Parser::parseUrl() - Code Metrics - voku/php-domain-parser - Measure and Improve Code Quality continuously with Scrutinizer

Parser::parseUrl() B
last analyzed 2018-10-19 00:23 UTC

↳ Parent: Parser

Complexity

Conditions	5
Paths	6

Size

Total Lines

Duplication

Lines	0
Ratio	0 %

Code Coverage

Tests	21
CRAP Score	5

Importance

Changes

Metric	Value
cc	5
nc	6
nop	1
dl	0
loc	46
ccs	21
cts	21
cp	1
crap	5
rs	8.867
c	0
b	0
f	0

<?php

declare(strict_types=1);

/**
 * PHP Domain Parser: Public Suffix List based URL parsing.
 *
 * @link      http://github.com/jeremykendall/php-domain-parser for the canonical source repository
 *
 * @copyright Copyright (c) 2014 Jeremy Kendall (http://about.me/jeremykendall)
 * @license   http://github.com/jeremykendall/php-domain-parser/blob/master/LICENSE MIT License
 */

namespace Pdp;

use Pdp\Exception\SeriouslyMalformedUrlException;
use Pdp\Uri\Url;
use Pdp\Uri\Url\Host;
use voku\helper\UTF8;

/**
 * Parser.
 *
 * This class is responsible for Public Suffix List based url parsing
 */
class Parser
{
  /**
   * @var string RFC 3986 compliant scheme regex pattern
   *
   * @see https://tools.ietf.org/html/rfc3986#section-3.1
   */
  const SCHEME_PATTERN = '#^([a-zA-Z][a-zA-Z0-9+\-.]*)://#';

  /**
   * @var string IP address regex pattern
   */
  const IP_ADDRESS_PATTERN = '/^(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)$/';

  /**
   * @var PublicSuffixList Public Suffix List
   */
  protected $publicSuffixList;

  /**
   * @var bool Whether or not a host part has been normalized
   */
  protected $isNormalized = false;

  /**
   * @var PunycodeWrapper
   */
  private $punycodeWrapper;

  /**
   * Public constructor.
   *
   * @codeCoverageIgnore
   *
   * @param PublicSuffixList $publicSuffixList Instance of PublicSuffixList
   */
  public function __construct(PublicSuffixList $publicSuffixList)
  {
    $this->publicSuffixList = $publicSuffixList;
    $this->punycodeWrapper = new PunycodeWrapper();
  }

  /**
   * Parses url.
   *
   * @param string $url Url to parse
   *
   * @return Url Object representation of url
   */
  public function parseUrl($url): Url
  {
    $rawUrl = $url;
    $elem = [
        'scheme'   => null,
        'user'     => null,
        'pass'     => null,
        'host'     => null,
        'port'     => null,
        'path'     => null,
        'query'    => null,
        'fragment' => null,
    ];

    if (\preg_match(self::SCHEME_PATTERN, $url) === 0) {
      // Wacky scheme required to overcome parse_url behavior in PHP
      // See https://github.com/jeremykendall/php-domain-parser/issues/49
      $url = 'php-hack://' . \preg_replace('#^//#', '', $url, 1);
    }

    $parts = pdp_parse_url($url);

    if ($parts === false || !isset($parts['host'])) {
      throw new SeriouslyMalformedUrlException($rawUrl);
    }

    if ($parts['scheme'] === 'php-hack') {
      $parts['scheme'] = null;
    }

    /** @noinspection AdditionOperationOnArraysInspection */
    $elem = (array)$parts + $elem;

    $host = $this->parseHost($parts['host']);

    return new Url(
        $elem['scheme'],
        $elem['user'],
        $elem['pass'],
        $host,
        $elem['port'],
        $elem['path'],
        $elem['query'],
        $elem['fragment']
    );
  }

  /**
   * Parses host part of url.
   *
   * @param string $host Host part of url
   *
   * @return Host Object representation of host portion of url
   */
  public function parseHost(string $host): Host
  {
    $host = UTF8::strtolower($host);

    return new Host(
        $this->getSubdomain($host),
        $this->getRegistrableDomain($host),
        $this->getPublicSuffix($host),
        $host
    );
  }

  /**
   * Get the raw public suffix based on the cached public suffix list file.
   * Return false if the provided suffix is not included in the PSL.
   *
   * @param string $host The host to process
   *
   * @return string|false The suffix or false if suffix not included in the PSL
   */
  protected function getRawPublicSuffix(string $host)
  {
    $host = $this->normalize($host);

    $parts = \array_reverse(\explode('.', $host));
    $publicSuffix = [];
    $publicSuffixList = $this->publicSuffixList;

    foreach ($parts as $part) {
      if (
          \array_key_exists($part, $publicSuffixList)
          &&
          \array_key_exists('!', $publicSuffixList[$part])
      ) {
        break;
      }

      if (\array_key_exists($part, $publicSuffixList)) {
        \array_unshift($publicSuffix, $part);
        $publicSuffixList = $publicSuffixList[$part];
        continue;
      }

      if (\array_key_exists('*', $publicSuffixList)) {
        \array_unshift($publicSuffix, $part);
        $publicSuffixList = $publicSuffixList['*'];
        continue;
      }

      // Avoids improper parsing when $host's subdomain + public suffix ===
      // a valid public suffix (e.g. host 'us.example.com' and public suffix 'us.com')
      //
      // Added by @goodhabit in https://github.com/jeremykendall/php-domain-parser/pull/15
      // Resolves https://github.com/jeremykendall/php-domain-parser/issues/16
      break;
    }

    // If empty, then the suffix is not included in the PSL and is
    // considered "invalid". This also triggers algorithm rule #2: If no
    // rules match, the prevailing rule is "*".
    if (empty($publicSuffix)) {
      return false;
    }

    $suffix = \implode('.', \array_filter($publicSuffix, '\strlen'));

    return $this->denormalize($suffix);
  }

  /**
   * Returns the public suffix portion of provided host.
   *
   * @param string $host host
   *
   * @return string|null public suffix or null if host does not contain a public suffix
   */
  public function getPublicSuffix(string $host)
  {
    if (\strpos($host, '.') === 0) {
      return null;
    }

    // Fixes #22: If a single label domain makes it this far (e.g.,
    // localhost, foo, etc.), this stops it from incorrectly being set as
    // the public suffix.
    if (!$this->isMultiLabelDomain($host)) {
      return null;
    }

    // Fixes #43
    if ($this->isIpv4Address($host)) {
      return null;
    }

    $suffix = $this->getRawPublicSuffix($host);

    // Apply algorithm rule #2: If no rules match, the prevailing rule is "*".
    if (false === $suffix) {
      $parts = \array_reverse(explode('.', $host));
      $suffix = \array_shift($parts);
    }

    return $suffix;
  }

  /**
   * Is suffix valid?
   *
   * Validity determined by whether or not the suffix is included in the PSL.
   *
   * @param string $host Host part
   *
   * @return bool True is suffix is valid, false otherwise
   */
  public function isSuffixValid(string $host): bool
  {
    return $this->getRawPublicSuffix($host) !== false;
  }

  /**
   * Returns registrable domain portion of provided host.
   *
   * Per the test cases provided by Mozilla
   * (http://mxr.mozilla.org/mozilla-central/source/netwerk/test/unit/data/test_psl.txt?raw=1),
   * this method should return null if the domain provided is a public suffix.
   *
   * @param string|null $host host
   *
   * @return string|null registrable domain
   */
  public function getRegistrableDomain($host)
  {
    if (!$this->isMultiLabelDomain($host)) {
      return null;
    }

    $publicSuffix = $this->getPublicSuffix($host);

    if ($publicSuffix === null || $host == $publicSuffix) {
      return null;
    }

    $publicSuffixParts = \array_reverse(\explode('.', $publicSuffix));
    $hostParts = \array_reverse(\explode('.', $host));
    $registrableDomainParts = $publicSuffixParts + \array_slice($hostParts, 0, \count($publicSuffixParts) + 1);

    return \implode('.', \array_reverse($registrableDomainParts));
  }

  /**
   * Returns the subdomain portion of provided host.
   *
   * @param string $host host
   *
   * @return string|null subdomain
   */
  public function getSubdomain(string $host)
  {
    $registrableDomain = $this->getRegistrableDomain($host);

    if ($registrableDomain === null || $host === $registrableDomain) {
      return null;
    }

    $registrableDomainParts = \array_reverse(\explode('.', $registrableDomain));

    $host = $this->normalize($host);

    $hostParts = \array_reverse(\explode('.', $host));
    $subdomainParts = \array_slice($hostParts, \count($registrableDomainParts));

    $subdomain = \implode('.', \array_reverse($subdomainParts));

    return $this->denormalize($subdomain);
  }

  /**
   * If a URL is not punycoded, then it may be an IDNA URL, so it must be
   * converted to ASCII. Performs conversion and sets flag.
   *
   * @param string $part Host part
   *
   * @return string Host part, transformed if not punycoded
   */
  protected function normalize(string $part): string
  {
    $punycoded = (\strpos($part, 'xn--') !== false);

    if ($punycoded === false) {
      $part = $this->punycodeWrapper->encode($part);
      $this->isNormalized = true;
    }

    return \strtolower($part);
  }

  /**
   * Converts any normalized part back to IDNA. Performs conversion and
   * resets flag.
   *
   * @param string $part Host part
   *
   * @return string Denormalized host part
   */
  protected function denormalize(string $part): string
  {
    if ($this->isNormalized === true) {
      $part = $this->punycodeWrapper->decode($part);
      $this->isNormalized = false;
    }

    return $part;
  }

  /**
   * Tests host for presence of '.'.
   *
   * Related to #22
   *
   * @param string|null $host Host part of url
   *
   * @return bool True if multi-label domain, false otherwise
   */
  protected function isMultiLabelDomain($host): bool
  {
    if (!$host) {
''   == false // true
''   == null  // true
'ab' == false // false
'ab' == null  // false

// It is often better to use strict comparison
'' === false // false
'' === null  // false
      return false;
    }

    return \strpos($host, '.') !== false;
  }

  /**
   * Tests host to determine if it is an IP address.
   *
   * Related to #43
   *
   * @param string $host Host part of url
   *
   * @return bool True if host is an ip address, false otherwise
   */
  protected function isIpv4Address(string $host): bool
  {
    return \preg_match(self::IP_ADDRESS_PATTERN, $host) === 1;
  }
}


1		<?php
2
3		declare(strict_types=1);
4
5		/**
6		* PHP Domain Parser: Public Suffix List based URL parsing.
7		*
8		* @link http://github.com/jeremykendall/php-domain-parser for the canonical source repository
9		*
10		* @copyright Copyright (c) 2014 Jeremy Kendall (http://about.me/jeremykendall)
11		* @license http://github.com/jeremykendall/php-domain-parser/blob/master/LICENSE MIT License
12		*/
13
14		namespace Pdp;
15
16		use Pdp\Exception\SeriouslyMalformedUrlException;
17		use Pdp\Uri\Url;
18		use Pdp\Uri\Url\Host;
19		use voku\helper\UTF8;
20
21		/**
22		* Parser.
23		*
24		* This class is responsible for Public Suffix List based url parsing
25		*/
26		class Parser
27		{
28		/**
29		* @var string RFC 3986 compliant scheme regex pattern
30		*
31		* @see https://tools.ietf.org/html/rfc3986#section-3.1
32		*/
33		const SCHEME_PATTERN = '#^([a-zA-Z][a-zA-Z0-9+\-.]*)://#';
34
35		/**
36		* @var string IP address regex pattern
37		*/
38		const IP_ADDRESS_PATTERN = '/^(?:(?:25[0-5]\|2[0-4][0-9]\|[01]?[0-9][0-9]?)\.){3}(?:25[0-5]\|2[0-4][0-9]\|[01]?[0-9][0-9]?)$/';
39
40		/**
41		* @var PublicSuffixList Public Suffix List
42		*/
43		protected $publicSuffixList;
44
45		/**
46		* @var bool Whether or not a host part has been normalized
47		*/
48		protected $isNormalized = false;
49
50		/**
51		* @var PunycodeWrapper
52		*/
53		private $punycodeWrapper;
54
55		/**
56		* Public constructor.
57		*
58		* @codeCoverageIgnore
59		*
60		* @param PublicSuffixList $publicSuffixList Instance of PublicSuffixList
61		*/
62		public function __construct(PublicSuffixList $publicSuffixList)
63		{
64		$this->publicSuffixList = $publicSuffixList;
65		$this->punycodeWrapper = new PunycodeWrapper();
66		}
67
68		/**
69		* Parses url.
70		*
71		* @param string $url Url to parse
72		*
73		* @return Url Object representation of url
74		*/
75	233	public function parseUrl($url): Url
76		{
77	233	$rawUrl = $url;
78		$elem = [
79	233	'scheme' => null,
80		'user' => null,
81		'pass' => null,
82		'host' => null,
83		'port' => null,
84		'path' => null,
85		'query' => null,
86		'fragment' => null,
87		];
88
89	233	if (\preg_match(self::SCHEME_PATTERN, $url) === 0) {
90		// Wacky scheme required to overcome parse_url behavior in PHP
91		// See https://github.com/jeremykendall/php-domain-parser/issues/49
92	129	$url = 'php-hack://' . \preg_replace('#^//#', '', $url, 1);
93		}
94
95	233	$parts = pdp_parse_url($url);
96
97	233	if ($parts === false \|\| !isset($parts['host'])) {
98	3	throw new SeriouslyMalformedUrlException($rawUrl);
99		}
100
101	230	if ($parts['scheme'] === 'php-hack') {
102	128	$parts['scheme'] = null;
103		}
104
105		/** @noinspection AdditionOperationOnArraysInspection */
106	230	$elem = (array)$parts + $elem;
107
108	230	$host = $this->parseHost($parts['host']);
109
110	230	return new Url(
111	230	$elem['scheme'],
112	230	$elem['user'],
113	230	$elem['pass'],
114	230	$host,
115	230	$elem['port'],
116	230	$elem['path'],
117	230	$elem['query'],
118	230	$elem['fragment']
119		);
120		}
121
122		/**
123		* Parses host part of url.
124		*
125		* @param string $host Host part of url
126		*
127		* @return Host Object representation of host portion of url
128		*/
129	54	public function parseHost(string $host): Host
130		{
131	54	$host = UTF8::strtolower($host);
132
133	54	return new Host(
134	54	$this->getSubdomain($host),
135	54	$this->getRegistrableDomain($host),
136	54	$this->getPublicSuffix($host),
137	54	$host
138		);
139		}
140
141		/**
142		* Get the raw public suffix based on the cached public suffix list file.
143		* Return false if the provided suffix is not included in the PSL.
144		*
145		* @param string $host The host to process
146		*
147		* @return string\|false The suffix or false if suffix not included in the PSL
148		*/
149	11	protected function getRawPublicSuffix(string $host)
150		{
151	11	$host = $this->normalize($host);
152
153	11	$parts = \array_reverse(\explode('.', $host));
154	11	$publicSuffix = [];
155	11	$publicSuffixList = $this->publicSuffixList;
156
157	11	foreach ($parts as $part) {
158		if (
159	11	\array_key_exists($part, $publicSuffixList)
160		&&
161	11	\array_key_exists('!', $publicSuffixList[$part])
162		) {
163	1	break;
164		}
165
166	11	if (\array_key_exists($part, $publicSuffixList)) {
167	11	\array_unshift($publicSuffix, $part);
168	11	$publicSuffixList = $publicSuffixList[$part];
169	11	continue;
170		}
171
172	11	if (\array_key_exists('*', $publicSuffixList)) {
173	1	\array_unshift($publicSuffix, $part);
174	1	$publicSuffixList = $publicSuffixList['*'];
175	1	continue;
176		}
177
178		// Avoids improper parsing when $host's subdomain + public suffix ===
179		// a valid public suffix (e.g. host 'us.example.com' and public suffix 'us.com')
180		//
181		// Added by @goodhabit in https://github.com/jeremykendall/php-domain-parser/pull/15
182		// Resolves https://github.com/jeremykendall/php-domain-parser/issues/16
183	11	break;
184		}
185
186		// If empty, then the suffix is not included in the PSL and is
187		// considered "invalid". This also triggers algorithm rule #2: If no
188		// rules match, the prevailing rule is "*".
189	11	if (empty($publicSuffix)) {
190	1	return false;
191		}
192
193	11	$suffix = \implode('.', \array_filter($publicSuffix, '\strlen'));
194
195	11	return $this->denormalize($suffix);
196		}
197
198		/**
199		* Returns the public suffix portion of provided host.
200		*
201		* @param string $host host
202		*
203		* @return string\|null public suffix or null if host does not contain a public suffix
204		*/
205	56	public function getPublicSuffix(string $host)
206		{
207	56	if (\strpos($host, '.') === 0) {
208	1	return null;
209		}
210
211		// Fixes #22: If a single label domain makes it this far (e.g.,
212		// localhost, foo, etc.), this stops it from incorrectly being set as
213		// the public suffix.
214	56	if (!$this->isMultiLabelDomain($host)) {
215	6	return null;
216		}
217
218		// Fixes #43
219	50	if ($this->isIpv4Address($host)) {
220	3	return null;
221		}
222
223	47	$suffix = $this->getRawPublicSuffix($host);
224
225		// Apply algorithm rule #2: If no rules match, the prevailing rule is "*".
226	47	if (false === $suffix) {
227	3	$parts = \array_reverse(explode('.', $host));
228	3	$suffix = \array_shift($parts);
229		}
230
231	47	return $suffix;
232		}
233
234		/**
235		* Is suffix valid?
236		*
237		* Validity determined by whether or not the suffix is included in the PSL.
238		*
239		* @param string $host Host part
240		*
241		* @return bool True is suffix is valid, false otherwise
242		*/
243	2	public function isSuffixValid(string $host): bool
244		{
245	2	return $this->getRawPublicSuffix($host) !== false;
246		}
247
248		/**
249		* Returns registrable domain portion of provided host.
250		*
251		* Per the test cases provided by Mozilla
252		* (http://mxr.mozilla.org/mozilla-central/source/netwerk/test/unit/data/test_psl.txt?raw=1),
253		* this method should return null if the domain provided is a public suffix.
254		*
255		* @param string\|null $host host
256		*
257		* @return string\|null registrable domain
258		*/
259	55	public function getRegistrableDomain($host)
260		{
261	55	if (!$this->isMultiLabelDomain($host)) {
262	7	return null;
263		}
264
265	49	$publicSuffix = $this->getPublicSuffix($host);
266
267	49	if ($publicSuffix === null \|\| $host == $publicSuffix) {
268	4	return null;
269		}
270
271	46	$publicSuffixParts = \array_reverse(\explode('.', $publicSuffix));
272	46	$hostParts = \array_reverse(\explode('.', $host));
273	46	$registrableDomainParts = $publicSuffixParts + \array_slice($hostParts, 0, \count($publicSuffixParts) + 1);
274
275	46	return \implode('.', \array_reverse($registrableDomainParts));
276		}
277
278		/**
279		* Returns the subdomain portion of provided host.
280		*
281		* @param string $host host
282		*
283		* @return string\|null subdomain
284		*/
285	55	public function getSubdomain(string $host)
286		{
287	55	$registrableDomain = $this->getRegistrableDomain($host);
288
289	55	if ($registrableDomain === null \|\| $host === $registrableDomain) {
290	30	return null;
291		}
292
293	28	$registrableDomainParts = \array_reverse(\explode('.', $registrableDomain));
294
295	28	$host = $this->normalize($host);
296
297	28	$hostParts = \array_reverse(\explode('.', $host));
298	28	$subdomainParts = \array_slice($hostParts, \count($registrableDomainParts));
299
300	28	$subdomain = \implode('.', \array_reverse($subdomainParts));
301
302	28	return $this->denormalize($subdomain);
303		}
304
305		/**
306		* If a URL is not punycoded, then it may be an IDNA URL, so it must be
307		* converted to ASCII. Performs conversion and sets flag.
308		*
309		* @param string $part Host part
310		*
311		* @return string Host part, transformed if not punycoded
312		*/
313	11	protected function normalize(string $part): string
314		{
315	11	$punycoded = (\strpos($part, 'xn--') !== false);
316
317	11	if ($punycoded === false) {
318	11	$part = $this->punycodeWrapper->encode($part);
319	11	$this->isNormalized = true;
320		}
321
322	11	return \strtolower($part);
323		}
324
325		/**
326		* Converts any normalized part back to IDNA. Performs conversion and
327		* resets flag.
328		*
329		* @param string $part Host part
330		*
331		* @return string Denormalized host part
332		*/
333	11	protected function denormalize(string $part): string
334		{
335	11	if ($this->isNormalized === true) {
336	11	$part = $this->punycodeWrapper->decode($part);
337	11	$this->isNormalized = false;
338		}
339
340	11	return $part;
341		}
342
343		/**
344		* Tests host for presence of '.'.
345		*
346		* Related to #22
347		*
348		* @param string\|null $host Host part of url
349		*
350		* @return bool True if multi-label domain, false otherwise
351		*/
352	11	protected function isMultiLabelDomain($host): bool
353		{
354	11	if (!$host) {
		0 ignored issues – show Bug Best Practice introduced 2018-03-19 12:58 UTC by Report Bug Copy Issue Report The expression `$host` of type `string\|null` is loosely compared to `false`; this is ambiguous if the string can be empty. You might want to explicitly use `=== null` instead. In PHP, under loose comparison (like `==`, or `!=`, or `switch` conditions), values of different types might be equal. For `string` values, the empty string `''` is a special case, in particular the following results might be unexpected: '' == false // true '' == null // true 'ab' == false // false 'ab' == null // false // It is often better to use strict comparison '' === false // false '' === null // false Loading history...
355	1	return false;
356		}
357
358	11	return \strpos($host, '.') !== false;
359		}
360
361		/**
362		* Tests host to determine if it is an IP address.
363		*
364		* Related to #43
365		*
366		* @param string $host Host part of url
367		*
368		* @return bool True if host is an ip address, false otherwise
369		*/
370	11	protected function isIpv4Address(string $host): bool
371		{
372	11	return \preg_match(self::IP_ADDRESS_PATTERN, $host) === 1;
373		}
374		}
375

voku / php-domain-parser

Parser::parseUrl() B last analyzed 2018-10-19 00:23 UTC

Complexity

Size

Duplication

Code Coverage

Importance

Duplication Side-by-Side

Filter issues like

Parser::parseUrl() B
last analyzed 2018-10-19 00:23 UTC