Imported Upstream version 0.5

[roundcube.git] / program / steps / utils / spell_pspell.inc

diff --git a/program/steps/utils/spell_pspell.inc b/program/steps/utils/spell_pspell.inc
new file mode 100644 (file)
index 0000000..f3178ec
--- /dev/null
+++ b/program/steps/utils/spell_pspell.inc
@@ -0,0 +1,75 @@
+<?php
+
+/*
+ +-----------------------------------------------------------------------+
+ | program/steps/utils/spell_pspell.inc                                  |
+ |                                                                       |
+ | This file is part of the Roundcube Webmail client                     |
+ | Licensed under the GNU GPL                                            |
+ |                                                                       |
+ | PURPOSE:                                                              |
+ |   Use the Pspell extension to check spelling, returns results         |
+ |   compatible with spell_googie.inc.                                   |
+ |                                                                       |
+ +-----------------------------------------------------------------------+
+ | Author: Kris Steinhoff <steinhof@umich.edu>                           |
+ +-----------------------------------------------------------------------+
+
+ $Id: spell_pspell.inc 3989 2010-09-25 13:03:53Z alec $
+
+*/
+
+if (!extension_loaded('pspell')) {
+    raise_error(array(
+      'code' => 500,
+      'type' => 'php',
+      'file' => __FILE__, 'line' => __LINE__,
+      'message' => "Pspell extension not available"), true, false);
+
+    header('HTTP/1.1 404 Not Found');
+    exit;
+}
+
+// read input
+$data = file_get_contents('php://input');
+
+// parse data (simplexml_load_string breaks CRLFs)
+$left = strpos($data, '<text>');
+$right = strrpos($data, '</text>');
+$text = substr($data, $left+6, $right-($left+6));
+$text = html_entity_decode($text, ENT_QUOTES, RCMAIL_CHARSET);
+
+// tokenize
+$words = preg_split('/[ !"#$%&()*+\\,\/\n:;<=>?@\[\]^_{|}-]+|\.[^\w]/', $text, NULL,  PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE );
+
+// init spellchecker
+$plink = pspell_new(get_input_value('lang', RCUBE_INPUT_GET), null, null, RCMAIL_CHARSET, PSPELL_FAST);
+
+// send output
+$out = '<?xml version="1.0" encoding="'.RCMAIL_CHARSET.'"?><spellresult charschecked="'.mb_strlen($text).'">';
+
+$diff = 0;
+foreach ($words as $w) {
+    $word = trim($w[0]);
+    $pos  = $w[1] - $diff;
+    $len  = mb_strlen($word);
+    if ($word && $plink && preg_match('/[^0-9\.]/', $word)
+       && !pspell_check($plink, $word)) {
+        $suggestions = pspell_suggest($plink, $word);
+       if (sizeof($suggestions)>MAX_SUGGESTIONS)
+         $suggestions = array_slice($suggestions, 0, MAX_SUGGESTIONS);
+
+        $out .= '<c o="'.$pos.'" l="'.$len.'">';
+        $out .= implode("\t", $suggestions);
+        $out .= '</c>';
+    }
+    $diff += (strlen($word) - $len);
+}
+
+$out .= '</spellresult>';
+
+header("Content-Type: text/xml; charset=".RCMAIL_CHARSET);
+echo $out;
+exit;
+
+