Avoid duplicated tags

[friendica.git] / include / bb2diaspora.php

diff --git a/include/bb2diaspora.php b/include/bb2diaspora.php
index 5a38852936dd40453fab0947ceb33d0e8e5b06e1..a8b39f741ac84a2d066066e129176cf0dff82b9a 100644 (file)
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@@ -18,7 +18,9 @@ function diaspora2bb($s) {
        $s = html_entity_decode($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);
 
        // The parser cannot handle paragraphs correctly
        $s = str_replace(array("</p>", "<p>", '<p dir="ltr">'),array("<br>", "<br>", "<br>"),$s);
@@ -32,6 +34,15 @@ function diaspora2bb($s) {
 
        $s = str_replace('&#35;','#',$s);
 
+       $search = array(" \n", "\n ");
+       $replace = array("\n", "\n");
+       do {
+               $oldtext = $s;
+               $s = str_replace($search, $replace, $s);
+       } while ($oldtext != $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
@@ -72,7 +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);
 
@@ -83,16 +93,20 @@ 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
                $Text = bbcode($Text, $preserve_nl, false, 4);
 
+    // mask some special HTML chars from conversation to markdown
+    $Text = str_replace(array('&lt;','&gt;','&amp;'),array('&_lt_;','&_gt_;','&_amp_;'),$Text);
+
        // If a link is followed by a quote then there should be a newline before it
        // Maybe we should make this newline at every time before a quote.
        $Text = str_replace(array("</a><blockquote>"), array("</a><br><blockquote>"), $Text);
@@ -102,6 +116,9 @@ function bb2diaspora($Text,$preserve_nl = false, $fordiaspora = true) {
        // Now convert HTML to Markdown
        $Text = new HTML_To_Markdown($Text);
 
+    // unmask the special chars back to HTML
+    $Text = str_replace(array('&_lt_;','&_gt_;','&_amp_;'),array('&lt;','&gt;','&amp;'),$Text);
+
        $a->save_timestamp($stamp1, "parser");
 
        // Libertree has a problem with escaped hashtags.
@@ -135,22 +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";