snac2

commit b262961c93d933b886db1fb1f7d0273e11b5d282
parent 8e1f4430ff0408fff1b25715d45065afb4f6bd10
Author: default <nobody@localhost>
Date:   Tue, 17 Jan 2023 09:37:41 +0100

Process HTML Unicode entities specially in process_tags().

Diffstat:
M
activitypub.c
 | 
9
++++++++-

1 file changed, 8 insertions(+), 1 deletion(-)
diff --git a/activitypub.c b/activitypub.c
@@ -272,7 +272,7 @@ void process_tags(snac *snac, const char *content, d_char **n_content, d_char **
     char *p, *v;
     int n = 0;
 
-    split = xs_regex_split(content, "(@[A-Za-z0-9_]+@[A-Za-z0-9\\.-]+|#[^ ,\\.:;<]+)");
+    split = xs_regex_split(content, "(@[A-Za-z0-9_]+@[A-Za-z0-9\\.-]+|&#[0-9]+;|#[^ ,\\.:;<]+)");
 
     p = split;
     while (xs_list_iter(&p, &v)) {
@@ -321,6 +321,13 @@ void process_tags(snac *snac, const char *content, d_char **n_content, d_char **
                 /* add the code */
                 nc = xs_str_cat(nc, l);
             }
+            else
+            if (*v == '&') {
+                /* HTML Unicode entity, probably part of an emoji */
+
+                /* write as is */
+                nc = xs_str_cat(nc, v);
+            }
         }
         else
             nc = xs_str_cat(nc, v);