1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
namespace XoopsModules\Publisher; |
4
|
|
|
|
5
|
|
|
/* |
6
|
|
|
You may not change or alter any portion of this comment or credits |
7
|
|
|
of supporting developers from this source code or any supporting source code |
8
|
|
|
which is considered copyrighted (c) material of the original comment or credit authors. |
9
|
|
|
|
10
|
|
|
This program is distributed in the hope that it will be useful, |
11
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of |
12
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. |
13
|
|
|
*/ |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* @copyright The XUUPS Project http://sourceforge.net/projects/xuups/ |
17
|
|
|
* @license http://www.fsf.org/copyleft/gpl.html GNU public license |
18
|
|
|
* @package Publisher |
19
|
|
|
* @since 1.0 |
20
|
|
|
* @author trabis <[email protected]> |
21
|
|
|
* @author The SmartFactory <www.smartfactory.ca> |
22
|
|
|
*/ |
23
|
|
|
|
24
|
|
|
use XoopsModules\Publisher; |
25
|
|
|
|
26
|
|
|
// defined('XOOPS_ROOT_PATH') || die('Restricted access'); |
27
|
|
|
|
28
|
|
|
require_once dirname(__DIR__) . '/include/common.php'; |
29
|
|
|
|
30
|
|
|
/** |
31
|
|
|
* Class Metagen |
32
|
|
|
*/ |
33
|
|
|
class Metagen |
34
|
|
|
{ |
35
|
|
|
/** |
36
|
|
|
* @var Publisher\Helper |
37
|
|
|
*/ |
38
|
|
|
public $helper; |
39
|
|
|
|
40
|
|
|
/** |
41
|
|
|
* @var \MyTextSanitizer |
42
|
|
|
*/ |
43
|
|
|
public $myts; |
44
|
|
|
|
45
|
|
|
/** |
46
|
|
|
* @var string |
47
|
|
|
*/ |
48
|
|
|
public $title; |
49
|
|
|
|
50
|
|
|
/** |
51
|
|
|
* @var string |
52
|
|
|
*/ |
53
|
|
|
public $originalTitle; |
54
|
|
|
|
55
|
|
|
/** |
56
|
|
|
* @var string |
57
|
|
|
*/ |
58
|
|
|
public $keywords; |
59
|
|
|
|
60
|
|
|
/** |
61
|
|
|
* @var string |
62
|
|
|
*/ |
63
|
|
|
public $categoryPath; |
64
|
|
|
|
65
|
|
|
/** |
66
|
|
|
* @var string |
67
|
|
|
*/ |
68
|
|
|
public $description; |
69
|
|
|
|
70
|
|
|
/** |
71
|
|
|
* @var int |
72
|
|
|
*/ |
73
|
|
|
public $minChar = 4; |
74
|
|
|
|
75
|
|
|
/** |
76
|
|
|
* @param string $title |
77
|
|
|
* @param string $keywords |
78
|
|
|
* @param string $description |
79
|
|
|
* @param string $categoryPath |
80
|
|
|
*/ |
81
|
|
|
public function __construct($title, $keywords = '', $description = '', $categoryPath = '') |
82
|
|
|
{ |
83
|
|
|
/** @var Publisher\Helper $this->helper */ |
84
|
|
|
$this->helper = \XoopsModules\Publisher\Helper::getInstance(); |
85
|
|
|
$this->myts = \MyTextSanitizer::getInstance(); |
86
|
|
|
$this->setCategoryPath($categoryPath); |
87
|
|
|
$this->setTitle($title); |
88
|
|
|
$this->setDescription($description); |
89
|
|
|
if ('' == $keywords) { |
90
|
|
|
$keywords = $this->createMetaKeywords(); |
91
|
|
|
} |
92
|
|
|
$this->setKeywords($keywords); |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* @param string $title |
97
|
|
|
*/ |
98
|
|
|
public function setTitle($title) |
99
|
|
|
{ |
100
|
|
|
$this->title = $this->html2text($title); |
101
|
|
|
$this->originalTitle = $this->title; |
102
|
|
|
$titleTag = []; |
103
|
|
|
$titleTag['module'] = $this->helper->getModule()->getVar('name'); |
104
|
|
|
if (isset($this->title) && ('' != $this->title) && (mb_strtoupper($this->title) != mb_strtoupper($titleTag['module']))) { |
|
|
|
|
105
|
|
|
$titleTag['title'] = $this->title; |
106
|
|
|
} |
107
|
|
|
if (isset($this->categoryPath) && ('' != $this->categoryPath)) { |
108
|
|
|
$titleTag['category'] = $this->categoryPath; |
109
|
|
|
} |
110
|
|
|
$ret = isset($titleTag['title']) ? $titleTag['title'] : ''; |
111
|
|
|
if (isset($titleTag['category']) && '' != $titleTag['category']) { |
112
|
|
|
if ('' != $ret) { |
113
|
|
|
$ret .= ' - '; |
114
|
|
|
} |
115
|
|
|
$ret .= $titleTag['category']; |
116
|
|
|
} |
117
|
|
|
if (isset($titleTag['module']) && '' != $titleTag['module']) { |
118
|
|
|
if ('' != $ret) { |
119
|
|
|
$ret .= ' - '; |
120
|
|
|
} |
121
|
|
|
$ret .= $titleTag['module']; |
122
|
|
|
} |
123
|
|
|
$this->title = $ret; |
124
|
|
|
} |
125
|
|
|
|
126
|
|
|
/** |
127
|
|
|
* @param string $keywords |
128
|
|
|
*/ |
129
|
|
|
public function setKeywords($keywords) |
130
|
|
|
{ |
131
|
|
|
$this->keywords = $keywords; |
132
|
|
|
} |
133
|
|
|
|
134
|
|
|
/** |
135
|
|
|
* @param string $categoryPath |
136
|
|
|
*/ |
137
|
|
|
public function setCategoryPath($categoryPath) |
138
|
|
|
{ |
139
|
|
|
$categoryPath = $this->html2text($categoryPath); |
140
|
|
|
$this->categoryPath = $categoryPath; |
141
|
|
|
} |
142
|
|
|
|
143
|
|
|
/** |
144
|
|
|
* @param string $description |
145
|
|
|
*/ |
146
|
|
|
public function setDescription($description) |
147
|
|
|
{ |
148
|
|
|
$description = $this->html2text($description); |
149
|
|
|
$description = $this->purifyText($description); |
150
|
|
|
$this->description = $description; |
151
|
|
|
} |
152
|
|
|
|
153
|
|
|
/** |
154
|
|
|
* Does nothing |
155
|
|
|
*/ |
156
|
|
|
public function createTitleTag() |
157
|
|
|
{ |
158
|
|
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* @param int $maxWords |
162
|
|
|
* |
163
|
|
|
* @return string |
164
|
|
|
*/ |
165
|
|
|
public function createMetaDescription($maxWords = 30) |
|
|
|
|
166
|
|
|
{ |
167
|
|
|
$description = $this->purifyText($this->description); |
168
|
|
|
$description = $this->html2text($description); |
169
|
|
|
$words = explode(' ', $description); |
170
|
|
|
$ret = ''; |
171
|
|
|
$i = 1; |
172
|
|
|
$wordCount = count($words); |
173
|
|
|
foreach ($words as $word) { |
174
|
|
|
$ret .= $word; |
175
|
|
|
if ($i < $wordCount) { |
176
|
|
|
$ret .= ' '; |
177
|
|
|
} |
178
|
|
|
++$i; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
return $ret; |
182
|
|
|
} |
183
|
|
|
|
184
|
|
|
/** |
185
|
|
|
* @param string $text |
186
|
|
|
* @param int $minChar |
187
|
|
|
* |
188
|
|
|
* @return array |
189
|
|
|
*/ |
190
|
|
|
public function findMetaKeywords($text, $minChar) |
191
|
|
|
{ |
192
|
|
|
$keywords = []; |
193
|
|
|
$text = $this->purifyText($text); |
194
|
|
|
$text = $this->html2text($text); |
195
|
|
|
$originalKeywords = explode(' ', $text); |
196
|
|
|
foreach ($originalKeywords as $originalKeyword) { |
197
|
|
|
$secondRoundKeywords = explode("'", $originalKeyword); |
198
|
|
|
foreach ($secondRoundKeywords as $secondRoundKeyword) { |
199
|
|
|
if (mb_strlen($secondRoundKeyword) >= $minChar) { |
200
|
|
|
if (!in_array($secondRoundKeyword, $keywords, true)) { |
201
|
|
|
$keywords[] = trim($secondRoundKeyword); |
202
|
|
|
} |
203
|
|
|
} |
204
|
|
|
} |
205
|
|
|
} |
206
|
|
|
|
207
|
|
|
return $keywords; |
208
|
|
|
} |
209
|
|
|
|
210
|
|
|
/** |
211
|
|
|
* @return string |
212
|
|
|
*/ |
213
|
|
|
public function createMetaKeywords() |
214
|
|
|
{ |
215
|
|
|
$keywords = $this->findMetaKeywords($this->originalTitle . ' ' . $this->description, $this->minChar); |
216
|
|
|
$moduleKeywords = $this->helper->getConfig('seo_meta_keywords'); |
217
|
|
|
if ('' != $moduleKeywords) { |
218
|
|
|
$moduleKeywords = explode(',', $moduleKeywords); |
219
|
|
|
$keywords = array_merge($keywords, array_map('trim', $moduleKeywords)); |
220
|
|
|
} |
221
|
|
|
$ret = implode(',', $keywords); |
222
|
|
|
|
223
|
|
|
return $ret; |
224
|
|
|
} |
225
|
|
|
|
226
|
|
|
/** |
227
|
|
|
* Does nothing |
228
|
|
|
*/ |
229
|
|
|
public function autoBuildMetaKeywords() |
230
|
|
|
{ |
231
|
|
|
} |
232
|
|
|
|
233
|
|
|
/** |
234
|
|
|
* Build Metatags |
235
|
|
|
*/ |
236
|
|
|
public function buildAutoMetaTags() |
237
|
|
|
{ |
238
|
|
|
$this->keywords = $this->createMetaKeywords(); |
239
|
|
|
$this->description = $this->createMetaDescription(); |
240
|
|
|
//$this->title = $this->createTitleTag(); |
241
|
|
|
} |
242
|
|
|
|
243
|
|
|
/** |
244
|
|
|
* Creates meta tags |
245
|
|
|
*/ |
246
|
|
|
public function createMetaTags() |
247
|
|
|
{ |
248
|
|
|
global $xoopsTpl, $xoTheme; |
249
|
|
|
if ('' != $this->keywords) { |
250
|
|
|
$xoTheme->addMeta('meta', 'keywords', $this->keywords); |
251
|
|
|
} |
252
|
|
|
if ('' != $this->description) { |
253
|
|
|
$xoTheme->addMeta('meta', 'description', $this->description); |
254
|
|
|
} |
255
|
|
|
if ('' != $this->title) { |
256
|
|
|
$xoopsTpl->assign('xoops_pagetitle', $this->title); |
257
|
|
|
} |
258
|
|
|
} |
259
|
|
|
|
260
|
|
|
/** |
261
|
|
|
* Return true if the string is length > 0 |
262
|
|
|
* |
263
|
|
|
* @credit psylove |
264
|
|
|
* @param mixed $var |
265
|
|
|
* @return bool |
266
|
|
|
*/ |
267
|
|
|
public static function emptyString($var) |
268
|
|
|
{ |
269
|
|
|
return ('' !== $var); |
270
|
|
|
} |
271
|
|
|
|
272
|
|
|
/** |
273
|
|
|
* Create a title for the short_url field of an article |
274
|
|
|
* |
275
|
|
|
* @credit psylove |
276
|
|
|
* |
277
|
|
|
* @param string|array $title title of the article |
278
|
|
|
* @param bool $withExt do we add an html extension or not |
279
|
|
|
* |
280
|
|
|
* @return string short url for article |
281
|
|
|
*/ |
282
|
|
|
public static function generateSeoTitle($title = '', $withExt = true) |
283
|
|
|
{ |
284
|
|
|
// Transformation de la chaine en minuscule |
285
|
|
|
// Codage de la chaine afin d'éviter les erreurs 500 en cas de caractères imprévus |
286
|
|
|
$title = rawurlencode(mb_strtolower($title)); |
|
|
|
|
287
|
|
|
// Transformation des ponctuations |
288
|
|
|
|
289
|
|
|
$pattern = [ |
290
|
|
|
'/%09/', // Tab |
291
|
|
|
'/%20/', // Space |
292
|
|
|
'/%21/', // ! |
293
|
|
|
'/%22/', // " |
294
|
|
|
'/%23/', // # |
295
|
|
|
'/%25/', // % |
296
|
|
|
'/%26/', // & |
297
|
|
|
'/%27/', // ' |
298
|
|
|
'/%28/', // ( |
299
|
|
|
'/%29/', // ) |
300
|
|
|
'/%2C/', // , |
301
|
|
|
'/%2F/', // / |
302
|
|
|
'/%3A/', // : |
303
|
|
|
'/%3B/', // ; |
304
|
|
|
'/%3C/', // < |
305
|
|
|
'/%3D/', // = |
306
|
|
|
'/%3E/', // > |
307
|
|
|
'/%3F/', // ? |
308
|
|
|
'/%40/', // @ |
309
|
|
|
'/%5B/', // [ |
310
|
|
|
'/%5C/', // \ |
311
|
|
|
'/%5D/', // ] |
312
|
|
|
'/%5E/', // ^ |
313
|
|
|
'/%7B/', // { |
314
|
|
|
'/%7C/', // | |
315
|
|
|
'/%7D/', // } |
316
|
|
|
'/%7E/', // ~ |
317
|
|
|
"/\./", // . |
318
|
|
|
]; |
319
|
|
|
$repPat = ['-', '-', '-', '-', '-', '-100', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-', '-at-', '-', '-', '-', '-', '-', '-', '-', '-', '-']; |
320
|
|
|
$title = str_replace($pattern, $repPat, $title); |
321
|
|
|
// Transformation des caractères accentués |
322
|
|
|
$pattern = [ |
323
|
|
|
'/%B0/', // ° |
324
|
|
|
'/%E8/', // è |
325
|
|
|
'/%E9/', // é |
326
|
|
|
'/%EA/', // ê |
327
|
|
|
'/%EB/', // ë |
328
|
|
|
'/%E7/', // ç |
329
|
|
|
'/%E0/', // à |
330
|
|
|
'/%E2/', // â |
331
|
|
|
'/%E4/', // ä |
332
|
|
|
'/%EE/', // î |
333
|
|
|
'/%EF/', // ï |
334
|
|
|
'/%F9/', // ù |
335
|
|
|
'/%FC/', // ü |
336
|
|
|
'/%FB/', // û |
337
|
|
|
'/%F4/', // ô |
338
|
|
|
'/%F6/', // ö |
339
|
|
|
]; |
340
|
|
|
$repPat = ['-', 'e', 'e', 'e', 'e', 'c', 'a', 'a', 'a', 'i', 'i', 'u', 'u', 'u', 'o', 'o']; |
341
|
|
|
$title = str_replace($pattern, $repPat, $title); |
342
|
|
|
$tableau = explode('-', $title); // Transforms the string in table //Transforme la chaine de caractères en tableau |
343
|
|
|
$tableau = array_filter($tableau, ['Metagen', 'emptyString']); // Remove empty strings of the table //Supprime les chaines vides du tableau |
344
|
|
|
$title = implode('-', $tableau); // Transforms a character string in table separated by a hyphen //Transforme un tableau en chaine de caractères séparé par un tiret |
345
|
|
|
if ($title && is_array($title)) { |
|
|
|
|
346
|
|
|
if ($withExt) { |
347
|
|
|
$title .= '.html'; |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
return $title; |
351
|
|
|
} |
352
|
|
|
|
353
|
|
|
return ''; |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* @param $text |
358
|
|
|
* @param bool $keyword |
359
|
|
|
* |
360
|
|
|
* @return mixed |
361
|
|
|
*/ |
362
|
|
|
public function purifyText($text, $keyword = false) |
363
|
|
|
{ |
364
|
|
|
// $text = str_replace([' ', ' '], ['<br>', ' '], $text); //for php 5.4 |
365
|
|
|
$text = str_replace(' ', ' ', $text); |
366
|
|
|
$text = str_replace('<br>', ' ', $text); |
367
|
|
|
$text = strip_tags($text); |
368
|
|
|
$text = html_entity_decode($text); |
369
|
|
|
$text = $this->myts->undoHtmlSpecialChars($text); |
370
|
|
|
|
371
|
|
|
$text = str_replace(')', ' ', $text); |
372
|
|
|
$text = str_replace('(', ' ', $text); |
373
|
|
|
$text = str_replace(':', ' ', $text); |
374
|
|
|
$text = str_replace('&euro', ' euro ', $text); |
375
|
|
|
$text = str_replace('&hellip', '...', $text); |
376
|
|
|
$text = str_replace('&rsquo', ' ', $text); |
377
|
|
|
$text = str_replace('!', ' ', $text); |
378
|
|
|
$text = str_replace('?', ' ', $text); |
379
|
|
|
$text = str_replace('"', ' ', $text); |
380
|
|
|
$text = str_replace('-', ' ', $text); |
381
|
|
|
$text = str_replace('\n', ' ', $text); |
382
|
|
|
|
383
|
|
|
// $text = str_replace([')','(',':','&euro','&hellip','&rsquo','!','?','"','-','\n'], [' ' , ' ', ' ', ' euro ', '...', ' ', ' ', ' ', ' ', ' ', ' '], $text); //for PHP 5.4 |
384
|
|
|
|
385
|
|
|
if ($keyword) { |
386
|
|
|
$text = str_replace('.', ' ', $text); |
387
|
|
|
$text = str_replace(',', ' ', $text); |
388
|
|
|
$text = str_replace('\'', ' ', $text); |
389
|
|
|
// $text = str_replace(['.', ' '], [',', ' '], ['\'', ' '], $text); //for PHP 5.4 |
390
|
|
|
} |
391
|
|
|
$text = str_replace(';', ' ', $text); |
392
|
|
|
|
393
|
|
|
return $text; |
394
|
|
|
} |
395
|
|
|
|
396
|
|
|
/** |
397
|
|
|
* @param string $document |
398
|
|
|
* |
399
|
|
|
* @return mixed |
400
|
|
|
*/ |
401
|
|
|
public function html2text($document) |
402
|
|
|
{ |
403
|
|
|
// PHP Manual:: function preg_replace |
404
|
|
|
// $document should contain an HTML document. |
405
|
|
|
// This will remove HTML tags, javascript sections |
406
|
|
|
// and white space. It will also convert some |
407
|
|
|
// common HTML entities to their text equivalent. |
408
|
|
|
// Credits : newbb2 |
409
|
|
|
$search = [ |
410
|
|
|
"'<script[^>]*?>.*?</script>'si", // Strip out javascript |
411
|
|
|
"'<img.*?>'si", // Strip out img tags |
412
|
|
|
"'<[\/\!]*?[^<>]*?>'si", // Strip out HTML tags |
413
|
|
|
"'([\r\n])[\s]+'", // Strip out white space |
414
|
|
|
"'&(quot|#34);'i", // Replace HTML entities |
415
|
|
|
"'&(amp|#38);'i", |
416
|
|
|
"'&(lt|#60);'i", |
417
|
|
|
"'&(gt|#62);'i", |
418
|
|
|
"'&(nbsp|#160);'i", |
419
|
|
|
"'&(iexcl|#161);'i", |
420
|
|
|
"'&(cent|#162);'i", |
421
|
|
|
"'&(pound|#163);'i", |
422
|
|
|
"'&(copy|#169);'i", |
423
|
|
|
]; // evaluate as php |
424
|
|
|
|
425
|
|
|
$replace = [ |
426
|
|
|
'', |
427
|
|
|
'', |
428
|
|
|
'', |
429
|
|
|
'\\1', |
430
|
|
|
'"', |
431
|
|
|
'&', |
432
|
|
|
'<', |
433
|
|
|
'>', |
434
|
|
|
' ', |
435
|
|
|
chr(161), |
436
|
|
|
chr(162), |
437
|
|
|
chr(163), |
438
|
|
|
chr(169), |
439
|
|
|
]; |
440
|
|
|
|
441
|
|
|
$text = preg_replace($search, $replace, $document); |
442
|
|
|
|
443
|
|
|
preg_replace_callback('/&#(\d+);/', function ($matches) { |
444
|
|
|
return chr($matches[1]); |
445
|
|
|
}, $document); |
446
|
|
|
|
447
|
|
|
return $text; |
448
|
|
|
} |
449
|
|
|
} |
450
|
|
|
|