Improve diaspora2bb autolinker for URLs

[friendica.git] / include / bb2diaspora.php

diff --git a/include/bb2diaspora.php b/include/bb2diaspora.php
index 69a87b3817a37e9640a38d4b2eb302ddd072b346..cd5424216487fcf8c9059d4f8d3db9e8ee9a42c5 100644 (file)
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@@ -15,24 +15,24 @@ require_once("library/html-to-markdown/HTML_To_Markdown.php");
 
 function diaspora2bb($s) {
 
-       $s = html_entity_decode($s,ENT_COMPAT,'UTF-8');
+       $s = html_entity_decode($s, ENT_COMPAT, 'UTF-8');
 
        // Remove CR to avoid problems with following code
-       $s = str_replace("\r","",$s);
+       $s = str_replace("\r", '', $s);
 
-       $s = str_replace("\n"," \n",$s);
+       $s = str_replace("\n", " \n", $s);
 
        // The parser cannot handle paragraphs correctly
-       $s = str_replace(array("</p>", "<p>", '<p dir="ltr">'),array("<br>", "<br>", "<br>"),$s);
+       $s = str_replace(array('</p>', '<p>', '<p dir="ltr">'), array('<br>', '<br>', '<br>'), $s);
 
        // Escaping the hash tags
-       $s = preg_replace('/\#([^\s\#])/','&#35;$1',$s);
+       $s = preg_replace('/\#([^\s\#])/', '&#35;$1', $s);
 
        $s = Markdown($s);
 
-       $s = preg_replace('/\@\{(.+?)\; (.+?)\@(.+?)\}/','@[url=https://$3/u/$2]$1[/url]',$s);
+       $s = preg_replace('/\@\{(.+?)\; (.+?)\@(.+?)\}/', '@[url=https://$3/u/$2]$1[/url]', $s);
 
-       $s = str_replace('&#35;','#',$s);
+       $s = str_replace('&#35;', '#', $s);
 
        $search = array(" \n", "\n ");
        $replace = array("\n", "\n");
@@ -41,23 +41,23 @@ function diaspora2bb($s) {
                $s = str_replace($search, $replace, $s);
        } while ($oldtext != $s);
 
-       $s = str_replace("\n\n", "<br>", $s);
+       $s = str_replace("\n\n", '<br>', $s);
 
        $s = html2bbcode($s);
 
        // protect the recycle symbol from turning into a tag, but without unescaping angles and naked ampersands
-       $s = str_replace('&#x2672;',html_entity_decode('&#x2672;',ENT_QUOTES,'UTF-8'),$s);
+       $s = str_replace('&#x2672;', html_entity_decode('&#x2672;', ENT_QUOTES, 'UTF-8'), $s);
 
        // Convert everything that looks like a link to a link
-       $s = preg_replace("/([^\]\=]|^)(https?\:\/\/)([a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url=$2$3]$2$3[/url]',$s);
+       $s = preg_replace('/([^\]=]|^)(https?\:\/\/)([a-zA-Z0-9:\/\-?&;.=_~#%$!+,@]+(?<!,))/ism', '$1[url=$2$3]$2$3[/url]', $s);
 
        //$s = preg_replace("/([^\]\=]|^)(https?\:\/\/)(vimeo|youtu|www\.youtube|soundcloud)([a-zA-Z0-9\:\/\-\?\&\;\.\=\_\~\#\%\$\!\+\,]+)/ism", '$1[url=$2$3$4]$2$3$4[/url]',$s);
-       $s = bb_tag_preg_replace("/\[url\=?(.*?)\]https?:\/\/www.youtube.com\/watch\?v\=(.*?)\[\/url\]/ism",'[youtube]$2[/youtube]','url',$s);
-       $s = bb_tag_preg_replace("/\[url\=https?:\/\/www.youtube.com\/watch\?v\=(.*?)\].*?\[\/url\]/ism",'[youtube]$1[/youtube]','url',$s);
-       $s = bb_tag_preg_replace("/\[url\=?(.*?)\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/url\]/ism",'[vimeo]$2[/vimeo]','url',$s);
-       $s = bb_tag_preg_replace("/\[url\=https?:\/\/vimeo.com\/([0-9]+)\](.*?)\[\/url\]/ism",'[vimeo]$1[/vimeo]','url',$s);
+       $s = bb_tag_preg_replace('/\[url\=?(.*?)\]https?:\/\/www.youtube.com\/watch\?v\=(.*?)\[\/url\]/ism', '[youtube]$2[/youtube]', 'url',$s);
+       $s = bb_tag_preg_replace('/\[url\=https?:\/\/www.youtube.com\/watch\?v\=(.*?)\].*?\[\/url\]/ism', '[youtube]$1[/youtube]', 'url',$s);
+       $s = bb_tag_preg_replace('/\[url\=?(.*?)\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/url\]/ism', '[vimeo]$2[/vimeo]', 'url', $s);
+       $s = bb_tag_preg_replace('/\[url\=https?:\/\/vimeo.com\/([0-9]+)\](.*?)\[\/url\]/ism', '[vimeo]$1[/vimeo]', 'url', $s);
        // remove duplicate adjacent code tags
-       $s = preg_replace("/(\[code\])+(.*?)(\[\/code\])+/ism","[code]$2[/code]", $s);
+       $s = preg_replace('/(\[code\])+(.*?)(\[\/code\])+/ism', '[code]$2[/code]', $s);
 
        // Don't show link to full picture (until it is fixed)
        $s = scale_external_images($s, false);
@@ -83,7 +83,6 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
                'return \'#\'. str_replace(\' \', \'_\', $match[2]);'
        ), $Text);
 
-
        // Converting images with size parameters to simple images. Markdown doesn't know it.
        $Text = preg_replace("/\[img\=([0-9]*)x([0-9]*)\](.*?)\[\/img\]/ism", '[img]$3[/img]', $Text);
 
@@ -94,11 +93,12 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
                // Add all tags that maybe were removed
                if (preg_match_all("/#\[url\=([$URLSearchString]*)\](.*?)\[\/url\]/ism",$OriginalText, $tags)) {
                        $tagline = "";
-                       foreach($tags[2] as $tag)
-                               if (!strpos($Text, "#".$tag))
+                       foreach($tags[2] as $tag) {
+                               $tag = html_entity_decode($tag, ENT_QUOTES, 'UTF-8');
+                               if (!strpos(html_entity_decode($Text, ENT_QUOTES, 'UTF-8'), "#".$tag))
                                        $tagline .= "#".$tag." ";
-
-                       $Text = $Text."<br />".$tagline;
+                       }
+                       $Text = $Text." ".$tagline;
                }
 
        } else
@@ -152,22 +152,22 @@ function format_event_diaspora($ev) {
        $o .= '**' . (($ev['summary']) ? bb2diaspora($ev['summary']) : bb2diaspora($ev['desc'])) .  '**' . "\n";
 
        $o .= t('Starts:') . ' ' . '['
-               . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC', 
+               . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC',
                        $ev['start'] , $bd_format ))
-                       :  day_translate(datetime_convert('UTC', 'UTC', 
+                       :  day_translate(datetime_convert('UTC', 'UTC',
                        $ev['start'] , $bd_format)))
                .  '](' . $a->get_baseurl() . '/localtime/?f=&time=' . urlencode(datetime_convert('UTC','UTC',$ev['start'])) . ")\n";
 
        if(! $ev['nofinish'])
-               $o .= t('Finishes:') . ' ' . '[' 
-                       . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC', 
+               $o .= t('Finishes:') . ' ' . '['
+                       . (($ev['adjust']) ? day_translate(datetime_convert('UTC', 'UTC',
                                $ev['finish'] , $bd_format ))
-                               :  day_translate(datetime_convert('UTC', 'UTC', 
+                               :  day_translate(datetime_convert('UTC', 'UTC',
                                $ev['finish'] , $bd_format )))
                        . '](' . $a->get_baseurl() . '/localtime/?f=&time=' . urlencode(datetime_convert('UTC','UTC',$ev['finish'])) . ")\n";
 
        if(strlen($ev['location']))
-               $o .= t('Location:') . bb2diaspora($ev['location']) 
+               $o .= t('Location:') . bb2diaspora($ev['location'])
                        . "\n";
 
        $o .= "\n";