functions/url_parser.php

   1 <?php
   2
   3 /**
   4  * url_parser.php
   5  *
   6  * Copyright (c) 1999-2004 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  * @version $Id$
  13  * @package squirrelmail
  14  */
  15
  16 /**
  17  * Undocumented - complain, then patch.
  18  */
  19 function replaceBlock (&$in, $replace, $start, $end) {
  20     $begin = substr($in,0,$start);
  21     $end   = substr($in,$end,strlen($in)-$end);
  22     $in    = $begin.$replace.$end;
  23 }
  24
  25 /* Having this defined in just one spot could help when changes need
  26  * to be made to the pattern
  27  * Make sure that the expression is evaluated case insensitively
  28  *
  29  * Here's pretty sophisticated IP matching:
  30  * $IPMatch = '(2[0-5][0-9]|1?[0-9]{1,2})';
  31  * $IPMatch = '\[?' . $IPMatch . '(\.' . $IPMatch . '){3}\]?';
  32  */
  33 /* Here's enough: */
  34 global $IP_RegExp_Match, $Host_RegExp_Match, $Email_RegExp_Match;
  35 $IP_RegExp_Match = '\\[?[0-9]{1,3}(\\.[0-9]{1,3}){3}\\]?';
  36 $Host_RegExp_Match = '(' . $IP_RegExp_Match .
  37     '|[0-9a-z]([-.]?[0-9a-z])*\\.[a-z][a-z]+)';
  38 $Email_RegExp_Match = '[0-9a-z]([-_.+]?[0-9a-z])*(%' . $Host_RegExp_Match .
  39     ')?@' . $Host_RegExp_Match;
  40
  41 /**
  42  * Parses a body and converts all found email addresses to clickable links.
  43  *
  44  * @param string body the body to process, by ref
  45  * @return int the number of unique addresses found
  46  */
  47 function parseEmail (&$body) {
  48     global $color, $Email_RegExp_Match;
  49     $sbody     = $body;
  50     $addresses = array();
  51
  52     /* Find all the email addresses in the body */
  53     while(eregi($Email_RegExp_Match, $sbody, $regs)) {
  54         $addresses[$regs[0]] = $regs[0];
  55         $start = strpos($sbody, $regs[0]) + strlen($regs[0]);
  56         $sbody = substr($sbody, $start);
  57     }
  58     /* Replace each email address with a compose URL */
  59     foreach ($addresses as $email) {
  60         $comp_uri = makeComposeLink('src/compose.php?send_to='.urlencode($email), $email);
  61         $body = str_replace($email, $comp_uri, $body);
  62     }
  63     /* Return number of unique addresses found */
  64     return count($addresses);
  65 }
  66
  67
  68 /* We don't want to re-initialize this stuff for every line.  Save work
  69  * and just do it once here.
  70  */
  71 global $url_parser_url_tokens;
  72 $url_parser_url_tokens = array(
  73     'http://',
  74     'https://',
  75     'ftp://',
  76     'telnet:',  // Special case -- doesn't need the slashes
  77     'mailto:',  // Special case -- doesn't use the slashes
  78     'gopher://',
  79     'news://');
  80
  81 global $url_parser_poss_ends;
  82 $url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
  83     '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
  84     ']', '[', '{', '}', "\240", ', ', '. ', ",\n", ",\r");
  85
  86
  87 /**
  88  * rfc 2368 (mailto URL) preg_match() regexp
  89  * @link http://www.ietf.org/rfc/rfc2368.txt
  90  * @global string MailTo_PReg_Match the encapsulated regexp for preg_match()
  91  */
  92 global $MailTo_PReg_Match;
  93 $Mailto_Email_RegExp = '[0-9a-z%]([-_.+%]?[0-9a-z])*(%' . $Host_RegExp_Match . ')?@' . $Host_RegExp_Match;
  94 $MailTo_PReg_Match = '/((?:' . $Mailto_Email_RegExp . ')*)((?:\?(?:to|cc|bcc|subject|body)=[^\s\?&=,()]+)?(?:&amp;(?:to|cc|bcc|subject|body)=[^\s\?&=,()]+)*)/i';
  95
  96 /**
  97  * Parses a body and converts all found URLs to clickable links.
  98  *
  99  * @param string body the body to process, by ref
 100  * @return void
 101  */
 102 function parseUrl (&$body) {
 103     global $url_parser_poss_ends, $url_parser_url_tokens;
 104     $start      = 0;
 105     $blength    = strlen($body);
 106
 107     while ($start < $blength) {
 108         $target_token = '';
 109         $target_pos = $blength;
 110
 111         /* Find the first token to replace */
 112         foreach ($url_parser_url_tokens as $the_token) {
 113             $pos = strpos(strtolower($body), $the_token, $start);
 114             if (is_int($pos) && $pos < $blength) {
 115                 $target_pos   = $pos;
 116                 $target_token = $the_token;
 117             }
 118         }
 119
 120         /* Look for email addresses between $start and $target_pos */
 121         $check_str = substr($body, $start, $target_pos-$start);
 122
 123         if (parseEmail($check_str)) {
 124             replaceBlock($body, $check_str, $start, $target_pos);
 125             $blength    = strlen($body);
 126             $target_pos = strlen($check_str) + $start;
 127         }
 128
 129         /* If there was a token to replace, replace it */
 130         if ($target_token == 'mailto:') {       // rfc 2368 (mailto URL)
 131             $target_pos += 7;   //skip mailto:
 132             $end = $blength;
 133
 134             $mailto = substr($body, $target_pos, $end-$target_pos);
 135
 136             global $MailTo_PReg_Match;
 137             if ((preg_match($MailTo_PReg_Match, $mailto, $regs)) && ($regs[0] != '')) {
 138                 //sm_print_r($regs);
 139                 $mailto_before = $target_token . $regs[0];
 140                 $mailto_params = $regs[10];
 141                 if ($regs[1]) { //if there is an email addr before '?', we need to merge it with the params
 142                     $to = 'to=' . $regs[1];
 143                     if (strpos($mailto_params, 'to=') > -1)     //already a 'to='
 144                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
 145                     else {
 146                         if ($mailto_params)     //already some params, append to them
 147                             $mailto_params .= '&amp;' . $to;
 148                         else
 149                             $mailto_params .= '?' . $to;
 150                     }
 151                 }
 152                 $url_str = str_replace(array('to=', 'cc=', 'bcc='), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
 153                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
 154                 replaceBlock($body, $comp_uri, $target_pos - 7, $target_pos + strlen($regs[0]));
 155                 $target_pos += strlen($comp_uri) - 7;
 156             }
 157         }
 158         else
 159         if ($target_token != '') {
 160             /* Find the end of the URL */
 161             $end = $blength;
 162             foreach ($url_parser_poss_ends as $val) {
 163                 $enda = strpos($body, $val, $target_pos);
 164                 if (is_int($enda) && $enda < $end) {
 165                     $end = $enda;
 166                 }
 167             }
 168
 169             /* Extract URL */
 170             $url = substr($body, $target_pos, $end-$target_pos);
 171
 172             /* Needed since lines are not passed with \n or \r */
 173             while ( ereg("[,\.]$", $url) ) {
 174                 $url = substr( $url, 0, -1 );
 175                 $end--;
 176             }
 177
 178             /* Replace URL with HyperLinked Url, requires 1 char in link */
 179             if ($url != '' && $url != $target_token) {
 180                 $url_str = "<a href=\"$url\" target=\"_blank\">$url</a>";
 181                 replaceBlock($body,$url_str,$target_pos,$end);
 182                 $target_pos += strlen($url_str);
 183             }
 184             else {
 185                 // Not quite a valid link, skip ahead to next chance
 186                 $target_pos += strlen($target_token);
 187             }
 188         }
 189
 190         /* Move forward */
 191         $start   = $target_pos;
 192         $blength = strlen($body);
 193     }
 194 }
 195 ?>