program/steps/utils/spell_pspell.inc

   1 <?php
   2
   3 /*
   4  +-----------------------------------------------------------------------+
   5  | program/steps/utils/spell_pspell.inc                                  |
   6  |                                                                       |
   7  | This file is part of the Roundcube Webmail client                     |
   8  | Licensed under the GNU GPL                                            |
   9  |                                                                       |
  10  | PURPOSE:                                                              |
  11  |   Use the Pspell extension to check spelling, returns results         |
  12  |   compatible with spell_googie.inc.                                   |
  13  |                                                                       |
  14  +-----------------------------------------------------------------------+
  15  | Author: Kris Steinhoff <steinhof@umich.edu>                           |
  16  +-----------------------------------------------------------------------+
  17
  18  $Id: spell_pspell.inc 3989 2010-09-25 13:03:53Z alec $
  19
  20 */
  21
  22 if (!extension_loaded('pspell')) {
  23     raise_error(array(
  24       'code' => 500,
  25       'type' => 'php',
  26       'file' => __FILE__, 'line' => __LINE__,
  27       'message' => "Pspell extension not available"), true, false);
  28
  29     header('HTTP/1.1 404 Not Found');
  30     exit;
  31 }
  32
  33 // read input
  34 $data = file_get_contents('php://input');
  35
  36 // parse data (simplexml_load_string breaks CRLFs)
  37 $left = strpos($data, '<text>');
  38 $right = strrpos($data, '</text>');
  39 $text = substr($data, $left+6, $right-($left+6));
  40 $text = html_entity_decode($text, ENT_QUOTES, RCMAIL_CHARSET);
  41
  42 // tokenize
  43 $words = preg_split('/[ !"#$%&()*+\\,\/\n:;<=>?@\[\]^_{|}-]+|\.[^\w]/', $text, NULL,  PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE );
  44
  45 // init spellchecker
  46 $plink = pspell_new(get_input_value('lang', RCUBE_INPUT_GET), null, null, RCMAIL_CHARSET, PSPELL_FAST);
  47
  48 // send output
  49 $out = '<?xml version="1.0" encoding="'.RCMAIL_CHARSET.'"?><spellresult charschecked="'.mb_strlen($text).'">';
  50
  51 $diff = 0;
  52 foreach ($words as $w) {
  53     $word = trim($w[0]);
  54     $pos  = $w[1] - $diff;
  55     $len  = mb_strlen($word);
  56     if ($word && $plink && preg_match('/[^0-9\.]/', $word)
  57         && !pspell_check($plink, $word)) {
  58         $suggestions = pspell_suggest($plink, $word);
  59         if (sizeof($suggestions)>MAX_SUGGESTIONS)
  60           $suggestions = array_slice($suggestions, 0, MAX_SUGGESTIONS);
  61
  62         $out .= '<c o="'.$pos.'" l="'.$len.'">';
  63         $out .= implode("\t", $suggestions);
  64         $out .= '</c>';
  65     }
  66     $diff += (strlen($word) - $len);
  67 }
  68
  69 $out .= '</spellresult>';
  70
  71 header("Content-Type: text/xml; charset=".RCMAIL_CHARSET);
  72 echo $out;
  73 exit;
  74
  75