 'abc'))
" preference for match after a separator (_ or space)
- call assert_equal([['xabx_ab'], [[5, 6]], [145]], matchfuzzypos(['xabx_ab'], 'ab'))
+ call assert_equal([['xabx_ab'], [[5, 6]], [195]], matchfuzzypos(['xabx_ab'], 'ab'))
" preference for leading letter match
- call assert_equal([['abcxabc'], [[0, 1]], [150]], matchfuzzypos(['abcxabc'], 'ab'))
+ call assert_equal([['abcxabc'], [[0, 1]], [200]], matchfuzzypos(['abcxabc'], 'ab'))
" preference for sequential match
- call assert_equal([['aobncedone'], [[7, 8, 9]], [158]], matchfuzzypos(['aobncedone'], 'one'))
+ call assert_equal([['aobncedone'], [[7, 8, 9]], [233]], matchfuzzypos(['aobncedone'], 'one'))
" best recursive match
- call assert_equal([['xoone'], [[2, 3, 4]], [168]], matchfuzzypos(['xoone'], 'one'))
+ call assert_equal([['xoone'], [[2, 3, 4]], [243]], matchfuzzypos(['xoone'], 'one'))
" match multiple words (separated by space)
- call assert_equal([['foo bar baz'], [[8, 9, 10, 0, 1, 2]], [369]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('baz foo'))
+ call assert_equal([['foo bar baz'], [[8, 9, 10, 0, 1, 2]], [519]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('baz foo'))
call assert_equal([[], [], []], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('baz foo', {'matchseq': 1}))
- call assert_equal([['foo bar baz'], [[0, 1, 2, 8, 9, 10]], [369]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('foo baz'))
- call assert_equal([['foo bar baz'], [[0, 1, 2, 3, 4, 5, 10]], [326]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('foo baz', {'matchseq': 1}))
+ call assert_equal([['foo bar baz'], [[0, 1, 2, 8, 9, 10]], [519]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('foo baz'))
+ call assert_equal([['foo bar baz'], [[0, 1, 2, 3, 4, 5, 10]], [476]], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('foo baz', {'matchseq': 1}))
call assert_equal([[], [], []], ['foo bar baz', 'foo', 'foo bar', 'baz bar']->matchfuzzypos('one two'))
call assert_equal([[], [], []], ['foo bar']->matchfuzzypos(" \t "))
- call assert_equal([['grace'], [[1, 2, 3, 4, 2, 3, 4, 0, 1, 2, 3, 4]], [757]], ['grace']->matchfuzzypos('race ace grace'))
+ call assert_equal([['grace'], [[1, 2, 3, 4, 2, 3, 4, 0, 1, 2, 3, 4]], [1057]], ['grace']->matchfuzzypos('race ace grace'))
let l = [{'id' : 5, 'val' : 'crayon'}, {'id' : 6, 'val' : 'camera'}]
- call assert_equal([[{'id' : 6, 'val' : 'camera'}], [[0, 1, 2]], [192]],
+ call assert_equal([[{'id' : 6, 'val' : 'camera'}], [[0, 1, 2]], [267]],
\ matchfuzzypos(l, 'cam', {'text_cb' : {v -> v.val}}))
- call assert_equal([[{'id' : 6, 'val' : 'camera'}], [[0, 1, 2]], [192]],
+ call assert_equal([[{'id' : 6, 'val' : 'camera'}], [[0, 1, 2]], [267]],
\ matchfuzzypos(l, 'cam', {'key' : 'val'}))
call assert_equal([[], [], []], matchfuzzypos(l, 'day', {'text_cb' : {v -> v.val}}))
call assert_equal([[], [], []], matchfuzzypos(l, 'day', {'key' : 'val'}))
@@ -154,6 +154,18 @@
call assert_fails("let x = matchfuzzypos(l, 'foo', {'key' : test_null_string()})", 'E475:')
call assert_fails("let x = matchfuzzypos(l, 'foo', {'text_cb' : test_null_function()})", 'E475:')
+ " case match
+ call assert_equal([['Match', 'match'], [[0, 1], [0, 1]], [202, 177]], matchfuzzypos(['match', 'Match'], 'Ma'))
+ call assert_equal([['match', 'Match'], [[0, 1], [0, 1]], [202, 177]], matchfuzzypos(['Match', 'match'], 'ma'))
+ " CamelCase has high weight even case match
+ call assert_equal(['MyTestCase', 'mytestcase'], matchfuzzy(['mytestcase', 'MyTestCase'], 'mtc'))
+ call assert_equal(['MyTestCase', 'mytestcase'], matchfuzzy(['MyTestCase', 'mytestcase'], 'mtc'))
+ call assert_equal(['MyTest', 'Mytest', 'mytest', ],matchfuzzy(['Mytest', 'mytest', 'MyTest'], 'MyT'))
+ call assert_equal(['CamelCaseMatchIngAlg', 'camelCaseMatchingAlg', 'camelcasematchingalg'],
+ \ matchfuzzy(['CamelCaseMatchIngAlg', 'camelcasematchingalg', 'camelCaseMatchingAlg'], 'CamelCase'))
+ call assert_equal(['CamelCaseMatchIngAlg', 'camelCaseMatchingAlg', 'camelcasematchingalg'],
+ \ matchfuzzy(['CamelCaseMatchIngAlg', 'camelcasematchingalg', 'camelCaseMatchingAlg'], 'CamelcaseM'))
+
let l = [{'id' : 5, 'name' : 'foo'}, {'id' : 6, 'name' : []}, {'id' : 7}]
call assert_fails("let x = matchfuzzypos(l, 'foo', {'key' : 'name'})", 'E730:')
endfunc
@@ -204,12 +216,12 @@
call assert_equal([['ンヹㄇヺヴ'], [[1, 3]], [88]], matchfuzzypos(['ンヹㄇヺヴ'], 'ヹヺ'))
" reverse the order of characters
call assert_equal([[], [], []], matchfuzzypos(['ンヹㄇヺヴ'], 'ヺヹ'))
- call assert_equal([['αβΩxxx', 'xαxβxΩx'], [[0, 1, 2], [1, 3, 5]], [222, 113]],
+ call assert_equal([['αβΩxxx', 'xαxβxΩx'], [[0, 1, 2], [1, 3, 5]], [252, 143]],
\ matchfuzzypos(['αβΩxxx', 'xαxβxΩx'], 'αβΩ'))
call assert_equal([['ππbbππ', 'πππbbbπππ', 'ππππbbbbππππ', 'πbπ'],
- \ [[0, 1], [0, 1], [0, 1], [0, 2]], [151, 148, 145, 110]],
+ \ [[0, 1], [0, 1], [0, 1], [0, 2]], [176, 173, 170, 135]],
\ matchfuzzypos(['πbπ', 'ππbbππ', 'πππbbbπππ', 'ππππbbbbππππ'], 'ππ'))
- call assert_equal([['ααααααα'], [[0, 1, 2]], [191]],
+ call assert_equal([['ααααααα'], [[0, 1, 2]], [216]],
\ matchfuzzypos(['ααααααα'], 'ααα'))
call assert_equal([[], [], []], matchfuzzypos(['ンヹㄇ', 'ŗŝţ'], 'fffifl'))
@@ -222,10 +234,10 @@
call assert_equal([[], [], []], ['세 마리의 작은 돼지', '마리의', '마리의 작은', '작은 돼지']->matchfuzzypos('파란 하늘'))
" match in a long string
- call assert_equal([[repeat('ぶ', 300) .. 'ẼẼẼ'], [[300, 301, 302]], [-135]],
+ call assert_equal([[repeat('ぶ', 300) .. 'ẼẼẼ'], [[300, 301, 302]], [-110]],
\ matchfuzzypos([repeat('ぶ', 300) .. 'ẼẼẼ'], 'ẼẼẼ'))
" preference for camel case match
- call assert_equal([['xѳѵҁxxѳѴҁ'], [[6, 7, 8]], [189]], matchfuzzypos(['xѳѵҁxxѳѴҁ'], 'ѳѵҁ'))
+ call assert_equal([['xѳѵҁxxѳѴҁ'], [[6, 7, 8]], [219]], matchfuzzypos(['xѳѵҁxxѳѴҁ'], 'ѳѵҁ'))
" preference for match after a separator (_ or space)
call assert_equal([['xちだx_ちだ'], [[5, 6]], [145]], matchfuzzypos(['xちだx_ちだ'], 'ちだ'))
" preference for leading letter match
diff --git a/src/version.c b/src/version.c
index fb5fd37..ad8a748 100644
--- a/src/version.c
+++ b/src/version.c
@@ -705,6 +705,8 @@
static int included_patches[] =
{ /* Add new patch number below this line */
/**/
+ 1046,
+/**/
1045,
/**/
1044,