6 * GET an HTTP URL to retrieve its content
7 * Uses the cURL library or a fallback method
9 * @param string $url URL to get (http://...)
10 * @param int $timeout network timeout (in seconds)
11 * @param int $maxBytes maximum downloaded bytes (default: 4 MiB)
12 * @param callable|string $curlWriteFunction Optional callback called during the download (cURL CURLOPT_WRITEFUNCTION).
13 * Can be used to add download conditions on the
14 * headers (response code, content type, etc.).
16 * @return array HTTP response headers, downloaded content
19 * [0] = associative array containing HTTP response headers
20 * [1] = URL content (downloaded data)
23 * list($headers, $data) = get_http_response('http://sebauvage.net/');
24 * if (strpos($headers[0], '200 OK') !== false) {
25 * echo 'Data type: '.htmlspecialchars($headers['Content-Type']);
27 * echo 'There was an error: '.htmlspecialchars($headers[0]);
30 * @see https://secure.php.net/manual/en/ref.curl.php
31 * @see https://secure.php.net/manual/en/functions.anonymous.php
32 * @see https://secure.php.net/manual/en/function.preg-split.php
33 * @see https://secure.php.net/manual/en/function.explode.php
34 * @see http://stackoverflow.com/q/17641073
35 * @see http://stackoverflow.com/q/9183178
36 * @see http://stackoverflow.com/q/1462720
38 function get_http_response($url, $timeout = 30, $maxBytes = 4194304, $curlWriteFunction = null)
40 $urlObj = new Url($url);
41 $cleanUrl = $urlObj->idnToAscii();
43 if (!filter_var($cleanUrl, FILTER_VALIDATE_URL
) || !$urlObj->isHttp()) {
44 return array(array(0 => 'Invalid HTTP UrlUtils'), false);
48 'Mozilla/5.0 (X11; Fedora; Linux x86_64; rv:45.0)'
49 . ' Gecko/20100101 Firefox/45.0';
51 substr(setlocale(LC_COLLATE
, 0), 0, 2) . ',en-US;q=0.7,en;q=0.3';
54 if (!function_exists('curl_init')) {
55 return get_http_response_fallback(
65 $ch = curl_init($cleanUrl);
67 return array(array(0 => 'curl_init() error'), false);
70 // General cURL settings
71 curl_setopt($ch, CURLOPT_AUTOREFERER
, true);
72 curl_setopt($ch, CURLOPT_FOLLOWLOCATION
, true);
73 curl_setopt($ch, CURLOPT_HEADER
, true);
77 array('Accept-Language: ' . $acceptLanguage)
79 curl_setopt($ch, CURLOPT_MAXREDIRS
, $maxRedirs);
80 curl_setopt($ch, CURLOPT_RETURNTRANSFER
, true);
81 curl_setopt($ch, CURLOPT_TIMEOUT
, $timeout);
82 curl_setopt($ch, CURLOPT_USERAGENT
, $userAgent);
84 if (is_callable($curlWriteFunction)) {
85 curl_setopt($ch, CURLOPT_WRITEFUNCTION
, $curlWriteFunction);
88 // Max download size management
89 curl_setopt($ch, CURLOPT_BUFFERSIZE
, 1024*16);
90 curl_setopt($ch, CURLOPT_NOPROGRESS
, false);
93 CURLOPT_PROGRESSFUNCTION
,
94 function ($arg0, $arg1, $arg2, $arg3, $arg4 = 0) use ($maxBytes) {
95 if (version_compare(phpversion(), '5.5', '<')) {
96 // PHP version lower than 5.5
97 // Callback has 4 arguments
100 // Callback has 5 arguments
103 // Non-zero return stops downloading
104 return ($downloaded > $maxBytes) ? 1 : 0;
108 $response = curl_exec($ch);
109 $errorNo = curl_errno($ch);
110 $errorStr = curl_error($ch);
111 $headSize = curl_getinfo($ch, CURLINFO_HEADER_SIZE
);
114 if ($response === false) {
115 if ($errorNo == CURLE_COULDNT_RESOLVE_HOST
) {
117 * Workaround to match fallback method behaviour
118 * Removing this would require updating
119 * GetHttpUrlTest::testGetInvalidRemoteUrl()
121 return array(false, false);
123 return array(array(0 => 'curl_exec() error: ' . $errorStr), false);
126 // Formatting output like the fallback method
127 $rawHeaders = substr($response, 0, $headSize);
129 // Keep only headers from latest redirection
130 $rawHeadersArrayRedirs = explode("\r\n\r\n", trim($rawHeaders));
131 $rawHeadersLastRedir = end($rawHeadersArrayRedirs);
133 $content = substr($response, $headSize);
135 foreach (preg_split('~[\r\n]+~', $rawHeadersLastRedir) as $line) {
136 if (empty($line) || ctype_space($line)) {
139 $splitLine = explode(': ', $line, 2);
140 if (count($splitLine) > 1) {
141 $key = $splitLine[0];
142 $value = $splitLine[1];
143 if (array_key_exists($key, $headers)) {
144 if (!is_array($headers[$key])) {
145 $headers[$key] = array(0 => $headers[$key]);
147 $headers[$key][] = $value;
149 $headers[$key] = $value;
152 $headers[] = $splitLine[0];
156 return array($headers, $content);
160 * GET an HTTP URL to retrieve its content (fallback method)
162 * @param string $cleanUrl URL to get (http://... valid and in ASCII form)
163 * @param int $timeout network timeout (in seconds)
164 * @param int $maxBytes maximum downloaded bytes
165 * @param string $userAgent "User-Agent" header
166 * @param string $acceptLanguage "Accept-Language" header
167 * @param int $maxRedr maximum amount of redirections followed
169 * @return array HTTP response headers, downloaded content
172 * [0] = associative array containing HTTP response headers
173 * [1] = URL content (downloaded data)
175 * @see http://php.net/manual/en/function.file-get-contents.php
176 * @see http://php.net/manual/en/function.stream-context-create.php
177 * @see http://php.net/manual/en/function.get-headers.php
179 function get_http_response_fallback(
190 'timeout' => $timeout,
191 'user_agent' => $userAgent,
192 'header' => "Accept: */*\r\n"
193 . 'Accept-Language: ' . $acceptLanguage
197 stream_context_set_default($options);
198 list($headers, $finalUrl) = get_redirected_headers($cleanUrl, $maxRedr);
199 if (! $headers || strpos($headers[0], '200 OK') === false) {
200 $options['http']['request_fulluri'] = true;
201 stream_context_set_default($options);
202 list($headers, $finalUrl) = get_redirected_headers($cleanUrl, $maxRedr);
206 return array($headers, false);
210 // TODO: catch Exception in calling code (thumbnailer)
211 $context = stream_context_create($options);
212 $content = file_get_contents($finalUrl, false, $context, -1, $maxBytes);
213 } catch (Exception
$exc) {
214 return array(array(0 => 'HTTP Error'), $exc->getMessage());
217 return array($headers, $content);
221 * Retrieve HTTP headers, following n redirections (temporary and permanent ones).
223 * @param string $url initial URL to reach.
224 * @param int $redirectionLimit max redirection follow.
226 * @return array HTTP headers, or false if it failed.
228 function get_redirected_headers($url, $redirectionLimit = 3)
230 $headers = get_headers($url, 1);
231 if (!empty($headers['location']) && empty($headers['Location'])) {
232 $headers['Location'] = $headers['location'];
235 // Headers found, redirection found, and limit not reached.
236 if ($redirectionLimit-- > 0
238 && (strpos($headers[0], '301') !== false || strpos($headers[0], '302') !== false)
239 && !empty($headers['Location'])) {
240 $redirection = is_array($headers['Location']) ? end($headers['Location']) : $headers['Location'];
241 if ($redirection != $url) {
242 $redirection = getAbsoluteUrl($url, $redirection);
243 return get_redirected_headers($redirection, $redirectionLimit);
247 return array($headers, $url);
251 * Get an absolute URL from a complete one, and another absolute/relative URL.
253 * @param string $originalUrl The original complete URL.
254 * @param string $newUrl The new one, absolute or relative.
256 * @return string Final URL:
257 * - $newUrl if it was already an absolute URL.
258 * - if it was relative, absolute URL from $originalUrl path.
260 function getAbsoluteUrl($originalUrl, $newUrl)
262 $newScheme = parse_url($newUrl, PHP_URL_SCHEME
);
263 // Already an absolute URL.
264 if (!empty($newScheme)) {
268 $parts = parse_url($originalUrl);
269 $final = $parts['scheme'] .'://'. $parts['host'];
270 $final .= (!empty($parts['port'])) ? $parts['port'] : '';
272 if ($newUrl[0] != '/') {
273 $final .= substr(ltrim($parts['path'], '/'), 0, strrpos($parts['path'], '/'));
275 $final .= ltrim($newUrl, '/');
280 * Returns the server's base URL: scheme://domain.tld[:port]
282 * @param array $server the $_SERVER array
284 * @return string the server's base URL
286 * @see http://www.ietf.org/rfc/rfc7239.txt
287 * @see http://www.ietf.org/rfc/rfc6648.txt
288 * @see http://stackoverflow.com/a/3561399
289 * @see http://stackoverflow.com/q/452375
291 function server_url($server)
296 // Shaarli is served behind a proxy
297 if (isset($server['HTTP_X_FORWARDED_PROTO'])) {
298 // Keep forwarded scheme
299 if (strpos($server['HTTP_X_FORWARDED_PROTO'], ',') !== false) {
300 $schemes = explode(',', $server['HTTP_X_FORWARDED_PROTO']);
301 $scheme = trim($schemes[0]);
303 $scheme = $server['HTTP_X_FORWARDED_PROTO'];
306 if (isset($server['HTTP_X_FORWARDED_PORT'])) {
307 // Keep forwarded port
308 if (strpos($server['HTTP_X_FORWARDED_PORT'], ',') !== false) {
309 $ports = explode(',', $server['HTTP_X_FORWARDED_PORT']);
310 $port = trim($ports[0]);
312 $port = $server['HTTP_X_FORWARDED_PORT'];
315 // This is a workaround for proxies that don't forward the scheme properly.
316 // Connecting over port 443 has to be in HTTPS.
317 // See https://github.com/shaarli/Shaarli/issues/1022
318 if ($port == '443') {
322 if (($scheme == 'http' && $port != '80')
323 || ($scheme == 'https' && $port != '443')
331 if (isset($server['HTTP_X_FORWARDED_HOST'])) {
332 // Keep forwarded host
333 if (strpos($server['HTTP_X_FORWARDED_HOST'], ',') !== false) {
334 $hosts = explode(',', $server['HTTP_X_FORWARDED_HOST']);
335 $host = trim($hosts[0]);
337 $host = $server['HTTP_X_FORWARDED_HOST'];
340 $host = $server['SERVER_NAME'];
343 return $scheme.'://'.$host.$port;
347 if ((! empty($server['HTTPS']) && strtolower($server['HTTPS']) == 'on')
348 || (isset($server['SERVER_PORT']) && $server['SERVER_PORT'] == '443')) {
352 // Do not append standard port values
353 if (($scheme == 'http' && $server['SERVER_PORT'] != '80')
354 || ($scheme == 'https' && $server['SERVER_PORT'] != '443')) {
355 $port = ':'.$server['SERVER_PORT'];
358 return $scheme.'://'.$server['SERVER_NAME'].$port;
362 * Returns the absolute URL of the current script, without the query
364 * If the resource is "index.php", then it is removed (for better-looking URLs)
366 * @param array $server the $_SERVER array
368 * @return string the absolute URL of the current script, without the query
370 function index_url($server)
372 if (defined('SHAARLI_ROOT_URL') && null !== SHAARLI_ROOT_URL
) {
373 return rtrim(SHAARLI_ROOT_URL
, '/') . '/';
376 $scriptname = !empty($server['SCRIPT_NAME']) ? $server['SCRIPT_NAME'] : '/';
377 if (endsWith($scriptname, 'index.php')) {
378 $scriptname = substr($scriptname, 0, -9);
380 return server_url($server) . $scriptname;
384 * Returns the absolute URL of the current script, with current route and query
386 * If the resource is "index.php", then it is removed (for better-looking URLs)
388 * @param array $server the $_SERVER array
390 * @return string the absolute URL of the current script, with the query
392 function page_url($server)
394 $scriptname = $server['SCRIPT_NAME'] ?? '';
395 if (endsWith($scriptname, 'index.php')) {
396 $scriptname = substr($scriptname, 0, -9);
399 $route = preg_replace('@^' . $scriptname . '@', '', $server['REQUEST_URI'] ?? '');
400 if (! empty($server['QUERY_STRING'])) {
401 return index_url($server) . $route . '?' . $server['QUERY_STRING'];
404 return index_url($server) . $route;
408 * Retrieve the initial IP forwarded by the reverse proxy.
410 * Inspired from: https://github.com/zendframework/zend-http/blob/master/src/PhpEnvironment/RemoteAddress.php
412 * @param array $server $_SERVER array which contains HTTP headers.
413 * @param array $trustedIps List of trusted IP from the configuration.
415 * @return string|bool The forwarded IP, or false if none could be extracted.
417 function getIpAddressFromProxy($server, $trustedIps)
419 $forwardedIpHeader = 'HTTP_X_FORWARDED_FOR';
420 if (empty($server[$forwardedIpHeader])) {
424 $ips = preg_split('/\s*,\s*/', $server[$forwardedIpHeader]);
425 $ips = array_diff($ips, $trustedIps);
430 return array_pop($ips);
435 * Return an identifier based on the advertised client IP address(es)
437 * This aims at preventing session hijacking from users behind the same proxy
438 * by relying on HTTP headers.
441 * - https://secure.php.net/manual/en/reserved.variables.server.php
442 * - https://stackoverflow.com/questions/3003145/how-to-get-the-client-ip-address-in-php
443 * - https://stackoverflow.com/questions/12233406/preventing-session-hijacking
444 * - https://stackoverflow.com/questions/21354859/trusting-x-forwarded-for-to-identify-a-visitor
446 * @param array $server The $_SERVER array
448 * @return string An identifier based on client IP address information
450 function client_ip_id($server)
452 $ip = $server['REMOTE_ADDR'];
454 if (isset($server['HTTP_X_FORWARDED_FOR'])) {
455 $ip = $ip . '_' . $server['HTTP_X_FORWARDED_FOR'];
457 if (isset($server['HTTP_CLIENT_IP'])) {
458 $ip = $ip . '_' . $server['HTTP_CLIENT_IP'];
465 * Returns true if Shaarli's currently browsed in HTTPS.
466 * Supports reverse proxies (if the headers are correctly set).
468 * @param array $server $_SERVER.
470 * @return bool true if HTTPS, false otherwise.
472 function is_https($server)
475 if (isset($server['HTTP_X_FORWARDED_PORT'])) {
476 // Keep forwarded port
477 if (strpos($server['HTTP_X_FORWARDED_PORT'], ',') !== false) {
478 $ports = explode(',', $server['HTTP_X_FORWARDED_PORT']);
479 $port = trim($ports[0]);
481 $port = $server['HTTP_X_FORWARDED_PORT'];
484 if ($port == '443') {
489 return ! empty($server['HTTPS']);
493 * Get cURL callback function for CURLOPT_WRITEFUNCTION
495 * @param string $charset to extract from the downloaded page (reference)
496 * @param string $title to extract from the downloaded page (reference)
497 * @param string $description to extract from the downloaded page (reference)
498 * @param string $keywords to extract from the downloaded page (reference)
499 * @param bool $retrieveDescription Automatically tries to retrieve description and keywords from HTML content
500 * @param string $curlGetInfo Optionally overrides curl_getinfo function
504 function get_curl_download_callback(
509 $retrieveDescription,
510 $curlGetInfo = 'curl_getinfo'
512 $isRedirected = false;
517 * cURL callback function for CURLOPT_WRITEFUNCTION (called during the download).
519 * While downloading the remote page, we check that the HTTP code is 200 and content type is 'html/text'
520 * Then we extract the title and the charset and stop the download when it's done.
522 * @param resource $ch cURL resource
523 * @param string $data chunk of data being downloaded
525 * @return int|bool length of $data or false if we need to stop the download
527 return function (&$ch, $data) use (
528 $retrieveDescription,
539 $responseCode = $curlGetInfo($ch, CURLINFO_RESPONSE_CODE
);
540 if (!empty($responseCode) && in_array($responseCode, [301, 302])) {
541 $isRedirected = true;
542 return strlen($data);
544 if (!empty($responseCode) && $responseCode !== 200) {
547 // After a redirection, the content type will keep the previous request value
548 // until it finds the next content-type header.
549 if (! $isRedirected || strpos(strtolower($data), 'content-type') !== false) {
550 $contentType = $curlGetInfo($ch, CURLINFO_CONTENT_TYPE
);
552 if (!empty($contentType) && strpos($contentType, 'text/html') === false) {
555 if (!empty($contentType) && empty($charset)) {
556 $charset = header_extract_charset($contentType);
558 if (empty($charset)) {
559 $charset = html_extract_charset($data);
562 $title = html_extract_title($data);
563 $foundChunk = ! empty($title) ? $currentChunk : $foundChunk;
565 if ($retrieveDescription && empty($description)) {
566 $description = html_extract_tag('description', $data);
567 $foundChunk = ! empty($description) ? $currentChunk : $foundChunk;
569 if ($retrieveDescription && empty($keywords)) {
570 $keywords = html_extract_tag('keywords', $data);
571 if (! empty($keywords)) {
572 $foundChunk = $currentChunk;
573 // Keywords use the format tag1, tag2 multiple words, tag
574 // So we format them to match Shaarli's separator and glue multiple words with '-'
575 $keywords = implode(' ', array_map(function($keyword) {
576 return implode('-', preg_split('/\s+/', trim($keyword)));
577 }, explode(',', $keywords)));
581 // We got everything we want, stop the download.
582 // If we already found either the title, description or keywords,
583 // it's highly unlikely that we'll found the other metas further than
584 // in the same chunk of data or the next one. So we also stop the download after that.
585 if ((!empty($responseCode) && !empty($contentType) && !empty($charset)) && $foundChunk !== null
586 && (! $retrieveDescription
587 || $foundChunk < $currentChunk
588 || (!empty($title) && !empty($description) && !empty($keywords))
594 return strlen($data);