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     $sbody     = $body;
  40     $addresses = array();
  41
  42     /* Find all the email addresses in the body */
  43     while(eregi($Email_RegExp_Match, $sbody, $regs)) {
  44         $addresses[$regs[0]] = $regs[0];
  45         $start = strpos($sbody, $regs[0]) + strlen($regs[0]);
  46         $sbody = substr($sbody, $start);
  47     }
  48     /* Replace each email address with a compose URL */
  49     foreach ($addresses as $email) {
  50         $comp_uri = '../src/compose.php?send_to='.urlencode($email);
  51         if ($compose_new_win == '1') {
  52             $comp_uri  = 'javascript:void(0)" onClick="comp_in_new('
  53                        . "'$comp_uri'" . ')';
  54         }
  55         $comp_uri = '<a href="'.$comp_uri.'">'.$email.'</a>';
  56         $body = str_replace($email, $comp_uri, $body);
  57     }
  58     /* Return number of unique addresses found */
  59     return count($addresses);
  60 }
  61
  62
  63 /* We don't want to re-initialize this stuff for every line.  Save work
  64  * and just do it once here.
  65  */
  66 global $url_parser_url_tokens;
  67 $url_parser_url_tokens = array(
  68     'http://',
  69     'https://',
  70     'ftp://',
  71     'telnet:',  // Special case -- doesn't need the slashes
  72     'gopher://',
  73     'news://');
  74
  75 global $url_parser_poss_ends;
  76 $url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
  77     '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
  78     ']', '[', '{', '}', "\240", ', ', '. ', ",\n", ",\r");
  79
  80
  81 function parseUrl (&$body) {
  82     global $url_parser_poss_ends, $url_parser_url_tokens;;
  83     $start      = 0;
  84     $blength    = strlen($body);
  85
  86     while ($start != $blength) {
  87         $target_token = '';
  88         $target_pos = $blength;
  89
  90         /* Find the first token to replace */
  91         foreach ($url_parser_url_tokens as $the_token) {
  92             $pos = strpos(strtolower($body), $the_token, $start);
  93             if (is_int($pos) && $pos < $blength) {
  94                 $target_pos   = $pos;
  95                 $target_token = $the_token;
  96             }
  97         }
  98
  99         /* Look for email addresses between $start and $target_pos */
 100         $check_str = substr($body, $start, $target_pos-$start);
 101
 102         if (parseEmail($check_str)) {
 103             replaceBlock($body, $check_str, $start, $target_pos);
 104             $blength    = strlen($body);
 105             $target_pos = strlen($check_str) + $start;
 106         }
 107
 108         /* If there was a token to replace, replace it */
 109         if ($target_token != '') {
 110             /* Find the end of the URL */
 111             $end = $blength;
 112             foreach ($url_parser_poss_ends as $val) {
 113                 $enda = strpos($body, $val, $target_pos);
 114                 if (is_int($enda) && $enda < $end) {
 115                     $end = $enda;
 116                 }
 117             }
 118
 119             /* Extract URL */
 120             $url = substr($body, $target_pos, $end-$target_pos);
 121
 122             /* Needed since lines are not passed with \n or \r */
 123             while ( ereg("[,\.]$", $url) ) {
 124                 $url = substr( $url, 0, -1 );
 125                 $end--;
 126             }
 127
 128             /* Replace URL with HyperLinked Url, requires 1 char in link */
 129             if ($url != '' && $url != $target_token) {
 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                 // Not quite a valid link, skip ahead to next chance
 136                 $target_pos += strlen($target_token);
 137             }
 138         }
 139
 140         /* Move forward */
 141         $start   = $target_pos;
 142         $blength = strlen($body);
 143     }
 144 }
 145 ?>