| 1441 |
ariadna |
1 |
<?php
|
|
|
2 |
|
|
|
3 |
/*
|
|
|
4 |
* Copyright (c) 2005-2007 Jon Abernathy <jon@chuggnutt.com>
|
|
|
5 |
*
|
|
|
6 |
* This script is free software; you can redistribute it and/or modify
|
|
|
7 |
* it under the terms of the GNU General Public License as published by
|
|
|
8 |
* the Free Software Foundation; either version 2 of the License, or
|
|
|
9 |
* (at your option) any later version.
|
|
|
10 |
*
|
|
|
11 |
* The GNU General Public License can be found at
|
|
|
12 |
* http://www.gnu.org/copyleft/gpl.html.
|
|
|
13 |
*
|
|
|
14 |
* This script is distributed in the hope that it will be useful,
|
|
|
15 |
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
16 |
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
17 |
* GNU General Public License for more details.
|
|
|
18 |
*/
|
|
|
19 |
|
|
|
20 |
namespace Html2Text;
|
|
|
21 |
|
|
|
22 |
class Html2Text
|
|
|
23 |
{
|
|
|
24 |
const ENCODING = 'UTF-8';
|
|
|
25 |
|
|
|
26 |
protected $htmlFuncFlags;
|
|
|
27 |
|
|
|
28 |
/**
|
|
|
29 |
* Contains the HTML content to convert.
|
|
|
30 |
*
|
|
|
31 |
* @var string $html
|
|
|
32 |
*/
|
|
|
33 |
protected $html;
|
|
|
34 |
|
|
|
35 |
/**
|
|
|
36 |
* Contains the converted, formatted text.
|
|
|
37 |
*
|
|
|
38 |
* @var string $text
|
|
|
39 |
*/
|
|
|
40 |
protected $text;
|
|
|
41 |
|
|
|
42 |
/**
|
|
|
43 |
* List of preg* regular expression patterns to search for,
|
|
|
44 |
* used in conjunction with $replace.
|
|
|
45 |
*
|
|
|
46 |
* @var array $search
|
|
|
47 |
* @see $replace
|
|
|
48 |
*/
|
|
|
49 |
protected $search = array(
|
|
|
50 |
"/\r/", // Non-legal carriage return
|
|
|
51 |
"/[\n\t]+/", // Newlines and tabs
|
|
|
52 |
'/<head\b[^>]*>.*?<\/head>/i', // <head>
|
|
|
53 |
'/<script\b[^>]*>.*?<\/script>/i', // <script>s -- which strip_tags supposedly has problems with
|
|
|
54 |
'/<style\b[^>]*>.*?<\/style>/i', // <style>s -- which strip_tags supposedly has problems with
|
|
|
55 |
'/<i\b[^>]*>(.*?)<\/i>/i', // <i>
|
|
|
56 |
'/<em\b[^>]*>(.*?)<\/em>/i', // <em>
|
|
|
57 |
'/<ins\b[^>]*>(.*?)<\/ins>/i', // <ins>
|
|
|
58 |
'/(<ul\b[^>]*>|<\/ul>)/i', // <ul> and </ul>
|
|
|
59 |
'/(<ol\b[^>]*>|<\/ol>)/i', // <ol> and </ol>
|
|
|
60 |
'/(<dl\b[^>]*>|<\/dl>)/i', // <dl> and </dl>
|
|
|
61 |
'/<li\b[^>]*>(.*?)<\/li>/i', // <li> and </li>
|
|
|
62 |
'/<dd\b[^>]*>(.*?)<\/dd>/i', // <dd> and </dd>
|
|
|
63 |
'/<dt\b[^>]*>(.*?)<\/dt>/i', // <dt> and </dt>
|
|
|
64 |
'/<li\b[^>]*>/i', // <li>
|
|
|
65 |
'/<hr\b[^>]*>/i', // <hr>
|
|
|
66 |
'/<div\b[^>]*>/i', // <div>
|
|
|
67 |
'/(<table\b[^>]*>|<\/table>)/i', // <table> and </table>
|
|
|
68 |
'/(<tr\b[^>]*>|<\/tr>)/i', // <tr> and </tr>
|
|
|
69 |
'/<td\b[^>]*>(.*?)<\/td>/i', // <td> and </td>
|
|
|
70 |
'/<span class="_html2text_ignore">.+?<\/span>/i', // <span class="_html2text_ignore">...</span>
|
|
|
71 |
'/<(img)\b[^>]*alt=\"([^>"]+)\"[^>]*>/i', // <img> with alt tag
|
|
|
72 |
);
|
|
|
73 |
|
|
|
74 |
/**
|
|
|
75 |
* List of pattern replacements corresponding to patterns searched.
|
|
|
76 |
*
|
|
|
77 |
* @var array $replace
|
|
|
78 |
* @see $search
|
|
|
79 |
*/
|
|
|
80 |
protected $replace = array(
|
|
|
81 |
'', // Non-legal carriage return
|
|
|
82 |
' ', // Newlines and tabs
|
|
|
83 |
'', // <head>
|
|
|
84 |
'', // <script>s -- which strip_tags supposedly has problems with
|
|
|
85 |
'', // <style>s -- which strip_tags supposedly has problems with
|
|
|
86 |
'_\\1_', // <i>
|
|
|
87 |
'_\\1_', // <em>
|
|
|
88 |
'_\\1_', // <ins>
|
|
|
89 |
"\n\n", // <ul> and </ul>
|
|
|
90 |
"\n\n", // <ol> and </ol>
|
|
|
91 |
"\n\n", // <dl> and </dl>
|
|
|
92 |
"\t* \\1\n", // <li> and </li>
|
|
|
93 |
" \\1\n", // <dd> and </dd>
|
|
|
94 |
"\t* \\1", // <dt> and </dt>
|
|
|
95 |
"\n\t* ", // <li>
|
|
|
96 |
"\n-------------------------\n", // <hr>
|
|
|
97 |
"<div>\n", // <div>
|
|
|
98 |
"\n\n", // <table> and </table>
|
|
|
99 |
"\n", // <tr> and </tr>
|
|
|
100 |
"\t\t\\1\n", // <td> and </td>
|
|
|
101 |
"", // <span class="_html2text_ignore">...</span>
|
|
|
102 |
'[\\2]', // <img> with alt tag
|
|
|
103 |
);
|
|
|
104 |
|
|
|
105 |
/**
|
|
|
106 |
* List of preg* regular expression patterns to search for,
|
|
|
107 |
* used in conjunction with $entReplace.
|
|
|
108 |
*
|
|
|
109 |
* @var array $entSearch
|
|
|
110 |
* @see $entReplace
|
|
|
111 |
*/
|
|
|
112 |
protected $entSearch = array(
|
|
|
113 |
'/™/i', // TM symbol in win-1252
|
|
|
114 |
'/—/i', // m-dash in win-1252
|
|
|
115 |
'/&(amp|#38);/i', // Ampersand: see converter()
|
|
|
116 |
'/[ ]{2,}/', // Runs of spaces, post-handling
|
|
|
117 |
'/'/i', // The apostrophe symbol
|
|
|
118 |
);
|
|
|
119 |
|
|
|
120 |
/**
|
|
|
121 |
* List of pattern replacements corresponding to patterns searched.
|
|
|
122 |
*
|
|
|
123 |
* @var array $entReplace
|
|
|
124 |
* @see $entSearch
|
|
|
125 |
*/
|
|
|
126 |
protected $entReplace = array(
|
|
|
127 |
'â„¢', // TM symbol
|
|
|
128 |
'—', // m-dash
|
|
|
129 |
'|+|amp|+|', // Ampersand: see converter()
|
|
|
130 |
' ', // Runs of spaces, post-handling
|
|
|
131 |
'\'', // Apostrophe
|
|
|
132 |
);
|
|
|
133 |
|
|
|
134 |
/**
|
|
|
135 |
* List of preg* regular expression patterns to search for
|
|
|
136 |
* and replace using callback function.
|
|
|
137 |
*
|
|
|
138 |
* @var array $callbackSearch
|
|
|
139 |
*/
|
|
|
140 |
protected $callbackSearch = array(
|
|
|
141 |
'/<(h)[123456]( [^>]*)?>(.*?)<\/h[123456]>/i', // h1 - h6
|
|
|
142 |
'/[ ]*<(p)( [^>]*)?>(.*?)<\/p>[ ]*/si', // <p> with surrounding whitespace.
|
|
|
143 |
'/<(br)[^>]*>[ ]*/i', // <br> with leading whitespace after the newline.
|
|
|
144 |
'/<(b)( [^>]*)?>(.*?)<\/b>/i', // <b>
|
|
|
145 |
'/<(strong)( [^>]*)?>(.*?)<\/strong>/i', // <strong>
|
|
|
146 |
'/<(del)( [^>]*)?>(.*?)<\/del>/i', // <del>
|
|
|
147 |
'/<(th)( [^>]*)?>(.*?)<\/th>/i', // <th> and </th>
|
|
|
148 |
'/<(a) [^>]*href=("|\')([^"\']+)\2([^>]*)>(.*?)<\/a>/i' // <a href="">
|
|
|
149 |
);
|
|
|
150 |
|
|
|
151 |
/**
|
|
|
152 |
* List of preg* regular expression patterns to search for in PRE body,
|
|
|
153 |
* used in conjunction with $preReplace.
|
|
|
154 |
*
|
|
|
155 |
* @var array $preSearch
|
|
|
156 |
* @see $preReplace
|
|
|
157 |
*/
|
|
|
158 |
protected $preSearch = array(
|
|
|
159 |
"/\n/",
|
|
|
160 |
"/\t/",
|
|
|
161 |
'/ /',
|
|
|
162 |
'/<pre[^>]*>/',
|
|
|
163 |
'/<\/pre>/'
|
|
|
164 |
);
|
|
|
165 |
|
|
|
166 |
/**
|
|
|
167 |
* List of pattern replacements corresponding to patterns searched for PRE body.
|
|
|
168 |
*
|
|
|
169 |
* @var array $preReplace
|
|
|
170 |
* @see $preSearch
|
|
|
171 |
*/
|
|
|
172 |
protected $preReplace = array(
|
|
|
173 |
'<br>',
|
|
|
174 |
' ',
|
|
|
175 |
' ',
|
|
|
176 |
'',
|
|
|
177 |
'',
|
|
|
178 |
);
|
|
|
179 |
|
|
|
180 |
/**
|
|
|
181 |
* Temporary workspace used during PRE processing.
|
|
|
182 |
*
|
|
|
183 |
* @var string $preContent
|
|
|
184 |
*/
|
|
|
185 |
protected $preContent = '';
|
|
|
186 |
|
|
|
187 |
/**
|
|
|
188 |
* Contains the base URL that relative links should resolve to.
|
|
|
189 |
*
|
|
|
190 |
* @var string $baseurl
|
|
|
191 |
*/
|
|
|
192 |
protected $baseurl = '';
|
|
|
193 |
|
|
|
194 |
/**
|
|
|
195 |
* Indicates whether content in the $html variable has been converted yet.
|
|
|
196 |
*
|
|
|
197 |
* @var boolean $converted
|
|
|
198 |
* @see $html, $text
|
|
|
199 |
*/
|
|
|
200 |
protected $converted = false;
|
|
|
201 |
|
|
|
202 |
/**
|
|
|
203 |
* Contains URL addresses from links to be rendered in plain text.
|
|
|
204 |
*
|
|
|
205 |
* @var array $linkList
|
|
|
206 |
* @see buildlinkList()
|
|
|
207 |
*/
|
|
|
208 |
protected $linkList = array();
|
|
|
209 |
|
|
|
210 |
/**
|
|
|
211 |
* Various configuration options (able to be set in the constructor)
|
|
|
212 |
*
|
|
|
213 |
* @var array $options
|
|
|
214 |
*/
|
|
|
215 |
protected $options = array(
|
|
|
216 |
'do_links' => 'inline', // 'none'
|
|
|
217 |
// 'inline' (show links inline)
|
|
|
218 |
// 'nextline' (show links on the next line)
|
|
|
219 |
// 'table' (if a table of link URLs should be listed after the text.
|
|
|
220 |
// 'bbcode' (show links as bbcode)
|
|
|
221 |
|
|
|
222 |
'width' => 70, // Maximum width of the formatted text, in columns.
|
|
|
223 |
// Set this value to 0 (or less) to ignore word wrapping
|
|
|
224 |
// and not constrain text to a fixed-width column.
|
|
|
225 |
);
|
|
|
226 |
|
|
|
227 |
private function legacyConstruct($html = '', $fromFile = false, array $options = array())
|
|
|
228 |
{
|
|
|
229 |
$this->set_html($html, $fromFile);
|
|
|
230 |
$this->options = array_merge($this->options, $options);
|
|
|
231 |
}
|
|
|
232 |
|
|
|
233 |
/**
|
|
|
234 |
* @param string $html Source HTML
|
|
|
235 |
* @param array $options Set configuration options
|
|
|
236 |
*/
|
|
|
237 |
public function __construct($html = '', $options = array())
|
|
|
238 |
{
|
|
|
239 |
$this->htmlFuncFlags = (PHP_VERSION_ID < 50400)
|
|
|
240 |
? ENT_QUOTES
|
|
|
241 |
: ENT_QUOTES | ENT_HTML5;
|
|
|
242 |
|
|
|
243 |
// for backwards compatibility
|
|
|
244 |
if (!is_array($options)) {
|
|
|
245 |
// phpcs:ignore (PHPCompatibility.FunctionUse.ArgumentFunctionsReportCurrentValue.NeedsInspection
|
|
|
246 |
call_user_func_array(array($this, 'legacyConstruct'), func_get_args());
|
|
|
247 |
return;
|
|
|
248 |
}
|
|
|
249 |
|
|
|
250 |
$this->html = $html;
|
|
|
251 |
$this->options = array_merge($this->options, $options);
|
|
|
252 |
}
|
|
|
253 |
|
|
|
254 |
/**
|
|
|
255 |
* Get the source HTML
|
|
|
256 |
*
|
|
|
257 |
* @return string
|
|
|
258 |
*/
|
|
|
259 |
public function getHtml()
|
|
|
260 |
{
|
|
|
261 |
return $this->html;
|
|
|
262 |
}
|
|
|
263 |
|
|
|
264 |
/**
|
|
|
265 |
* Set the source HTML
|
|
|
266 |
*
|
|
|
267 |
* @param string $html HTML source content
|
|
|
268 |
*/
|
|
|
269 |
public function setHtml($html)
|
|
|
270 |
{
|
|
|
271 |
$this->html = $html;
|
|
|
272 |
$this->converted = false;
|
|
|
273 |
}
|
|
|
274 |
|
|
|
275 |
/**
|
|
|
276 |
* @deprecated
|
|
|
277 |
*/
|
|
|
278 |
public function set_html($html, $from_file = false)
|
|
|
279 |
{
|
|
|
280 |
if ($from_file) {
|
|
|
281 |
throw new \InvalidArgumentException("Argument from_file no longer supported");
|
|
|
282 |
}
|
|
|
283 |
|
|
|
284 |
return $this->setHtml($html);
|
|
|
285 |
}
|
|
|
286 |
|
|
|
287 |
/**
|
|
|
288 |
* Returns the text, converted from HTML.
|
|
|
289 |
*
|
|
|
290 |
* @return string Plain text
|
|
|
291 |
*/
|
|
|
292 |
public function getText()
|
|
|
293 |
{
|
|
|
294 |
if (!$this->converted) {
|
|
|
295 |
$this->convert();
|
|
|
296 |
}
|
|
|
297 |
|
|
|
298 |
return $this->text;
|
|
|
299 |
}
|
|
|
300 |
|
|
|
301 |
/**
|
|
|
302 |
* @deprecated
|
|
|
303 |
*/
|
|
|
304 |
public function get_text()
|
|
|
305 |
{
|
|
|
306 |
return $this->getText();
|
|
|
307 |
}
|
|
|
308 |
|
|
|
309 |
/**
|
|
|
310 |
* @deprecated
|
|
|
311 |
*/
|
|
|
312 |
public function print_text()
|
|
|
313 |
{
|
|
|
314 |
print $this->getText();
|
|
|
315 |
}
|
|
|
316 |
|
|
|
317 |
/**
|
|
|
318 |
* @deprecated
|
|
|
319 |
*/
|
|
|
320 |
public function p()
|
|
|
321 |
{
|
|
|
322 |
return $this->print_text();
|
|
|
323 |
}
|
|
|
324 |
|
|
|
325 |
/**
|
|
|
326 |
* Sets a base URL to handle relative links.
|
|
|
327 |
*
|
|
|
328 |
* @param string $baseurl
|
|
|
329 |
*/
|
|
|
330 |
public function setBaseUrl($baseurl)
|
|
|
331 |
{
|
|
|
332 |
$this->baseurl = $baseurl;
|
|
|
333 |
}
|
|
|
334 |
|
|
|
335 |
/**
|
|
|
336 |
* @deprecated
|
|
|
337 |
*/
|
|
|
338 |
public function set_base_url($baseurl)
|
|
|
339 |
{
|
|
|
340 |
return $this->setBaseUrl($baseurl);
|
|
|
341 |
}
|
|
|
342 |
|
|
|
343 |
protected function convert()
|
|
|
344 |
{
|
|
|
345 |
$origEncoding = mb_internal_encoding();
|
|
|
346 |
mb_internal_encoding(self::ENCODING);
|
|
|
347 |
|
|
|
348 |
$this->doConvert();
|
|
|
349 |
|
|
|
350 |
mb_internal_encoding($origEncoding);
|
|
|
351 |
}
|
|
|
352 |
|
|
|
353 |
protected function doConvert()
|
|
|
354 |
{
|
|
|
355 |
$this->linkList = array();
|
|
|
356 |
|
|
|
357 |
if ($this->html === null) {
|
|
|
358 |
$text = '';
|
|
|
359 |
} else {
|
|
|
360 |
$text = trim($this->html);
|
|
|
361 |
}
|
|
|
362 |
|
|
|
363 |
$this->converter($text);
|
|
|
364 |
|
|
|
365 |
if ($this->linkList) {
|
|
|
366 |
$text .= "\n\nLinks:\n------\n";
|
|
|
367 |
foreach ($this->linkList as $i => $url) {
|
|
|
368 |
$text .= '[' . ($i + 1) . '] ' . $url . "\n";
|
|
|
369 |
}
|
|
|
370 |
}
|
|
|
371 |
|
|
|
372 |
$this->text = $text;
|
|
|
373 |
|
|
|
374 |
$this->converted = true;
|
|
|
375 |
}
|
|
|
376 |
|
|
|
377 |
protected function converter(&$text)
|
|
|
378 |
{
|
|
|
379 |
$this->convertBlockquotes($text);
|
|
|
380 |
$this->convertPre($text);
|
|
|
381 |
$text = preg_replace($this->search, $this->replace, $text);
|
|
|
382 |
$text = preg_replace_callback($this->callbackSearch, array($this, 'pregCallback'), $text);
|
|
|
383 |
$text = strip_tags($text);
|
|
|
384 |
$text = preg_replace($this->entSearch, $this->entReplace, $text);
|
|
|
385 |
$text = html_entity_decode($text, $this->htmlFuncFlags, self::ENCODING);
|
|
|
386 |
|
|
|
387 |
// Remove unknown/unhandled entities (this cannot be done in search-and-replace block)
|
|
|
388 |
$text = preg_replace('/&([a-zA-Z0-9]{2,6}|#[0-9]{2,4});/', '', $text);
|
|
|
389 |
|
|
|
390 |
// Convert "|+|amp|+|" into "&", need to be done after handling of unknown entities
|
|
|
391 |
// This properly handles situation of "&quot;" in input string
|
|
|
392 |
$text = str_replace('|+|amp|+|', '&', $text);
|
|
|
393 |
|
|
|
394 |
// Normalise empty lines
|
|
|
395 |
$text = preg_replace("/\n\s+\n/", "\n\n", $text);
|
|
|
396 |
$text = preg_replace("/[\n]{3,}/", "\n\n", $text);
|
|
|
397 |
|
|
|
398 |
// remove leading empty lines (can be produced by eg. P tag on the beginning)
|
|
|
399 |
if ($text === null) {
|
|
|
400 |
$text = '';
|
|
|
401 |
}
|
|
|
402 |
$text = ltrim($text, "\n");
|
|
|
403 |
|
|
|
404 |
if ($this->options['width'] > 0) {
|
|
|
405 |
$text = wordwrap($text, $this->options['width']);
|
|
|
406 |
}
|
|
|
407 |
}
|
|
|
408 |
|
|
|
409 |
/**
|
|
|
410 |
* Helper function called by preg_replace() on link replacement.
|
|
|
411 |
*
|
|
|
412 |
* Maintains an internal list of links to be displayed at the end of the
|
|
|
413 |
* text, with numeric indices to the original point in the text they
|
|
|
414 |
* appeared. Also makes an effort at identifying and handling absolute
|
|
|
415 |
* and relative links.
|
|
|
416 |
*
|
|
|
417 |
* @param string $link URL of the link
|
|
|
418 |
* @param string $display Part of the text to associate number with
|
|
|
419 |
* @param null $linkOverride
|
|
|
420 |
* @return string
|
|
|
421 |
*/
|
|
|
422 |
protected function buildlinkList($link, $display, $linkOverride = null)
|
|
|
423 |
{
|
|
|
424 |
$linkMethod = ($linkOverride) ? $linkOverride : $this->options['do_links'];
|
|
|
425 |
if ($linkMethod == 'none') {
|
|
|
426 |
return $display;
|
|
|
427 |
}
|
|
|
428 |
|
|
|
429 |
// Ignored link types
|
|
|
430 |
if (preg_match('!^(javascript:|mailto:|#)!i', html_entity_decode($link, $this->htmlFuncFlags, self::ENCODING))) {
|
|
|
431 |
return $display;
|
|
|
432 |
}
|
|
|
433 |
|
|
|
434 |
if (preg_match('!^([a-z][a-z0-9.+-]+:)!i', $link)) {
|
|
|
435 |
$url = $link;
|
|
|
436 |
} else {
|
|
|
437 |
$url = $this->baseurl;
|
|
|
438 |
if (mb_substr($link, 0, 1) != '/') {
|
|
|
439 |
$url .= '/';
|
|
|
440 |
}
|
|
|
441 |
$url .= $link;
|
|
|
442 |
}
|
|
|
443 |
|
|
|
444 |
if ($linkMethod == 'table') {
|
|
|
445 |
if (($index = array_search($url, $this->linkList)) === false) {
|
|
|
446 |
$index = count($this->linkList);
|
|
|
447 |
$this->linkList[] = $url;
|
|
|
448 |
}
|
|
|
449 |
|
|
|
450 |
return $display . ' [' . ($index + 1) . ']';
|
|
|
451 |
} elseif ($linkMethod == 'nextline') {
|
|
|
452 |
if ($url === $display) {
|
|
|
453 |
return $display;
|
|
|
454 |
}
|
|
|
455 |
return $display . "\n[" . $url . ']';
|
|
|
456 |
} elseif ($linkMethod == 'bbcode') {
|
|
|
457 |
return sprintf('[url=%s]%s[/url]', $url, $display);
|
|
|
458 |
} else { // link_method defaults to inline
|
|
|
459 |
if ($url === $display) {
|
|
|
460 |
return $display;
|
|
|
461 |
}
|
|
|
462 |
return $display . ' [' . $url . ']';
|
|
|
463 |
}
|
|
|
464 |
}
|
|
|
465 |
|
|
|
466 |
/**
|
|
|
467 |
* Helper function for PRE body conversion.
|
|
|
468 |
*
|
|
|
469 |
* @param string &$text HTML content
|
|
|
470 |
*/
|
|
|
471 |
protected function convertPre(&$text)
|
|
|
472 |
{
|
|
|
473 |
// get the content of PRE element
|
|
|
474 |
while (preg_match('/<pre[^>]*>(.*)<\/pre>/ismU', $text, $matches)) {
|
|
|
475 |
// Replace br tags with newlines to prevent the search-and-replace callback from killing whitespace
|
|
|
476 |
$this->preContent = preg_replace('/(<br\b[^>]*>)/i', "\n", $matches[1]);
|
|
|
477 |
|
|
|
478 |
// Run our defined tags search-and-replace with callback
|
|
|
479 |
$this->preContent = preg_replace_callback(
|
|
|
480 |
$this->callbackSearch,
|
|
|
481 |
array($this, 'pregCallback'),
|
|
|
482 |
$this->preContent
|
|
|
483 |
);
|
|
|
484 |
|
|
|
485 |
// convert the content
|
|
|
486 |
$this->preContent = sprintf(
|
|
|
487 |
'<div><br>%s<br></div>',
|
|
|
488 |
preg_replace($this->preSearch, $this->preReplace, $this->preContent)
|
|
|
489 |
);
|
|
|
490 |
|
|
|
491 |
// replace the content (use callback because content can contain $0 variable)
|
|
|
492 |
$text = preg_replace_callback(
|
|
|
493 |
'/<pre[^>]*>.*<\/pre>/ismU',
|
|
|
494 |
array($this, 'pregPreCallback'),
|
|
|
495 |
$text,
|
|
|
496 |
1
|
|
|
497 |
);
|
|
|
498 |
|
|
|
499 |
// free memory
|
|
|
500 |
$this->preContent = '';
|
|
|
501 |
}
|
|
|
502 |
}
|
|
|
503 |
|
|
|
504 |
/**
|
|
|
505 |
* Helper function for BLOCKQUOTE body conversion.
|
|
|
506 |
*
|
|
|
507 |
* @param string &$text HTML content
|
|
|
508 |
*/
|
|
|
509 |
protected function convertBlockquotes(&$text)
|
|
|
510 |
{
|
|
|
511 |
if (preg_match_all('/<\/*blockquote[^>]*>/i', $text, $matches, PREG_OFFSET_CAPTURE)) {
|
|
|
512 |
$originalText = $text;
|
|
|
513 |
$start = 0;
|
|
|
514 |
$taglen = 0;
|
|
|
515 |
$level = 0;
|
|
|
516 |
$diff = 0;
|
|
|
517 |
foreach ($matches[0] as $m) {
|
|
|
518 |
$m[1] = mb_strlen(substr($originalText, 0, $m[1]));
|
|
|
519 |
if ($m[0][0] == '<' && $m[0][1] == '/') {
|
|
|
520 |
$level--;
|
|
|
521 |
if ($level < 0) {
|
|
|
522 |
$level = 0; // malformed HTML: go to next blockquote
|
|
|
523 |
} elseif ($level > 0) {
|
|
|
524 |
// skip inner blockquote
|
|
|
525 |
} else {
|
|
|
526 |
$end = $m[1];
|
|
|
527 |
$len = $end - $taglen - $start;
|
|
|
528 |
// Get blockquote content
|
|
|
529 |
$body = mb_substr($text, $start + $taglen - $diff, $len);
|
|
|
530 |
|
|
|
531 |
// Set text width
|
|
|
532 |
$pWidth = $this->options['width'];
|
|
|
533 |
if ($this->options['width'] > 0) $this->options['width'] -= 2;
|
|
|
534 |
// Convert blockquote content
|
|
|
535 |
$body = trim($body);
|
|
|
536 |
$this->converter($body);
|
|
|
537 |
// Add citation markers and create PRE block
|
|
|
538 |
$body = preg_replace('/((^|\n)>*)/', '\\1> ', trim($body));
|
|
|
539 |
$body = '<pre>' . htmlspecialchars($body, $this->htmlFuncFlags, self::ENCODING) . '</pre>';
|
|
|
540 |
// Re-set text width
|
|
|
541 |
$this->options['width'] = $pWidth;
|
|
|
542 |
// Replace content
|
|
|
543 |
$text = mb_substr($text, 0, $start - $diff)
|
|
|
544 |
. $body
|
|
|
545 |
. mb_substr($text, $end + mb_strlen($m[0]) - $diff);
|
|
|
546 |
|
|
|
547 |
$diff += $len + $taglen + mb_strlen($m[0]) - mb_strlen($body);
|
|
|
548 |
unset($body);
|
|
|
549 |
}
|
|
|
550 |
} else {
|
|
|
551 |
if ($level == 0) {
|
|
|
552 |
$start = $m[1];
|
|
|
553 |
$taglen = mb_strlen($m[0]);
|
|
|
554 |
}
|
|
|
555 |
$level++;
|
|
|
556 |
}
|
|
|
557 |
}
|
|
|
558 |
}
|
|
|
559 |
}
|
|
|
560 |
|
|
|
561 |
/**
|
|
|
562 |
* Callback function for preg_replace_callback use.
|
|
|
563 |
*
|
|
|
564 |
* @param array $matches PREG matches
|
|
|
565 |
* @return string
|
|
|
566 |
*/
|
|
|
567 |
protected function pregCallback($matches)
|
|
|
568 |
{
|
|
|
569 |
switch (mb_strtolower($matches[1])) {
|
|
|
570 |
case 'p':
|
|
|
571 |
// Replace newlines with spaces.
|
|
|
572 |
$para = str_replace("\n", " ", $matches[3]);
|
|
|
573 |
|
|
|
574 |
// Trim trailing and leading whitespace within the tag.
|
|
|
575 |
$para = trim($para);
|
|
|
576 |
|
|
|
577 |
// Add trailing newlines for this para.
|
|
|
578 |
return "\n" . $para . "\n";
|
|
|
579 |
case 'br':
|
|
|
580 |
return "\n";
|
|
|
581 |
case 'b':
|
|
|
582 |
case 'strong':
|
|
|
583 |
return $this->toupper($matches[3]);
|
|
|
584 |
case 'del':
|
|
|
585 |
return $this->tostrike($matches[3]);
|
|
|
586 |
case 'th':
|
|
|
587 |
return $this->toupper("\t\t" . $matches[3] . "\n");
|
|
|
588 |
case 'h':
|
|
|
589 |
return $this->toupper("\n\n" . $matches[3] . "\n\n");
|
|
|
590 |
case 'a':
|
|
|
591 |
// override the link method
|
|
|
592 |
$linkOverride = null;
|
|
|
593 |
if (preg_match('/_html2text_link_(\w+)/', $matches[4], $linkOverrideMatch)) {
|
|
|
594 |
$linkOverride = $linkOverrideMatch[1];
|
|
|
595 |
}
|
|
|
596 |
// Remove spaces in URL (#1487805)
|
|
|
597 |
$url = str_replace(' ', '', $matches[3]);
|
|
|
598 |
|
|
|
599 |
return $this->buildlinkList($url, $matches[5], $linkOverride);
|
|
|
600 |
}
|
|
|
601 |
|
|
|
602 |
return '';
|
|
|
603 |
}
|
|
|
604 |
|
|
|
605 |
/**
|
|
|
606 |
* Callback function for preg_replace_callback use in PRE content handler.
|
|
|
607 |
*
|
|
|
608 |
* @param array $matches PREG matches
|
|
|
609 |
* @return string
|
|
|
610 |
*/
|
|
|
611 |
protected function pregPreCallback(/** @noinspection PhpUnusedParameterInspection */ $matches)
|
|
|
612 |
{
|
|
|
613 |
return $this->preContent;
|
|
|
614 |
}
|
|
|
615 |
|
|
|
616 |
/**
|
|
|
617 |
* Strtoupper function with HTML tags and entities handling.
|
|
|
618 |
*
|
|
|
619 |
* @param string $str Text to convert
|
|
|
620 |
* @return string Converted text
|
|
|
621 |
*/
|
|
|
622 |
protected function toupper($str)
|
|
|
623 |
{
|
|
|
624 |
// string can contain HTML tags
|
|
|
625 |
$chunks = preg_split('/(<[^>]*>)/', $str, -1, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_DELIM_CAPTURE);
|
|
|
626 |
|
|
|
627 |
// convert toupper only the text between HTML tags
|
|
|
628 |
foreach ($chunks as $i => $chunk) {
|
|
|
629 |
if ($chunk[0] != '<') {
|
|
|
630 |
$chunks[$i] = $this->strtoupper($chunk);
|
|
|
631 |
}
|
|
|
632 |
}
|
|
|
633 |
|
|
|
634 |
return implode($chunks);
|
|
|
635 |
}
|
|
|
636 |
|
|
|
637 |
/**
|
|
|
638 |
* Strtoupper multibyte wrapper function with HTML entities handling.
|
|
|
639 |
*
|
|
|
640 |
* @param string $str Text to convert
|
|
|
641 |
* @return string Converted text
|
|
|
642 |
*/
|
|
|
643 |
protected function strtoupper($str)
|
|
|
644 |
{
|
|
|
645 |
$str = html_entity_decode($str, $this->htmlFuncFlags, self::ENCODING);
|
|
|
646 |
$str = mb_strtoupper($str);
|
|
|
647 |
$str = htmlspecialchars($str, $this->htmlFuncFlags, self::ENCODING);
|
|
|
648 |
|
|
|
649 |
return $str;
|
|
|
650 |
}
|
|
|
651 |
|
|
|
652 |
/**
|
|
|
653 |
* Helper function for DEL conversion.
|
|
|
654 |
*
|
|
|
655 |
* @param string $text HTML content
|
|
|
656 |
* @return string Converted text
|
|
|
657 |
*/
|
|
|
658 |
protected function tostrike($str)
|
|
|
659 |
{
|
|
|
660 |
$rtn = '';
|
|
|
661 |
for ($i = 0; $i < mb_strlen($str); $i++) {
|
|
|
662 |
$chr = mb_substr($str, $i, 1);
|
|
|
663 |
$combiningChr = chr(0xC0 | 0x336 >> 6). chr(0x80 | 0x336 & 0x3F);
|
|
|
664 |
$rtn .= $chr . $combiningChr;
|
|
|
665 |
}
|
|
|
666 |
return $rtn;
|
|
|
667 |
}
|
|
|
668 |
}
|