3 * @file include/html2bbcode.php
4 * @brief Converter for HTML to BBCode
6 * Made by: ike@piratenpartei.de
7 * Originally made for the syncom project: http://wiki.piratenpartei.de/Syncom
8 * https://github.com/annando/Syncom
11 use Friendica\Core\Addon;
12 use Friendica\Util\XML;
14 function node2bbcode(&$doc, $oldnode, $attributes, $startbb, $endbb)
17 $done = node2bbcodesub($doc, $oldnode, $attributes, $startbb, $endbb);
21 function node2bbcodesub(&$doc, $oldnode, $attributes, $startbb, $endbb)
23 $savestart = str_replace('$', '\x01', $startbb);
26 $xpath = new DomXPath($doc);
28 $list = $xpath->query("//".$oldnode);
29 foreach ($list as $oldNode) {
31 if ($oldNode->attributes->length) {
32 foreach ($oldNode->attributes as $attribute) {
33 $attr[$attribute->name] = $attribute->value;
39 $startbb = $savestart;
43 foreach ($attributes as $attribute => $value) {
44 $startbb = str_replace('\x01'.++$i, '$1', $startbb);
45 if (strpos('*'.$startbb, '$1') > 0) {
46 if ($replace && (@$attr[$attribute] != '')) {
47 $startbb = preg_replace($value, $startbb, $attr[$attribute], -1, $count);
49 // If nothing could be changed
57 if (@$attr[$attribute] != $value) {
64 $StartCode = $oldNode->ownerDocument->createTextNode($startbb);
65 $EndCode = $oldNode->ownerDocument->createTextNode($endbb);
67 $oldNode->parentNode->insertBefore($StartCode, $oldNode);
69 if ($oldNode->hasChildNodes()) {
70 foreach ($oldNode->childNodes as $child) {
71 $newNode = $child->cloneNode(true);
72 $oldNode->parentNode->insertBefore($newNode, $oldNode);
76 $oldNode->parentNode->insertBefore($EndCode, $oldNode);
77 $oldNode->parentNode->removeChild($oldNode);
83 function html2bbcode($message, $basepath = '')
86 $message = str_replace("\r", "", $message);
88 // Removing code blocks before the whitespace removal processing below
90 $message = preg_replace_callback(
91 '#<pre><code(?: class="([^"]*)")?>(.*)</code></pre>#iUs',
92 function ($matches) use (&$codeblocks) {
93 $return = '[codeblock-' . count($codeblocks) . ']';
96 if ($matches[1] != '') {
97 $prefix = '[code=' . $matches[1] . ']';
99 $codeblocks[] = $prefix . trim($matches[2]) . '[/code]';
105 $message = str_replace(
118 $message = preg_replace('=<(\w+):(.+?)>=', '<removeme>', $message);
119 $message = preg_replace('=</(\w+):(.+?)>=', '</removeme>', $message);
121 $doc = new DOMDocument();
122 $doc->preserveWhiteSpace = false;
124 $message = mb_convert_encoding($message, 'HTML-ENTITIES', "UTF-8");
126 @$doc->loadHTML($message);
128 XML::deleteNode($doc, 'style');
129 XML::deleteNode($doc, 'head');
130 XML::deleteNode($doc, 'title');
131 XML::deleteNode($doc, 'meta');
132 XML::deleteNode($doc, 'xml');
133 XML::deleteNode($doc, 'removeme');
135 $xpath = new DomXPath($doc);
136 $list = $xpath->query("//pre");
137 foreach ($list as $node) {
138 $node->nodeValue = str_replace("\n", "\r", $node->nodeValue);
141 $message = $doc->saveHTML();
142 $message = str_replace(["\n<", ">\n", "\r", "\n", "\xC3\x82\xC2\xA0"], ["<", ">", "<br />", " ", ""], $message);
143 $message = preg_replace('= [\s]*=i', " ", $message);
144 @$doc->loadHTML($message);
146 node2bbcode($doc, 'html', [], "", "");
147 node2bbcode($doc, 'body', [], "", "");
149 // Outlook-Quote - Variant 1
150 node2bbcode($doc, 'p', ['class'=>'MsoNormal', 'style'=>'margin-left:35.4pt'], '[quote]', '[/quote]');
152 // Outlook-Quote - Variant 2
153 node2bbcode($doc, 'div', ['style'=>'border:none;border-left:solid blue 1.5pt;padding:0cm 0cm 0cm 4.0pt'], '[quote]', '[/quote]');
156 node2bbcode($doc, 'span', ['style'=>'text-decoration: underline;'], '[u]', '[/u]');
157 node2bbcode($doc, 'span', ['style'=>'font-style: italic;'], '[i]', '[/i]');
158 node2bbcode($doc, 'span', ['style'=>'font-weight: bold;'], '[b]', '[/b]');
160 /*node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[font=$1][size=$2][color=$3]', '[/color][/size][/font]');
161 node2bbcode($doc, 'font', array('size'=>'/(\d+)/', 'color'=>'/(.+)/'), '[size=$1][color=$2]', '[/color][/size]');
162 node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'size'=>'/(.+)/'), '[font=$1][size=$2]', '[/size][/font]');
163 node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/', 'color'=>'/(.+)/'), '[font=$1][color=$3]', '[/color][/font]');
164 node2bbcode($doc, 'font', array('face'=>'/([\w ]+)/'), '[font=$1]', '[/font]');
165 node2bbcode($doc, 'font', array('size'=>'/(\d+)/'), '[size=$1]', '[/size]');
166 node2bbcode($doc, 'font', array('color'=>'/(.+)/'), '[color=$1]', '[/color]');
169 //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*font-family:\s*(.+?)[,;].*color:\s*(.+?)[,;].*/'), '[size=$1][font=$2][color=$3]', '[/color][/font][/size]');
170 //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(\d+)[,;].*/'), '[size=$1]', '[/size]');
171 //node2bbcode($doc, 'span', array('style'=>'/.*font-size:\s*(.+?)[,;].*/'), '[size=$1]', '[/size]');
173 node2bbcode($doc, 'span', ['style'=>'/.*color:\s*(.+?)[,;].*/'], '[color="$1"]', '[/color]');
175 //node2bbcode($doc, 'span', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
177 //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)pt.*/'), '[font=$1][size=$2]', '[/size][/font]');
178 //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*font-size:\s*(\d+?)px.*/'), '[font=$1][size=$2]', '[/size][/font]');
179 //node2bbcode($doc, 'div', array('style'=>'/.*font-family:\s*(.+?)[,;].*/'), '[font=$1]', '[/font]');
181 // Importing the classes - interesting for importing of posts from third party networks that were exported from friendica
183 //node2bbcode($doc, 'span', array('class'=>'/([\w ]+)/'), '[class=$1]', '[/class]');
184 node2bbcode($doc, 'span', ['class'=>'type-link'], '[class=type-link]', '[/class]');
185 node2bbcode($doc, 'span', ['class'=>'type-video'], '[class=type-video]', '[/class]');
187 node2bbcode($doc, 'strong', [], '[b]', '[/b]');
188 node2bbcode($doc, 'em', [], '[i]', '[/i]');
189 node2bbcode($doc, 'b', [], '[b]', '[/b]');
190 node2bbcode($doc, 'i', [], '[i]', '[/i]');
191 node2bbcode($doc, 'u', [], '[u]', '[/u]');
193 node2bbcode($doc, 'big', [], "[size=large]", "[/size]");
194 node2bbcode($doc, 'small', [], "[size=small]", "[/size]");
196 node2bbcode($doc, 'blockquote', [], '[quote]', '[/quote]');
198 node2bbcode($doc, 'br', [], "\n", '');
200 node2bbcode($doc, 'p', ['class'=>'MsoNormal'], "\n", "");
201 node2bbcode($doc, 'div', ['class'=>'MsoNormal'], "\r", "");
203 node2bbcode($doc, 'span', [], "", "");
205 node2bbcode($doc, 'span', [], "", "");
206 node2bbcode($doc, 'pre', [], "", "");
208 node2bbcode($doc, 'div', [], "\r", "\r");
209 node2bbcode($doc, 'p', [], "\n", "\n");
211 node2bbcode($doc, 'ul', [], "[list]", "[/list]");
212 node2bbcode($doc, 'ol', [], "[list=1]", "[/list]");
213 node2bbcode($doc, 'li', [], "[*]", "");
215 node2bbcode($doc, 'hr', [], "[hr]", "");
217 node2bbcode($doc, 'table', [], "", "");
218 node2bbcode($doc, 'tr', [], "\n", "");
219 node2bbcode($doc, 'td', [], "\t", "");
220 //node2bbcode($doc, 'table', array(), "[table]", "[/table]");
221 //node2bbcode($doc, 'th', array(), "[th]", "[/th]");
222 //node2bbcode($doc, 'tr', array(), "[tr]", "[/tr]");
223 //node2bbcode($doc, 'td', array(), "[td]", "[/td]");
225 //node2bbcode($doc, 'h1', array(), "\n\n[size=xx-large][b]", "[/b][/size]\n");
226 //node2bbcode($doc, 'h2', array(), "\n\n[size=x-large][b]", "[/b][/size]\n");
227 //node2bbcode($doc, 'h3', array(), "\n\n[size=large][b]", "[/b][/size]\n");
228 //node2bbcode($doc, 'h4', array(), "\n\n[size=medium][b]", "[/b][/size]\n");
229 //node2bbcode($doc, 'h5', array(), "\n\n[size=small][b]", "[/b][/size]\n");
230 //node2bbcode($doc, 'h6', array(), "\n\n[size=x-small][b]", "[/b][/size]\n");
232 node2bbcode($doc, 'h1', [], "\n\n[h1]", "[/h1]\n");
233 node2bbcode($doc, 'h2', [], "\n\n[h2]", "[/h2]\n");
234 node2bbcode($doc, 'h3', [], "\n\n[h3]", "[/h3]\n");
235 node2bbcode($doc, 'h4', [], "\n\n[h4]", "[/h4]\n");
236 node2bbcode($doc, 'h5', [], "\n\n[h5]", "[/h5]\n");
237 node2bbcode($doc, 'h6', [], "\n\n[h6]", "[/h6]\n");
239 node2bbcode($doc, 'a', ['href'=>'/mailto:(.+)/'], '[mail=$1]', '[/mail]');
240 node2bbcode($doc, 'a', ['href'=>'/(.+)/'], '[url=$1]', '[/url]');
242 node2bbcode($doc, 'img', ['src'=>'/(.+)/', 'width'=>'/(\d+)/', 'height'=>'/(\d+)/'], '[img=$2x$3]$1', '[/img]');
243 node2bbcode($doc, 'img', ['src'=>'/(.+)/'], '[img]$1', '[/img]');
246 node2bbcode($doc, 'video', ['src'=>'/(.+)/'], '[video]$1', '[/video]');
247 node2bbcode($doc, 'audio', ['src'=>'/(.+)/'], '[audio]$1', '[/audio]');
248 node2bbcode($doc, 'iframe', ['src'=>'/(.+)/'], '[iframe]$1', '[/iframe]');
250 node2bbcode($doc, 'key', [], '[code]', '[/code]');
251 node2bbcode($doc, 'code', [], '[code]', '[/code]');
253 $message = $doc->saveHTML();
255 // I'm removing something really disturbing
256 // Don't know exactly what it is
257 $message = str_replace(chr(194).chr(160), ' ', $message);
259 $message = str_replace(" ", " ", $message);
261 // removing multiple DIVs
262 $message = preg_replace('=\r *\r=i', "\n", $message);
263 $message = str_replace("\r", "\n", $message);
265 Addon::callHooks('html2bbcode', $message);
267 $message = strip_tags($message);
269 $message = html_entity_decode($message, ENT_QUOTES, 'UTF-8');
271 $message = str_replace(["<"], ["<"], $message);
273 // remove quotes if they don't make sense
274 $message = preg_replace('=\[/quote\][\s]*\[quote\]=i', "\n", $message);
276 $message = preg_replace('=\[quote\]\s*=i', "[quote]", $message);
277 $message = preg_replace('=\s*\[/quote\]=i', "[/quote]", $message);
280 $oldmessage = $message;
281 $message = str_replace("\n \n", "\n\n", $message);
282 } while ($oldmessage != $message);
285 $oldmessage = $message;
286 $message = str_replace("\n\n\n", "\n\n", $message);
287 } while ($oldmessage != $message);
290 $oldmessage = $message;
291 $message = str_replace(
314 } while ($message != $oldmessage);
316 $message = str_replace(
317 ['[b][b]', '[/b][/b]', '[i][i]', '[/i][/i]'],
318 ['[b]', '[/b]', '[i]', '[/i]'],
322 // Handling Yahoo style of mails
323 $message = str_replace('[hr][b]From:[/b]', '[quote][b]From:[/b]', $message);
325 // Restore code blocks
326 $message = preg_replace_callback(
327 '#\[codeblock-([0-9]+)\]#iU',
328 function ($matches) use ($codeblocks) {
330 if (isset($codeblocks[intval($matches[1])])) {
331 $return = $codeblocks[$matches[1]];
338 $message = trim($message);
340 if ($basepath != '') {
341 $message = addHostname($message, $basepath);
348 * @brief Sub function to complete incomplete URL
350 * @param array $matches Result of preg_replace_callback
351 * @param string $basepath Basepath that is used to complete the URL
353 * @return string The expanded URL
355 function addHostnameSub($matches, $basepath)
357 $base = parse_url($basepath);
358 unset($base['query']);
359 unset($base['fragment']);
364 $parts = array_merge($base, parse_url($url));
365 $url2 = unParseUrl($parts);
367 return str_replace($url, $url2, $link);
371 * @brief Complete incomplete URLs in BBCode
373 * @param string $body Body with URLs
374 * @param string $basepath Basepath that is used to complete the URL
376 * @return string Body with expanded URLs
378 function addHostname($body, $basepath)
380 $URLSearchString = "^\[\]";
382 $matches = ["/\[url\=([$URLSearchString]*)\].*?\[\/url\]/ism",
383 "/\[url\]([$URLSearchString]*)\[\/url\]/ism",
384 "/\[img\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
385 "/\[img\](.*?)\[\/img\]/ism",
386 "/\[zmg\=[0-9]*x[0-9]*\](.*?)\[\/img\]/ism",
387 "/\[zmg\](.*?)\[\/zmg\]/ism",
388 "/\[video\](.*?)\[\/video\]/ism",
389 "/\[audio\](.*?)\[\/audio\]/ism",
392 foreach ($matches as $match) {
393 $body = preg_replace_callback(
395 function ($match) use ($basepath) {
396 return addHostnameSub($match, $basepath);