functions/url_parser.php

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