HttpUtil::parseHeaders() - Code Metrics - Inspection of "Fix error logged on some internet radio stations" - owncloud/music - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 1fcd2e...458bc4 )

by Pauli

created 2025-03-27 21:12 UTC

HttpUtil::parseHeaders() B

↳ Parent: HttpUtil

Complexity

Conditions	7
Paths	6

Size

Total Lines	31
Code Lines	18

Duplication

Lines	0
Ratio	0 %

Importance

Changes	2
Bugs	0	Features	0

Metric	Value
cc	7
eloc	18
c	2
b	0
f	0
nc	6
nop	2
dl	0
loc	31
rs	8.8333

<?php declare(strict_types=1);

/**
 * ownCloud - Music app
 *
 * This file is licensed under the Affero General Public License version 3 or
 * later. See the COPYING file.
 *
 * @author Pauli Järvinen <[email protected]>
 * @copyright Pauli Järvinen 2022 - 2025
 */

namespace OCA\Music\Utility;

use OCP\AppFramework\Http;
use OCP\AppFramework\Http\Response;

/**
 * Static utility functions to work with HTTP requests
 */
class HttpUtil {

	private const ALLOWED_SCHEMES = ['http', 'https', 'feed', 'podcast', 'pcast', 'podcasts', 'itms-pcast', 'itms-pcasts', 'itms-podcast', 'itms-podcasts'];

	/**
	 * Use HTTP GET to load the requested URL
	 * @return array with three keys: ['content' => string|false, 'status_code' => int, 'message' => string, 'content_type' => string]
	 */
	public static function loadFromUrl(string $url, ?int $maxLength=null, ?int $timeout_s=null) : array {
		$status_code = 0;
		$content_type = null;

		if (!self::isUrlSchemeOneOf($url, self::ALLOWED_SCHEMES)) {
			$content = false;
			$message = 'URL scheme must be one of ' . \json_encode(self::ALLOWED_SCHEMES);
		} else {
			$context = self::createContext($timeout_s);

			// The length parameter of file_get_contents isn't nullable prior to PHP8.0
			if ($maxLength === null) {
				$content = @\file_get_contents($url, false, $context);
			} else {
				$content = @\file_get_contents($url, false, $context, 0, $maxLength);
			}

			// It's some PHP magic that calling file_get_contents creates and populates also a local
			// variable array $http_response_header, provided that the server could be reached.
			if (!empty($http_response_header)) {
				$parsedHeaders = self::parseHeaders($http_response_header, true);
				$status_code = $parsedHeaders['status_code'];
				$message = $parsedHeaders['status_msg'];
				$content_type = $parsedHeaders['content-type'] ?? null;
			} else {
				$message = 'The requested URL did not respond';
			}
		}

		return \compact('content', 'status_code', 'message', 'content_type');
	}

	/**
	 * @return resource
	 */
	public static function createContext(?int $timeout_s = null, array $extraHeaders = [], ?int $maxRedirects = null) {
		$opts = self::contextOptions($extraHeaders);
		if ($timeout_s !== null) {
			$opts['http']['timeout'] = $timeout_s;
		}
		if ($maxRedirects !== null) {
			$opts['http']['max_redirects'] = $maxRedirects;
		}
		return \stream_context_create($opts);
	}

	/**
	 * @param resource $context
	 * @param bool $convertKeysToLower When true, the header names used as keys of the result array are
	 * 				converted to lower case. According to RFC 2616, HTTP headers are case-insensitive.
	 * @return array The headers from the URL, after any redirections. The header names will be array keys.
	 * 					In addition to the named headers from the server, the key 'status_code' will contain
	 * 					the status code number of the HTTP request (like 200, 302, 404) and 'status_msg'
	 * 					the textual status following the code (like 'OK' or 'Not Found').
	 */
	public static function getUrlHeaders(string $url, $context, bool $convertKeysToLower=false) : array {
		$result = null;
		if (self::isUrlSchemeOneOf($url, self::ALLOWED_SCHEMES)) {
			// Don't use the built-in associative mode of get_headers because it mixes up the headers from the redirection
			// responses with those of the last response after all the redirections, making it impossible to know,
			// what is the source of each header. Hence, we roll out our own parsing logic which discards all the
			// headers from the intermediate redirection responses.

			// the type of the second parameter of get_header has changed in PHP 8.0
			$associative = \version_compare(\phpversion(), '8.0', '<') ? 0 : false;
			$rawHeaders = @\get_headers($url, /** @scrutinizer ignore-type */ $associative, $context);

			if ($rawHeaders !== false) {
				$result = self::parseHeaders($rawHeaders, $convertKeysToLower);
			} else {
				$result = ['status_code' => Http::STATUS_SERVICE_UNAVAILABLE, 'status_msg' => 'Error connecting the URL', 'content-length' => '0'];
			}
		} else {
			$result = ['status_code' => Http::STATUS_FORBIDDEN, 'status_msg' => 'URL scheme not allowed', 'content-length' => '0'];
		}
		return $result;
	}

	private static function parseHeaders(array $rawHeaders, bool $convertKeysToLower) : array {
		$result = [];

		foreach ($rawHeaders as $row) {
			// The response usually starts with a header like "HTTP/1.1 200 OK". However, some shoutcast streams
			// may instead use "ICY 200 OK".
			if (Util::startsWith($row, 'HTTP/', /*ignoreCase=*/true) || Util::startsWith($row, 'ICY ', /*ignoreCase=*/true)) {
				// Start of new response. If we have already parsed some headers, then those are from some
				// intermediate redirect response and those should be discarded.
				$parts = \explode(' ', $row, 3);
				if (\count($parts) == 3) {
					list(, $status_code, $status_msg) = $parts;
				} else {
					$status_code = Http::STATUS_INTERNAL_SERVER_ERROR;
					$status_msg = 'Bad response status header';
				}
				$result = ['status_code' => (int)$status_code, 'status_msg' => $status_msg];
			} else {
				// All other lines besides the initial status line should have the format "key: value"
				$parts = \explode(':', $row, 2);
				if (\count($parts) == 2) {
					list($key, $value) = $parts;
					if ($convertKeysToLower) {
						$key = \mb_strtolower($key);
					}
					$result[\trim($key)] = \trim($value);
				}
			}
		}

		return $result;
	}

	public static function userAgentHeader() : string {
		return 'User-Agent: OCMusic/' . AppInfo::getVersion();
	}

	private static function contextOptions(array $extraHeaders = []) : array {
		$opts = [
			'http' => [
				'header' => self::userAgentHeader(),	// some servers don't allow requests without a user agent header
				'ignore_errors' => true,				// don't emit warnings for bad/unavailable URL, we handle errors manually
				'max_redirects' => 20
			]
		];

		foreach ($extraHeaders as $key => $value) {
			$opts['http']['header'] .= "\r\n$key: $value";
		}

		return $opts;
	}

	private static function isUrlSchemeOneOf(string $url, array $schemes) : bool {
		$url = \mb_strtolower($url);

		foreach ($schemes as $scheme) {
			if (Util::startsWith($url, $scheme . '://')) {
				return true;
			}
		}

		return false;
	}

	public static function setClientCachingDays(Response &$httpResponse, int $days) : void {
		$httpResponse->cacheFor($days * 24 * 60 * 60);
		$httpResponse->addHeader('Pragma', 'cache');
	}
}


1			<?php declare(strict_types=1);
2
3			/**
4			* ownCloud - Music app
5			*
6			* This file is licensed under the Affero General Public License version 3 or
7			* later. See the COPYING file.
8			*
9			* @author Pauli Järvinen <[email protected]>
10			* @copyright Pauli Järvinen 2022 - 2025
11			*/
12
13			namespace OCA\Music\Utility;
14
15			use OCP\AppFramework\Http;
16			use OCP\AppFramework\Http\Response;
17
18			/**
19			* Static utility functions to work with HTTP requests
20			*/
21			class HttpUtil {
22
23			private const ALLOWED_SCHEMES = ['http', 'https', 'feed', 'podcast', 'pcast', 'podcasts', 'itms-pcast', 'itms-pcasts', 'itms-podcast', 'itms-podcasts'];
24
25			/**
26			* Use HTTP GET to load the requested URL
27			* @return array with three keys: ['content' => string\|false, 'status_code' => int, 'message' => string, 'content_type' => string]
28			*/
29			public static function loadFromUrl(string $url, ?int $maxLength=null, ?int $timeout_s=null) : array {
30			$status_code = 0;
31			$content_type = null;
32
33			if (!self::isUrlSchemeOneOf($url, self::ALLOWED_SCHEMES)) {
34			$content = false;
35			$message = 'URL scheme must be one of ' . \json_encode(self::ALLOWED_SCHEMES);
36			} else {
37			$context = self::createContext($timeout_s);
38
39			// The length parameter of file_get_contents isn't nullable prior to PHP8.0
40			if ($maxLength === null) {
41			$content = @\file_get_contents($url, false, $context);
42			} else {
43			$content = @\file_get_contents($url, false, $context, 0, $maxLength);
44			}
45
46			// It's some PHP magic that calling file_get_contents creates and populates also a local
47			// variable array $http_response_header, provided that the server could be reached.
48			if (!empty($http_response_header)) {
49			$parsedHeaders = self::parseHeaders($http_response_header, true);
50			$status_code = $parsedHeaders['status_code'];
51			$message = $parsedHeaders['status_msg'];
52			$content_type = $parsedHeaders['content-type'] ?? null;
53			} else {
54			$message = 'The requested URL did not respond';
55			}
56			}
57
58			return \compact('content', 'status_code', 'message', 'content_type');
59			}
60
61			/**
62			* @return resource
63			*/
64			public static function createContext(?int $timeout_s = null, array $extraHeaders = [], ?int $maxRedirects = null) {
65			$opts = self::contextOptions($extraHeaders);
66			if ($timeout_s !== null) {
67			$opts['http']['timeout'] = $timeout_s;
68			}
69			if ($maxRedirects !== null) {
70			$opts['http']['max_redirects'] = $maxRedirects;
71			}
72			return \stream_context_create($opts);
73			}
74
75			/**
76			* @param resource $context
77			* @param bool $convertKeysToLower When true, the header names used as keys of the result array are
78			* converted to lower case. According to RFC 2616, HTTP headers are case-insensitive.
79			* @return array The headers from the URL, after any redirections. The header names will be array keys.
80			* In addition to the named headers from the server, the key 'status_code' will contain
81			* the status code number of the HTTP request (like 200, 302, 404) and 'status_msg'
82			* the textual status following the code (like 'OK' or 'Not Found').
83			*/
84			public static function getUrlHeaders(string $url, $context, bool $convertKeysToLower=false) : array {
85			$result = null;
86			if (self::isUrlSchemeOneOf($url, self::ALLOWED_SCHEMES)) {
87			// Don't use the built-in associative mode of get_headers because it mixes up the headers from the redirection
88			// responses with those of the last response after all the redirections, making it impossible to know,
89			// what is the source of each header. Hence, we roll out our own parsing logic which discards all the
90			// headers from the intermediate redirection responses.
91
92			// the type of the second parameter of get_header has changed in PHP 8.0
93			$associative = \version_compare(\phpversion(), '8.0', '<') ? 0 : false;
94			$rawHeaders = @\get_headers($url, /** @scrutinizer ignore-type */ $associative, $context);
95
96			if ($rawHeaders !== false) {
97			$result = self::parseHeaders($rawHeaders, $convertKeysToLower);
98			} else {
99			$result = ['status_code' => Http::STATUS_SERVICE_UNAVAILABLE, 'status_msg' => 'Error connecting the URL', 'content-length' => '0'];
100			}
101			} else {
102			$result = ['status_code' => Http::STATUS_FORBIDDEN, 'status_msg' => 'URL scheme not allowed', 'content-length' => '0'];
103			}
104			return $result;
105			}
106
107			private static function parseHeaders(array $rawHeaders, bool $convertKeysToLower) : array {
108			$result = [];
109
110			foreach ($rawHeaders as $row) {
111			// The response usually starts with a header like "HTTP/1.1 200 OK". However, some shoutcast streams
112			// may instead use "ICY 200 OK".
113			if (Util::startsWith($row, 'HTTP/', /ignoreCase=/true) \|\| Util::startsWith($row, 'ICY ', /ignoreCase=/true)) {
114			// Start of new response. If we have already parsed some headers, then those are from some
115			// intermediate redirect response and those should be discarded.
116			$parts = \explode(' ', $row, 3);
117			if (\count($parts) == 3) {
118			list(, $status_code, $status_msg) = $parts;
119			} else {
120			$status_code = Http::STATUS_INTERNAL_SERVER_ERROR;
121			$status_msg = 'Bad response status header';
122			}
123			$result = ['status_code' => (int)$status_code, 'status_msg' => $status_msg];
124			} else {
125			// All other lines besides the initial status line should have the format "key: value"
126			$parts = \explode(':', $row, 2);
127			if (\count($parts) == 2) {
128			list($key, $value) = $parts;
129			if ($convertKeysToLower) {
130			$key = \mb_strtolower($key);
131			}
132			$result[\trim($key)] = \trim($value);
133			}
134			}
135			}
136
137			return $result;
138			}
139
140			public static function userAgentHeader() : string {
141			return 'User-Agent: OCMusic/' . AppInfo::getVersion();
142			}
143
144			private static function contextOptions(array $extraHeaders = []) : array {
145			$opts = [
146			'http' => [
147			'header' => self::userAgentHeader(), // some servers don't allow requests without a user agent header
148			'ignore_errors' => true, // don't emit warnings for bad/unavailable URL, we handle errors manually
149			'max_redirects' => 20
150			]
151			];
152
153			foreach ($extraHeaders as $key => $value) {
154			$opts['http']['header'] .= "\r\n$key: $value";
155			}
156
157			return $opts;
158			}
159
160			private static function isUrlSchemeOneOf(string $url, array $schemes) : bool {
161			$url = \mb_strtolower($url);
162
163			foreach ($schemes as $scheme) {
164			if (Util::startsWith($url, $scheme . '://')) {
165			return true;
166			}
167			}
168
169			return false;
170			}
171
172			public static function setClientCachingDays(Response &$httpResponse, int $days) : void {
173			$httpResponse->cacheFor($days * 24 * 60 * 60);
174			$httpResponse->addHeader('Pragma', 'cache');
175			}
176			}
177

owncloud / music

Push — master ( 1fcd2e...458bc4 )

HttpUtil::parseHeaders() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like