Take t.co URL lengthe^Wshortening into account when counting message length

before posting a Twitter message. Disable this functionality by default for
identi.ca accounts. Hardcode post-t.co URL length to 20 since it's probably
not going to change any time soon. Bug #855. Patch by Artem Savkov, thanks!

2 files changed, 50 insertions, 1 deletions

diff --git a/doc/user-guide/commands.xml b/doc/user-guide/commands.xml
index 81eb7452..95a127cf 100644
--- a/doc/user-guide/commands.xml
+++ b/doc/user-guide/commands.xml
@@ -1031,6 +1031,21 @@
 		</description>
 
 	</bitlbee-setting>
+	
+	<bitlbee-setting name="target_url_length" type="integer" scope="account">
+		<default>20</default>
+
+		<description>
+			<para>
+				Twitter replaces every URL with fixed-length t.co URLs. BitlBee is able to take t.co urls into account when calculating <emphasis>message_length</emphasis> replacing the actual URL length with target_url_length. Setting target_url_length to 0 disables this feature.
+			</para>
+
+			<para>
+				This setting is disabled for identica accounts by default and will not affect anything other than message safety checks (i.e. Twitter will still replace your URLs with t.co links, even if that makes them longer).
+			</para>
+		</description>
+
+	</bitlbee-setting>
 
 	<bitlbee-setting name="mode" type="string" scope="account">
 		<possible-values>one, many, chat</possible-values>
diff --git a/protocols/twitter/twitter.c b/protocols/twitter/twitter.c
index 91b6e28c..c389677f 100644
--- a/protocols/twitter/twitter.c
+++ b/protocols/twitter/twitter.c
@@ -196,11 +196,40 @@ static char *set_eval_mode(set_t * set, char *value)
 		return NULL;
 }
 
+int twitter_url_len_diff(gchar *msg, unsigned int target_len)
+{
+	int url_len_diff = 0;
+
+	static GRegex *regex = NULL;
+	GMatchInfo *match_info;
+
+	if (regex == NULL)
+		regex = g_regex_new("(^|\\s)(http(s)?://[^\\s$]+)", 0, 0, NULL);
+	
+	g_regex_match(regex, msg, 0, &match_info);
+	while (g_match_info_matches(match_info)) {
+		gchar *url = g_match_info_fetch(match_info, 2);
+		url_len_diff += target_len - g_utf8_strlen(url, -1);
+		if (g_match_info_fetch(match_info, 3) != NULL)
+			url_len_diff += 1;
+		g_free(url);
+		g_match_info_next(match_info, NULL);
+	}
+	g_match_info_free(match_info);
+
+	return url_len_diff;
+}
+
 static gboolean twitter_length_check(struct im_connection *ic, gchar * msg)
 {
 	int max = set_getint(&ic->acc->set, "message_length"), len;
+	int target_len = set_getint(&ic->acc->set, "target_url_length");
+	int url_len_diff = 0;
+    
+	if (target_len > 0)
+		url_len_diff = twitter_url_len_diff(msg, target_len);
 
-	if (max == 0 || (len = g_utf8_strlen(msg, -1)) <= max)
+	if (max == 0 || (len = g_utf8_strlen(msg, -1) + url_len_diff) <= max)
 		return TRUE;
 
 	imcb_error(ic, "Maximum message length exceeded: %d > %d", len, max);
@@ -213,13 +242,16 @@ static void twitter_init(account_t * acc)
 	set_t *s;
 	char *def_url;
 	char *def_oauth;
+	char *def_tul;
 
 	if (strcmp(acc->prpl->name, "twitter") == 0) {
 		def_url = TWITTER_API_URL;
 		def_oauth = "true";
+		def_tul = "20";
 	} else {		/* if( strcmp( acc->prpl->name, "identica" ) == 0 ) */
 		def_url = IDENTICA_API_URL;
 		def_oauth = "true";
+		def_tul = "0";
 	}
 
 	s = set_add(&acc->set, "auto_reply_timeout", "10800", set_eval_int, acc);
@@ -236,6 +268,8 @@ static void twitter_init(account_t * acc)
 
 	s = set_add(&acc->set, "message_length", "140", set_eval_int, acc);
 
+	s = set_add(&acc->set, "target_url_length", def_tul, set_eval_int, acc);
+
 	s = set_add(&acc->set, "mode", "chat", set_eval_mode, acc);
 	s->flags |= ACC_SET_OFFLINE_ONLY;