Add Trending Tags widget + template

[friendica.git] / src / Content / Text / HTML.php

diff --git a/src/Content/Text/HTML.php b/src/Content/Text/HTML.php
index 47463bdd091d1b665c07a116bcdaf042ffc843c4..c127973b13b8493837844b770c8762694f28785f 100644 (file)
--- a/src/Content/Text/HTML.php
+++ b/src/Content/Text/HTML.php
@@ -56,6 +56,7 @@ class HTML
 
                $xpath = new DOMXPath($doc);
 
+               /** @var \DOMNode[] $list */
                $list = $xpath->query("//" . $tag);
                foreach ($list as $node) {
                        $attr = [];
@@ -98,9 +99,12 @@ class HTML
                                $node->parentNode->insertBefore($StartCode, $node);
 
                                if ($node->hasChildNodes()) {
+                                       /** @var \DOMNode $child */
                                        foreach ($node->childNodes as $child) {
-                                               $newNode = $child->cloneNode(true);
-                                               $node->parentNode->insertBefore($newNode, $node);
+                                               if (trim($child->nodeValue)) {
+                                                       $newNode = $child->cloneNode(true);
+                                                       $node->parentNode->insertBefore($newNode, $node);
+                                               }
                                        }
                                }
 
@@ -290,6 +294,7 @@ class HTML
                self::tagToBBCode($doc, 'a', ['href' => '/mailto:(.+)/'], '[mail=$1]', '[/mail]');
                self::tagToBBCode($doc, 'a', ['href' => '/(.+)/'], '[url=$1]', '[/url]');
 
+               self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'alt' => '/(.+)/'], '[img=$1]$2', '[/img]');
                self::tagToBBCode($doc, 'img', ['src' => '/(.+)/', 'width' => '/(\d+)/', 'height' => '/(\d+)/'], '[img=$2x$3]$1', '[/img]');
                self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], '[img]$1', '[/img]');
 
@@ -559,6 +564,8 @@ class HTML
                                $ignore = false;
                        }
 
+                       $ignore = $ignore || strpos($treffer[1], '#') === 0;
+
                        if (!$ignore) {
                                $urls[$treffer[1]] = $treffer[1];
                        }
@@ -567,7 +574,13 @@ class HTML
                return $urls;
        }
 
-       public static function toPlaintext($html, $wraplength = 75, $compact = false)
+       /**
+        * @param string $html
+        * @param int    $wraplength Ensures individual lines aren't longer than this many characters. Doesn't break words.
+        * @param bool   $compact    True: Completely strips image tags; False: Keeps image URLs
+        * @return string
+        */
+       public static function toPlaintext(string $html, $wraplength = 75, $compact = false)
        {
                $message = str_replace("\r", "", $html);
 
@@ -576,38 +589,20 @@ class HTML
 
                $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
 
-               @$doc->loadHTML($message);
-
-               $xpath = new DOMXPath($doc);
-               $list = $xpath->query("//pre");
-               foreach ($list as $node) {
-                       $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
-               }
+               @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 
                $message = $doc->saveHTML();
-               $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br>", " ", ""], $message);
-               $message = preg_replace('= [\s]*=i', " ", $message);
+               // Remove eventual UTF-8 BOM
+               $message = str_replace("\xC3\x82\xC2\xA0", "", $message);
 
                // Collecting all links
                $urls = self::collectURLs($message);
 
-               @$doc->loadHTML($message);
+               @$doc->loadHTML($message, LIBXML_HTML_NODEFDTD);
 
                self::tagToBBCode($doc, 'html', [], '', '');
                self::tagToBBCode($doc, 'body', [], '', '');
 
-               // MyBB-Auszeichnungen
-               /*
-                 self::node2BBCode($doc, 'span', array('style'=>'text-decoration: underline;'), '_', '_');
-                 self::node2BBCode($doc, 'span', array('style'=>'font-style: italic;'), '/', '/');
-                 self::node2BBCode($doc, 'span', array('style'=>'font-weight: bold;'), '*', '*');
-
-                 self::node2BBCode($doc, 'strong', array(), '*', '*');
-                 self::node2BBCode($doc, 'b', array(), '*', '*');
-                 self::node2BBCode($doc, 'i', array(), '/', '/');
-                 self::node2BBCode($doc, 'u', array(), '_', '_');
-                */
-
                if ($compact) {
                        self::tagToBBCode($doc, 'blockquote', [], "»", "«");
                } else {
@@ -621,8 +616,6 @@ class HTML
                self::tagToBBCode($doc, 'div', [], "\r", "\r");
                self::tagToBBCode($doc, 'p', [], "\n", "\n");
 
-               //self::node2BBCode($doc, 'ul', array(), "\n[list]", "[/list]\n");
-               //self::node2BBCode($doc, 'ol', array(), "\n[list=1]", "[/list]\n");
                self::tagToBBCode($doc, 'li', [], "\n* ", "\n");
 
                self::tagToBBCode($doc, 'hr', [], "\n" . str_repeat("-", 70) . "\n", "");
@@ -637,12 +630,6 @@ class HTML
                self::tagToBBCode($doc, 'h5', [], "\n\n*", "*\n");
                self::tagToBBCode($doc, 'h6', [], "\n\n*", "*\n");
 
-               // Problem: there is no reliable way to detect if it is a link to a tag or profile
-               //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/'), ' $1 ', ' ', true);
-               //self::node2BBCode($doc, 'a', array('href'=>'/(.+)/', 'rel'=>'oembed'), ' $1 ', '', true);
-               //self::node2BBCode($doc, 'img', array('alt'=>'/(.+)/'), '$1', '');
-               //self::node2BBCode($doc, 'img', array('title'=>'/(.+)/'), '$1', '');
-               //self::node2BBCode($doc, 'img', array(), '', '');
                if (!$compact) {
                        self::tagToBBCode($doc, 'img', ['src' => '/(.+)/'], ' [img]$1', '[/img] ');
                } else {