| commit | author | age | ||
| 5d2b7f | 1 | <?php |
| T | 2 | |
| 3 | /* | |
| 4 | +-----------------------------------------------------------------------+ | |
| 677e1f | 5 | | program/steps/utils/spell_pspell.inc | |
| 5d2b7f | 6 | | | |
| T | 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 | $Id$ | |
| 19 | ||
| 20 | */ | |
| 21 | ||
| 22 | if (!extension_loaded('pspell')) { | |
| 23 | raise_error(array( | |
| 24 | 'code' => 500, | |
| 6710a6 | 25 | 'type' => 'php', |
| 10eedb | 26 | 'file' => __FILE__, 'line' => __LINE__, |
| 5d2b7f | 27 | 'message' => "Pspell extension not available"), true, false); |
| 677e1f | 28 | |
| 5d2b7f | 29 | header('HTTP/1.1 404 Not Found'); |
| T | 30 | exit; |
| 31 | } | |
| 32 | ||
| 4e65a3 | 33 | // max. number of suggestions for one word |
| A | 34 | define('MAX_SUGGESTIONS', 10); |
| 35 | ||
| b214f8 | 36 | // read input |
| 5d2b7f | 37 | $data = file_get_contents('php://input'); |
| b214f8 | 38 | |
| A | 39 | // parse data (simplexml_load_string breaks CRLFs) |
| 40 | $left = strpos($data, '<text>'); | |
| 41 | $right = strrpos($data, '</text>'); | |
| 42 | $text = substr($data, $left+6, $right-($left+6)); | |
| f9a4bc | 43 | $text = html_entity_decode($text, ENT_QUOTES, RCMAIL_CHARSET); |
| b214f8 | 44 | |
| A | 45 | // tokenize |
| 16dd37 | 46 | $words = preg_split('/[ !"#$%&()*+\\,\/\n:;<=>?@\[\]^_{|}-]+|\.[^\w]/', $text, NULL, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE ); |
| b214f8 | 47 | |
| A | 48 | // init spellchecker |
| f9a4bc | 49 | $plink = pspell_new(get_input_value('lang', RCUBE_INPUT_GET), null, null, RCMAIL_CHARSET, PSPELL_FAST); |
| b214f8 | 50 | |
| A | 51 | // send output |
| ee258c | 52 | $out = '<?xml version="1.0" encoding="'.RCMAIL_CHARSET.'"?><spellresult charschecked="'.mb_strlen($text).'">'; |
| 81308b | 53 | |
| A | 54 | $diff = 0; |
| 5d2b7f | 55 | foreach ($words as $w) { |
| b214f8 | 56 | $word = trim($w[0]); |
| 81308b | 57 | $pos = $w[1] - $diff; |
| ee258c | 58 | $len = mb_strlen($word); |
| 16dd37 | 59 | if ($word && $plink && preg_match('/[^0-9\.]/', $word) |
| A | 60 | && !pspell_check($plink, $word)) { |
| 5d2b7f | 61 | $suggestions = pspell_suggest($plink, $word); |
| 4e65a3 | 62 | if (sizeof($suggestions)>10) |
| A | 63 | $suggestions = array_slice($suggestions, 0, MAX_SUGGESTIONS); |
| 64 | ||
| 5d2b7f | 65 | $out .= '<c o="'.$pos.'" l="'.$len.'">'; |
| T | 66 | $out .= implode("\t", $suggestions); |
| 67 | $out .= '</c>'; | |
| 68 | } | |
| 81308b | 69 | $diff += (strlen($word) - $len); |
| 5d2b7f | 70 | } |
| b214f8 | 71 | |
| 5d2b7f | 72 | $out .= '</spellresult>'; |
| T | 73 | |
| 57837f | 74 | header("Content-Type: text/xml; charset=".RCMAIL_CHARSET); |
| 5d2b7f | 75 | echo $out; |
| T | 76 | exit; |
| 77 | ||
| 78 | ?> | |
