tyiu/damus
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix issue with emojis next to hashtags and urls

Browse Source 
Treat utf8 bytes next to hashtags and urls as boundary conditions

Changelog-Fixed: Fix issue with emojis next to hashtags and urls
This commit is contained in:
William Casarin
2023-07-16 11:00:48 -07:00
parent 8d14fdffb5
commit 77331644cb
2 changed files with 11 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
damus-c/cursor.h
View File

@@ -431,10 +431,18 @@ static inline int is_whitespace(char c) {
return c == ' ' || c == '\t' || c == '\n' || c == '\v' || c == '\f' || c == '\r';
}
static inline int is_boundary(char c) {
static inline int is_utf8_byte(u8 c) {
return c & 0x80;
}
static inline int is_right_boundary(char c) {
return is_whitespace(c) || ispunct(c);
}
static inline int is_left_boundary(char c) {
return is_right_boundary(c) || is_utf8_byte(c);
}
static inline int is_invalid_url_ending(char c) {
return c == '!' || c == '?' || c == ')' || c == '.' || c == ',' || c == ';';
}
@@ -449,7 +457,7 @@ static inline int consume_until_boundary(struct cursor *cur) {
while (cur->p < cur->end) {
c = *cur->p;
if (is_boundary(c))
if (is_right_boundary(c))
return 1;
cur->p++;