Merge pull request 'use utf-8 lowercase function for tags #396' (#400) from postscriptum/snac2:utf8-lower-tags into master

Reviewed-on: https://codeberg.org/grunfink/snac2/pulls/400

4 files changed, 13 insertions, 11 deletions

diff --git a/activitypub.c b/activitypub.c
index 120b4a1..c06d8df 100644
--- a/activitypub.c
+++ b/activitypub.c
@@ -903,7 +903,7 @@ xs_str *process_tags(snac *snac, const char *content, xs_list **tag)
             if (*v == '#') {
                 /* hashtag */
                 xs *d = xs_dict_new();
-                xs *n = xs_tolower_i(xs_dup(v));
+                xs *n = xs_utf8_to_lower(xs_dup(v));
                 xs *h = xs_fmt("%s?t=%s", srv_baseurl, n + 1);
                 xs *l = xs_fmt("<a href=\"%s\" class=\"mention hashtag\" rel=\"tag\">%s</a>", h, v);
 
diff --git a/data.c b/data.c
index f9d27f9..631a68b 100644
--- a/data.c
+++ b/data.c
@@ -2217,7 +2217,7 @@ void tag_index(const char *id, const xs_dict *obj)
                 if (*name == '\0')
                     continue;
 
-                name = xs_tolower_i((xs_str *)name);
+                name = xs_utf8_to_lower((xs_str *)name);
 
                 xs *md5_tag   = xs_md5_hex(name, strlen(name));
                 xs *tag_dir   = xs_fmt("%s/%c%c", g_tag_dir, md5_tag[0], md5_tag[1]);
@@ -2247,7 +2247,7 @@ xs_str *tag_fn(const char *tag)
     if (*tag == '#')
         tag++;
 
-    xs *lw_tag = xs_tolower_i(xs_dup(tag));
+    xs *lw_tag = xs_utf8_to_lower(xs_dup(tag));
     xs *md5    = xs_md5_hex(lw_tag, strlen(lw_tag));
 
     return xs_fmt("%s/tag/%c%c/%s.idx", srv_basedir, md5[0], md5[1], md5);
@@ -2834,7 +2834,7 @@ int content_match(const char *file, const xs_dict *msg)
             /* massage content (strip HTML tags, etc.) */
             xs *c = xs_regex_replace(v, "<[^>]+>", " ");
             c = xs_regex_replace_i(c, " {2,}", " ");
-            c = xs_tolower_i(c);
+            c = xs_utf8_to_lower(c);
 
             while (!r && !feof(f)) {
                 xs *rx = xs_strip_i(xs_readline(f));
diff --git a/format.c b/format.c
index 2f30a0d..0f844cd 100644
--- a/format.c
+++ b/format.c
@@ -8,6 +8,7 @@
 #include "xs_json.h"
 #include "xs_time.h"
 #include "xs_match.h"
+#include "xs_unicode.h"
 
 #include "snac.h"
 
@@ -443,7 +444,7 @@ xs_str *sanitize(const char *content)
         if (n & 0x1) {
             xs *s1  = xs_strip_i(xs_crop_i(xs_dup(v), v[1] == '/' ? 2 : 1, -1));
             xs *l1  = xs_split_n(s1, " ", 1);
-            xs *tag = xs_tolower_i(xs_dup(xs_list_get(l1, 0)));
+            xs *tag = xs_utf8_to_lower(xs_dup(xs_list_get(l1, 0)));
             xs *s2  = NULL;
             int i;
 
diff --git a/mastoapi.c b/mastoapi.c
index a7d9c34..c46a971 100644
--- a/mastoapi.c
+++ b/mastoapi.c
@@ -15,6 +15,7 @@
 #include "xs_url.h"
 #include "xs_mime.h"
 #include "xs_match.h"
+#include "xs_unicode.h"
 
 #include "snac.h"
 
@@ -381,7 +382,7 @@ int oauth_post_handler(const xs_dict *req, const char *q_path,
             }
         }
 
-        /* no code? 
+        /* no code?
            I'm not sure of the impacts of this right now, but Subway Tooter does not
            provide a code so one must be generated */
         if (xs_is_null(code)){
@@ -1637,7 +1638,7 @@ int mastoapi_get_handler(const xs_dict *req, const char *q_path,
                 const char *aq = xs_dict_get(args, "q");
 
                 if (!xs_is_null(aq)) {
-                    xs *q    = xs_tolower_i(xs_dup(aq));
+                    xs *q    = xs_utf8_to_lower(xs_dup(aq));
                     out      = xs_list_new();
                     xs *wing = following_list(&snac1);
                     xs *wers = follower_list(&snac1);
@@ -1780,7 +1781,7 @@ int mastoapi_get_handler(const xs_dict *req, const char *q_path,
                     }
                     else
                     if (strcmp(opt, "statuses") == 0) {
-                        /* we don't serve statuses of others; return the empty list */ 
+                        /* we don't serve statuses of others; return the empty list */
                         out = xs_list_new();
                     }
                     else
@@ -1999,7 +2000,7 @@ int mastoapi_get_handler(const xs_dict *req, const char *q_path,
     }
     else
     if (strcmp(cmd, "/v2/filters") == 0) { /** **/
-        /* snac will never have filters 
+        /* snac will never have filters
          * but still, without a v2 endpoint a short delay is introduced
          * in some apps */
         *body  = xs_dup("[]");
@@ -2459,7 +2460,7 @@ int mastoapi_get_handler(const xs_dict *req, const char *q_path,
         if (logged_in) {
             const xs_list *timeline = xs_dict_get(args, "timeline[]");
             xs_str *json = NULL;
-            if (!xs_is_null(timeline)) 
+            if (!xs_is_null(timeline))
                 json = xs_json_dumps(markers_get(&snac1, timeline), 4);
 
             if (!xs_is_null(json))
@@ -3227,7 +3228,7 @@ int mastoapi_post_handler(const xs_dict *req, const char *q_path,
                 if (!xs_is_null(home))
                     home_marker = xs_dict_get(home, "last_read_id");
             }
-            
+
             const xs_str *notify_marker = xs_dict_get(args, "notifications[last_read_id]");
             if (xs_is_null(notify_marker)) {
                 const xs_dict *notify = xs_dict_get(args, "notifications");