[empathy: 1/10] Use /\b<nickname>\b/ to decide whether to highlight

[Will Thompson <wjt src gnome org>]

commit aded85c91fb409bafb06492ddb85aa1224178d4c
Author: Will Thompson <will willthompson co uk>
Date:   Wed Jan 18 15:41:07 2012 +0000

    Use /\b<nickname>\b/ to decide whether to highlight
    
    The current highlighting code finds the first occurrence of the
    nickname, then checks whether the characters before or after are a
    space, a comma, a colon or a full stop (or the start or end of the
    string). This means that someone saying âno! That's wjtâs coffee!â
    didnât highlight me, because the apostrophe isn't in the whitelist. It
    also means that saying âborrow some Sudafed from dafâ would not
    highlight daf, since the first match is in the middle of a word.
    
    Weâre trying to check whether the nickname occurs as a complete word
    within the message. The regex metacharacter \b matches word boundaries,
    so /\b<nickname>\b/ is what we actually want. It gets the above cases
    right, including Unicode punctuation.
    
    https://bugzilla.gnome.org/show_bug.cgi?id=591667

 libempathy/empathy-message.c |   67 +++++++++++++++++++++--------------------
 1 files changed, 34 insertions(+), 33 deletions(-)
---
diff --git a/libempathy/empathy-message.c b/libempathy/empathy-message.c
index 995f49a..9841cbb 100644
--- a/libempathy/empathy-message.c
+++ b/libempathy/empathy-message.c
@@ -39,6 +39,9 @@
 # include <telepathy-logger/call-event.h>
 #endif
 
+#define DEBUG_FLAG EMPATHY_DEBUG_CHAT
+#include "empathy-debug.h"
+
 #include "empathy-client-factory.h"
 #include "empathy-message.h"
 #include "empathy-utils.h"
@@ -633,21 +636,42 @@ empathy_message_is_backlog (EmpathyMessage *message)
 	return priv->is_backlog;
 }
 
-#define IS_SEPARATOR(ch) (ch == ' ' || ch == ',' || ch == '.' || ch == ':')
+static GRegex *
+get_highlight_regex_for (const gchar *name)
+{
+	GRegex *regex;
+	gchar *name_esc, *pattern;
+	GError *error = NULL;
+
+	name_esc = g_regex_escape_string (name, -1);
+	pattern = g_strdup_printf ("\\b%s\\b", name_esc);
+	regex = g_regex_new (pattern, G_REGEX_CASELESS | G_REGEX_OPTIMIZE, 0,
+		&error);
+
+	if (regex == NULL) {
+		DEBUG ("couldn't compile regex /%s/: %s", pattern,
+			error->message);
+
+		g_error_free (error);
+	}
+
+	g_free (pattern);
+	g_free (name_esc);
+
+	return regex;
+}
+
 gboolean
 empathy_message_should_highlight (EmpathyMessage *message)
 {
 	EmpathyContact *contact;
 	const gchar   *msg, *to;
-	gchar         *cf_msg, *cf_to;
-	gchar         *ch;
-	gboolean       ret_val;
+	gboolean       ret_val = FALSE;
 	TpChannelTextMessageFlags flags;
+	GRegex *regex;
 
 	g_return_val_if_fail (EMPATHY_IS_MESSAGE (message), FALSE);
 
-	ret_val = FALSE;
-
 	msg = empathy_message_get_body (message);
 	if (!msg) {
 		return FALSE;
@@ -670,34 +694,11 @@ empathy_message_should_highlight (EmpathyMessage *message)
 		return FALSE;
 	}
 
-	cf_msg = g_utf8_casefold (msg, -1);
-	cf_to = g_utf8_casefold (to, -1);
-
-	ch = strstr (cf_msg, cf_to);
-	if (ch == NULL) {
-		goto finished;
+	regex = get_highlight_regex_for (to);
+	if (regex != NULL) {
+		ret_val = g_regex_match (regex, msg, 0, NULL);
+		g_regex_unref (regex);
 	}
-	if (ch != cf_msg) {
-		/* Not first in the message */
-		if (!IS_SEPARATOR (*(ch - 1))) {
-			goto finished;
-		}
-	}
-
-	ch = ch + strlen (cf_to);
-	if (ch >= cf_msg + strlen (cf_msg)) {
-		ret_val = TRUE;
-		goto finished;
-	}
-
-	if (IS_SEPARATOR (*ch)) {
-		ret_val = TRUE;
-		goto finished;
-	}
-
-finished:
-	g_free (cf_msg);
-	g_free (cf_to);
 
 	return ret_val;
 }