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.$end;
  23 }
  24
  25 /* Having this defined in just one spot could help when changes need
  26  * to be made to the pattern
  27  * Make sure that the expression is evaluated case insensitively
  28  *
  29  * RFC2822 (and RFC822) defines the left side of an email address as (roughly):
  30  *  1*atext *("." 1*atext)
  31  * where atext is: a-zA-Z0-9!#$%&'*+-/=?^_`{|}~
  32  *
  33  * Here's pretty sophisticated IP matching:
  34  * $IPMatch = '(2[0-5][0-9]|1?[0-9]{1,2})';
  35  * $IPMatch = '\[?' . $IPMatch . '(\.' . $IPMatch . '){3}\]?';
  36  */
  37 /* Here's enough: */
  38 global $IP_RegExp_Match, $Host_RegExp_Match, $Email_RegExp_Match;
  39 $IP_RegExp_Match = '\\[?[0-9]{1,3}(\\.[0-9]{1,3}){3}\\]?';
  40 $Host_RegExp_Match = '(' . $IP_RegExp_Match .
  41     '|[0-9a-z]([-.]?[0-9a-z])*\\.[a-z][a-z]+)';
  42 $atext = '([a-z0-9!#$&%*+/=?^_`{|}~-]|&amp;)';
  43 $dot_atom = $atext . '+(\.' . $atext . '+)*';
  44 $Email_RegExp_Match = $dot_atom . '(%' . $Host_RegExp_Match . ')?@' .
  45                       $Host_RegExp_Match;
  46
  47 /**
  48  * Parses a body and converts all found email addresses to clickable links.
  49  *
  50  * @param string body the body to process, by ref
  51  * @return int the number of unique addresses found
  52  */
  53 function parseEmail (&$body) {
  54     global $Email_RegExp_Match;
  55     $sbody     = $body;
  56     $addresses = array();
  57
  58     /* Find all the email addresses in the body */
  59     while(eregi($Email_RegExp_Match, $sbody, $regs)) {
  60         $addresses[$regs[0]] = strtr($regs[0], array('&amp;' => '&'));
  61         $start = strpos($sbody, $regs[0]) + strlen($regs[0]);
  62         $sbody = substr($sbody, $start);
  63     }
  64
  65     /* Replace each email address with a compose URL */
  66     foreach ($addresses as $text => $email) {
  67         $comp_uri = makeComposeLink('src/compose.php?send_to='.urlencode($email), $text);
  68         $body = str_replace($text, $comp_uri, $body);
  69     }
  70
  71     /* Return number of unique addresses found */
  72     return count($addresses);
  73 }
  74
  75
  76 /* We don't want to re-initialize this stuff for every line.  Save work
  77  * and just do it once here.
  78  */
  79 global $url_parser_url_tokens;
  80 $url_parser_url_tokens = array(
  81     'http://',
  82     'https://',
  83     'ftp://',
  84     'telnet:',  // Special case -- doesn't need the slashes
  85     'mailto:',  // Special case -- doesn't use the slashes
  86     'gopher://',
  87     'news://');
  88
  89 global $url_parser_poss_ends;
  90 $url_parser_poss_ends = array(' ', "\n", "\r", '<', '>', ".\r", ".\n",
  91     '.&nbsp;', '&nbsp;', ')', '(', '&quot;', '&lt;', '&gt;', '.<',
  92     ']', '[', '{', '}', "\240", ', ', '. ', ",\n", ",\r");
  93
  94
  95 /**
  96  * rfc 2368 (mailto URL) preg_match() regexp
  97  * @link http://www.ietf.org/rfc/rfc2368.txt
  98  * @global string MailTo_PReg_Match the encapsulated regexp for preg_match()
  99  */
 100 global $MailTo_PReg_Match;
 101 $Mailto_Email_RegExp = '[0-9a-z%]([-_.+%]?[0-9a-z])*(%' . $Host_RegExp_Match . ')?@' . $Host_RegExp_Match;
 102 $MailTo_PReg_Match = '/((?:' . $Mailto_Email_RegExp . ')*)((?:\?(?:to|cc|bcc|subject|body)=[^\s\?&=,()]+)?(?:&amp;(?:to|cc|bcc|subject|body)=[^\s\?&=,()]+)*)/i';
 103
 104 /**
 105  * Parses a body and converts all found URLs to clickable links.
 106  *
 107  * @param string body the body to process, by ref
 108  * @return void
 109  */
 110 function parseUrl (&$body) {
 111     global $url_parser_poss_ends, $url_parser_url_tokens;
 112     $start      = 0;
 113     $blength    = strlen($body);
 114
 115     while ($start < $blength) {
 116         $target_token = '';
 117         $target_pos = $blength;
 118
 119         /* Find the first token to replace */
 120         foreach ($url_parser_url_tokens as $the_token) {
 121             $pos = strpos(strtolower($body), $the_token, $start);
 122             if (is_int($pos) && $pos < $target_pos) {
 123                 $target_pos   = $pos;
 124                 $target_token = $the_token;
 125             }
 126         }
 127
 128         /* Look for email addresses between $start and $target_pos */
 129         $check_str = substr($body, $start, $target_pos-$start);
 130
 131         if (parseEmail($check_str)) {
 132             replaceBlock($body, $check_str, $start, $target_pos);
 133             $blength    = strlen($body);
 134             $target_pos = strlen($check_str) + $start;
 135         }
 136
 137         /* If there was a token to replace, replace it */
 138         if ($target_token == 'mailto:') {    // rfc 2368 (mailto URL)
 139             $target_pos += 7;    //skip mailto:
 140             $end = $blength;
 141
 142             $mailto = substr($body, $target_pos, $end-$target_pos);
 143
 144             global $MailTo_PReg_Match;
 145             if ((preg_match($MailTo_PReg_Match, $mailto, $regs)) && ($regs[0] != '')) {
 146                 //sm_print_r($regs);
 147                 $mailto_before = $target_token . $regs[0];
 148                 $mailto_params = $regs[10];
 149                 if ($regs[1]) {    //if there is an email addr before '?', we need to merge it with the params
 150                     $to = 'to=' . $regs[1];
 151                     if (strpos($mailto_params, 'to=') > -1)    //already a 'to='
 152                         $mailto_params = str_replace('to=', $to . '%2C%20', $mailto_params);
 153                     else {
 154                         if ($mailto_params)    //already some params, append to them
 155                             $mailto_params .= '&amp;' . $to;
 156                         else
 157                             $mailto_params .= '?' . $to;
 158                     }
 159                 }
 160                 $url_str = str_replace(array('to=', 'cc=', 'bcc='), array('send_to=', 'send_to_cc=', 'send_to_bcc='), $mailto_params);
 161                 $comp_uri = makeComposeLink('src/compose.php' . $url_str, $mailto_before);
 162                 replaceBlock($body, $comp_uri, $target_pos - 7, $target_pos + strlen($regs[0]));
 163                 $target_pos += strlen($comp_uri) - 7;
 164             }
 165         }
 166         else
 167         if ($target_token != '') {
 168             /* Find the end of the URL */
 169             $end = $blength;
 170             foreach ($url_parser_poss_ends as $val) {
 171                 $enda = strpos($body, $val, $target_pos);
 172                 if (is_int($enda) && $enda < $end) {
 173                     $end = $enda;
 174                 }
 175             }
 176
 177             /* Extract URL */
 178             $url = substr($body, $target_pos, $end-$target_pos);
 179
 180             /* Needed since lines are not passed with \n or \r */
 181             while ( ereg("[,\.]$", $url) ) {
 182                 $url = substr( $url, 0, -1 );
 183                 $end--;
 184             }
 185
 186             /* Replace URL with HyperLinked Url, requires 1 char in link */
 187             if ($url != '' && $url != $target_token) {
 188                 $url_str = "<a href=\"$url\" target=\"_blank\">$url</a>";
 189                 replaceBlock($body,$url_str,$target_pos,$end);
 190                 $target_pos += strlen($url_str);
 191             }
 192             else {
 193                 // Not quite a valid link, skip ahead to next chance
 194                 $target_pos += strlen($target_token);
 195             }
 196         }
 197
 198         /* Move forward */
 199         $start   = $target_pos;
 200         $blength = strlen($body);
 201     }
 202 }
 203
 204 /**
 205  * Parses a string and returns the first e-mail address found.
 206  *
 207  * @param string string the string to process
 208  * @return string the first e-mail address found
 209  */
 210 function getEmail($string) {
 211     global $Email_RegExp_Match;
 212     $addresses = array();
 213
 214     /* Find all the email addresses in the body */
 215     while (eregi($Email_RegExp_Match, $string, $regs)) {
 216         $addresses[$regs[0]] = strtr($regs[0], array('&amp;' => '&'));
 217         $start = strpos($string, $regs[0]) + strlen($regs[0]);
 218         $string = substr($string, $start);
 219     }
 220
 221     /* Return the first address, or an empty string if no address was found */
 222     $addresses = array_values($addresses);
 223     return (array_key_exists(0, $addresses) ? $addresses[0] : '');
 224 }
 225
 226 ?>