Use GRegex instead of custom code and use a new regex to detect URIs

svn path=/trunk/; revision=1932

3 files changed, 53 insertions, 196 deletions

diff --git a/libempathy-gtk/empathy-theme.c b/libempathy-gtk/empathy-theme.c
index e2da4e494..d68b72f02 100644
--- a/libempathy-gtk/empathy-theme.c
+++ b/libempathy-gtk/empathy-theme.c
@@ -34,7 +34,17 @@
 /* Number of seconds between timestamps when using normal mode, 5 minutes. */
 #define TIMESTAMP_INTERVAL 300
 
+#define SHEMES "(https?|ftps?|nntp|news|javascript|about|ghelp|apt|telnet|"\
+	       "file|webcal|mailto)"
+#define SEPARATOR "([^,;\?><()\\ ])"
+#define BODY "([^\\ ]*(\\\\ )?)+"
+#define URI_REGEX "("SHEMES"://"BODY SEPARATOR")" \
+		  "|((mailto:)?"BODY"@"BODY"."BODY SEPARATOR")"\
+		  "|((www|ftp)."BODY SEPARATOR")"
+static GRegex *uri_regex = NULL;
+
 #define GET_PRIV(obj) EMPATHY_GET_PRIV (obj, EmpathyTheme)
+
 typedef struct {
 	EmpathySmileyManager *smiley_manager;
 	gboolean show_avatars;
@@ -254,8 +264,11 @@ empathy_theme_append_text (EmpathyTheme        *theme,
 	GtkTextIter      start_iter, end_iter;
 	GtkTextMark     *mark;
 	GtkTextIter      iter;
-	gint             num_matches, i;
-	GArray          *start, *end;
+	GMatchInfo      *match_info;
+	gboolean         match;
+	gint             last = 0;
+	gint             s = 0, e = 0;
+	gchar           *tmp;
 
 	priv = GET_PRIV (theme);
 	buffer = gtk_text_view_get_buffer (GTK_TEXT_VIEW (view));
@@ -263,57 +276,17 @@ empathy_theme_append_text (EmpathyTheme        *theme,
 	gtk_text_buffer_get_end_iter (buffer, &start_iter);
 	mark = gtk_text_buffer_create_mark (buffer, NULL, &start_iter, TRUE);
 
-	start = g_array_new (FALSE, FALSE, sizeof (gint));
-	end = g_array_new (FALSE, FALSE, sizeof (gint));
-
-	num_matches = empathy_regex_match (EMPATHY_REGEX_ALL, body, start, end);
-
-	if (num_matches == 0) {
-		gtk_text_buffer_get_end_iter (buffer, &iter);
-		theme_insert_text_with_emoticons (buffer, &iter, body, priv->smiley_manager);
-	} else {
-		gint   last = 0;
-		gint   s = 0, e = 0;
-		gchar *tmp;
-
-		for (i = 0; i < num_matches; i++) {
-			s = g_array_index (start, gint, i);
-			e = g_array_index (end, gint, i);
-
-			if (s > last) {
-				tmp = empathy_substring (body, last, s);
-
-				gtk_text_buffer_get_end_iter (buffer, &iter);
-				theme_insert_text_with_emoticons (buffer,
-								  &iter,
-								  tmp,
-								  priv->smiley_manager);
-				g_free (tmp);
-			}
-
-			tmp = empathy_substring (body, s, e);
-
-			gtk_text_buffer_get_end_iter (buffer, &iter);
-			if (!link_tag) {
-				gtk_text_buffer_insert (buffer, &iter,
-							tmp, -1);
-			} {
-				gtk_text_buffer_insert_with_tags_by_name (buffer,
-									  &iter,
-									  tmp,
-									  -1,
-									  link_tag,
-									  "link",
-									  NULL);
-			}
-
-			g_free (tmp);
+	if (!uri_regex) {
+		uri_regex = g_regex_new (URI_REGEX, 0, 0, NULL);
+	}
 
-			last = e;
-		}
+	for (match = g_regex_match (uri_regex, body, 0, &match_info); match;
+	     match = g_match_info_next (match_info, NULL)) {
+		if (!g_match_info_fetch_pos (match_info, 0, &s, &e))
+			continue;
 
-		if (e < strlen (body)) {
-			tmp = empathy_substring (body, e, strlen (body));
+		if (s > last) {
+			tmp = empathy_substring (body, last, s);
 
 			gtk_text_buffer_get_end_iter (buffer, &iter);
 			theme_insert_text_with_emoticons (buffer,
@@ -322,10 +295,35 @@ empathy_theme_append_text (EmpathyTheme        *theme,
 							  priv->smiley_manager);
 			g_free (tmp);
 		}
+
+		tmp = empathy_substring (body, s, e);
+
+		gtk_text_buffer_get_end_iter (buffer, &iter);
+		if (!link_tag) {
+			gtk_text_buffer_insert (buffer, &iter,
+						tmp, -1);
+		} else {
+			gtk_text_buffer_insert_with_tags_by_name (buffer,
+								  &iter,
+								  tmp,
+								  -1,
+								  link_tag,
+								  "link",
+								  NULL);
+		}
+
+		g_free (tmp);
+		last = e;
 	}
+	g_match_info_free (match_info);
 
-	g_array_free (start, TRUE);
-	g_array_free (end, TRUE);
+	if (last < strlen (body)) {
+		gtk_text_buffer_get_end_iter (buffer, &iter);
+		theme_insert_text_with_emoticons (buffer,
+						  &iter,
+						  body + last,
+						  priv->smiley_manager);
+	}
 
 	gtk_text_buffer_get_end_iter (buffer, &iter);
 	gtk_text_buffer_insert (buffer, &iter, "\n", 1);
diff --git a/libempathy/empathy-utils.c b/libempathy/empathy-utils.c
index b5bdb9ff7..671595e68 100644
--- a/libempathy/empathy-utils.c
+++ b/libempathy/empathy-utils.c
@@ -28,7 +28,6 @@
 #include <string.h>
 #include <time.h>
 #include <sys/types.h>
-#include <regex.h>
 
 #include <glib/gi18n.h>
 
@@ -44,8 +43,6 @@
 #define DEBUG_FLAG EMPATHY_DEBUG_OTHER
 #include "empathy-debug.h"
 
-static void regex_init (void);
-
 gchar *
 empathy_substring (const gchar *str,
 		  gint         start,
@@ -54,129 +51,6 @@ empathy_substring (const gchar *str,
 	return g_strndup (str + start, end - start);
 }
 
-/*
- * Regular Expression code to match urls.
- */
-#define APTCHARS  "-A-Za-z0-9,-."
-#define USERCHARS "-A-Za-z0-9"
-#define PASSCHARS "-A-Za-z0-9,?;.:/!%$^*&~\"#'"
-#define HOSTCHARS "-A-Za-z0-9_"
-#define PATHCHARS "-A-Za-z0-9_$.+!*(),;:@&=?/~#%"
-#define SCHEME    "(news:|telnet:|nntp:|file:/|https?:|ftps?:|webcal:)"
-#define USER      "[" USERCHARS "]+(:["PASSCHARS "]+)?"
-#define URLPATH   "/[" PATHCHARS "]*[^]'.}>) \t\r\n,\\\"]"
-
-static regex_t dingus[EMPATHY_REGEX_ALL];
-
-static void
-regex_init (void)
-{
-	static gboolean  inited = FALSE;
-	const gchar     *expression;
-	gint             i;
-
-	if (inited) {
-		return;
-	}
-
-	for (i = 0; i < EMPATHY_REGEX_ALL; i++) {
-		switch (i) {
-		case EMPATHY_REGEX_AS_IS:
-			expression =
-				SCHEME "//(" USER "@)?[" HOSTCHARS ".]+"
-				"(:[0-9]+)?(" URLPATH ")?";
-			break;
-		case EMPATHY_REGEX_BROWSER:
-			expression =
-				"(www|ftp)[" HOSTCHARS "]*\\.[" HOSTCHARS ".]+"
-				"(:[0-9]+)?(" URLPATH ")?";
-			break;
-		case EMPATHY_REGEX_APT:
-			expression =
-				"apt://[" APTCHARS "]*";
-			break;
-		case EMPATHY_REGEX_EMAIL:
-			expression =
-				"(mailto:)?[a-z0-9][a-z0-9._-]*@[a-z0-9]"
-				"[a-z0-9-]*(\\.[a-z0-9][a-z0-9-]*)+";
-			break;
-		case EMPATHY_REGEX_OTHER:
-			expression =
-				"news:[-A-Z\\^_a-z{|}~!\"#$%&'()*+,./0-9;:=?`]+"
-				"@[" HOSTCHARS ".]+(:[0-9]+)?";
-			break;
-		default:
-			/* Silence the compiler. */
-			expression = NULL;
-			continue;
-		}
-
-		memset (&dingus[i], 0, sizeof (regex_t));
-		regcomp (&dingus[i], expression, REG_EXTENDED | REG_ICASE);
-	}
-
-	inited = TRUE;
-}
-
-gint
-empathy_regex_match (EmpathyRegExType  type,
-		    const gchar     *msg,
-		    GArray          *start,
-		    GArray          *end)
-{
-	regmatch_t matches[1];
-	gint       ret = 0;
-	gint       num_matches = 0;
-	gint       offset = 0;
-	gint       i;
-
-	g_return_val_if_fail (type >= 0 || type <= EMPATHY_REGEX_ALL, 0);
-
-	regex_init ();
-
-	while (!ret && type != EMPATHY_REGEX_ALL) {
-		ret = regexec (&dingus[type], msg + offset, 1, matches, 0);
-		if (ret == 0) {
-			gint s;
-
-			num_matches++;
-
-			s = matches[0].rm_so + offset;
-			offset = matches[0].rm_eo + offset;
-
-			g_array_append_val (start, s);
-			g_array_append_val (end, offset);
-		}
-	}
-
-	if (type != EMPATHY_REGEX_ALL) {
-		DEBUG ("Found %d matches for regex type:%d", num_matches, type);
-		return num_matches;
-	}
-
-	/* If EMPATHY_REGEX_ALL then we run ALL regex's on the string. */
-	for (i = 0; i < EMPATHY_REGEX_ALL; i++, ret = 0) {
-		while (!ret) {
-			ret = regexec (&dingus[i], msg + offset, 1, matches, 0);
-			if (ret == 0) {
-				gint s;
-
-				num_matches++;
-
-				s = matches[0].rm_so + offset;
-				offset = matches[0].rm_eo + offset;
-
-				g_array_append_val (start, s);
-				g_array_append_val (end, offset);
-			}
-		}
-	}
-
-	DEBUG ("Found %d matches for ALL regex types", num_matches);
-
-	return num_matches;
-}
-
 gint
 empathy_strcasecmp (const gchar *s1,
 		   const gchar *s2)
diff --git a/libempathy/empathy-utils.h b/libempathy/empathy-utils.h
index a320c6246..51ddd231f 100644
--- a/libempathy/empathy-utils.h
+++ b/libempathy/empathy-utils.h
@@ -37,30 +37,15 @@
 
 #include "empathy-contact.h"
 
-G_BEGIN_DECLS
-
 #define EMPATHY_GET_PRIV(obj,type) ((type##Priv*) ((type*)obj)->priv)
 #define G_STR_EMPTY(x) ((x) == NULL || (x)[0] == '\0')
 
-typedef enum {
-	EMPATHY_REGEX_AS_IS,
-	EMPATHY_REGEX_BROWSER,
-	EMPATHY_REGEX_APT,
-	EMPATHY_REGEX_EMAIL,
-	EMPATHY_REGEX_OTHER,
-	EMPATHY_REGEX_ALL,
-} EmpathyRegExType;
+G_BEGIN_DECLS
 
-/* Regular expressions */
+/* Strings */
 gchar *      empathy_substring                      (const gchar     *str,
 						    gint             start,
 						    gint             end);
-gint         empathy_regex_match                    (EmpathyRegExType  type,
-						    const gchar     *msg,
-						    GArray          *start,
-						    GArray          *end);
-
-/* Strings */
 gint         empathy_strcasecmp                     (const gchar     *s1,
 						    const gchar     *s2);
 gint         empathy_strncasecmp                    (const gchar     *s1,