damus

commit 77331644cbad717d7d3bc50c1c638707ac96cea5
parent 8d14fdffb576e4f9b635ec5f8899dd354b9e5441
Author: William Casarin <jb55@jb55.com>
Date:   Sun, 16 Jul 2023 11:00:48 -0700

Fix issue with emojis next to hashtags and urls

Treat utf8 bytes next to hashtags and urls as boundary conditions

Changelog-Fixed: Fix issue with emojis next to hashtags and urls

Diffstat:
M
damus-c/cursor.h
 | 
12
++++++++++--
M
damus-c/damus.c
 | 
2
+-

2 files changed, 11 insertions(+), 3 deletions(-)
diff --git a/damus-c/cursor.h b/damus-c/cursor.h
@@ -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++;
diff --git a/damus-c/damus.c b/damus-c/damus.c
@@ -231,7 +231,7 @@ int damus_parse_content(struct note_blocks *blocks, const char *content) {
         }
         
         pre_mention = cur.p;
-        if (cp == -1 || is_boundary(cp) || c == '#') {
+        if (cp == -1 || is_left_boundary(cp) || c == '#') {
             if (c == '#' && (parse_mention_index(&cur, &block) || parse_hashtag(&cur, &block))) {
                 if (!add_text_then_block(&cur, blocks, block, &start, pre_mention))
                     return 0;