functions/url_parser.php

   1 <?php
   2
   3 /**
   4  * url_parser.php
   5  *
   6  * Copyright (c) 1999-2002 The SquirrelMail Project 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, $compose_new_win;
  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>", $body);
  54      * }
  55      */
  56
  57     if( eregi($Email_RegExp_Match, $body, $regs) ) {
  58                 if ($compose_new_win == '1') {
  59                 $body = str_replace($regs[0],  '<a href="../src/compose.php?send_to='.urlencode($regs[0]).'" target="compose_window" onClick="comp_in_new()">'.$regs[0].'</a>', $body);
  60                 }
  61                 else {
  62                 $body = str_replace($regs[0],  '<a href="../src/compose.php?send_to='.
  63             urlencode($regs[0]).'">'.$regs[0].'</a>', $body);
  64                 }
  65     }
  66
  67     /* If there are any changes, it'll just get bigger. */
  68     if ($Size != strlen($body)) {
  69         return 1;
  70     }
  71     return 0;
  72 }
  73
  74
  75 /* We don't want to re-initialize this stuff for every line.  Save work
  76  * and just do it once here.
  77  */
  78 global $url_parser_url_tokens;
  79 $url_parser_url_tokens = array(
  80     'http://',
  81     'https://',
  82     'ftp://',
  83     'telnet:',  // Special case -- doesn't need the slashes
  84     'gopher://',
  85     'news://');
  86
  87 global $url_parser_poss_ends;
  88 $url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
  89     '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
  90     ']', '[', '{', '}', "\240", ', ', '. ', ",\n", ",\r");
  91
  92
  93 function parseUrl (&$body) {
  94     global $url_parser_poss_ends, $url_parser_url_tokens;;
  95     $start = 0;
  96     $target_pos = strlen($body);
  97
  98     while ($start != $target_pos) {
  99         $target_token = '';
 100
 101         /* Find the first token to replace */
 102         foreach ($url_parser_url_tokens as $the_token) {
 103             $pos = strpos(strtolower($body), $the_token, $start);
 104             if (is_int($pos) && $pos < $target_pos) {
 105                 $target_pos = $pos;
 106                 $target_token = $the_token;
 107             }
 108         }
 109
 110         /* Look for email addresses between $start and $target_pos */
 111         $check_str = substr($body, $start, $target_pos);
 112
 113         if (parseEmail($check_str)) {
 114             replaceBlock($body, $check_str, $start, $target_pos);
 115             $target_pos = strlen($check_str) + $start;
 116         }
 117
 118         /* If there was a token to replace, replace it */
 119         if ($target_token != '') {
 120             /* Find the end of the URL */
 121             $end=strlen($body);
 122             foreach ($url_parser_poss_ends as $key => $val) {
 123                 $enda = strpos($body,$val,$target_pos);
 124                 if (is_int($enda) && $enda < $end) {
 125                     $end = $enda;
 126                 }
 127             }
 128
 129             /* Extract URL */
 130             $url = substr($body, $target_pos, $end-$target_pos);
 131
 132             /* Needed since lines are not passed with \n or \r */
 133             while ( ereg("[,\.]$", $url) ) {
 134                 $url = substr( $url, 0, -1 );
 135                 $end--;
 136             }
 137
 138             /* Replace URL with HyperLinked Url, requires 1 char in link */
 139             if ($url != '' && $url != $target_token) {
 140                 $url_str = "<a href=\"$url\" target=\"_blank\">$url</a>";
 141                 replaceBlock($body,$url_str,$target_pos,$end);
 142                 $target_pos += strlen($url_str);
 143             }
 144             else {
 145                  // Not quite a valid link, skip ahead to next chance
 146                  $target_pos += strlen($target_token);
 147             }
 148         }
 149
 150         /* Move forward */
 151         $start = $target_pos;
 152         $target_pos = strlen($body);
 153     }
 154 }
 155 ?>