wp-includes/class-snoopy.php

   1 <?php
   2 /**
   3  * Snoopy - the PHP net client
   4  * @author Monte Ohrt <monte@ispi.net>
   5  * @copyright 1999-2000 ispi, all rights reserved
   6  * @version 1.01
   7  * @license GNU Lesser GPL
   8  * @link http://snoopy.sourceforge.net/
   9  * @package Snoopy
  10  */
  11
  12 if ( !in_array('Snoopy', get_declared_classes() ) ) :
  13 /**
  14  * Snoopy - the PHP net client
  15  *
  16  * @author Monte Ohrt <monte@ispi.net>
  17  * @copyright (c): 1999-2000 ispi, all rights reserved
  18  * @version 1.01
  19  *
  20  * This library is free software; you can redistribute it and/or
  21  * modify it under the terms of the GNU Lesser General Public
  22  * License as published by the Free Software Foundation; either
  23  * version 2.1 of the License, or (at your option) any later version.
  24  *
  25  * This library is distributed in the hope that it will be useful,
  26  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  27  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  28  * Lesser General Public License for more details.
  29  *
  30  * You should have received a copy of the GNU Lesser General Public
  31  * License along with this library; if not, write to the Free Software
  32  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
  33  *
  34  * You may contact the author of Snoopy by e-mail at:
  35  * monte@ispi.net
  36  *
  37  * Or, write to:
  38  * Monte Ohrt
  39  * CTO, ispi
  40  * 237 S. 70th suite 220
  41  * Lincoln, NE 68510
  42  *
  43  * @link http://snoopy.sourceforge.net/ The latest version of Snoopy can be
  44  *              obtained
  45  */
  46 class Snoopy
  47 {
  48         /**** Public variables ****/
  49
  50         /* user definable vars */
  51
  52         var $host                       =       "www.php.net";          // host name we are connecting to
  53         var $port                       =       80;                                     // port we are connecting to
  54         var $proxy_host         =       "";                                     // proxy host to use
  55         var $proxy_port         =       "";                                     // proxy port to use
  56         var $proxy_user         =       "";                                     // proxy user to use
  57         var $proxy_pass         =       "";                                     // proxy password to use
  58
  59         var $agent                      =       "Snoopy v1.2.3";        // agent we masquerade as
  60         var     $referer                =       "";                                     // referer info to pass
  61         var $cookies            =       array();                        // array of cookies to pass
  62                                                                                                 // $cookies["username"]="joe";
  63         var     $rawheaders             =       array();                        // array of raw headers to send
  64                                                                                                 // $rawheaders["Content-type"]="text/html";
  65
  66         var $maxredirs          =       5;                                      // http redirection depth maximum. 0 = disallow
  67         var $lastredirectaddr   =       "";                             // contains address of last redirected address
  68         var     $offsiteok              =       true;                           // allows redirection off-site
  69         var $maxframes          =       0;                                      // frame content depth maximum. 0 = disallow
  70         var $expandlinks        =       true;                           // expand links to fully qualified URLs.
  71                                                                                                 // this only applies to fetchlinks()
  72                                                                                                 // submitlinks(), and submittext()
  73         var $passcookies        =       true;                           // pass set cookies back through redirects
  74                                                                                                 // NOTE: this currently does not respect
  75                                                                                                 // dates, domains or paths.
  76
  77         var     $user                   =       "";                                     // user for http authentication
  78         var     $pass                   =       "";                                     // password for http authentication
  79
  80         // http accept types
  81         var $accept                     =       "image/gif, image/x-xbitmap, image/jpeg, image/pjpeg, */*";
  82
  83         var $results            =       "";                                     // where the content is put
  84
  85         var $error                      =       "";                                     // error messages sent here
  86         var     $response_code  =       "";                                     // response code returned from server
  87         var     $headers                =       array();                        // headers returned from server sent here
  88         var     $maxlength              =       8192;                           // max return data length (body)
  89         var $read_timeout       =       0;                                      // timeout on read operations, in seconds
  90                                                                                                 // supported only since PHP 4 Beta 4
  91                                                                                                 // set to 0 to disallow timeouts
  92         var $timed_out          =       false;                          // if a read operation timed out
  93         var     $status                 =       0;                                      // http request status
  94
  95         var $temp_dir           =       "/tmp";                         // temporary directory that the webserver
  96                                                                                                 // has permission to write to.
  97                                                                                                 // under Windows, this should be C:\temp
  98
  99         var     $curl_path              =       "/usr/local/bin/curl";
 100                                                                                                 // Snoopy will use cURL for fetching
 101                                                                                                 // SSL content if a full system path to
 102                                                                                                 // the cURL binary is supplied here.
 103                                                                                                 // set to false if you do not have
 104                                                                                                 // cURL installed. See http://curl.haxx.se
 105                                                                                                 // for details on installing cURL.
 106                                                                                                 // Snoopy does *not* use the cURL
 107                                                                                                 // library functions built into php,
 108                                                                                                 // as these functions are not stable
 109                                                                                                 // as of this Snoopy release.
 110
 111         /**** Private variables ****/
 112
 113         var     $_maxlinelen    =       4096;                           // max line length (headers)
 114
 115         var $_httpmethod        =       "GET";                          // default http request method
 116         var $_httpversion       =       "HTTP/1.0";                     // default http request version
 117         var $_submit_method     =       "POST";                         // default submit method
 118         var $_submit_type       =       "application/x-www-form-urlencoded";    // default submit type
 119         var $_mime_boundary     =   "";                                 // MIME boundary for multipart/form-data submit type
 120         var $_redirectaddr      =       false;                          // will be set if page fetched is a redirect
 121         var $_redirectdepth     =       0;                                      // increments on an http redirect
 122         var $_frameurls         =       array();                        // frame src urls
 123         var $_framedepth        =       0;                                      // increments on frame depth
 124
 125         var $_isproxy           =       false;                          // set if using a proxy server
 126         var $_fp_timeout        =       30;                                     // timeout for socket connection
 127
 128 /*======================================================================*\
 129         Function:       fetch
 130         Purpose:        fetch the contents of a web page
 131                                 (and possibly other protocols in the
 132                                 future like ftp, nntp, gopher, etc.)
 133         Input:          $URI    the location of the page to fetch
 134         Output:         $this->results  the output text from the fetch
 135 \*======================================================================*/
 136
 137         function fetch($URI)
 138         {
 139
 140                 //preg_match("|^([^:]+)://([^:/]+)(:[\d]+)*(.*)|",$URI,$URI_PARTS);
 141                 $URI_PARTS = parse_url($URI);
 142                 if (!empty($URI_PARTS["user"]))
 143                         $this->user = $URI_PARTS["user"];
 144                 if (!empty($URI_PARTS["pass"]))
 145                         $this->pass = $URI_PARTS["pass"];
 146                 if (empty($URI_PARTS["query"]))
 147                         $URI_PARTS["query"] = '';
 148                 if (empty($URI_PARTS["path"]))
 149                         $URI_PARTS["path"] = '';
 150
 151                 switch(strtolower($URI_PARTS["scheme"]))
 152                 {
 153                         case "http":
 154                                 $this->host = $URI_PARTS["host"];
 155                                 if(!empty($URI_PARTS["port"]))
 156                                         $this->port = $URI_PARTS["port"];
 157                                 if($this->_connect($fp))
 158                                 {
 159                                         if($this->_isproxy)
 160                                         {
 161                                                 // using proxy, send entire URI
 162                                                 $this->_httprequest($URI,$fp,$URI,$this->_httpmethod);
 163                                         }
 164                                         else
 165                                         {
 166                                                 $path = $URI_PARTS["path"].($URI_PARTS["query"] ? "?".$URI_PARTS["query"] : "");
 167                                                 // no proxy, send only the path
 168                                                 $this->_httprequest($path, $fp, $URI, $this->_httpmethod);
 169                                         }
 170
 171                                         $this->_disconnect($fp);
 172
 173                                         if($this->_redirectaddr)
 174                                         {
 175                                                 /* url was redirected, check if we've hit the max depth */
 176                                                 if($this->maxredirs > $this->_redirectdepth)
 177                                                 {
 178                                                         // only follow redirect if it's on this site, or offsiteok is true
 179                                                         if(preg_match("|^http://".preg_quote($this->host)."|i",$this->_redirectaddr) || $this->offsiteok)
 180                                                         {
 181                                                                 /* follow the redirect */
 182                                                                 $this->_redirectdepth++;
 183                                                                 $this->lastredirectaddr=$this->_redirectaddr;
 184                                                                 $this->fetch($this->_redirectaddr);
 185                                                         }
 186                                                 }
 187                                         }
 188
 189                                         if($this->_framedepth < $this->maxframes && count($this->_frameurls) > 0)
 190                                         {
 191                                                 $frameurls = $this->_frameurls;
 192                                                 $this->_frameurls = array();
 193
 194                                                 while(list(,$frameurl) = each($frameurls))
 195                                                 {
 196                                                         if($this->_framedepth < $this->maxframes)
 197                                                         {
 198                                                                 $this->fetch($frameurl);
 199                                                                 $this->_framedepth++;
 200                                                         }
 201                                                         else
 202                                                                 break;
 203                                                 }
 204                                         }
 205                                 }
 206                                 else
 207                                 {
 208                                         return false;
 209                                 }
 210                                 return true;
 211                                 break;
 212                         case "https":
 213                                 if(!$this->curl_path)
 214                                         return false;
 215                                 if(function_exists("is_executable"))
 216                                     if (!is_executable($this->curl_path))
 217                                         return false;
 218                                 $this->host = $URI_PARTS["host"];
 219                                 if(!empty($URI_PARTS["port"]))
 220                                         $this->port = $URI_PARTS["port"];
 221                                 if($this->_isproxy)
 222                                 {
 223                                         // using proxy, send entire URI
 224                                         $this->_httpsrequest($URI,$URI,$this->_httpmethod);
 225                                 }
 226                                 else
 227                                 {
 228                                         $path = $URI_PARTS["path"].($URI_PARTS["query"] ? "?".$URI_PARTS["query"] : "");
 229                                         // no proxy, send only the path
 230                                         $this->_httpsrequest($path, $URI, $this->_httpmethod);
 231                                 }
 232
 233                                 if($this->_redirectaddr)
 234                                 {
 235                                         /* url was redirected, check if we've hit the max depth */
 236                                         if($this->maxredirs > $this->_redirectdepth)
 237                                         {
 238                                                 // only follow redirect if it's on this site, or offsiteok is true
 239                                                 if(preg_match("|^http://".preg_quote($this->host)."|i",$this->_redirectaddr) || $this->offsiteok)
 240                                                 {
 241                                                         /* follow the redirect */
 242                                                         $this->_redirectdepth++;
 243                                                         $this->lastredirectaddr=$this->_redirectaddr;
 244                                                         $this->fetch($this->_redirectaddr);
 245                                                 }
 246                                         }
 247                                 }
 248
 249                                 if($this->_framedepth < $this->maxframes && count($this->_frameurls) > 0)
 250                                 {
 251                                         $frameurls = $this->_frameurls;
 252                                         $this->_frameurls = array();
 253
 254                                         while(list(,$frameurl) = each($frameurls))
 255                                         {
 256                                                 if($this->_framedepth < $this->maxframes)
 257                                                 {
 258                                                         $this->fetch($frameurl);
 259                                                         $this->_framedepth++;
 260                                                 }
 261                                                 else
 262                                                         break;
 263                                         }
 264                                 }
 265                                 return true;
 266                                 break;
 267                         default:
 268                                 // not a valid protocol
 269                                 $this->error    =       'Invalid protocol "'.$URI_PARTS["scheme"].'"\n';
 270                                 return false;
 271                                 break;
 272                 }
 273                 return true;
 274         }
 275
 276 /*======================================================================*\
 277         Function:       submit
 278         Purpose:        submit an http form
 279         Input:          $URI    the location to post the data
 280                                 $formvars       the formvars to use.
 281                                         format: $formvars["var"] = "val";
 282                                 $formfiles  an array of files to submit
 283                                         format: $formfiles["var"] = "/dir/filename.ext";
 284         Output:         $this->results  the text output from the post
 285 \*======================================================================*/
 286
 287         function submit($URI, $formvars="", $formfiles="")
 288         {
 289                 unset($postdata);
 290
 291                 $postdata = $this->_prepare_post_body($formvars, $formfiles);
 292
 293                 $URI_PARTS = parse_url($URI);
 294                 if (!empty($URI_PARTS["user"]))
 295                         $this->user = $URI_PARTS["user"];
 296                 if (!empty($URI_PARTS["pass"]))
 297                         $this->pass = $URI_PARTS["pass"];
 298                 if (empty($URI_PARTS["query"]))
 299                         $URI_PARTS["query"] = '';
 300                 if (empty($URI_PARTS["path"]))
 301                         $URI_PARTS["path"] = '';
 302
 303                 switch(strtolower($URI_PARTS["scheme"]))
 304                 {
 305                         case "http":
 306                                 $this->host = $URI_PARTS["host"];
 307                                 if(!empty($URI_PARTS["port"]))
 308                                         $this->port = $URI_PARTS["port"];
 309                                 if($this->_connect($fp))
 310                                 {
 311                                         if($this->_isproxy)
 312                                         {
 313                                                 // using proxy, send entire URI
 314                                                 $this->_httprequest($URI,$fp,$URI,$this->_submit_method,$this->_submit_type,$postdata);
 315                                         }
 316                                         else
 317                                         {
 318                                                 $path = $URI_PARTS["path"].($URI_PARTS["query"] ? "?".$URI_PARTS["query"] : "");
 319                                                 // no proxy, send only the path
 320                                                 $this->_httprequest($path, $fp, $URI, $this->_submit_method, $this->_submit_type, $postdata);
 321                                         }
 322
 323                                         $this->_disconnect($fp);
 324
 325                                         if($this->_redirectaddr)
 326                                         {
 327                                                 /* url was redirected, check if we've hit the max depth */
 328                                                 if($this->maxredirs > $this->_redirectdepth)
 329                                                 {
 330                                                         if(!preg_match("|^".$URI_PARTS["scheme"]."://|", $this->_redirectaddr))
 331                                                                 $this->_redirectaddr = $this->_expandlinks($this->_redirectaddr,$URI_PARTS["scheme"]."://".$URI_PARTS["host"]);
 332
 333                                                         // only follow redirect if it's on this site, or offsiteok is true
 334                                                         if(preg_match("|^http://".preg_quote($this->host)."|i",$this->_redirectaddr) || $this->offsiteok)
 335                                                         {
 336                                                                 /* follow the redirect */
 337                                                                 $this->_redirectdepth++;
 338                                                                 $this->lastredirectaddr=$this->_redirectaddr;
 339                                                                 if( strpos( $this->_redirectaddr, "?" ) > 0 )
 340                                                                         $this->fetch($this->_redirectaddr); // the redirect has changed the request method from post to get
 341                                                                 else
 342                                                                         $this->submit($this->_redirectaddr,$formvars, $formfiles);
 343                                                         }
 344                                                 }
 345                                         }
 346
 347                                         if($this->_framedepth < $this->maxframes && count($this->_frameurls) > 0)
 348                                         {
 349                                                 $frameurls = $this->_frameurls;
 350                                                 $this->_frameurls = array();
 351
 352                                                 while(list(,$frameurl) = each($frameurls))
 353                                                 {
 354                                                         if($this->_framedepth < $this->maxframes)
 355                                                         {
 356                                                                 $this->fetch($frameurl);
 357                                                                 $this->_framedepth++;
 358                                                         }
 359                                                         else
 360                                                                 break;
 361                                                 }
 362                                         }
 363
 364                                 }
 365                                 else
 366                                 {
 367                                         return false;
 368                                 }
 369                                 return true;
 370                                 break;
 371                         case "https":
 372                                 if(!$this->curl_path)
 373                                         return false;
 374                                 if(function_exists("is_executable"))
 375                                     if (!is_executable($this->curl_path))
 376                                         return false;
 377                                 $this->host = $URI_PARTS["host"];
 378                                 if(!empty($URI_PARTS["port"]))
 379                                         $this->port = $URI_PARTS["port"];
 380                                 if($this->_isproxy)
 381                                 {
 382                                         // using proxy, send entire URI
 383                                         $this->_httpsrequest($URI, $URI, $this->_submit_method, $this->_submit_type, $postdata);
 384                                 }
 385                                 else
 386                                 {
 387                                         $path = $URI_PARTS["path"].($URI_PARTS["query"] ? "?".$URI_PARTS["query"] : "");
 388                                         // no proxy, send only the path
 389                                         $this->_httpsrequest($path, $URI, $this->_submit_method, $this->_submit_type, $postdata);
 390                                 }
 391
 392                                 if($this->_redirectaddr)
 393                                 {
 394                                         /* url was redirected, check if we've hit the max depth */
 395                                         if($this->maxredirs > $this->_redirectdepth)
 396                                         {
 397                                                 if(!preg_match("|^".$URI_PARTS["scheme"]."://|", $this->_redirectaddr))
 398                                                         $this->_redirectaddr = $this->_expandlinks($this->_redirectaddr,$URI_PARTS["scheme"]."://".$URI_PARTS["host"]);
 399
 400                                                 // only follow redirect if it's on this site, or offsiteok is true
 401                                                 if(preg_match("|^http://".preg_quote($this->host)."|i",$this->_redirectaddr) || $this->offsiteok)
 402                                                 {
 403                                                         /* follow the redirect */
 404                                                         $this->_redirectdepth++;
 405                                                         $this->lastredirectaddr=$this->_redirectaddr;
 406                                                         if( strpos( $this->_redirectaddr, "?" ) > 0 )
 407                                                                 $this->fetch($this->_redirectaddr); // the redirect has changed the request method from post to get
 408                                                         else
 409                                                                 $this->submit($this->_redirectaddr,$formvars, $formfiles);
 410                                                 }
 411                                         }
 412                                 }
 413
 414                                 if($this->_framedepth < $this->maxframes && count($this->_frameurls) > 0)
 415                                 {
 416                                         $frameurls = $this->_frameurls;
 417                                         $this->_frameurls = array();
 418
 419                                         while(list(,$frameurl) = each($frameurls))
 420                                         {
 421                                                 if($this->_framedepth < $this->maxframes)
 422                                                 {
 423                                                         $this->fetch($frameurl);
 424                                                         $this->_framedepth++;
 425                                                 }
 426                                                 else
 427                                                         break;
 428                                         }
 429                                 }
 430                                 return true;
 431                                 break;
 432
 433                         default:
 434                                 // not a valid protocol
 435                                 $this->error    =       'Invalid protocol "'.$URI_PARTS["scheme"].'"\n';
 436                                 return false;
 437                                 break;
 438                 }
 439                 return true;
 440         }
 441
 442 /*======================================================================*\
 443         Function:       fetchlinks
 444         Purpose:        fetch the links from a web page
 445         Input:          $URI    where you are fetching from
 446         Output:         $this->results  an array of the URLs
 447 \*======================================================================*/
 448
 449         function fetchlinks($URI)
 450         {
 451                 if ($this->fetch($URI))
 452                 {
 453                         if($this->lastredirectaddr)
 454                                 $URI = $this->lastredirectaddr;
 455                         if(is_array($this->results))
 456                         {
 457                                 for($x=0;$x<count($this->results);$x++)
 458                                         $this->results[$x] = $this->_striplinks($this->results[$x]);
 459                         }
 460                         else
 461                                 $this->results = $this->_striplinks($this->results);
 462
 463                         if($this->expandlinks)
 464                                 $this->results = $this->_expandlinks($this->results, $URI);
 465                         return true;
 466                 }
 467                 else
 468                         return false;
 469         }
 470
 471 /*======================================================================*\
 472         Function:       fetchform
 473         Purpose:        fetch the form elements from a web page
 474         Input:          $URI    where you are fetching from
 475         Output:         $this->results  the resulting html form
 476 \*======================================================================*/
 477
 478         function fetchform($URI)
 479         {
 480
 481                 if ($this->fetch($URI))
 482                 {
 483
 484                         if(is_array($this->results))
 485                         {
 486                                 for($x=0;$x<count($this->results);$x++)
 487                                         $this->results[$x] = $this->_stripform($this->results[$x]);
 488                         }
 489                         else
 490                                 $this->results = $this->_stripform($this->results);
 491
 492                         return true;
 493                 }
 494                 else
 495                         return false;
 496         }
 497
 498
 499 /*======================================================================*\
 500         Function:       fetchtext
 501         Purpose:        fetch the text from a web page, stripping the links
 502         Input:          $URI    where you are fetching from
 503         Output:         $this->results  the text from the web page
 504 \*======================================================================*/
 505
 506         function fetchtext($URI)
 507         {
 508                 if($this->fetch($URI))
 509                 {
 510                         if(is_array($this->results))
 511                         {
 512                                 for($x=0;$x<count($this->results);$x++)
 513                                         $this->results[$x] = $this->_striptext($this->results[$x]);
 514                         }
 515                         else
 516                                 $this->results = $this->_striptext($this->results);
 517                         return true;
 518                 }
 519                 else
 520                         return false;
 521         }
 522
 523 /*======================================================================*\
 524         Function:       submitlinks
 525         Purpose:        grab links from a form submission
 526         Input:          $URI    where you are submitting from
 527         Output:         $this->results  an array of the links from the post
 528 \*======================================================================*/
 529
 530         function submitlinks($URI, $formvars="", $formfiles="")
 531         {
 532                 if($this->submit($URI,$formvars, $formfiles))
 533                 {
 534                         if($this->lastredirectaddr)
 535                                 $URI = $this->lastredirectaddr;
 536                         if(is_array($this->results))
 537                         {
 538                                 for($x=0;$x<count($this->results);$x++)
 539                                 {
 540                                         $this->results[$x] = $this->_striplinks($this->results[$x]);
 541                                         if($this->expandlinks)
 542                                                 $this->results[$x] = $this->_expandlinks($this->results[$x],$URI);
 543                                 }
 544                         }
 545                         else
 546                         {
 547                                 $this->results = $this->_striplinks($this->results);
 548                                 if($this->expandlinks)
 549                                         $this->results = $this->_expandlinks($this->results,$URI);
 550                         }
 551                         return true;
 552                 }
 553                 else
 554                         return false;
 555         }
 556
 557 /*======================================================================*\
 558         Function:       submittext
 559         Purpose:        grab text from a form submission
 560         Input:          $URI    where you are submitting from
 561         Output:         $this->results  the text from the web page
 562 \*======================================================================*/
 563
 564         function submittext($URI, $formvars = "", $formfiles = "")
 565         {
 566                 if($this->submit($URI,$formvars, $formfiles))
 567                 {
 568                         if($this->lastredirectaddr)
 569                                 $URI = $this->lastredirectaddr;
 570                         if(is_array($this->results))
 571                         {
 572                                 for($x=0;$x<count($this->results);$x++)
 573                                 {
 574                                         $this->results[$x] = $this->_striptext($this->results[$x]);
 575                                         if($this->expandlinks)
 576                                                 $this->results[$x] = $this->_expandlinks($this->results[$x],$URI);
 577                                 }
 578                         }
 579                         else
 580                         {
 581                                 $this->results = $this->_striptext($this->results);
 582                                 if($this->expandlinks)
 583                                         $this->results = $this->_expandlinks($this->results,$URI);
 584                         }
 585                         return true;
 586                 }
 587                 else
 588                         return false;
 589         }
 590
 591
 592
 593 /*======================================================================*\
 594         Function:       set_submit_multipart
 595         Purpose:        Set the form submission content type to
 596                                 multipart/form-data
 597 \*======================================================================*/
 598         function set_submit_multipart()
 599         {
 600                 $this->_submit_type = "multipart/form-data";
 601         }
 602
 603
 604 /*======================================================================*\
 605         Function:       set_submit_normal
 606         Purpose:        Set the form submission content type to
 607                                 application/x-www-form-urlencoded
 608 \*======================================================================*/
 609         function set_submit_normal()
 610         {
 611                 $this->_submit_type = "application/x-www-form-urlencoded";
 612         }
 613
 614
 615
 616
 617 /*======================================================================*\
 618         Private functions
 619 \*======================================================================*/
 620
 621
 622 /*======================================================================*\
 623         Function:       _striplinks
 624         Purpose:        strip the hyperlinks from an html document
 625         Input:          $document       document to strip.
 626         Output:         $match          an array of the links
 627 \*======================================================================*/
 628
 629         function _striplinks($document)
 630         {
 631                 preg_match_all("'<\s*a\s.*?href\s*=\s*                  # find <a href=
 632                                                 ([\"\'])?                                       # find single or double quote
 633                                                 (?(1) (.*?)\\1 | ([^\s\>]+))            # if quote found, match up to next matching
 634                                                                                                         # quote, otherwise match up to next space
 635                                                 'isx",$document,$links);
 636
 637
 638                 // catenate the non-empty matches from the conditional subpattern
 639
 640                 while(list($key,$val) = each($links[2]))
 641                 {
 642                         if(!empty($val))
 643                                 $match[] = $val;
 644                 }
 645
 646                 while(list($key,$val) = each($links[3]))
 647                 {
 648                         if(!empty($val))
 649                                 $match[] = $val;
 650                 }
 651
 652                 // return the links
 653                 return $match;
 654         }
 655
 656 /*======================================================================*\
 657         Function:       _stripform
 658         Purpose:        strip the form elements from an html document
 659         Input:          $document       document to strip.
 660         Output:         $match          an array of the links
 661 \*======================================================================*/
 662
 663         function _stripform($document)
 664         {
 665                 preg_match_all("'<\/?(FORM|INPUT|SELECT|TEXTAREA|(OPTION))[^<>]*>(?(2)(.*(?=<\/?(option|select)[^<>]*>[\r\n]*)|(?=[\r\n]*))|(?=[\r\n]*))'Usi",$document,$elements);
 666
 667                 // catenate the matches
 668                 $match = implode("\r\n",$elements[0]);
 669
 670                 // return the links
 671                 return $match;
 672         }
 673
 674
 675
 676 /*======================================================================*\
 677         Function:       _striptext
 678         Purpose:        strip the text from an html document
 679         Input:          $document       document to strip.
 680         Output:         $text           the resulting text
 681 \*======================================================================*/
 682
 683         function _striptext($document)
 684         {
 685
 686                 // I didn't use preg eval (//e) since that is only available in PHP 4.0.
 687                 // so, list your entities one by one here. I included some of the
 688                 // more common ones.
 689
 690                 $search = array("'<script[^>]*?>.*?</script>'si",       // strip out javascript
 691                                                 "'<[\/\!]*?[^<>]*?>'si",                        // strip out html tags
 692                                                 "'([\r\n])[\s]+'",                                      // strip out white space
 693                                                 "'&(quot|#34|#034|#x22);'i",            // replace html entities
 694                                                 "'&(amp|#38|#038|#x26);'i",                     // added hexadecimal values
 695                                                 "'&(lt|#60|#060|#x3c);'i",
 696                                                 "'&(gt|#62|#062|#x3e);'i",
 697                                                 "'&(nbsp|#160|#xa0);'i",
 698                                                 "'&(iexcl|#161);'i",
 699                                                 "'&(cent|#162);'i",
 700                                                 "'&(pound|#163);'i",
 701                                                 "'&(copy|#169);'i",
 702                                                 "'&(reg|#174);'i",
 703                                                 "'&(deg|#176);'i",
 704                                                 "'&(#39|#039|#x27);'",
 705                                                 "'&(euro|#8364);'i",                            // europe
 706                                                 "'&a(uml|UML);'",                                       // german
 707                                                 "'&o(uml|UML);'",
 708                                                 "'&u(uml|UML);'",
 709                                                 "'&A(uml|UML);'",
 710                                                 "'&O(uml|UML);'",
 711                                                 "'&U(uml|UML);'",
 712                                                 "'&szlig;'i",
 713                                                 );
 714                 $replace = array(       "",
 715                                                         "",
 716                                                         "\\1",
 717                                                         "\"",
 718                                                         "&",
 719                                                         "<",
 720                                                         ">",
 721                                                         " ",
 722                                                         chr(161),
 723                                                         chr(162),
 724                                                         chr(163),
 725                                                         chr(169),
 726                                                         chr(174),
 727                                                         chr(176),
 728                                                         chr(39),
 729                                                         chr(128),
 730                                                         "ä",
 731                                                         "ö",
 732                                                         "ü",
 733                                                         "Ä",
 734                                                         "Ö",
 735                                                         "Ü",
 736                                                         "ß",
 737                                                 );
 738
 739                 $text = preg_replace($search,$replace,$document);
 740
 741                 return $text;
 742         }
 743
 744 /*======================================================================*\
 745         Function:       _expandlinks
 746         Purpose:        expand each link into a fully qualified URL
 747         Input:          $links                  the links to qualify
 748                                 $URI                    the full URI to get the base from
 749         Output:         $expandedLinks  the expanded links
 750 \*======================================================================*/
 751
 752         function _expandlinks($links,$URI)
 753         {
 754
 755                 preg_match("/^[^\?]+/",$URI,$match);
 756
 757                 $match = preg_replace("|/[^\/\.]+\.[^\/\.]+$|","",$match[0]);
 758                 $match = preg_replace("|/$|","",$match);
 759                 $match_part = parse_url($match);
 760                 $match_root =
 761                 $match_part["scheme"]."://".$match_part["host"];
 762
 763                 $search = array(        "|^http://".preg_quote($this->host)."|i",
 764                                                         "|^(\/)|i",
 765                                                         "|^(?!http://)(?!mailto:)|i",
 766                                                         "|/\./|",
 767                                                         "|/[^\/]+/\.\./|"
 768                                                 );
 769
 770                 $replace = array(       "",
 771                                                         $match_root."/",
 772                                                         $match."/",
 773                                                         "/",
 774                                                         "/"
 775                                                 );
 776
 777                 $expandedLinks = preg_replace($search,$replace,$links);
 778
 779                 return $expandedLinks;
 780         }
 781
 782 /*======================================================================*\
 783         Function:       _httprequest
 784         Purpose:        go get the http data from the server
 785         Input:          $url            the url to fetch
 786                                 $fp                     the current open file pointer
 787                                 $URI            the full URI
 788                                 $body           body contents to send if any (POST)
 789         Output:
 790 \*======================================================================*/
 791
 792         function _httprequest($url,$fp,$URI,$http_method,$content_type="",$body="")
 793         {
 794                 $cookie_headers = '';
 795                 if($this->passcookies && $this->_redirectaddr)
 796                         $this->setcookies();
 797
 798                 $URI_PARTS = parse_url($URI);
 799                 if(empty($url))
 800                         $url = "/";
 801                 $headers = $http_method." ".$url." ".$this->_httpversion."\r\n";
 802                 if(!empty($this->agent))
 803                         $headers .= "User-Agent: ".$this->agent."\r\n";
 804                 if(!empty($this->host) && !isset($this->rawheaders['Host'])) {
 805                         $headers .= "Host: ".$this->host;
 806                         if(!empty($this->port) && $this->port != 80)
 807                                 $headers .= ":".$this->port;
 808                         $headers .= "\r\n";
 809                 }
 810                 if(!empty($this->accept))
 811                         $headers .= "Accept: ".$this->accept."\r\n";
 812                 if(!empty($this->referer))
 813                         $headers .= "Referer: ".$this->referer."\r\n";
 814                 if(!empty($this->cookies))
 815                 {
 816                         if(!is_array($this->cookies))
 817                                 $this->cookies = (array)$this->cookies;
 818
 819                         reset($this->cookies);
 820                         if ( count($this->cookies) > 0 ) {
 821                                 $cookie_headers .= 'Cookie: ';
 822                                 foreach ( $this->cookies as $cookieKey => $cookieVal ) {
 823                                 $cookie_headers .= $cookieKey."=".urlencode($cookieVal)."; ";
 824                                 }
 825                                 $headers .= substr($cookie_headers,0,-2) . "\r\n";
 826                         }
 827                 }
 828                 if(!empty($this->rawheaders))
 829                 {
 830                         if(!is_array($this->rawheaders))
 831                                 $this->rawheaders = (array)$this->rawheaders;
 832                         while(list($headerKey,$headerVal) = each($this->rawheaders))
 833                                 $headers .= $headerKey.": ".$headerVal."\r\n";
 834                 }
 835                 if(!empty($content_type)) {
 836                         $headers .= "Content-type: $content_type";
 837                         if ($content_type == "multipart/form-data")
 838                                 $headers .= "; boundary=".$this->_mime_boundary;
 839                         $headers .= "\r\n";
 840                 }
 841                 if(!empty($body))
 842                         $headers .= "Content-length: ".strlen($body)."\r\n";
 843                 if(!empty($this->user) || !empty($this->pass))
 844                         $headers .= "Authorization: Basic ".base64_encode($this->user.":".$this->pass)."\r\n";
 845
 846                 //add proxy auth headers
 847                 if(!empty($this->proxy_user))
 848                         $headers .= 'Proxy-Authorization: ' . 'Basic ' . base64_encode($this->proxy_user . ':' . $this->proxy_pass)."\r\n";
 849
 850
 851                 $headers .= "\r\n";
 852
 853                 // set the read timeout if needed
 854                 if ($this->read_timeout > 0)
 855                         socket_set_timeout($fp, $this->read_timeout);
 856                 $this->timed_out = false;
 857
 858                 fwrite($fp,$headers.$body,strlen($headers.$body));
 859
 860                 $this->_redirectaddr = false;
 861                 unset($this->headers);
 862
 863                 while($currentHeader = fgets($fp,$this->_maxlinelen))
 864                 {
 865                         if ($this->read_timeout > 0 && $this->_check_timeout($fp))
 866                         {
 867                                 $this->status=-100;
 868                                 return false;
 869                         }
 870
 871                         if($currentHeader == "\r\n")
 872                                 break;
 873
 874                         // if a header begins with Location: or URI:, set the redirect
 875                         if(preg_match("/^(Location:|URI:)/i",$currentHeader))
 876                         {
 877                                 // get URL portion of the redirect
 878                                 preg_match("/^(Location:|URI:)[ ]+(.*)/i",chop($currentHeader),$matches);
 879                                 // look for :// in the Location header to see if hostname is included
 880                                 if(!preg_match("|\:\/\/|",$matches[2]))
 881                                 {
 882                                         // no host in the path, so prepend
 883                                         $this->_redirectaddr = $URI_PARTS["scheme"]."://".$this->host.":".$this->port;
 884                                         // eliminate double slash
 885                                         if(!preg_match("|^/|",$matches[2]))
 886                                                         $this->_redirectaddr .= "/".$matches[2];
 887                                         else
 888                                                         $this->_redirectaddr .= $matches[2];
 889                                 }
 890                                 else
 891                                         $this->_redirectaddr = $matches[2];
 892                         }
 893
 894                         if(preg_match("|^HTTP/|",$currentHeader))
 895                         {
 896                 if(preg_match("|^HTTP/[^\s]*\s(.*?)\s|",$currentHeader, $status))
 897                                 {
 898                                         $this->status= $status[1];
 899                 }
 900                                 $this->response_code = $currentHeader;
 901                         }
 902
 903                         $this->headers[] = $currentHeader;
 904                 }
 905
 906                 $results = '';
 907                 do {
 908                 $_data = fread($fp, $this->maxlength);
 909                 if (strlen($_data) == 0) {
 910                         break;
 911                 }
 912                 $results .= $_data;
 913                 } while(true);
 914
 915                 if ($this->read_timeout > 0 && $this->_check_timeout($fp))
 916                 {
 917                         $this->status=-100;
 918                         return false;
 919                 }
 920
 921                 // check if there is a a redirect meta tag
 922
 923                 if(preg_match("'<meta[\s]*http-equiv[^>]*?content[\s]*=[\s]*[\"\']?\d+;[\s]*URL[\s]*=[\s]*([^\"\']*?)[\"\']?>'i",$results,$match))
 924
 925                 {
 926                         $this->_redirectaddr = $this->_expandlinks($match[1],$URI);
 927                 }
 928
 929                 // have we hit our frame depth and is there frame src to fetch?
 930                 if(($this->_framedepth < $this->maxframes) && preg_match_all("'<frame\s+.*src[\s]*=[\'\"]?([^\'\"\>]+)'i",$results,$match))
 931                 {
 932                         $this->results[] = $results;
 933                         for($x=0; $x<count($match[1]); $x++)
 934                                 $this->_frameurls[] = $this->_expandlinks($match[1][$x],$URI_PARTS["scheme"]."://".$this->host);
 935                 }
 936                 // have we already fetched framed content?
 937                 elseif(is_array($this->results))
 938                         $this->results[] = $results;
 939                 // no framed content
 940                 else
 941                         $this->results = $results;
 942
 943                 return true;
 944         }
 945
 946 /*======================================================================*\
 947         Function:       _httpsrequest
 948         Purpose:        go get the https data from the server using curl
 949         Input:          $url            the url to fetch
 950                                 $URI            the full URI
 951                                 $body           body contents to send if any (POST)
 952         Output:
 953 \*======================================================================*/
 954
 955         function _httpsrequest($url,$URI,$http_method,$content_type="",$body="")
 956         {
 957                 if($this->passcookies && $this->_redirectaddr)
 958                         $this->setcookies();
 959
 960                 $headers = array();
 961
 962                 $URI_PARTS = parse_url($URI);
 963                 if(empty($url))
 964                         $url = "/";
 965                 // GET ... header not needed for curl
 966                 //$headers[] = $http_method." ".$url." ".$this->_httpversion;
 967                 if(!empty($this->agent))
 968                         $headers[] = "User-Agent: ".$this->agent;
 969                 if(!empty($this->host))
 970                         if(!empty($this->port))
 971                                 $headers[] = "Host: ".$this->host.":".$this->port;
 972                         else
 973                                 $headers[] = "Host: ".$this->host;
 974                 if(!empty($this->accept))
 975                         $headers[] = "Accept: ".$this->accept;
 976                 if(!empty($this->referer))
 977                         $headers[] = "Referer: ".$this->referer;
 978                 if(!empty($this->cookies))
 979                 {
 980                         if(!is_array($this->cookies))
 981                                 $this->cookies = (array)$this->cookies;
 982
 983                         reset($this->cookies);
 984                         if ( count($this->cookies) > 0 ) {
 985                                 $cookie_str = 'Cookie: ';
 986                                 foreach ( $this->cookies as $cookieKey => $cookieVal ) {
 987                                 $cookie_str .= $cookieKey."=".urlencode($cookieVal)."; ";
 988                                 }
 989                                 $headers[] = substr($cookie_str,0,-2);
 990                         }
 991                 }
 992                 if(!empty($this->rawheaders))
 993                 {
 994                         if(!is_array($this->rawheaders))
 995                                 $this->rawheaders = (array)$this->rawheaders;
 996                         while(list($headerKey,$headerVal) = each($this->rawheaders))
 997                                 $headers[] = $headerKey.": ".$headerVal;
 998                 }
 999                 if(!empty($content_type)) {
1000                         if ($content_type == "multipart/form-data")
1001                                 $headers[] = "Content-type: $content_type; boundary=".$this->_mime_boundary;
1002                         else
1003                                 $headers[] = "Content-type: $content_type";
1004                 }
1005                 if(!empty($body))
1006                         $headers[] = "Content-length: ".strlen($body);
1007                 if(!empty($this->user) || !empty($this->pass))
1008                         $headers[] = "Authorization: BASIC ".base64_encode($this->user.":".$this->pass);
1009
1010                 for($curr_header = 0; $curr_header < count($headers); $curr_header++) {
1011                         $safer_header = strtr( $headers[$curr_header], "\"", " " );
1012                         $cmdline_params .= " -H \"".$safer_header."\"";
1013                 }
1014
1015                 if(!empty($body))
1016                         $cmdline_params .= " -d \"$body\"";
1017
1018                 if($this->read_timeout > 0)
1019                         $cmdline_params .= " -m ".$this->read_timeout;
1020
1021                 $headerfile = tempnam($temp_dir, "sno");
1022
1023                 $safer_URI = strtr( $URI, "\"", " " ); // strip quotes from the URI to avoid shell access
1024                 exec(escapeshellcmd($this->curl_path." -D \"$headerfile\"".$cmdline_params." \"".$safer_URI."\""),$results,$return);
1025
1026                 if($return)
1027                 {
1028                         $this->error = "Error: cURL could not retrieve the document, error $return.";
1029                         return false;
1030                 }
1031
1032
1033                 $results = implode("\r\n",$results);
1034
1035                 $result_headers = file("$headerfile");
1036
1037                 $this->_redirectaddr = false;
1038                 unset($this->headers);
1039
1040                 for($currentHeader = 0; $currentHeader < count($result_headers); $currentHeader++)
1041                 {
1042
1043                         // if a header begins with Location: or URI:, set the redirect
1044                         if(preg_match("/^(Location: |URI: )/i",$result_headers[$currentHeader]))
1045                         {
1046                                 // get URL portion of the redirect
1047                                 preg_match("/^(Location: |URI:)\s+(.*)/",chop($result_headers[$currentHeader]),$matches);
1048                                 // look for :// in the Location header to see if hostname is included
1049                                 if(!preg_match("|\:\/\/|",$matches[2]))
1050                                 {
1051                                         // no host in the path, so prepend
1052                                         $this->_redirectaddr = $URI_PARTS["scheme"]."://".$this->host.":".$this->port;
1053                                         // eliminate double slash
1054                                         if(!preg_match("|^/|",$matches[2]))
1055                                                         $this->_redirectaddr .= "/".$matches[2];
1056                                         else
1057                                                         $this->_redirectaddr .= $matches[2];
1058                                 }
1059                                 else
1060                                         $this->_redirectaddr = $matches[2];
1061                         }
1062
1063                         if(preg_match("|^HTTP/|",$result_headers[$currentHeader]))
1064                                 $this->response_code = $result_headers[$currentHeader];
1065
1066                         $this->headers[] = $result_headers[$currentHeader];
1067                 }
1068
1069                 // check if there is a a redirect meta tag
1070
1071                 if(preg_match("'<meta[\s]*http-equiv[^>]*?content[\s]*=[\s]*[\"\']?\d+;[\s]*URL[\s]*=[\s]*([^\"\']*?)[\"\']?>'i",$results,$match))
1072                 {
1073                         $this->_redirectaddr = $this->_expandlinks($match[1],$URI);
1074                 }
1075
1076                 // have we hit our frame depth and is there frame src to fetch?
1077                 if(($this->_framedepth < $this->maxframes) && preg_match_all("'<frame\s+.*src[\s]*=[\'\"]?([^\'\"\>]+)'i",$results,$match))
1078                 {
1079                         $this->results[] = $results;
1080                         for($x=0; $x<count($match[1]); $x++)
1081                                 $this->_frameurls[] = $this->_expandlinks($match[1][$x],$URI_PARTS["scheme"]."://".$this->host);
1082                 }
1083                 // have we already fetched framed content?
1084                 elseif(is_array($this->results))
1085                         $this->results[] = $results;
1086                 // no framed content
1087                 else
1088                         $this->results = $results;
1089
1090                 unlink("$headerfile");
1091
1092                 return true;
1093         }
1094
1095 /*======================================================================*\
1096         Function:       setcookies()
1097         Purpose:        set cookies for a redirection
1098 \*======================================================================*/
1099
1100         function setcookies()
1101         {
1102                 for($x=0; $x<count($this->headers); $x++)
1103                 {
1104                 if(preg_match('/^set-cookie:[\s]+([^=]+)=([^;]+)/i', $this->headers[$x],$match))
1105                         $this->cookies[$match[1]] = urldecode($match[2]);
1106                 }
1107         }
1108
1109
1110 /*======================================================================*\
1111         Function:       _check_timeout
1112         Purpose:        checks whether timeout has occurred
1113         Input:          $fp     file pointer
1114 \*======================================================================*/
1115
1116         function _check_timeout($fp)
1117         {
1118                 if ($this->read_timeout > 0) {
1119                         $fp_status = socket_get_status($fp);
1120                         if ($fp_status["timed_out"]) {
1121                                 $this->timed_out = true;
1122                                 return true;
1123                         }
1124                 }
1125                 return false;
1126         }
1127
1128 /*======================================================================*\
1129         Function:       _connect
1130         Purpose:        make a socket connection
1131         Input:          $fp     file pointer
1132 \*======================================================================*/
1133
1134         function _connect(&$fp)
1135         {
1136                 if(!empty($this->proxy_host) && !empty($this->proxy_port))
1137                         {
1138                                 $this->_isproxy = true;
1139
1140                                 $host = $this->proxy_host;
1141                                 $port = $this->proxy_port;
1142                         }
1143                 else
1144                 {
1145                         $host = $this->host;
1146                         $port = $this->port;
1147                 }
1148
1149                 $this->status = 0;
1150
1151                 if($fp = fsockopen(
1152                                         $host,
1153                                         $port,
1154                                         $errno,
1155                                         $errstr,
1156                                         $this->_fp_timeout
1157                                         ))
1158                 {
1159                         // socket connection succeeded
1160
1161                         return true;
1162                 }
1163                 else
1164                 {
1165                         // socket connection failed
1166                         $this->status = $errno;
1167                         switch($errno)
1168                         {
1169                                 case -3:
1170                                         $this->error="socket creation failed (-3)";
1171                                 case -4:
1172                                         $this->error="dns lookup failure (-4)";
1173                                 case -5:
1174                                         $this->error="connection refused or timed out (-5)";
1175                                 default:
1176                                         $this->error="connection failed (".$errno.")";
1177                         }
1178                         return false;
1179                 }
1180         }
1181 /*======================================================================*\
1182         Function:       _disconnect
1183         Purpose:        disconnect a socket connection
1184         Input:          $fp     file pointer
1185 \*======================================================================*/
1186
1187         function _disconnect($fp)
1188         {
1189                 return(fclose($fp));
1190         }
1191
1192
1193 /*======================================================================*\
1194         Function:       _prepare_post_body
1195         Purpose:        Prepare post body according to encoding type
1196         Input:          $formvars  - form variables
1197                                 $formfiles - form upload files
1198         Output:         post body
1199 \*======================================================================*/
1200
1201         function _prepare_post_body($formvars, $formfiles)
1202         {
1203                 settype($formvars, "array");
1204                 settype($formfiles, "array");
1205                 $postdata = '';
1206
1207                 if (count($formvars) == 0 && count($formfiles) == 0)
1208                         return;
1209
1210                 switch ($this->_submit_type) {
1211                         case "application/x-www-form-urlencoded":
1212                                 reset($formvars);
1213                                 while(list($key,$val) = each($formvars)) {
1214                                         if (is_array($val) || is_object($val)) {
1215                                                 while (list($cur_key, $cur_val) = each($val)) {
1216                                                         $postdata .= urlencode($key)."[]=".urlencode($cur_val)."&";
1217                                                 }
1218                                         } else
1219                                                 $postdata .= urlencode($key)."=".urlencode($val)."&";
1220                                 }
1221                                 break;
1222
1223                         case "multipart/form-data":
1224                                 $this->_mime_boundary = "Snoopy".md5(uniqid(microtime()));
1225
1226                                 reset($formvars);
1227                                 while(list($key,$val) = each($formvars)) {
1228                                         if (is_array($val) || is_object($val)) {
1229                                                 while (list($cur_key, $cur_val) = each($val)) {
1230                                                         $postdata .= "--".$this->_mime_boundary."\r\n";
1231                                                         $postdata .= "Content-Disposition: form-data; name=\"$key\[\]\"\r\n\r\n";
1232                                                         $postdata .= "$cur_val\r\n";
1233                                                 }
1234                                         } else {
1235                                                 $postdata .= "--".$this->_mime_boundary."\r\n";
1236                                                 $postdata .= "Content-Disposition: form-data; name=\"$key\"\r\n\r\n";
1237                                                 $postdata .= "$val\r\n";
1238                                         }
1239                                 }
1240
1241                                 reset($formfiles);
1242                                 while (list($field_name, $file_names) = each($formfiles)) {
1243                                         settype($file_names, "array");
1244                                         while (list(, $file_name) = each($file_names)) {
1245                                                 if (!is_readable($file_name)) continue;
1246
1247                                                 $fp = fopen($file_name, "r");
1248                                                 while (!feof($fp)) {
1249                                                         $file_content .= fread($fp, filesize($file_name));
1250                                                 }
1251                                                 fclose($fp);
1252                                                 $base_name = basename($file_name);
1253
1254                                                 $postdata .= "--".$this->_mime_boundary."\r\n";
1255                                                 $postdata .= "Content-Disposition: form-data; name=\"$field_name\"; filename=\"$base_name\"\r\n\r\n";
1256                                                 $postdata .= "$file_content\r\n";
1257                                         }
1258                                 }
1259                                 $postdata .= "--".$this->_mime_boundary."--\r\n";
1260                                 break;
1261                 }
1262
1263                 return $postdata;
1264         }
1265 }
1266 endif;
1267
1268 ?>