/soc/2015/igor.gajowiak/chatlog: 48b7ccc5444a: Fixed parsing mes...

[Igor Gajowiak]

Changeset: 48b7ccc5444ac72c1539676158fdf1453ba8322b
Author:	 Igor Gajowiak <igor.gajowiak at gmail.com>
Date:	 2015-08-06 20:27 +0200
Branch:	 default
URL: https://hg.pidgin.im/soc/2015/igor.gajowiak/chatlog/rev/48b7ccc5444a

Description:

Fixed parsing message timestamp from the legacy log.

diffstat:

 libpurple/plugins/log/legacylog.c |  108 +++++++++++++++++++++++++------------
 1 files changed, 72 insertions(+), 36 deletions(-)

diffs (181 lines):

diff --git a/libpurple/plugins/log/legacylog.c b/libpurple/plugins/log/legacylog.c
--- a/libpurple/plugins/log/legacylog.c
+++ b/libpurple/plugins/log/legacylog.c
@@ -37,6 +37,10 @@
 /* PurpleGenericLog API implementation                                    */
 /**************************************************************************/
 
+static GRegex *msg_regex = NULL;
+static GRegex *date_regex = NULL;
+static GRegex *time_regex = NULL;
+
 typedef enum
 {
 	DIR_DEPTH_PROTOCOL  = 1,
@@ -302,37 +306,54 @@ normalize_protocol_name(const gchar *pro
 		return protocol;
 }
 
-/**
- * TODO: Use regex
- *
- * date "YYYY-MM-DD.hhmmss"
- * time "HH:MM:SS (A|P)M"
- */
-static guint64
-make_timestamp(const gchar *date, const gchar *msg_time)
+static gboolean
+parse_timestamp(const gchar *date, const gchar *time, guint64 *result)
 {
-	int year = atoi(date);
-	date += 5;
-	int month = atoi(date);
-	date += 3;
-	int day = atoi(date);
-	date += 3;
+	g_assert(date);
+	g_assert(time);
+	g_assert(result);
 
-	int hour = 0;
-	for(int i = 0; i < 2; ++i, ++msg_time)
-		hour = hour * 10 + (*msg_time - '0');
-	++msg_time;
-	int min = 0;
-	for(int i = 0; i < 2; ++i, ++msg_time)
-		min = min * 10 + (*msg_time - '0');
-	++msg_time;
-	int sec = 0;
-	for(int i = 0; i < 2; ++i, ++msg_time)
-		sec = sec * 10 + (*msg_time - '0');
-	if(*(++msg_time) == 'P')
+	GMatchInfo *info = NULL;
+	gint begin, end;
+
+	if (!g_regex_match(date_regex, date, 0, &info)) {
+		g_match_info_free(info);
+		return FALSE;
+	}
+
+	g_match_info_fetch_pos(info, 1, &begin, &end);
+	guint64 year = g_ascii_strtoull(&date[begin], NULL, 10);
+
+	g_match_info_fetch_pos(info, 2, &begin, &end);
+	guint64 month = g_ascii_strtoull(&date[begin], NULL, 10);
+
+	g_match_info_fetch_pos(info, 3, &begin, &end);
+	guint64 day = g_ascii_strtoull(&date[begin], NULL, 10);
+
+	g_match_info_free(info);
+
+	if (!g_regex_match(time_regex, time, 0, &info)) {
+		g_match_info_free(info);
+		return FALSE;
+	}
+
+	g_match_info_fetch_pos(info, 1, &begin, &end);
+	guint64 hour = g_ascii_strtoull(&time[begin], NULL, 10);
+
+	g_match_info_fetch_pos(info, 2, &begin, &end);
+	guint64 min = g_ascii_strtoull(&time[begin], NULL, 10);
+
+	g_match_info_fetch_pos(info, 3, &begin, &end);
+	guint64 sec = g_ascii_strtoull(&time[begin], NULL, 10);
+
+	g_match_info_fetch_pos(info, 4, &begin, &end);
+	if (g_strcmp0(&time[begin], "PM") == 0)
 		hour += 12;
+	
+	g_match_info_free(info);
 
-	return purple_time_build(year, month, day, hour, min, sec);
+	*result = purple_time_build(year, month, day, hour, min, sec);
+	return TRUE;
 }
 
 static PurpleAccount*
@@ -353,6 +374,10 @@ make_message(PurpleAccount *account, con
 	if (!buddy)
 		return NULL;
 
+	guint64 timestamp;
+	if(!parse_timestamp(iter->log_file_name, msg_time, &timestamp))
+		return NULL;
+
 	PurpleMessage *msg;
 	if (strcmp(color, "#A82F2F") == 0) {
 		msg = purple_message_new_incoming(purple_buddy_get_name(buddy),
@@ -363,7 +388,6 @@ make_message(PurpleAccount *account, con
 			contents, PURPLE_MESSAGE_SEND);
 	}
 
-	guint64 timestamp = make_timestamp(iter->log_file_name, msg_time);
 	purple_message_set_time(msg, timestamp);
 
 	return msg;
@@ -377,7 +401,6 @@ remove_divs(gchar *text)
 
 	for (size_t i = 0; i < len && text[i] != '\0';) {
 		if (len - i >= 5 && g_str_has_prefix(&text[i], "<div>")) {
-			g_string_append_c(result, '\n');
 			i += 5;
 		}
 		else if (len - i >= 6 && g_str_has_prefix(&text[i], "</div>")) {
@@ -396,13 +419,8 @@ remove_divs(gchar *text)
 static int
 fill_arrays(gchar *text, GArray *accounts, GArray *messages, const Iter *iter)
 {
-	GRegex *regex = g_regex_new(
-		"<font color=\"(.*?)\"><font size=.*?>\\((.*?)\\)</font> .*?"
-		"</b></font>(.*?)<br/>",
-		0, 0, NULL);
-
 	GMatchInfo *info;
-	g_regex_match(regex, text, 0, &info);
+	g_regex_match(msg_regex, text, 0, &info);
 
 	int msgs_count = 0;
 	while (g_match_info_matches(info)) {
@@ -435,7 +453,6 @@ fill_arrays(gchar *text, GArray *account
     }
 
 	g_match_info_free(info);
-	g_regex_unref(regex);
 
 	return msgs_count;
 }
@@ -630,15 +647,34 @@ plugin_load(PurplePlugin *plugin, GError
 		PURPLE_GENERICLOG_PROPERTY_NAME, LEGACYLOG_NAME,
 		NULL);
 	purple_genericlog_register(PURPLE_GENERICLOG(legacylog_instance));
+
+	msg_regex = g_regex_new(
+		"<font color=\"(.*?)\"><font size=.*?>\\((.*?)\\)</font> .*?"
+		"</b></font>(.*?)<br/>",
+		0, 0, NULL);
+	g_assert(msg_regex);
+
+	date_regex = g_regex_new("^([\\d]{4})-([\\d]{2})-([\\d]{2})\\.",0, 0, NULL);
+	g_assert(date_regex);
+
+	time_regex = g_regex_new("^([\\d]{2}):([\\d]{2}):([\\d]{2}) (AM|PM)",
+		0, 0, NULL);
+	g_assert(time_regex);
+
 	return TRUE;
 }
 
 static gboolean
 plugin_unload(PurplePlugin *plugin, GError **error)
 {
+	g_regex_unref(msg_regex);
+	g_regex_unref(date_regex);
+	g_regex_unref(time_regex);
+
 	purple_genericlog_unregister(PURPLE_GENERICLOG(legacylog_instance));
 	g_object_unref(legacylog_instance);
 	legacylog_instance = NULL;
+
 	return TRUE;
 }