1
|
|
|
<?php |
2
|
|
|
|
3
|
|
|
/** |
4
|
|
|
* PHP Domain Parser: Public Suffix List based URL parsing. |
5
|
|
|
* |
6
|
|
|
* @link http://github.com/jeremykendall/php-domain-parser for the canonical source repository |
7
|
|
|
* |
8
|
|
|
* @copyright Copyright (c) 2014 Jeremy Kendall (http://about.me/jeremykendall) |
9
|
|
|
* @license http://github.com/jeremykendall/php-domain-parser/blob/master/LICENSE MIT License |
10
|
|
|
*/ |
11
|
|
|
namespace Pdp; |
12
|
|
|
|
13
|
|
|
use Pdp\HttpAdapter\HttpAdapterInterface; |
14
|
|
|
|
15
|
|
|
/** |
16
|
|
|
* Public Suffix List Manager. |
17
|
|
|
* |
18
|
|
|
* This class obtains, writes, caches, and returns text and PHP representations |
19
|
|
|
* of the Public Suffix List |
20
|
|
|
*/ |
21
|
|
|
class PublicSuffixListManager |
22
|
|
|
{ |
23
|
|
|
const ALL_DOMAINS = 'ALL'; |
24
|
|
|
|
25
|
|
|
const PDP_PSL_TEXT_FILE = 'public-suffix-list.txt'; |
26
|
|
|
const PDP_PSL_PHP_FILE = 'public-suffix-list.php'; |
27
|
|
|
|
28
|
|
|
const ICANN_DOMAINS = 'ICANN'; |
29
|
|
|
const ICANN_PSL_PHP_FILE = 'icann-public-suffix-list.php'; |
30
|
|
|
|
31
|
|
|
const PRIVATE_DOMAINS = 'PRIVATE'; |
32
|
|
|
const PRIVATE_PSL_PHP_FILE = 'private-public-suffix-list.php'; |
33
|
|
|
|
34
|
|
|
/** |
35
|
|
|
* @var string Public Suffix List URL |
36
|
|
|
*/ |
37
|
|
|
protected $publicSuffixListUrl = 'https://publicsuffix.org/list/effective_tld_names.dat'; |
38
|
|
|
|
39
|
|
|
/** |
40
|
|
|
* @var string Directory where text and php versions of list will be cached |
41
|
|
|
*/ |
42
|
|
|
protected $cacheDir; |
43
|
|
|
|
44
|
|
|
/** |
45
|
|
|
* @var PublicSuffixList Public Suffix List |
46
|
|
|
*/ |
47
|
|
|
protected static $domainList = array( |
48
|
|
|
self::ALL_DOMAINS => self::PDP_PSL_PHP_FILE, |
49
|
|
|
self::ICANN_DOMAINS => self::ICANN_PSL_PHP_FILE, |
50
|
|
|
self::PRIVATE_DOMAINS => self::PRIVATE_PSL_PHP_FILE, |
51
|
11 |
|
); |
52
|
|
|
|
53
|
11 |
|
/** |
54
|
1 |
|
* @var \Pdp\HttpAdapter\HttpAdapterInterface Http adapter |
55
|
1 |
|
*/ |
56
|
1 |
|
protected $httpAdapter; |
57
|
1 |
|
|
58
|
|
|
/** |
59
|
11 |
|
* Public constructor. |
60
|
11 |
|
* |
61
|
|
|
* @param string $cacheDir Optional cache directory |
62
|
|
|
*/ |
63
|
|
|
public function __construct($cacheDir = null) |
64
|
|
|
{ |
65
|
|
|
if (null === $cacheDir) { |
66
|
1 |
|
$cacheDir = realpath( |
67
|
|
|
dirname(dirname(__DIR__)) . DIRECTORY_SEPARATOR . 'data' |
68
|
1 |
|
); |
69
|
1 |
|
} |
70
|
1 |
|
|
71
|
1 |
|
$this->cacheDir = $cacheDir; |
72
|
1 |
|
} |
73
|
1 |
|
|
74
|
|
|
/** |
75
|
|
|
* Downloads Public Suffix List and writes text cache and PHP cache. If these files |
76
|
|
|
* already exist, they will be overwritten. |
77
|
|
|
*/ |
78
|
|
|
public function refreshPublicSuffixList() |
79
|
|
|
{ |
80
|
2 |
|
$this->fetchListFromSource(); |
81
|
|
|
$cacheFile = $this->cacheDir . '/' . self::PDP_PSL_TEXT_FILE; |
82
|
2 |
|
$publicSuffixListArray = $this->convertListToArray($cacheFile); |
83
|
|
|
foreach ($publicSuffixListArray as $domain => $data) { |
84
|
2 |
|
// do not empty existing PHP cache file if source TXT is empty |
85
|
|
|
if ( |
86
|
|
|
is_array($data) |
87
|
|
|
&& |
88
|
2 |
|
!empty($data) |
89
|
|
|
) { |
90
|
|
|
$this->varExportToFile(self::$domainList[$domain], $data); |
91
|
|
|
} |
92
|
|
|
} |
93
|
|
|
} |
94
|
|
|
|
95
|
|
|
/** |
96
|
|
|
* Obtain Public Suffix List from its online source and write to cache dir. |
97
|
|
|
* |
98
|
|
|
* @return int|bool Number of bytes that were written to the file OR false in case of error |
99
|
|
|
*/ |
100
|
|
|
public function fetchListFromSource() |
101
|
|
|
{ |
102
|
|
|
$publicSuffixList = $this->getHttpAdapter()->getContent($this->publicSuffixListUrl); |
103
|
|
|
|
104
|
|
|
if ($publicSuffixList === false) { |
105
|
|
|
return 0; |
106
|
|
|
} |
107
|
4 |
|
|
108
|
|
|
return $this->write(self::PDP_PSL_TEXT_FILE, $publicSuffixList); |
109
|
|
|
} |
110
|
4 |
|
|
111
|
4 |
|
/** |
112
|
1 |
|
* Parses text representation of list to associative, multidimensional array. |
113
|
|
|
* |
114
|
|
|
* This method is based heavily on the code found in generateEffectiveTLDs.php |
115
|
3 |
|
* |
116
|
3 |
|
* @link https://github.com/usrflo/registered-domain-libs/blob/master/generateEffectiveTLDs.php |
117
|
3 |
|
* A copy of the Apache License, Version 2.0, is provided with this |
118
|
3 |
|
* distribution |
119
|
|
|
* |
120
|
3 |
|
* @param string $textFile Public Suffix List text filename |
121
|
3 |
|
* |
122
|
|
|
* @return array Associative, multidimensional array representation of the |
123
|
3 |
|
* public suffx list |
124
|
3 |
|
* |
125
|
3 |
|
* @throws \Exception Throws \Exception if unable to read file |
126
|
3 |
|
*/ |
127
|
|
|
public function parseListToArray($textFile) |
128
|
3 |
|
{ |
129
|
|
|
/** @noinspection PhpUsageOfSilenceOperatorInspection */ |
130
|
3 |
|
$fp = @fopen($textFile, 'r'); |
131
|
|
|
if (!$fp || !flock($fp, LOCK_SH)) { |
132
|
3 |
|
throw new \Exception("Cannot read '$textFile'"); |
133
|
3 |
|
} |
134
|
3 |
|
|
135
|
3 |
|
$data = file( |
136
|
|
|
$textFile, |
137
|
3 |
|
FILE_IGNORE_NEW_LINES | FILE_SKIP_EMPTY_LINES |
138
|
|
|
); |
139
|
|
|
|
140
|
|
|
flock($fp, LOCK_UN); |
141
|
|
|
fclose($fp); |
142
|
|
|
|
143
|
|
|
$data = array_filter( |
144
|
|
|
$data, |
145
|
|
|
function ($line) { |
146
|
|
|
return false === strpos($line, '//'); |
147
|
|
|
} |
148
|
|
|
); |
149
|
|
|
|
150
|
|
|
$publicSuffixListArray = array(); |
151
|
|
|
|
152
|
|
|
foreach ($data as $line) { |
153
|
|
|
$ruleParts = explode('.', $line); |
154
|
3 |
|
$this->buildArray($publicSuffixListArray, $ruleParts); |
155
|
|
|
} |
156
|
3 |
|
|
157
|
|
|
return $publicSuffixListArray; |
158
|
3 |
|
} |
159
|
|
|
|
160
|
|
|
/** |
161
|
|
|
* Recursive method to build the array representation of the Public Suffix List. |
162
|
|
|
* |
163
|
|
|
* This method is based heavily on the code found in generateEffectiveTLDs.php |
164
|
3 |
|
* |
165
|
3 |
|
* @link https://github.com/usrflo/registered-domain-libs/blob/master/generateEffectiveTLDs.php |
166
|
|
|
* A copy of the Apache License, Version 2.0, is provided with this |
167
|
3 |
|
* distribution |
168
|
3 |
|
* |
169
|
3 |
|
* @param array $publicSuffixListArray Initially an empty array, this eventually |
170
|
3 |
|
* becomes the array representation of the Public Suffix List |
171
|
|
|
* @param array $ruleParts One line (rule) from the Public Suffix List |
172
|
3 |
|
* exploded on '.', or the remaining portion of that array during recursion |
173
|
3 |
|
*/ |
174
|
3 |
|
public function buildArray(array &$publicSuffixListArray, array $ruleParts) |
175
|
3 |
|
{ |
176
|
3 |
|
$isDomain = true; |
177
|
|
|
|
178
|
3 |
|
$part = array_pop($ruleParts); |
179
|
|
|
|
180
|
3 |
|
// Adheres to canonicalization rule from the "Formal Algorithm" section |
181
|
3 |
|
// of https://publicsuffix.org/list/ |
182
|
3 |
|
// "The domain and all rules must be canonicalized in the normal way |
183
|
3 |
|
// for hostnames - lower-case, Punycode (RFC 3492)." |
184
|
|
|
$punycode = new PunycodeWrapper(); |
185
|
|
|
$part = $punycode->encode($part); |
186
|
|
|
|
187
|
|
|
if (strpos($part, '!') === 0) { |
188
|
|
|
$part = substr($part, 1); |
189
|
|
|
$isDomain = false; |
190
|
|
|
} |
191
|
|
|
|
192
|
3 |
|
if (!isset($publicSuffixListArray[$part])) { |
193
|
|
|
if ($isDomain) { |
194
|
3 |
|
$publicSuffixListArray[$part] = array(); |
195
|
|
|
} else { |
196
|
3 |
|
$publicSuffixListArray[$part] = array('!' => ''); |
197
|
|
|
} |
198
|
|
|
} |
199
|
|
|
|
200
|
|
|
if ($isDomain && count($ruleParts) > 0) { |
201
|
|
|
$this->buildArray($publicSuffixListArray[$part], $ruleParts); |
202
|
|
|
} |
203
|
|
|
} |
204
|
|
|
|
205
|
|
|
/** |
206
|
3 |
|
* Writes php array representation of the Public Suffix List to disk. |
207
|
|
|
* |
208
|
3 |
|
* @param array $publicSuffixList Array representation of the Public Suffix List |
209
|
|
|
* |
210
|
3 |
|
* @return int Number of bytes that were written to the file |
211
|
1 |
|
*/ |
212
|
1 |
|
public function writePhpCache(array $publicSuffixList) |
213
|
|
|
{ |
214
|
3 |
|
$data = '<?php' . PHP_EOL . 'static $data = ' . var_export($publicSuffixList, true) . '; $result =& $data; unset($data); return $result;'; |
215
|
|
|
|
216
|
3 |
|
return $this->write(self::PDP_PSL_PHP_FILE, $data); |
217
|
|
|
} |
218
|
|
|
|
219
|
|
|
/** |
220
|
|
|
* Writes php array representation to disk. |
221
|
|
|
* |
222
|
|
|
* @param string $basename file path |
223
|
|
|
* @param array $input input data |
224
|
|
|
* |
225
|
|
|
* @return int Number of bytes that were written to the file |
226
|
4 |
|
*/ |
227
|
|
|
protected function varExportToFile($basename, array $input) |
228
|
|
|
{ |
229
|
4 |
|
$data = '<?php' . PHP_EOL . 'static $data = ' . var_export($input, true) . '; $result =& $data; unset($data); return $result;'; |
230
|
4 |
|
|
231
|
1 |
|
return $this->write($basename, $data); |
232
|
|
|
} |
233
|
|
|
|
234
|
|
|
/** |
235
|
3 |
|
* Gets Public Suffix List. |
236
|
|
|
* |
237
|
3 |
|
* @param string $list the Public Suffix List type |
238
|
|
|
* @param bool $withStaticCache |
239
|
3 |
|
* |
240
|
3 |
|
* @return PublicSuffixList Instance of Public Suffix List |
241
|
|
|
* |
242
|
3 |
|
* @throws \Exception Throws \Exception if unable to read file |
243
|
|
|
*/ |
244
|
|
|
public function getList($list = self::ALL_DOMAINS, $withStaticCache = true) |
245
|
|
|
{ |
246
|
|
|
// init |
247
|
|
|
static $LIST_STATIC = array(); |
248
|
|
|
|
249
|
|
|
$cacheBasename = isset(self::$domainList[$list]) ? self::$domainList[$list] : self::PDP_PSL_PHP_FILE; |
250
|
|
|
$cacheFile = $this->cacheDir . '/' . $cacheBasename; |
251
|
|
|
$cacheKey = md5($cacheFile); |
252
|
|
|
|
253
|
|
|
if ($withStaticCache === true && isset($LIST_STATIC[$cacheKey])) { |
254
|
|
|
return $LIST_STATIC[$cacheKey]; |
255
|
4 |
|
} |
256
|
|
|
|
257
|
4 |
|
if (!file_exists($cacheFile)) { |
258
|
|
|
$this->refreshPublicSuffixList(); |
259
|
|
|
} |
260
|
|
|
|
261
|
4 |
|
if (!isset($LIST_STATIC[$cacheKey])) { |
262
|
|
|
$LIST_STATIC[$cacheKey] = new PublicSuffixList($cacheFile); |
263
|
4 |
|
} |
264
|
4 |
|
|
265
|
4 |
|
return $LIST_STATIC[$cacheKey]; |
266
|
4 |
|
} |
267
|
|
|
|
268
|
4 |
|
/** |
269
|
1 |
|
* Retrieves public suffix list from file after obtaining a shared lock. |
270
|
1 |
|
* |
271
|
|
|
* @param string $phpFile |
272
|
|
|
* |
273
|
3 |
|
* @return PublicSuffixList Instance of Public Suffix List |
274
|
3 |
|
* |
275
|
|
|
* @throws \Exception Throws \Exception if unable to read file |
276
|
3 |
|
*/ |
277
|
|
|
public function getListFromFile($phpFile) |
278
|
|
|
{ |
279
|
|
|
/** @noinspection PhpUsageOfSilenceOperatorInspection */ |
280
|
|
|
$fp = @fopen($phpFile, 'r'); |
281
|
|
|
if (!$fp || !flock($fp, LOCK_SH)) { |
282
|
|
|
throw new \Exception("Cannot read '$phpFile'"); |
283
|
|
|
} |
284
|
3 |
|
|
285
|
|
|
/** @noinspection PhpIncludeInspection */ |
286
|
3 |
|
$list = new PublicSuffixList( |
287
|
1 |
|
require $phpFile |
288
|
1 |
|
); |
289
|
1 |
|
|
290
|
|
|
flock($fp, LOCK_UN); |
291
|
|
|
fclose($fp); |
292
|
1 |
|
|
293
|
|
|
return $list; |
294
|
3 |
|
} |
295
|
|
|
|
296
|
|
|
/** |
297
|
|
|
* Parses text representation of list to associative, multidimensional array. |
298
|
|
|
* |
299
|
|
|
* @param string $textFile Public Suffix List text filename |
300
|
|
|
* |
301
|
|
|
* @return array Associative, multidimensional array representation of the |
302
|
11 |
|
* public suffx list |
303
|
|
|
*/ |
304
|
11 |
|
protected function convertListToArray($textFile) |
305
|
11 |
|
{ |
306
|
|
|
$addDomain = array( |
307
|
|
|
self::ICANN_DOMAINS => false, |
308
|
|
|
self::PRIVATE_DOMAINS => false, |
309
|
|
|
); |
310
|
|
|
|
311
|
|
|
$publicSuffixListArray = array( |
312
|
|
|
self::ALL_DOMAINS => array(), |
313
|
|
|
self::ICANN_DOMAINS => array(), |
314
|
|
|
self::PRIVATE_DOMAINS => array(), |
315
|
|
|
); |
316
|
|
|
|
317
|
|
|
$data = new \SplFileObject($textFile); |
318
|
|
|
$data->setFlags(\SplFileObject::DROP_NEW_LINE | \SplFileObject::READ_AHEAD | \SplFileObject::SKIP_EMPTY); |
319
|
|
|
foreach ($data as $line) { |
320
|
|
|
$addDomain = $this->validateDomainAddition($line, $addDomain); |
|
|
|
|
321
|
|
|
if (strstr($line, '//') !== false) { |
322
|
|
|
continue; |
323
|
|
|
} |
324
|
|
|
$publicSuffixListArray = $this->convertLineToArray($line, $publicSuffixListArray, $addDomain); |
|
|
|
|
325
|
|
|
} |
326
|
|
|
|
327
|
|
|
return $publicSuffixListArray; |
328
|
|
|
} |
329
|
|
|
|
330
|
|
|
/** |
331
|
|
|
* Convert a line from the Public Suffix list. |
332
|
|
|
* |
333
|
|
|
* @param string $textLine Public Suffix List text line |
334
|
|
|
* @param array $publicSuffixListArray Associative, multidimensional array representation of the |
335
|
|
|
* public suffx list |
336
|
|
|
* @param array $addDomain Tell which section should be converted |
337
|
|
|
* |
338
|
|
|
* @return array Associative, multidimensional array representation of the |
339
|
|
|
* public suffx list |
340
|
|
|
*/ |
341
|
|
|
protected function convertLineToArray($textLine, array $publicSuffixListArray, array $addDomain) |
342
|
|
|
{ |
343
|
|
|
$ruleParts = explode('.', $textLine); |
344
|
|
|
$this->buildArray($publicSuffixListArray[self::ALL_DOMAINS], $ruleParts); |
345
|
|
|
$domainNames = array_keys(array_filter($addDomain)); |
346
|
|
|
foreach ($domainNames as $domainName) { |
347
|
|
|
$this->buildArray($publicSuffixListArray[$domainName], $ruleParts); |
348
|
|
|
} |
349
|
|
|
|
350
|
|
|
return $publicSuffixListArray; |
351
|
|
|
} |
352
|
|
|
|
353
|
|
|
/** |
354
|
|
|
* Update the addition status for a given line against the domain list (ICANN and PRIVATE). |
355
|
|
|
* |
356
|
|
|
* @param string $line the current file line |
357
|
|
|
* @param array $addDomain the domain addition status |
358
|
|
|
* |
359
|
|
|
* @return array |
360
|
|
|
*/ |
361
|
|
|
protected function validateDomainAddition($line, array $addDomain) |
362
|
|
|
{ |
363
|
|
|
foreach ($addDomain as $section => $status) { |
364
|
|
|
$addDomain[$section] = $this->isValidSection($status, $line, $section); |
365
|
|
|
} |
366
|
|
|
|
367
|
|
|
return $addDomain; |
368
|
|
|
} |
369
|
|
|
|
370
|
|
|
/** |
371
|
|
|
* Tell whether the line can be converted for a given domain. |
372
|
|
|
* |
373
|
|
|
* @param bool $previousStatus the previous status |
374
|
|
|
* @param string $line the current file line |
375
|
|
|
* @param string $section the section to be considered |
376
|
|
|
* |
377
|
|
|
* @return bool |
378
|
|
|
*/ |
379
|
|
|
protected function isValidSection($previousStatus, $line, $section) |
380
|
|
|
{ |
381
|
|
View Code Duplication |
if (!$previousStatus && 0 === strpos($line, '// ===BEGIN ' . $section . ' DOMAINS===')) { |
|
|
|
|
382
|
|
|
return true; |
383
|
|
|
} |
384
|
|
|
|
385
|
|
View Code Duplication |
if ($previousStatus && 0 === strpos($line, '// ===END ' . $section . ' DOMAINS===')) { |
|
|
|
|
386
|
|
|
return false; |
387
|
|
|
} |
388
|
|
|
|
389
|
|
|
return $previousStatus; |
390
|
|
|
} |
391
|
|
|
|
392
|
|
|
/** |
393
|
|
|
* Writes to file after obtaining an exclusive lock. |
394
|
|
|
* |
395
|
|
|
* @param string $filename Filename in cache dir where data will be written |
396
|
|
|
* @param mixed $data Data to write |
397
|
|
|
* |
398
|
|
|
* @return int Number of bytes that were written to the file |
399
|
|
|
* |
400
|
|
|
* @throws \Exception Throws \Exception if unable to write file |
401
|
|
|
*/ |
402
|
|
|
protected function write($filename, $data) |
403
|
|
|
{ |
404
|
|
|
$data = trim($data); |
405
|
|
|
$filePath = $this->cacheDir . '/' . $filename; |
406
|
|
|
|
407
|
|
|
if (empty($data)) { |
408
|
|
|
throw new \Exception("No data to write into '{$filePath}'"); |
409
|
|
|
} |
410
|
|
|
|
411
|
|
|
// open with 'c' and truncate file only after obtaining a lock |
412
|
|
|
/** @noinspection PhpUsageOfSilenceOperatorInspection */ |
413
|
|
|
$fp = @fopen($filePath, 'c'); |
414
|
|
|
$result = $fp |
415
|
|
|
&& flock($fp, LOCK_EX) |
416
|
|
|
&& ftruncate($fp, 0) |
417
|
|
|
&& fwrite($fp, $data) !== false |
418
|
|
|
&& fflush($fp); |
419
|
|
|
|
420
|
|
|
if (!$result) { |
421
|
|
|
$fp && fclose($fp); |
422
|
|
|
throw new \Exception("Cannot write to '$filePath'"); |
423
|
|
|
} |
424
|
|
|
|
425
|
|
|
flock($fp, LOCK_UN); |
426
|
|
|
fclose($fp); |
427
|
|
|
|
428
|
|
|
return $result; |
429
|
|
|
} |
430
|
|
|
|
431
|
|
|
/** |
432
|
|
|
* Returns http adapter. Returns default http adapter if one is not set. |
433
|
|
|
* |
434
|
|
|
* @return \Pdp\HttpAdapter\HttpAdapterInterface Http adapter |
435
|
|
|
*/ |
436
|
|
|
public function getHttpAdapter() |
437
|
|
|
{ |
438
|
|
|
if (!$this->httpAdapter instanceof HttpAdapterInterface) { |
439
|
|
|
if (extension_loaded('curl')) { |
440
|
|
|
$this->httpAdapter = new HttpAdapter\CurlHttpAdapter(); |
441
|
|
|
} else { |
442
|
|
|
$this->httpAdapter = new HttpAdapter\PhpHttpAdapter(); |
443
|
|
|
} |
444
|
|
|
} |
445
|
|
|
|
446
|
|
|
return $this->httpAdapter; |
447
|
|
|
} |
448
|
|
|
|
449
|
|
|
/** |
450
|
|
|
* Sets http adapter. |
451
|
|
|
* |
452
|
|
|
* @param \Pdp\HttpAdapter\HttpAdapterInterface $httpAdapter |
453
|
|
|
*/ |
454
|
|
|
public function setHttpAdapter(HttpAdapter\HttpAdapterInterface $httpAdapter) |
455
|
|
|
{ |
456
|
|
|
$this->httpAdapter = $httpAdapter; |
457
|
|
|
} |
458
|
|
|
} |
459
|
|
|
|
If a method or function can return multiple different values and unless you are sure that you only can receive a single value in this context, we recommend to add an additional type check:
If this a common case that PHP Analyzer should handle natively, please let us know by opening an issue.