* make the hyperlinkify regexp code a bit faster, also don't try

  to hyperlinkify really big parts, because it will take very, very
  long even in sbox. limit is 100KB now.

pmo-trunk-r2822

diff --git a/src/modest-text-utils.c b/src/modest-text-utils.c
index 3c32ada..6122f69 100644 (file)
--- a/src/modest-text-utils.c
+++ b/src/modest-text-utils.c
@@ -51,6 +51,14 @@
 #define EMPTY_STRING ""
 
 /*
+ * do the hyperlinkification only for texts < 50 Kb,
+ * as it's quite slow. Without this, e.g. mail with
+ * an uuencoded part (which is not recognized as attachment,
+ * will hang modest
+ */
+#define HYPERLINKIFY_MAX_LENGTH (1024*50)
+
+/*
  * we need these regexps to find URLs in plain text e-mails
  */
 typedef struct _url_match_pattern_t url_match_pattern_t;
@@ -86,7 +94,6 @@ struct _url_match_t {
 static gchar*   cite                    (const time_t sent_date, const gchar *from);
 static void     hyperlinkify_plain_text (GString *txt);
 static gint     cmp_offsets_reverse     (const url_match_t *match1, const url_match_t *match2);
-static void     chk_partial_match       (const url_match_t *match, guint* offset);
 static GSList*  get_url_matches         (GString *txt);
 
 static GString* get_next_line           (const char *b, const gsize blen, const gchar * iter);
@@ -366,7 +373,9 @@ modest_text_utils_convert_to_html (const gchar *data)
        modest_text_utils_convert_buffer_to_html (html, data);
        
        g_string_append (html, "</body></html>");
-       hyperlinkify_plain_text (html);
+
+       if (len <= HYPERLINKIFY_MAX_LENGTH)
+               hyperlinkify_plain_text (html);
 
        return g_string_free (html, FALSE);
 }
@@ -385,7 +394,8 @@ modest_text_utils_convert_to_html_body (const gchar *data)
 
        modest_text_utils_convert_buffer_to_html (html, data);
 
-       hyperlinkify_plain_text (html);
+       if (len < HYPERLINKIFY_MAX_LENGTH)
+               hyperlinkify_plain_text (html);
 
        return g_string_free (html, FALSE);
 }
@@ -814,16 +824,6 @@ cmp_offsets_reverse (const url_match_t *match1, const url_match_t *match2)
 }
 
 
-
-/*
- * check if the match is inside an existing match... */
-static void
-chk_partial_match (const url_match_t *match, guint* offset)
-{
-       if (*offset >= match->offset && *offset < match->offset + match->len)
-               *offset = -1;
-}
-
 static GSList*
 get_url_matches (GString *txt)
 {
@@ -846,28 +846,39 @@ get_url_matches (GString *txt)
        for (i = 0; i != pattern_num; ++i) {
                offset     = 0; 
                while (1) {
-                       int test_offset;
+                       url_match_t *match;
+                       gboolean is_submatch;
+                       GSList *cursor;
+                       
                        if ((rv = regexec (patterns[i].preg, txt->str + offset, 1, &rm, 0)) != 0) {
                                g_return_val_if_fail (rv == REG_NOMATCH, NULL); /* this should not happen */
                                break; /* try next regexp */ 
                        }
                        if (rm.rm_so == -1)
                                break;
-
-                       /* FIXME: optimize this */
-                       /* to avoid partial matches on something that was already found... */
-                       /* check_partial_match will put -1 in the data ptr if that is the case */
-                       test_offset = offset + rm.rm_so;
-                       g_slist_foreach (match_list, (GFunc)chk_partial_match, &test_offset);
                        
-                       /* make a list of our matches (<offset, len, prefix> tupels)*/
-                       if (test_offset != -1) {
-                               url_match_t *match = g_slice_new (url_match_t);
+                       is_submatch = FALSE;
+                       /* check  old matches to see if this has already been matched */
+                       cursor = match_list;
+                       while (cursor && !is_submatch) {
+                               const url_match_t *old_match =
+                                       (const url_match_t *) cursor->data;
+                               guint new_offset = offset + rm.rm_so;
+                               is_submatch = (new_offset >  old_match->offset &&
+                                              new_offset <  old_match->offset + old_match->len);
+                               cursor = g_slist_next (cursor);
+                       }
+
+                       if (!is_submatch) {
+                               /* make a list of our matches (<offset, len, prefix> tupels)*/
+                               match = g_slice_new (url_match_t);
                                match->offset = offset + rm.rm_so;
                                match->len    = rm.rm_eo - rm.rm_so;
                                match->prefix = patterns[i].prefix;
+                               g_warning ("<%d, %d, %s>",  match->offset, match->len, match->prefix);
                                match_list = g_slist_prepend (match_list, match);
                        }
+                               
                        offset += rm.rm_eo;
                }
        }