functions/url_parser.php

   1 <?php
   2
   3    /**
   4     **  url_parser.php
   5     **
   6     **  Copyright (c) 1999-2001 The Squirrelmail Development Team
   7     **  Licensed under the GNU GPL. For full terms see the file COPYING.
   8     **
   9     **  This code provides various string manipulation functions that are
  10     **  used by the rest of the Squirrelmail code.
  11     **
  12     **  $Id$
  13     **/
  14
  15    function replaceBlock (&$in, $replace, $start, $end) {
  16       $begin = substr($in,0,$start);
  17       $end   = substr($in,$end,strlen($in)-$end);
  18       $in    = $begin.$replace.$end;
  19    }
  20
  21    // Having this defined in just one spot could help when changes need
  22    // to be made to the pattern
  23    // Make sure that the expression is evaluated case insensitively
  24    //
  25    // Here's pretty sophisticated IP matching:
  26    // $IPMatch = '(2[0-5][0-9]|1?[0-9]{1,2})';
  27    // $IPMatch = '\[?' . $IPMatch . '(\.' . $IPMatch . '){3}\]?';
  28    //
  29    // Here's enough:
  30    global $IP_RegExp_Match, $Host_RegExp_Match, $Email_RegExp_Match;
  31    $IP_RegExp_Match = '\\[?[0-9]{1,3}(\\.[0-9]{1,3}){3}\\]?';
  32    $Host_RegExp_Match = '(' . $IP_RegExp_Match .
  33        '|[0-9a-z]([-.]?[0-9a-z])*\\.[a-z][a-z]+)';
  34    $Email_RegExp_Match = '[0-9a-z]([-_.]?[0-9a-z])*(%' . $Host_RegExp_Match .
  35        ')?@' . $Host_RegExp_Match;
  36
  37    function parseEmail (&$body) {
  38       global $color, $Email_RegExp_Match;
  39       $Size = strlen($body);
  40
  41       /*
  42         This is here in case we ever decide to use highlighting of searched
  43         text.  this does it for email addresses
  44
  45       if ($what && ($where == "BODY" || $where == "TEXT")) {
  46          eregi ($Email_RegExp_Match, $body, $regs);
  47          $oldaddr = $regs[0];
  48          if ($oldaddr) {
  49             $newaddr = eregi_replace ($what, "<b><font color=\"$color[2]\">$what</font></font></b>", $oldaddr);
  50             $body = str_replace ($oldaddr, "<a href=\"../src/compose.php?send_to=$oldaddr\">$newaddr</a>", $body);
  51          }
  52       } else {
  53          $body = eregi_replace ($Email_RegExp_Match, "<a href=\"../src/compose.php?send_to=\\0\">\\0</a>", $body);
  54       }
  55       */
  56
  57       $body = eregi_replace ($Email_RegExp_Match, "<a href=\"../src/compose.php?send_to=\\0\">\\0</a>", $body);
  58
  59       // If there are any changes, it'll just get bigger.
  60       if ($Size != strlen($body))
  61           return 1;
  62       return 0;
  63    }
  64
  65
  66    // We don't want to re-initialize this stuff for every line.  Save work
  67    // and just do it once here.
  68    global $url_parser_url_tokens;
  69    $url_parser_url_tokens = array(
  70        'http://',
  71        'https://',
  72        'ftp://',
  73        'telnet:',  // Special case -- doesn't need the slashes
  74        'gopher://',
  75        'news://');
  76
  77    global $url_parser_poss_ends;
  78    $url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
  79        '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
  80        ']', '[', '{', '}', "\240");
  81
  82
  83    function parseUrl (&$body)
  84    {
  85       global $url_parser_poss_ends, $url_parser_url_tokens;;
  86       $start = 0;
  87       $target_pos = strlen($body);
  88
  89       while ($start != $target_pos)
  90       {
  91         $target_token = '';
  92
  93         // Find the first token to replace
  94         foreach ($url_parser_url_tokens as $the_token)
  95         {
  96           $pos = strpos(strtolower($body), $the_token, $start);
  97           if (is_int($pos) && $pos < $target_pos)
  98           {
  99             $target_pos = $pos;
 100             $target_token = $the_token;
 101           }
 102         }
 103
 104         // Look for email addresses between $start and $target_pos
 105         $check_str = substr($body, $start, $target_pos);
 106
 107         if (parseEmail($check_str))
 108         {
 109           replaceBlock($body, $check_str, $start, $target_pos);
 110           $target_pos = strlen($check_str) + $start;
 111         }
 112
 113         // If there was a token to replace, replace it
 114         if ($target_token != '')
 115         {
 116           // Find the end of the URL
 117           $end=strlen($body);
 118           foreach ($url_parser_poss_ends as $key => $val)
 119           {
 120             $enda = strpos($body,$val,$target_pos);
 121             if (is_int($enda) && $enda < $end)
 122               $end = $enda;
 123           }
 124
 125           // Extract URL
 126           $url = substr($body, $target_pos, $end-$target_pos);
 127
 128           // Replace URL with HyperLinked Url, requires 1 char in link
 129           if ($url != '' && $url != $target_token)
 130           {
 131             $url_str = "<a href=\"$url\" target=\"_blank\">$url</a>";
 132             replaceBlock($body,$url_str,$target_pos,$end);
 133             $target_pos += strlen($url_str);
 134           }
 135           else
 136           {
 137              // Not quite a valid link, skip ahead to next chance
 138              $target_pos += strlen($target_token);
 139           }
 140         }
 141
 142         // Move forward
 143         $start = $target_pos;
 144         $target_pos = strlen($body);
 145       }
 146    }
 147
 148 ?>