[Sommaire] [Imprimer]

   1  <?php
   2  /**

   3   *

   4   *

   5   * This file is part of the b2evolution/evocms project - {@link http://b2evolution.net/}.

   6   * See also {@link http://sourceforge.net/projects/evocms/}.

   7   *

   8   * @copyright (c)2003-2007 by Francois PLANQUE - {@link http://fplanque.net/}.

   9   * Parts of this file are copyright (c)2006 by Daniel HAHLER - {@link http://daniel.hahler.de/}.

  10   *

  11   * @license http://b2evolution.net/about/license.html GNU General Public License (GPL)

  12   *

  13   * {@internal Open Source relicensing agreement:

  14   * Daniel HAHLER grants Francois PLANQUE the right to license

  15   * Daniel HAHLER's contributions to this file and the b2evolution project

  16   * under any OSI approved OSS license (http://www.opensource.org/licenses/).

  17   * }}

  18   *

  19   * @package evocore

  20   *

  21   * @author blueyed: Daniel HAHLER

  22   * @author Danny Ferguson

  23   *

  24   * @version $Id: _url.funcs.php,v 1.2 2007/07/03 22:16:15 blueyed Exp $

  25   */
  26  if( !defined('EVO_MAIN_INIT') ) die( 'Please, do not access this page directly.' );
  27  
  28  
  29  /**

  30   * Fetch remote page

  31   *

  32   * Attempt to retrieve a remote page, first with cURL, then fsockopen, then fopen.

  33   *

  34   * @todo dh> Should we try remaining methods, if the previous one(s) failed?

  35   * @param string URL

  36   * @param array Info (by reference)

  37   *        'error': holds error message, if any

  38   *        'status': HTTP status (e.g. 200 or 404)

  39   *        'used_method': Used method ("curl", "fopen", "fsockopen" or null if no method

  40   *                       is available)

  41   * @return string|false The remote page as a string; false in case of error

  42   */
  43  function fetch_remote_page( $url, & $info )
  44  {
  45      $info = array(
  46          'error' => '',
  47          'status' => NULL );
  48  
  49      // CURL:

  50      if( extension_loaded('curl') )
  51      {
  52          $info['used_method'] = 'curl';
  53  
  54          $ch = curl_init();
  55          curl_setopt($ch, CURLOPT_URL, $url);
  56          curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);
  57          if( ! empty($params['method']) && $params['method'] == 'HEAD'  )
  58          {
  59              curl_setopt($ch, CURLOPT_NOBODY, true);
  60          }
  61          $r = curl_exec($ch);
  62          $info['status'] = curl_getinfo($ch, CURLINFO_HTTP_CODE);
  63          $info['error'] = curl_error($ch);
  64          if( curl_errno($ch) )
  65          {
  66              $info['error'] .= '(#'.curl_errno($ch).')';
  67          }
  68          curl_close($ch);
  69  
  70          return $r;
  71      }
  72  
  73  
  74      // FSOCKOPEN:

  75      if( function_exists('fsockopen') ) // may have been disabled
  76      {
  77          $info['used_method'] = 'fsockopen';
  78          $url_parsed = parse_url($url);
  79          if( empty($url_parsed['scheme']) ) {
  80              $url_parsed = parse_url('http://'.$url);
  81          }
  82  
  83          $host = $url_parsed['host'];
  84          $port = ( empty($url_parsed['port']) ? 80 : $url_parsed['port'] );
  85          $path = empty($url_parsed['path']) ? '/' : $url_parsed['path'];
  86          if( ! empty($url_parsed['query']) )
  87          {
  88              $path .= '?'.$url_parsed['query'];
  89          }
  90  
  91          $out = "GET $path HTTP/1.0\r\n";
  92          $out .= "Host: $host:$port\r\n";
  93          $out .= "Connection: Close\r\n\r\n";
  94  
  95          $fp = @fsockopen($host, $port, $errno, $errstr, 30);
  96          if( ! $fp )
  97          {
  98              $info['error'] = $errstr.' (#'.$errno.')';
  99              return false;
 100          }
 101  
 102          // Set timeout for data:

 103          if( function_exists('stream_set_timeout') )
 104              stream_set_timeout( $fp, 20 ); // PHP 4.3.0

 105          else
 106              socket_set_timeout( $fp, 20 ); // PHP 4

 107  
 108          // Send request:

 109          fwrite($fp, $out);
 110  
 111          // Read response:

 112          $r = '';
 113          // First line:

 114          $s = fgets($fp, 4096);
 115          if( ! preg_match( '~^HTTP/\d+\.\d+ (\d+)~', $s, $match ) )
 116          {
 117              $info['error'] = 'Invalid response.';
 118              $r = false;
 119          }
 120          else
 121          {
 122              $info['status'] = $match[1];
 123  
 124              $foundBody = false;
 125              while( ! feof($fp) )
 126              {
 127                  $s = fgets($fp, 4096);
 128                  if( $s == "\r\n" )
 129                  {
 130                      $foundBody = true;
 131                      continue;
 132                  }
 133                  if( $foundBody )
 134                  {
 135                      $r .= $s;
 136                  }
 137              }
 138          }
 139          fclose($fp);
 140          return $r;
 141      }
 142  
 143  
 144      // URL FOPEN:

 145      if( ini_get('allow_url_fopen') && function_exists('stream_get_meta_data') /* PHP 4.3, may also be disabled!? */ )
 146      {
 147          $info['used_method'] = 'fopen';
 148  
 149          $fp = @fopen($url, 'r');
 150          if( $fp )
 151          { // this will be false e.g. for "404", but it's not trivial to get the status error for this, so we retry with fsockopen further down
 152              // headers:

 153              $meta = stream_get_meta_data($fp);
 154              if( ! $meta || ! preg_match( '~^HTTP/\d+\.\d+ (\d+)~', $meta['wrapper_data'][0], $match ) )
 155              {
 156                  $info['error'] = 'Invalid response.';
 157                  $r = false;
 158              }
 159              else
 160              {
 161                  $info['status'] = $match[1];
 162                  $r = '';
 163                  while( $buf = fread($fp, 4096) )
 164                  { //read the complete file (binary safe)
 165                      $r .= $buf;
 166                  }
 167              }
 168              fclose($fp);
 169  
 170              return $r;
 171          }
 172      }
 173  
 174  
 175      // All failed:

 176      $info['used_method'] = null;
 177      $info['error'] = 'No method available to access URL!';
 178      return false;
 179  }
 180  
 181  
 182  /**

 183   * Get $url with the same protocol (http/https) as $other_url.

 184   *

 185   * @param string URL

 186   * @param string other URL (defaults to {@link $ReqHost})

 187   * @return string

 188   */
 189  function url_same_protocol( $url, $other_url = NULL )
 190  {
 191      if( is_null($other_url) )
 192      {
 193          global $ReqHost;
 194  
 195          $other_url = $ReqHost;
 196      }
 197  
 198      // change protocol of $url to same of admin ('https' <=> 'http')

 199      if( substr( $url, 0, 7 ) == 'http://' )
 200      {
 201          if( substr( $other_url, 0, 8 ) == 'https://' )
 202          {
 203              $url = 'https://'.substr( $url, 7 );
 204          }
 205      }
 206      elseif( substr( $url, 0, 8 ) == 'https://' )
 207      {
 208          if( substr( $other_url, 0, 7 ) == 'http://' )
 209          {
 210              $url = 'http://'.substr( $url, 8 );
 211          }
 212      }
 213  
 214      return $url;
 215  }
 216  
 217  
 218  /* {{{ Revision log:

 219   * $Log: _url.funcs.php,v $

 220   * Revision 1.2  2007/07/03 22:16:15  blueyed

 221   * Solved todo: changed order to "cURL, fsockopen, fopen" for fetch_remote_page

 222   *

 223   * Revision 1.1  2007/06/25 10:58:54  fplanque

 224   * MODULES (refactored MVC)

 225   *

 226   * Revision 1.10  2007/04/26 00:11:08  fplanque

 227   * (c) 2007

 228   *

 229   * Revision 1.9  2007/03/12 22:39:42  blueyed

 230   * Fixed just introduced E_PARSE

 231   *

 232   * Revision 1.8  2007/03/12 22:12:58  blueyed

 233   * fetch_remote_page(): handle case when fsockopen is disabled

 234   *

 235   * Revision 1.7  2007/03/08 22:55:21  blueyed

 236   * fetch_remote_page: Added "used_method" to $info and errno to "curl" method.

 237   *

 238   * Revision 1.6  2006/12/22 00:25:15  blueyed

 239   * Do not send URL fragment in GET

 240   *

 241   * Revision 1.5  2006/12/21 21:50:52  fplanque

 242   * doc

 243   *

 244   * Revision 1.4  2006/12/19 21:35:26  blueyed

 245   * fetch_remote_page(): if url_fopen fails, use fsockopen-fallback (for errstr and status). Plus minor fixes.

 246   *

 247   * Revision 1.3  2006/12/01 17:31:38  blueyed

 248   * Fixed url_fopen method for fetch_remote_page

 249   *

 250   * Revision 1.2  2006/11/29 20:48:46  blueyed

 251   * Moved url_rel_to_same_host() from _misc.funcs.php to _url.funcs.php

 252   *

 253   * Revision 1.1  2006/11/25 23:00:39  blueyed

 254   * Added file for URL handling. Includes fetch_remote_page()

 255   *

 256   * }}}

 257   */
 258  ?>