| commit | author | age | ||
| b4edf7 | 1 | <?php |
| A | 2 | |
| 3 | /* | |
| 4 | +-----------------------------------------------------------------------+ | |
| 5 | | program/include/rcube_spellchecker.php | | |
| 6 | | | | |
| 7 | | This file is part of the Roundcube Webmail client | | |
| 8 | | Copyright (C) 2011, Kolab Systems AG | | |
| 9 | | Copyright (C) 2008-2011, The Roundcube Dev Team | | |
| 10 | | Licensed under the GNU GPL | | |
| 11 | | | | |
| 12 | | PURPOSE: | | |
| 13 | | Spellchecking using different backends | | |
| 14 | | | | |
| 15 | +-----------------------------------------------------------------------+ | |
| 16 | | Author: Aleksander Machniak <machniak@kolabsys.com> | | |
| 17 | | Author: Thomas Bruederli <roundcube@gmail.com> | | |
| 18 | +-----------------------------------------------------------------------+ | |
| 19 | ||
| 20 | $Id$ | |
| 21 | ||
| 22 | */ | |
| 23 | ||
| 24 | ||
| 25 | /** | |
| 26 | * Helper class for spellchecking with Googielspell and PSpell support. | |
| 27 | * | |
| 28 | * @package Core | |
| 29 | */ | |
| 30 | class rcube_spellchecker | |
| 31 | { | |
| 32 | private $matches = array(); | |
| 33 | private $engine; | |
| 34 | private $lang; | |
| 35 | private $rc; | |
| 36 | private $error; | |
| 37 | private $separator = '/[ !"#$%&()*+\\,\/\n:;<=>?@\[\]^_{|}-]+|\.[^\w]/'; | |
| 38 | ||
| 39 | ||
| 40 | // default settings | |
| 41 | const GOOGLE_HOST = 'ssl://www.google.com'; | |
| 42 | const GOOGLE_PORT = 443; | |
| 43 | const MAX_SUGGESTIONS = 10; | |
| 44 | ||
| 45 | ||
| 46 | /** | |
| 47 | * Constructor | |
| 48 | * | |
| 49 | * @param string $lang Language code | |
| 50 | */ | |
| 51 | function __construct($lang = 'en') | |
| 52 | { | |
| 53 | $this->rc = rcmail::get_instance(); | |
| 54 | $this->engine = $this->rc->config->get('spellcheck_engine', 'googie'); | |
| 644e3a | 55 | $this->lang = $lang ? $lang : 'en'; |
| b4edf7 | 56 | |
| A | 57 | if ($this->engine == 'pspell' && !extension_loaded('pspell')) { |
| 58 | raise_error(array( | |
| 59 | 'code' => 500, 'type' => 'php', | |
| 60 | 'file' => __FILE__, 'line' => __LINE__, | |
| 61 | 'message' => "Pspell extension not available"), true, true); | |
| 62 | } | |
| 63 | } | |
| 64 | ||
| 65 | ||
| 66 | /** | |
| 67 | * Set content and check spelling | |
| 68 | * | |
| 69 | * @param string $text Text content for spellchecking | |
| 70 | * @param bool $is_html Enables HTML-to-Text conversion | |
| 71 | * | |
| 72 | * @return bool True when no mispelling found, otherwise false | |
| 73 | */ | |
| 74 | function check($text, $is_html=false) | |
| 75 | { | |
| 76 | // convert to plain text | |
| 77 | if ($is_html) { | |
| 78 | $this->content = $this->html2text($text); | |
| 79 | } | |
| 80 | else { | |
| 81 | $this->content = $text; | |
| 82 | } | |
| 83 | ||
| 84 | if ($this->engine == 'pspell') { | |
| 85 | $this->matches = $this->_pspell_check($this->content); | |
| 86 | } | |
| 87 | else { | |
| 88 | $this->matches = $this->_googie_check($this->content); | |
| 89 | } | |
| 90 | ||
| 91 | return $this->found() == 0; | |
| 92 | } | |
| 93 | ||
| 94 | ||
| 95 | /** | |
| 96 | * Number of mispellings found (after check) | |
| 97 | * | |
| 98 | * @return int Number of mispellings | |
| 99 | */ | |
| 100 | function found() | |
| 101 | { | |
| 102 | return count($this->matches); | |
| 103 | } | |
| 104 | ||
| 105 | ||
| 106 | /** | |
| 107 | * Returns suggestions for the specified word | |
| 108 | * | |
| 109 | * @param string $word The word | |
| 110 | * | |
| 111 | * @return array Suggestions list | |
| 112 | */ | |
| 113 | function get_suggestions($word) | |
| 114 | { | |
| 115 | if ($this->engine == 'pspell') { | |
| 116 | return $this->_pspell_suggestions($word); | |
| 117 | } | |
| 118 | ||
| 119 | return $this->_googie_suggestions($word); | |
| 120 | } | |
| 121 | ||
| 122 | ||
| 123 | /** | |
| 124 | * Returns mispelled words | |
| 125 | * | |
| 126 | * @param string $text The content for spellchecking. If empty content | |
| 127 | * used for check() method will be used. | |
| 128 | * | |
| 129 | * @return array List of mispelled words | |
| 130 | */ | |
| 131 | function get_words($text = null, $is_html=false) | |
| 132 | { | |
| 133 | if ($this->engine == 'pspell') { | |
| 134 | return $this->_pspell_words($text, $is_html); | |
| 135 | } | |
| 136 | ||
| 137 | return $this->_googie_words($text, $is_html); | |
| 138 | } | |
| 139 | ||
| 140 | ||
| 141 | /** | |
| 142 | * Returns checking result in XML (Googiespell) format | |
| 143 | * | |
| 144 | * @return string XML content | |
| 145 | */ | |
| 146 | function get_xml() | |
| 147 | { | |
| 148 | // send output | |
| 149 | $out = '<?xml version="1.0" encoding="'.RCMAIL_CHARSET.'"?><spellresult charschecked="'.mb_strlen($this->content).'">'; | |
| 150 | ||
| 151 | foreach ($this->matches as $item) { | |
| 152 | $out .= '<c o="'.$item[1].'" l="'.$item[2].'">'; | |
| 153 | $out .= is_array($item[4]) ? implode("\t", $item[4]) : $item[4]; | |
| 154 | $out .= '</c>'; | |
| 155 | } | |
| 156 | ||
| 157 | $out .= '</spellresult>'; | |
| 158 | ||
| 159 | return $out; | |
| 160 | } | |
| 161 | ||
| 162 | ||
| 163 | /** | |
| 644e3a | 164 | * Returns checking result (mispelled words with suggestions) |
| A | 165 | * |
| 166 | * @return array Spellchecking result. An array indexed by word. | |
| 167 | */ | |
| 168 | function get() | |
| 169 | { | |
| 170 | $result = array(); | |
| 171 | ||
| 172 | foreach ($this->matches as $item) { | |
| 173 | if ($this->engine == 'pspell') { | |
| 174 | $word = $item[0]; | |
| 175 | } | |
| 176 | else { | |
| 177 | $word = mb_substr($this->content, $item[1], $item[2], RCMAIL_CHARSET); | |
| 178 | } | |
| 179 | $result[$word] = is_array($item[4]) ? implode("\t", $item[4]) : $item[4]; | |
| 180 | } | |
| 181 | ||
| 182 | return $out; | |
| 183 | } | |
| 184 | ||
| 185 | ||
| 186 | /** | |
| b4edf7 | 187 | * Returns error message |
| A | 188 | * |
| 189 | * @return string Error message | |
| 190 | */ | |
| 191 | function error() | |
| 192 | { | |
| 193 | return $this->error; | |
| 194 | } | |
| 195 | ||
| 196 | ||
| 197 | /** | |
| 198 | * Checks the text using pspell | |
| 199 | * | |
| 200 | * @param string $text Text content for spellchecking | |
| 201 | */ | |
| 202 | private function _pspell_check($text) | |
| 203 | { | |
| 204 | // init spellchecker | |
| 205 | $this->_pspell_init(); | |
| 206 | ||
| 207 | if (!$this->plink) { | |
| 208 | return array(); | |
| 209 | } | |
| 210 | ||
| 211 | // tokenize | |
| 212 | $text = preg_split($this->separator, $text, NULL, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE); | |
| 213 | ||
| 214 | $diff = 0; | |
| 215 | $matches = array(); | |
| 216 | ||
| 217 | foreach ($text as $w) { | |
| 218 | $word = trim($w[0]); | |
| 219 | $pos = $w[1] - $diff; | |
| 220 | $len = mb_strlen($word); | |
| 221 | ||
| 222 | if ($word && preg_match('/[^0-9\.]/', $word) && !pspell_check($this->plink, $word)) { | |
| 223 | $suggestions = pspell_suggest($this->plink, $word); | |
| 224 | ||
| 225 | if (sizeof($suggestions) > self::MAX_SUGGESTIONS) | |
| 226 | $suggestions = array_slice($suggestions, 0, self::MAX_SUGGESTIONS); | |
| 227 | ||
| 228 | $matches[] = array($word, $pos, $len, null, $suggestions); | |
| 229 | } | |
| 230 | ||
| 231 | $diff += (strlen($word) - $len); | |
| 232 | } | |
| 233 | ||
| 234 | return $matches; | |
| 235 | } | |
| 236 | ||
| 237 | ||
| 238 | /** | |
| 239 | * Returns the mispelled words | |
| 240 | */ | |
| 241 | private function _pspell_words($text = null, $is_html=false) | |
| 242 | { | |
| 243 | if ($text) { | |
| 244 | // init spellchecker | |
| 245 | $this->_pspell_init(); | |
| 246 | ||
| 247 | if (!$this->plink) { | |
| 248 | return array(); | |
| 249 | } | |
| 250 | ||
| 251 | // With PSpell we don't need to get suggestions to return mispelled words | |
| 252 | if ($is_html) { | |
| 253 | $text = $this->html2text($text); | |
| 254 | } | |
| 255 | ||
| 256 | $text = preg_split($this->separator, $text, NULL, PREG_SPLIT_NO_EMPTY | PREG_SPLIT_OFFSET_CAPTURE); | |
| 257 | ||
| 258 | foreach ($text as $w) { | |
| 259 | $word = trim($w[0]); | |
| 260 | if ($word && preg_match('/[^0-9\.]/', $word) && !pspell_check($this->plink, $word)) { | |
| 261 | $result[] = $word; | |
| 262 | } | |
| 263 | } | |
| 264 | ||
| 265 | return $result; | |
| 266 | } | |
| 267 | ||
| 268 | $result = array(); | |
| 269 | ||
| 270 | foreach ($this->matches as $m) { | |
| 271 | $result[] = $m[0]; | |
| 272 | } | |
| 273 | ||
| 274 | return $result; | |
| 275 | } | |
| 276 | ||
| 277 | ||
| 278 | /** | |
| 279 | * Returns suggestions for mispelled word | |
| 280 | */ | |
| 281 | private function _pspell_suggestions($word) | |
| 282 | { | |
| 283 | // init spellchecker | |
| 284 | $this->_pspell_init(); | |
| 285 | ||
| 286 | if (!$this->plink) { | |
| 287 | return array(); | |
| 288 | } | |
| 289 | ||
| 290 | $suggestions = pspell_suggest($this->plink, $word); | |
| 291 | ||
| 292 | if (sizeof($suggestions) > self::MAX_SUGGESTIONS) | |
| 293 | $suggestions = array_slice($suggestions, 0, self::MAX_SUGGESTIONS); | |
| 294 | ||
| 295 | return is_array($suggestions) ? $suggestions : array(); | |
| 296 | } | |
| 297 | ||
| 298 | ||
| 299 | /** | |
| 300 | * Initializes PSpell dictionary | |
| 301 | */ | |
| 302 | private function _pspell_init() | |
| 303 | { | |
| 304 | if (!$this->plink) { | |
| 305 | $this->plink = pspell_new($this->lang, null, null, RCMAIL_CHARSET, PSPELL_FAST); | |
| 306 | } | |
| 307 | ||
| 308 | if (!$this->plink) { | |
| 309 | $this->error = "Unable to load Pspell engine for selected language"; | |
| 310 | } | |
| 311 | } | |
| 312 | ||
| 313 | ||
| 314 | private function _googie_check($text) | |
| 315 | { | |
| 316 | // spell check uri is configured | |
| 317 | $url = $this->rc->config->get('spellcheck_uri'); | |
| 318 | ||
| 319 | if ($url) { | |
| 320 | $a_uri = parse_url($url); | |
| 321 | $ssl = ($a_uri['scheme'] == 'https' || $a_uri['scheme'] == 'ssl'); | |
| 322 | $port = $a_uri['port'] ? $a_uri['port'] : ($ssl ? 443 : 80); | |
| 323 | $host = ($ssl ? 'ssl://' : '') . $a_uri['host']; | |
| 324 | $path = $a_uri['path'] . ($a_uri['query'] ? '?'.$a_uri['query'] : '') . $this->lang; | |
| 325 | } | |
| 326 | else { | |
| 327 | $host = self::GOOGLE_HOST; | |
| 328 | $port = self::GOOGLE_PORT; | |
| 329 | $path = '/tbproxy/spell?lang=' . $this->lang; | |
| 330 | } | |
| 331 | ||
| 332 | // Google has some problem with spaces, use \n instead | |
| 333 | $text = str_replace(' ', "\n", $text); | |
| 334 | ||
| 335 | $text = '<?xml version="1.0" encoding="utf-8" ?>' | |
| 336 | .'<spellrequest textalreadyclipped="0" ignoredups="0" ignoredigits="1" ignoreallcaps="1">' | |
| 337 | .'<text>' . $text . '</text>' | |
| 338 | .'</spellrequest>'; | |
| 339 | ||
| 340 | $store = ''; | |
| 341 | if ($fp = fsockopen($host, $port, $errno, $errstr, 30)) { | |
| 342 | $out = "POST $path HTTP/1.0\r\n"; | |
| 343 | $out .= "Host: " . str_replace('ssl://', '', $host) . "\r\n"; | |
| 344 | $out .= "Content-Length: " . strlen($text) . "\r\n"; | |
| 345 | $out .= "Content-Type: application/x-www-form-urlencoded\r\n"; | |
| 346 | $out .= "Connection: Close\r\n\r\n"; | |
| 347 | $out .= $text; | |
| 348 | fwrite($fp, $out); | |
| 349 | ||
| 350 | while (!feof($fp)) | |
| 351 | $store .= fgets($fp, 128); | |
| 352 | fclose($fp); | |
| 353 | } | |
| 354 | ||
| 355 | if (!$store) { | |
| 356 | $this->error = "Empty result from spelling engine"; | |
| 357 | } | |
| 358 | ||
| 359 | preg_match_all('/<c o="([^"]*)" l="([^"]*)" s="([^"]*)">([^<]*)<\/c>/', $store, $matches, PREG_SET_ORDER); | |
| 360 | ||
| 361 | return $matches; | |
| 362 | } | |
| 363 | ||
| 364 | ||
| 365 | private function _googie_words($text = null, $is_html=false) | |
| 366 | { | |
| 367 | if ($text) { | |
| 368 | if ($is_html) { | |
| 369 | $text = $this->html2text($text); | |
| 370 | } | |
| 371 | ||
| 372 | $matches = $this->_googie_check($text); | |
| 373 | } | |
| 374 | else { | |
| 375 | $matches = $this->matches; | |
| 376 | $text = $this->content; | |
| 377 | } | |
| 378 | ||
| 379 | $result = array(); | |
| 380 | ||
| 381 | foreach ($matches as $m) { | |
| 382 | $result[] = mb_substr($text, $m[1], $m[2], RCMAIL_CHARSET); | |
| 383 | } | |
| 384 | ||
| 385 | return $result; | |
| 386 | } | |
| 387 | ||
| 388 | ||
| 389 | private function _googie_suggestions($word) | |
| 390 | { | |
| 391 | if ($word) { | |
| 392 | $matches = $this->_googie_check($word); | |
| 393 | } | |
| 394 | else { | |
| 395 | $matches = $this->matches; | |
| 396 | } | |
| 397 | ||
| 398 | if ($matches[0][4]) { | |
| 399 | $suggestions = explode("\t", $matches[0][4]); | |
| 400 | if (sizeof($suggestions) > self::MAX_SUGGESTIONS) { | |
| 401 | $suggestions = array_slice($suggestions, 0, MAX_SUGGESTIONS); | |
| 402 | } | |
| 403 | ||
| 404 | return $suggestions; | |
| 405 | } | |
| 406 | ||
| 407 | return array(); | |
| 408 | } | |
| 409 | ||
| 410 | ||
| 411 | private function html2text($text) | |
| 412 | { | |
| 413 | $h2t = new html2text($text, false, true, 0); | |
| 414 | return $h2t->get_text(); | |
| 415 | } | |
| 416 | } | |
