Statusnet: New intelligent shortening enabled.

statusnet/statusnet.php

@@ -404,6 +404,107 @@ function short_link($url) {
     return $slinky->short();
 } };
 
+function statusnet_shortenmsg($b, $max_char) {
+	require_once("include/bbcode.php");
+	require_once("include/html2plain.php");
+
+	// Looking for the first image
+	$image = '';
+	if(preg_match("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/is",$b['body'],$matches))
+		$image = $matches[3];
+
+	if ($image == '')
+		if(preg_match("/\[img\](.*?)\[\/img\]/is",$b['body'],$matches))
+			$image = $matches[1];
+
+	$multipleimages = (strpos($b['body'], "[img") != strrpos($b['body'], "[img"));
+
+	// When saved into the database the content is sent through htmlspecialchars
+	// That means that we have to decode all image-urls
+	$image = htmlspecialchars_decode($image);
+
+	$body = $b["body"];
+	if ($b["title"] != "")
+		$body = $b["title"]."\n\n".$body;
+
+	// remove the recycle signs and the names since they aren't helpful on twitter
+	// recycle 1
+	$recycle = html_entity_decode("♲ ", ENT_QUOTES, 'UTF-8');
+	$body = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', "\n", $body);
+	// recycle 2 (Test)
+	$recycle = html_entity_decode("◌ ", ENT_QUOTES, 'UTF-8');
+	$body = preg_replace( '/'.$recycle.'\[url\=(\w+.*?)\](\w+.*?)\[\/url\]/i', "\n", $body);
+
+	// At first convert the text to html
+	$html = bbcode($body, false, false);
+
+	// Then convert it to plain text
+	//$msg = trim($b['title']." \n\n".html2plain($html, 0, true));
+	$msg = trim(html2plain($html, 0, true));
+	$msg = html_entity_decode($msg,ENT_QUOTES,'UTF-8');
+
+	// Removing multiple newlines
+	while (strpos($msg, "\n\n\n") !== false)
+		$msg = str_replace("\n\n\n", "\n\n", $msg);
+
+	// Removing multiple spaces
+	while (strpos($msg, "  ") !== false)
+		$msg = str_replace("  ", " ", $msg);
+
+	$msg = trim($msg);
+
+	$link = '';
+	// look for bookmark-bbcode and handle it with priority
+	if(preg_match("/\[bookmark\=([^\]]*)\](.*?)\[\/bookmark\]/is",$b['body'],$matches))
+		$link = $matches[1];
+
+	$multiplelinks = (strpos($b['body'], "[bookmark") != strrpos($b['body'], "[bookmark"));
+
+	// If there is no bookmark element then take the first link
+	if ($link == '') {
+		$links = collecturls($html);
+		if (sizeof($links) > 0) {
+			reset($links);
+			$link = current($links);
+		}
+		$multiplelinks = (sizeof($links) > 1);
+	}
+
+	$msglink = "";
+	if ($multiplelinks)
+		$msglink = $b["plink"];
+	else if ($link != "")
+		$msglink = $link;
+	else if ($multipleimages)
+		$msglink = $b["plink"];
+	else if ($image != "")
+		$msglink = $image;
+
+	if (($msglink == "") and strlen($msg) > $max_char)
+		$msglink = $b["plink"];
+
+	if (strlen($msglink) > 20)
+		$msglink = short_link($msglink);
+
+	if (strlen(trim($msg." ".$msglink)) > $max_char) {
+		$msg = substr($msg, 0, $max_char - (strlen($msglink)));
+		$lastchar = substr($msg, -1);
+		$msg = substr($msg, 0, -1);
+		$pos = strrpos($msg, "\n");
+		if ($pos > 0)
+			$msg = substr($msg, 0, $pos-1);
+		else if ($lastchar != "\n")
+			$msg = substr($msg, 0, -3)."...";
+	}
+	$msg = str_replace("\n", " ", $msg);
+
+	// Removing multiple spaces - again
+	while (strpos($msg, "  ") !== false)
+		$msg = str_replace("  ", " ", $msg);
+
+	return(trim($msg." ".$msglink));
+}
+
 function statusnet_post_hook(&$a,&$b) {
 
 	/**
@@ -433,9 +534,12 @@ function statusnet_post_hook(&$a,&$b) {
                 // we can later send to StatusNet. This way we can "gain" some
                 // information during shortening of potential links but do not
                 // shorten all the links in a 200000 character long essay.
+
+		$intelligent_shortening = get_config('statusnet','intelligent_shortening');
+		if (!$intelligent_shortening) {
 	                if (! $b['title']=='') {
 				$tmp = $b['title'].": \n".$b['body'];
-//                    $tmp = substr($tmp, 0, 4*$max_char);
+	//                    $tmp = substr($tmp, 0, 4*$max_char);
         	        } else {
                 	    $tmp = $b['body']; // substr($b['body'], 0, 3*$max_char);
 	                }
@@ -505,6 +609,8 @@ function statusnet_post_hook(&$a,&$b) {
 			}
 
 			$msg = trim($msg);
+		} else
+			$msg = statusnet_shortenmsg($b, $max_char);
 
 		// and now dent it :-)
 		if(strlen($msg)) {

twitter/twitter.php

@@ -282,10 +282,9 @@ function twitter_shortenmsg($b) {
 	// That means that we have to decode all image-urls
 	$image = htmlspecialchars_decode($image);
 
-	if ($b["title"] == "")
 	$body = $b["body"];
-	else
+	if ($b["title"] != "")
-		$body = $b["title"];
+		$body = $b["title"]."\n\n".$body;
 
 	// remove the recycle signs and the names since they aren't helpful on twitter
 	// recycle 1