tommy/libpostal
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

[utils] require at least one non-space/non-hyphen match in utf8_common_prefix_len_ignore_separators

Browse Source
This commit is contained in:
Al
2015-06-12 11:13:49 -04:00
parent 6841ed8fb3
commit 3442b9ad92
1 changed files with 9 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

10
src/string_utils.c
View File

@@ -236,6 +236,8 @@ size_t utf8_common_prefix_len_ignore_separators(const char *str1, const char *st
size_t match_len = 0; size_t match_len = 0;
bool one_char_match = false;
while (1) { while (1) {
len1 = utf8proc_iterate(ptr1, -1, &c1); len1 = utf8proc_iterate(ptr1, -1, &c1);
len2 = utf8proc_iterate(ptr2, -1, &c2); len2 = utf8proc_iterate(ptr2, -1, &c2);
@@ -247,9 +249,14 @@ size_t utf8_common_prefix_len_ignore_separators(const char *str1, const char *st
ptr2 += len2; ptr2 += len2;
remaining -= len1; remaining -= len1;
match_len += len1; match_len += len1;
one_char_match = true;
} else if (utf8_is_hyphen(c1) || utf8_is_separator(utf8proc_category(c1))) { } else if (utf8_is_hyphen(c1) || utf8_is_separator(utf8proc_category(c1))) {
ptr1 += len1; ptr1 += len1;
match_len += len1; match_len += len1;
if (utf8_is_hyphen(c2) || utf8_is_separator(utf8proc_category(c2))) {
ptr2 += len2;
remaining -= len2;
}
} else if (utf8_is_hyphen(c2) || utf8_is_separator(utf8proc_category(c2))) { } else if (utf8_is_hyphen(c2) || utf8_is_separator(utf8proc_category(c2))) {
ptr2 += len2; ptr2 += len2;
remaining -= len2; remaining -= len2;
@@ -261,7 +268,8 @@ size_t utf8_common_prefix_len_ignore_separators(const char *str1, const char *st
} }
return match_len; return one_char_match ? match_len : 0;
} }
inline size_t utf8_common_prefix_ignore_separators(const char *str1, const char *str2) { inline size_t utf8_common_prefix_ignore_separators(const char *str1, const char *str2) {