[1] | 1 | <?php |
---|
| 2 | /** |
---|
[362] | 3 | * The Strangecode Codebase - a general application development framework for PHP |
---|
| 4 | * For details visit the project site: <http://trac.strangecode.com/codebase/> |
---|
[396] | 5 | * Copyright 2001-2012 Strangecode, LLC |
---|
[362] | 6 | * |
---|
| 7 | * This file is part of The Strangecode Codebase. |
---|
| 8 | * |
---|
| 9 | * The Strangecode Codebase is free software: you can redistribute it and/or |
---|
| 10 | * modify it under the terms of the GNU General Public License as published by the |
---|
| 11 | * Free Software Foundation, either version 3 of the License, or (at your option) |
---|
| 12 | * any later version. |
---|
| 13 | * |
---|
| 14 | * The Strangecode Codebase is distributed in the hope that it will be useful, but |
---|
| 15 | * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or |
---|
| 16 | * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more |
---|
| 17 | * details. |
---|
| 18 | * |
---|
| 19 | * You should have received a copy of the GNU General Public License along with |
---|
| 20 | * The Strangecode Codebase. If not, see <http://www.gnu.org/licenses/>. |
---|
| 21 | */ |
---|
| 22 | |
---|
| 23 | /** |
---|
[1] | 24 | * SpellCheck.inc.php |
---|
| 25 | * |
---|
| 26 | * Interface to PHP's pspell functions. |
---|
| 27 | * |
---|
| 28 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
[15] | 29 | * @version 1.1 |
---|
[1] | 30 | */ |
---|
[42] | 31 | |
---|
[1] | 32 | /* Implementation example: |
---|
| 33 | -------------------------------------------------------------------------------- |
---|
[15] | 34 | include_once dirname(__FILE__) . '/_config.inc.php'; |
---|
[1] | 35 | include 'codebase/lib/SpellCheck.inc.php'; |
---|
| 36 | |
---|
[15] | 37 | // Instantiate with parameters. In this example we'll set the language and the path to the personal wordlist file. |
---|
| 38 | $spell = new SpellCheck(array( |
---|
[42] | 39 | 'language' => 'en', |
---|
[15] | 40 | 'personal_wordlist' => '/tmp/my_custom_dict' |
---|
| 41 | )); |
---|
[1] | 42 | |
---|
[15] | 43 | // Just for the heck of it add a new word to persistent personal wordlist file. |
---|
| 44 | $spell->add('mealworm'); |
---|
| 45 | |
---|
[1] | 46 | $text_to_check = 'donky rinds taste like mealworm paste'; |
---|
| 47 | |
---|
| 48 | if (!$spell->checkString($text_to_check)) { |
---|
| 49 | $suggestions = $spell->getStringSuggestions($text_to_check); |
---|
[15] | 50 | echo 'Spelling errors! Here are suggested alternatives:'; |
---|
[1] | 51 | print_r($suggestions); |
---|
| 52 | } else { |
---|
| 53 | echo 'No spelling errors'; |
---|
| 54 | } |
---|
| 55 | |
---|
| 56 | // Save added words to persistent custom wordlist file. |
---|
| 57 | $spell->save(); |
---|
| 58 | -------------------------------------------------------------------------------- |
---|
| 59 | */ |
---|
| 60 | |
---|
| 61 | class SpellCheck { |
---|
| 62 | |
---|
| 63 | var $_params = array( |
---|
[15] | 64 | 'language' => 'en', |
---|
| 65 | 'personal_wordlist' => '', // Text file to save custom words to. |
---|
| 66 | 'skip_length' => 3, // Words with this many chars or less will not be checked. |
---|
[1] | 67 | 'mode' => PSPELL_NORMAL, // PSPELL_FAST, PSPELL_NORMAL, or PSPELL_BAD_SPELLERS. |
---|
| 68 | 'highlight_start' => '<strong style="color:red;">', |
---|
| 69 | 'highlight_end' => '</strong>', |
---|
| 70 | ); |
---|
[42] | 71 | |
---|
[1] | 72 | var $_pspell_cfg_handle; |
---|
| 73 | var $_pspell_handle; |
---|
| 74 | var $_use_personal_wordlist = false; |
---|
| 75 | var $_errors = array(); |
---|
[42] | 76 | |
---|
[1] | 77 | /** |
---|
| 78 | * Constructor. |
---|
[42] | 79 | * |
---|
[15] | 80 | * @param array $params Array of parameters (key => val pairs). |
---|
[1] | 81 | */ |
---|
[15] | 82 | function SpellCheck($params) |
---|
[1] | 83 | { |
---|
[136] | 84 | $app =& App::getInstance(); |
---|
| 85 | |
---|
[275] | 86 | if (!extension_loaded('pspell')) { |
---|
| 87 | trigger_error('Pspell module not installed', E_USER_ERROR); |
---|
| 88 | } |
---|
| 89 | |
---|
[42] | 90 | if (!is_array($params) || empty($params)) { |
---|
| 91 | trigger_error('SpellCheck parameters not set properly', E_USER_ERROR); |
---|
| 92 | } |
---|
[1] | 93 | |
---|
[42] | 94 | $this->setParam($params); |
---|
[1] | 95 | |
---|
[15] | 96 | $this->_pspell_cfg_handle = pspell_config_create($this->getParam('language')); |
---|
| 97 | |
---|
| 98 | pspell_config_ignore($this->_pspell_cfg_handle, $this->getParam('skip_length')); |
---|
| 99 | pspell_config_mode($this->_pspell_cfg_handle, $this->getParam('mode')); |
---|
| 100 | |
---|
| 101 | if ('' != $this->getParam('personal_wordlist')) { |
---|
| 102 | if (!is_writable(dirname($this->getParam('personal_wordlist'))) || !is_writable($this->getParam('personal_wordlist'))) { |
---|
[136] | 103 | $app->logMsg(sprintf('Personal wordlist file not writable: %s', $this->getParam('personal_wordlist')), LOG_WARNING, __FILE__, __LINE__); |
---|
[1] | 104 | } else { |
---|
[15] | 105 | pspell_config_personal($this->_pspell_cfg_handle, $this->getParam('personal_wordlist')); |
---|
[1] | 106 | $this->_use_personal_wordlist = true; |
---|
[136] | 107 | $app->logMsg(sprintf('Using personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
[1] | 108 | } |
---|
| 109 | } |
---|
| 110 | |
---|
| 111 | $this->_pspell_handle = pspell_new_config($this->_pspell_cfg_handle); |
---|
| 112 | } |
---|
| 113 | |
---|
| 114 | /** |
---|
| 115 | * Set (or overwrite existing) parameters by passing an array of new parameters. |
---|
| 116 | * |
---|
| 117 | * @access public |
---|
| 118 | * @param array $params Array of parameters (key => val pairs). |
---|
| 119 | */ |
---|
| 120 | function setParam($params) |
---|
| 121 | { |
---|
[136] | 122 | $app =& App::getInstance(); |
---|
| 123 | |
---|
[1] | 124 | if (isset($params) && is_array($params)) { |
---|
| 125 | // Merge new parameters with old overriding only those passed. |
---|
| 126 | $this->_params = array_merge($this->_params, $params); |
---|
| 127 | } else { |
---|
[136] | 128 | $app->logMsg(sprintf('Parameters are not an array: %s', $params), LOG_ERR, __FILE__, __LINE__); |
---|
[1] | 129 | } |
---|
| 130 | } |
---|
| 131 | |
---|
| 132 | /** |
---|
| 133 | * Return the value of a parameter, if it exists. |
---|
| 134 | * |
---|
| 135 | * @access public |
---|
| 136 | * @param string $param Which parameter to return. |
---|
| 137 | * @return mixed Configured parameter value. |
---|
| 138 | */ |
---|
| 139 | function getParam($param) |
---|
| 140 | { |
---|
[136] | 141 | $app =& App::getInstance(); |
---|
| 142 | |
---|
[1] | 143 | if (isset($this->_params[$param])) { |
---|
| 144 | return $this->_params[$param]; |
---|
| 145 | } else { |
---|
[146] | 146 | $app->logMsg(sprintf('Parameter is not set: %s', $param), LOG_DEBUG, __FILE__, __LINE__); |
---|
[1] | 147 | return null; |
---|
| 148 | } |
---|
| 149 | } |
---|
[42] | 150 | |
---|
[1] | 151 | /** |
---|
| 152 | * Check whether any errors have been triggered. |
---|
| 153 | * |
---|
| 154 | * @return bool True if any errors were found, false otherwise. |
---|
| 155 | */ |
---|
| 156 | function anyErrors() |
---|
| 157 | { |
---|
| 158 | return (sizeof($this->_errors) > 0); |
---|
| 159 | } |
---|
| 160 | |
---|
| 161 | /** |
---|
| 162 | * Reset the error list. |
---|
| 163 | */ |
---|
| 164 | function resetErrorList() |
---|
| 165 | { |
---|
| 166 | $this->_errors = array(); |
---|
| 167 | } |
---|
[42] | 168 | |
---|
[1] | 169 | /** |
---|
| 170 | * Check one word. |
---|
| 171 | * |
---|
| 172 | * @access public |
---|
| 173 | * @param string $word |
---|
| 174 | * @return bool True if word is correct. |
---|
| 175 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 176 | * @version 1.0 |
---|
| 177 | * @since 09 Jun 2005 18:23:51 |
---|
| 178 | */ |
---|
| 179 | function check($word) |
---|
| 180 | { |
---|
| 181 | if (pspell_check($this->_pspell_handle, $word)) { |
---|
| 182 | return true; |
---|
| 183 | } else { |
---|
| 184 | $this->_errors[] = $word; |
---|
| 185 | return false; |
---|
| 186 | } |
---|
| 187 | } |
---|
[42] | 188 | |
---|
[1] | 189 | /** |
---|
| 190 | * Suggest the correct spelling for one misspelled word. |
---|
| 191 | * |
---|
| 192 | * @access public |
---|
| 193 | * @param string $word |
---|
| 194 | * @return array Word suggestions. |
---|
| 195 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 196 | * @version 1.0 |
---|
| 197 | * @since 09 Jun 2005 18:23:51 |
---|
| 198 | */ |
---|
| 199 | function suggest($word) |
---|
| 200 | { |
---|
| 201 | return pspell_suggest($this->_pspell_handle, $word); |
---|
| 202 | } |
---|
[42] | 203 | |
---|
[1] | 204 | /** |
---|
| 205 | * Add a word to a personal list. |
---|
| 206 | * |
---|
| 207 | * @access public |
---|
| 208 | * @param string $word |
---|
| 209 | * @return array Word suggestions. |
---|
| 210 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 211 | * @version 1.0 |
---|
| 212 | * @since 09 Jun 2005 18:23:51 |
---|
| 213 | */ |
---|
| 214 | function add($word) |
---|
| 215 | { |
---|
[136] | 216 | $app =& App::getInstance(); |
---|
| 217 | |
---|
[1] | 218 | if ($this->_use_personal_wordlist) { |
---|
[15] | 219 | if (pspell_add_to_personal($this->_pspell_handle, $word)) { |
---|
[136] | 220 | $app->logMsg(sprintf('Added "%s" to personal wordlist: %s', $word, $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
[42] | 221 | return true; |
---|
| 222 | } else { |
---|
[136] | 223 | $app->logMsg(sprintf('Failed adding "%s" to personal wordlist: %s', $word, $this->getParam('personal_wordlist')), LOG_WARNING, __FILE__, __LINE__); |
---|
[42] | 224 | return false; |
---|
| 225 | } |
---|
[1] | 226 | } |
---|
| 227 | } |
---|
[42] | 228 | |
---|
[1] | 229 | /** |
---|
| 230 | * Save personal list to file. |
---|
| 231 | * |
---|
| 232 | * @access public |
---|
| 233 | * @param string $word |
---|
| 234 | * @return array Word suggestions. |
---|
| 235 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 236 | * @version 1.0 |
---|
| 237 | * @since 09 Jun 2005 18:23:51 |
---|
| 238 | */ |
---|
| 239 | function save() |
---|
| 240 | { |
---|
[136] | 241 | $app =& App::getInstance(); |
---|
| 242 | |
---|
[1] | 243 | if ($this->_use_personal_wordlist) { |
---|
| 244 | if (pspell_save_wordlist($this->_pspell_handle)) { |
---|
[136] | 245 | $app->logMsg(sprintf('Saved personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
[1] | 246 | return true; |
---|
| 247 | } else { |
---|
[136] | 248 | $app->logMsg(sprintf('Failed saving personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_ERR, __FILE__, __LINE__); |
---|
[1] | 249 | return false; |
---|
| 250 | } |
---|
| 251 | } |
---|
| 252 | } |
---|
[42] | 253 | |
---|
[1] | 254 | /** |
---|
[334] | 255 | * Returns an array of suggested words for each misspelled word in the given text. |
---|
[1] | 256 | * The first word of the returned array is the (possibly) misspelled word. |
---|
| 257 | * |
---|
| 258 | * @access public |
---|
| 259 | * @param string $string String to get suggestions for. |
---|
| 260 | * @return mixed Array of suggested words or false if none. |
---|
| 261 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 262 | * @version 1.0 |
---|
| 263 | * @since 09 Jun 2005 21:29:49 |
---|
| 264 | */ |
---|
| 265 | function getStringSuggestions($string) |
---|
| 266 | { |
---|
| 267 | $corrections = array(); |
---|
[425] | 268 | // Split words on punctuation except apostrophes |
---|
| 269 | // http://stackoverflow.com/questions/790596/split-a-text-into-single-words |
---|
| 270 | $words = preg_split("/((^\p{P}+)|(\p{P}*\s+\p{P}*)|[\p{Pd}-ââ]+|(\+|(\p{P}+$))/", $string); |
---|
[1] | 271 | // Remove non-word elements. |
---|
| 272 | $words = preg_grep('/\w+/', $words); |
---|
| 273 | |
---|
| 274 | if (is_array($words) && !empty($words)) { |
---|
| 275 | foreach ($words as $i => $word) { |
---|
| 276 | if (!$this->check($word)) { |
---|
| 277 | $corrections[$i] = $this->suggest($word); |
---|
| 278 | // Keep the original spelling as one of the suggestions. |
---|
| 279 | array_unshift($corrections[$i], $word); |
---|
| 280 | array_unique($corrections[$i]); |
---|
| 281 | } |
---|
| 282 | } |
---|
| 283 | } |
---|
| 284 | if (is_array($corrections) && !empty($corrections)) { |
---|
| 285 | return $corrections; |
---|
| 286 | } else { |
---|
| 287 | return false; |
---|
| 288 | } |
---|
| 289 | } |
---|
[42] | 290 | |
---|
[1] | 291 | /** |
---|
| 292 | * Checks all words in a given string. |
---|
| 293 | * |
---|
| 294 | * @access public |
---|
| 295 | * @param string $string String to check. |
---|
| 296 | * @return void |
---|
| 297 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 298 | * @version 1.0 |
---|
| 299 | * @since 09 Jun 2005 22:11:27 |
---|
| 300 | */ |
---|
| 301 | function checkString($string) |
---|
| 302 | { |
---|
| 303 | $errors = array(); |
---|
| 304 | $words = preg_split('/([\W]+?)/', $string, -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
| 305 | // Remove non-word elements. |
---|
| 306 | $check_words = preg_grep('/\w+/', $words); |
---|
| 307 | |
---|
| 308 | if (is_array($check_words) && !empty($check_words)) { |
---|
| 309 | foreach ($check_words as $i => $word) { |
---|
| 310 | if (!$this->check($word)) { |
---|
| 311 | $errors[] = $word; |
---|
| 312 | } |
---|
| 313 | } |
---|
| 314 | } |
---|
| 315 | if (empty($errors)) { |
---|
| 316 | return true; |
---|
| 317 | } else { |
---|
| 318 | $this->_errors = $errors + $this->_errors; |
---|
| 319 | return false; |
---|
| 320 | } |
---|
| 321 | } |
---|
[42] | 322 | |
---|
[1] | 323 | /** |
---|
| 324 | * Returns a given string with misspelled words highlighted. |
---|
| 325 | * |
---|
| 326 | * @access public |
---|
| 327 | * @param string $string Text to highlight. |
---|
| 328 | * @return string Highlighted text. |
---|
| 329 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 330 | * @version 1.0 |
---|
| 331 | * @since 09 Jun 2005 21:29:49 |
---|
| 332 | */ |
---|
| 333 | function getStringHighlighted($string, $show_footnote=false) |
---|
| 334 | { |
---|
| 335 | $words = preg_split('/([\W]+?)/', $string, -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
| 336 | $check_words = preg_grep('/\w+/', $words); |
---|
| 337 | $cnt = 0; |
---|
| 338 | if (is_array($check_words) && !empty($check_words)) { |
---|
| 339 | foreach ($check_words as $i => $word) { |
---|
| 340 | if (!$this->check($word)) { |
---|
| 341 | $footnote = $show_footnote ? '<sup>' . ++$cnt . '</sup>' : ''; |
---|
| 342 | $words[$i] = $this->getParam('highlight_start') . $word . $this->getParam('highlight_end') . $footnote; |
---|
| 343 | } |
---|
| 344 | } |
---|
| 345 | } |
---|
| 346 | return join('', $words); |
---|
| 347 | } |
---|
[42] | 348 | |
---|
[1] | 349 | /** |
---|
[334] | 350 | * Prints the HTML for correcting all misspellings found in the text of one $_FORM element. |
---|
[1] | 351 | * |
---|
| 352 | * @access public |
---|
| 353 | * @param string $form_name Name of the form to check. |
---|
| 354 | * @return void |
---|
| 355 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 356 | * @version 1.0 |
---|
| 357 | * @since 09 Jun 2005 21:29:49 |
---|
| 358 | */ |
---|
| 359 | function printCorrectionForm($form_name) |
---|
| 360 | { |
---|
| 361 | ?> |
---|
| 362 | <input name="<?php echo $form_name ?>" type="hidden" value="<?php echo getFormData($form_name) ?>" /> |
---|
| 363 | <?php |
---|
[42] | 364 | |
---|
[1] | 365 | $form_words = $this->getStringSuggestions(getFormData($form_name)); |
---|
| 366 | if (is_array($form_words) && !empty($form_words)) { |
---|
| 367 | ?><ol><?php |
---|
| 368 | foreach ($form_words as $i => $words) { |
---|
| 369 | ?> |
---|
| 370 | <li> |
---|
[270] | 371 | <select name="spelling_suggestions[<?php echo $form_name ?>][<?php echo $i ?>]" onchange"document.forms[0].elements['spelling_corrections[<?php echo $form_name ?>][<?php echo $i ?>]'].value = this.value;"> |
---|
[1] | 372 | <?php $original_word = array_shift($words); ?> |
---|
| 373 | <option value="<?php echo $original_word ?>">(<?php echo $original_word ?>)</option> |
---|
| 374 | <?php |
---|
[42] | 375 | |
---|
[1] | 376 | foreach ($words as $suggestion) { |
---|
| 377 | ?> |
---|
| 378 | <option value="<?php echo $suggestion ?>"><?php echo $suggestion ?></option> |
---|
| 379 | <?php |
---|
| 380 | } |
---|
[42] | 381 | |
---|
[1] | 382 | ?> |
---|
| 383 | </select> |
---|
[176] | 384 | <input type="text" name="spelling_corrections[<?php echo $form_name ?>][<?php echo $i ?>]" value="<?php echo $original_word ?>" class="sc-small" /> |
---|
[1] | 385 | <?php if ($this->_use_personal_wordlist) { ?> |
---|
| 386 | <input name="save_to_personal_wordlist[]" type="checkbox" value="<?php echo $i ?>" /><?php echo _("Learn spelling") ?> |
---|
| 387 | <?php } ?> |
---|
| 388 | </li> |
---|
| 389 | <?php |
---|
| 390 | } |
---|
| 391 | ?></ol><?php |
---|
| 392 | } |
---|
| 393 | } |
---|
[42] | 394 | |
---|
[1] | 395 | /** |
---|
| 396 | * Tests if any form spelling corrections have been submitted. |
---|
| 397 | * |
---|
| 398 | * @access public |
---|
| 399 | * @return bool True if form spelling has been checked. |
---|
| 400 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 401 | * @version 1.0 |
---|
| 402 | * @since 09 Jun 2005 23:15:35 |
---|
| 403 | */ |
---|
| 404 | function anyFormCorrections() |
---|
| 405 | { |
---|
| 406 | return (false !== getFormData('spelling_suggestions', false)) || (false !== getFormData('spelling_corrections', false)); |
---|
| 407 | } |
---|
[42] | 408 | |
---|
[1] | 409 | /** |
---|
| 410 | * Replace the misspelled words in the text of a specified form with the corrections. |
---|
| 411 | * |
---|
| 412 | * @access public |
---|
| 413 | * @param string $form_name Name of form to apply corrections to. |
---|
| 414 | * @return string Corrected form text. |
---|
| 415 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
| 416 | * @version 1.0 |
---|
| 417 | * @since 09 Jun 2005 23:18:51 |
---|
| 418 | */ |
---|
| 419 | function applyFormCorrections($form_name) |
---|
| 420 | { |
---|
| 421 | $form_words = preg_split('/([\W]+?)/', getFormData($form_name), -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
| 422 | $suggestions = getFormData('spelling_suggestions'); |
---|
| 423 | $corrections = getFormData('spelling_corrections'); |
---|
| 424 | |
---|
| 425 | $form_words = array_diff($corrections[$form_name], array('')) + $suggestions[$form_name] + $form_words; |
---|
| 426 | ksort($form_words); |
---|
| 427 | |
---|
| 428 | if ($this->_use_personal_wordlist) { |
---|
| 429 | $save_to_personal_wordlist = getFormData('save_to_personal_wordlist'); |
---|
| 430 | if (is_array($save_to_personal_wordlist) && !empty($save_to_personal_wordlist)) { |
---|
| 431 | foreach ($save_to_personal_wordlist as $cust) { |
---|
| 432 | $this->add($form_words[$cust]); |
---|
| 433 | } |
---|
| 434 | } |
---|
| 435 | $this->save(); |
---|
| 436 | } |
---|
| 437 | |
---|
| 438 | if (is_array($form_words) && !empty($form_words)) { |
---|
| 439 | return join('', $form_words); |
---|
| 440 | } else { |
---|
| 441 | return getFormData($form_name); |
---|
| 442 | } |
---|
| 443 | } |
---|
| 444 | |
---|
| 445 | } // End. |
---|
| 446 | |
---|
| 447 | ?> |
---|