1 | <?php |
---|
2 | /** |
---|
3 | * SpellCheck.inc.php |
---|
4 | * code by strangecode :: www.strangecode.com :: this document contains copyrighted information |
---|
5 | * |
---|
6 | * Interface to PHP's pspell functions. |
---|
7 | * |
---|
8 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
9 | * @version 1.0 |
---|
10 | */ |
---|
11 | class SpellCheck { |
---|
12 | |
---|
13 | // Default parameters. |
---|
14 | var $_params = array( |
---|
15 | 'language' => 'en', |
---|
16 | 'personal_wordlist' => '', |
---|
17 | 'skip_length' => 3, |
---|
18 | 'mode' => PSPELL_NORMAL, // PSPELL_FAST, PSPELL_NORMAL, or PSPELL_BAD_SPELLERS. |
---|
19 | 'highlight_start' => '<strong style="color:red;">', |
---|
20 | 'highlight_end' => '</strong>', |
---|
21 | ); |
---|
22 | |
---|
23 | var $_pspell_cfg_handle; |
---|
24 | var $_pspell_handle; |
---|
25 | var $_use_personal_wordlist = false; |
---|
26 | var $_errors = array(); |
---|
27 | |
---|
28 | /** |
---|
29 | * Constructor. |
---|
30 | * |
---|
31 | * @param array $params Array of parameters (key => val pairs). |
---|
32 | */ |
---|
33 | function SpellCheck($params) |
---|
34 | { |
---|
35 | if (!is_array($params) || empty($params)) { |
---|
36 | trigger_error('SpellCheck parameters not set properly', E_USER_ERROR); |
---|
37 | } |
---|
38 | |
---|
39 | $this->setParam($params); |
---|
40 | |
---|
41 | $this->_pspell_cfg_handle = pspell_config_create($this->getParam('language')); |
---|
42 | |
---|
43 | pspell_config_ignore($this->_pspell_cfg_handle, $this->getParam('skip_length')); |
---|
44 | pspell_config_mode($this->_pspell_cfg_handle, $this->getParam('mode')); |
---|
45 | |
---|
46 | if ('' != $this->getParam('personal_wordlist')) { |
---|
47 | if (!file_exists($this->getParam('personal_wordlist')) || !is_writable($this->getParam('personal_wordlist'))) { |
---|
48 | logMsg(sprintf('Personal wordlist file does not exist or is not writable: %s', $this->getParam('personal_wordlist')), LOG_WARNING, __FILE__, __LINE__); |
---|
49 | } else { |
---|
50 | pspell_config_personal($this->_pspell_cfg_handle, $this->getParam('personal_wordlist')); |
---|
51 | $this->_use_personal_wordlist = true; |
---|
52 | logMsg(sprintf('Using personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
53 | } |
---|
54 | } |
---|
55 | |
---|
56 | $this->_pspell_handle = pspell_new_config($this->_pspell_cfg_handle); |
---|
57 | } |
---|
58 | |
---|
59 | /** |
---|
60 | * Set (or overwrite existing) parameters by passing an array of new parameters. |
---|
61 | * |
---|
62 | * @access public |
---|
63 | * |
---|
64 | * @param array $params Array of parameters (key => val pairs). |
---|
65 | */ |
---|
66 | function setParam($params=null) |
---|
67 | { |
---|
68 | if (isset($params) && is_array($params)) { |
---|
69 | // Merge new parameters with old overriding only those passed. |
---|
70 | $this->_params = array_merge($this->_params, $params); |
---|
71 | } else { |
---|
72 | logMsg(sprintf('Supplied argument is not an array: %s', $params), LOG_WARNING, __FILE__, __LINE__); |
---|
73 | } |
---|
74 | } |
---|
75 | |
---|
76 | /** |
---|
77 | * Return the value of a parameter. |
---|
78 | * |
---|
79 | * @access public |
---|
80 | * @param string $param The key of the parameter to return. |
---|
81 | * @return mixed Parameter value. |
---|
82 | */ |
---|
83 | function getParam($param) |
---|
84 | { |
---|
85 | return $this->_params[$param]; |
---|
86 | } |
---|
87 | |
---|
88 | /** |
---|
89 | * Check whether any errors have been triggered. |
---|
90 | * |
---|
91 | * @return bool True if any errors were found, false otherwise. |
---|
92 | */ |
---|
93 | function anyErrors() |
---|
94 | { |
---|
95 | return (sizeof($this->_errors) > 0); |
---|
96 | } |
---|
97 | |
---|
98 | /** |
---|
99 | * Reset the error list. |
---|
100 | */ |
---|
101 | function resetErrorList() |
---|
102 | { |
---|
103 | $this->_errors = array(); |
---|
104 | } |
---|
105 | |
---|
106 | /** |
---|
107 | * Check one word. |
---|
108 | * |
---|
109 | * @access public |
---|
110 | * @param string $word |
---|
111 | * @return bool True if word is correct. |
---|
112 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
113 | * @version 1.0 |
---|
114 | * @since 09 Jun 2005 18:23:51 |
---|
115 | */ |
---|
116 | function check($word) |
---|
117 | { |
---|
118 | if (pspell_check($this->_pspell_handle, $word)) { |
---|
119 | return true; |
---|
120 | } else { |
---|
121 | $this->_errors[] = $word; |
---|
122 | return false; |
---|
123 | } |
---|
124 | } |
---|
125 | |
---|
126 | /** |
---|
127 | * Suggest the correct spelling for one misspelled word. |
---|
128 | * |
---|
129 | * @access public |
---|
130 | * @param string $word |
---|
131 | * @return array Word suggestions. |
---|
132 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
133 | * @version 1.0 |
---|
134 | * @since 09 Jun 2005 18:23:51 |
---|
135 | */ |
---|
136 | function suggest($word) |
---|
137 | { |
---|
138 | return pspell_suggest($this->_pspell_handle, $word); |
---|
139 | } |
---|
140 | |
---|
141 | /** |
---|
142 | * Add a word to a personal list. |
---|
143 | * |
---|
144 | * @access public |
---|
145 | * @param string $word |
---|
146 | * @return array Word suggestions. |
---|
147 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
148 | * @version 1.0 |
---|
149 | * @since 09 Jun 2005 18:23:51 |
---|
150 | */ |
---|
151 | function add($word) |
---|
152 | { |
---|
153 | if ($this->_use_personal_wordlist) { |
---|
154 | if (pspell_add_to_personal($this->_pspell_handle, $word)) { |
---|
155 | logMsg(sprintf('Added "%s" to personal wordlist: %s', $word, $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
156 | return true; |
---|
157 | } else { |
---|
158 | logMsg(sprintf('Failed adding "%s" to personal wordlist: %s', $word, $this->getParam('personal_wordlist')), LOG_ERR, __FILE__, __LINE__); |
---|
159 | return false; |
---|
160 | } |
---|
161 | } |
---|
162 | } |
---|
163 | |
---|
164 | /** |
---|
165 | * Save personal list to file. |
---|
166 | * |
---|
167 | * @access public |
---|
168 | * @param string $word |
---|
169 | * @return array Word suggestions. |
---|
170 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
171 | * @version 1.0 |
---|
172 | * @since 09 Jun 2005 18:23:51 |
---|
173 | */ |
---|
174 | function save() |
---|
175 | { |
---|
176 | if ($this->_use_personal_wordlist) { |
---|
177 | if (pspell_save_wordlist($this->_pspell_handle)) { |
---|
178 | logMsg(sprintf('Saved personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
179 | return true; |
---|
180 | } else { |
---|
181 | logMsg(sprintf('Failed saving personal wordlist: %s', $this->getParam('personal_wordlist')), LOG_DEBUG, __FILE__, __LINE__); |
---|
182 | return false; |
---|
183 | } |
---|
184 | } |
---|
185 | } |
---|
186 | |
---|
187 | /** |
---|
188 | * Returns an array of suggested words for each misspelled word in the given text. |
---|
189 | * The first word of the returned array is the (possibly) misspelled word. |
---|
190 | * |
---|
191 | * @access public |
---|
192 | * @param string $string String to get suggestions for. |
---|
193 | * @return mixed Array of suggested words or false if none. |
---|
194 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
195 | * @version 1.0 |
---|
196 | * @since 09 Jun 2005 21:29:49 |
---|
197 | */ |
---|
198 | function getStringSuggestions($string) |
---|
199 | { |
---|
200 | $corrections = array(); |
---|
201 | // Split words on punctuation except apostrophes (this regex is used in several places in this class). |
---|
202 | // http://stackoverflow.com/questions/790596/split-a-text-into-single-words |
---|
203 | $words = preg_split("/((?:^\p{P}+)|(?:\p{P}*\s+\p{P}*)|[\p{Pd}ââ-]+|(?:\p{P}+$))/", $string, -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
204 | if (is_array($words) && !empty($words)) { |
---|
205 | // Remove non-word elements. |
---|
206 | $words = preg_grep('/\w+/', $words); |
---|
207 | $words = array_map('strip_tags', $words); |
---|
208 | foreach ($words as $i => $word) { |
---|
209 | if (!$this->check($word)) { |
---|
210 | $corrections[$i] = $this->suggest($word); |
---|
211 | // Keep the original spelling as one of the suggestions. |
---|
212 | array_unshift($corrections[$i], $word); |
---|
213 | array_unique($corrections[$i]); |
---|
214 | } |
---|
215 | } |
---|
216 | } |
---|
217 | if (is_array($corrections) && !empty($corrections)) { |
---|
218 | return $corrections; |
---|
219 | } else { |
---|
220 | return false; |
---|
221 | } |
---|
222 | } |
---|
223 | |
---|
224 | /** |
---|
225 | * Checks all words in a given string. |
---|
226 | * |
---|
227 | * @access public |
---|
228 | * @param string $string String to check. |
---|
229 | * @return void |
---|
230 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
231 | * @version 1.0 |
---|
232 | * @since 09 Jun 2005 22:11:27 |
---|
233 | */ |
---|
234 | function checkString($string) |
---|
235 | { |
---|
236 | $errors = array(); |
---|
237 | // Split words on punctuation except apostrophes (this regex is used in several places in this class). |
---|
238 | // http://stackoverflow.com/questions/790596/split-a-text-into-single-words |
---|
239 | $words = preg_split("/((?:^\p{P}+)|(?:\p{P}*\s+\p{P}*)|[\p{Pd}ââ-]+|(?:\p{P}+$))/", $string, -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
240 | if (is_array($words) && !empty($words)) { |
---|
241 | // Remove non-word elements. |
---|
242 | $words = preg_grep('/\w+/', $words); |
---|
243 | $words = array_map('strip_tags', $words); |
---|
244 | foreach ($words as $i => $word) { |
---|
245 | if (!$this->check($word)) { |
---|
246 | $errors[] = $word; |
---|
247 | } |
---|
248 | } |
---|
249 | } |
---|
250 | if (empty($errors)) { |
---|
251 | return true; |
---|
252 | } else { |
---|
253 | $this->_errors = $errors + $this->_errors; |
---|
254 | return false; |
---|
255 | } |
---|
256 | } |
---|
257 | |
---|
258 | /** |
---|
259 | * Returns a given string with misspelled words highlighted. |
---|
260 | * |
---|
261 | * @access public |
---|
262 | * @param string $string Text to highlight. |
---|
263 | * @return string Highlighted text. |
---|
264 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
265 | * @version 1.0 |
---|
266 | * @since 09 Jun 2005 21:29:49 |
---|
267 | */ |
---|
268 | function getStringHighlighted($string, $show_footnote=false) |
---|
269 | { |
---|
270 | // Split words on punctuation except apostrophes (this regex is used in several places in this class). |
---|
271 | // http://stackoverflow.com/questions/790596/split-a-text-into-single-words |
---|
272 | $words = preg_split("/((?:^\p{P}+)|(?:\p{P}*\s+\p{P}*)|[\p{Pd}ââ-]+|(?:\p{P}+$))/", $string, -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
273 | $cnt = 0; |
---|
274 | if (is_array($words) && !empty($words)) { |
---|
275 | $words = preg_grep('/\w+/', $words); |
---|
276 | $words = array_map('strip_tags', $words); |
---|
277 | foreach ($words as $i => $word) { |
---|
278 | if (!$this->check($word)) { |
---|
279 | $footnote = $show_footnote ? '<sup style="color:#999;">' . ++$cnt . '</sup>' : ''; |
---|
280 | $words[$i] = $this->getParam('highlight_start') . $word . $this->getParam('highlight_end') . $footnote; |
---|
281 | $string = preg_replace("/\b$word\b/", $words[$i], $string); |
---|
282 | } |
---|
283 | } |
---|
284 | } |
---|
285 | return $string; |
---|
286 | // return join(' ', $words); |
---|
287 | } |
---|
288 | |
---|
289 | /** |
---|
290 | * Prints the HTML for correcting all misspellings found in the text of one $_FORM element. |
---|
291 | * |
---|
292 | * @access public |
---|
293 | * @param string $form_name Name of the form to check. |
---|
294 | * @return void |
---|
295 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
296 | * @version 1.0 |
---|
297 | * @since 09 Jun 2005 21:29:49 |
---|
298 | */ |
---|
299 | function printCorrectionForm($form_name) |
---|
300 | { |
---|
301 | ?> |
---|
302 | <input name="<?php echo $form_name ?>" type="hidden" value="<?php echo oTxt(getFormData($form_name)) ?>" /> |
---|
303 | <?php |
---|
304 | |
---|
305 | $form_words = $this->getStringSuggestions(getFormData($form_name)); |
---|
306 | if (is_array($form_words) && !empty($form_words)) { |
---|
307 | ?><ol><?php |
---|
308 | $j = 1; |
---|
309 | foreach ($form_words as $i => $words) { |
---|
310 | ?> |
---|
311 | <li> |
---|
312 | <label style="color:#999;"><sub style="vertical-align:text-top;"><?php echo $j++; ?></sub></label> |
---|
313 | <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;"> |
---|
314 | <?php $original_word = array_shift($words); ?> |
---|
315 | <option value="<?php echo $original_word ?>">(<?php echo $original_word ?>)</option> |
---|
316 | <?php |
---|
317 | |
---|
318 | foreach ($words as $suggestion) { |
---|
319 | ?> |
---|
320 | <option value="<?php echo $suggestion ?>"><?php echo $suggestion ?></option> |
---|
321 | <?php |
---|
322 | } |
---|
323 | |
---|
324 | ?> |
---|
325 | </select> |
---|
326 | <input type="text" name="spelling_corrections[<?php echo $form_name ?>][<?php echo $i ?>]" value="<?php echo oTxt($original_word) ?>" size="20"> |
---|
327 | <?php if ($this->_use_personal_wordlist) { ?> |
---|
328 | <input name="save_to_personal_wordlist[]" type="checkbox" value="<?php echo $i ?>" /><?php echo _("Learn spelling") ?> |
---|
329 | <?php } ?> |
---|
330 | </li> |
---|
331 | <?php |
---|
332 | } |
---|
333 | ?></ol><?php |
---|
334 | } |
---|
335 | } |
---|
336 | |
---|
337 | /** |
---|
338 | * Tests if any form spelling corrections have been submitted. |
---|
339 | * |
---|
340 | * @access public |
---|
341 | * @return bool True if form spelling has been checked. |
---|
342 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
343 | * @version 1.0 |
---|
344 | * @since 09 Jun 2005 23:15:35 |
---|
345 | */ |
---|
346 | function anyFormCorrections() |
---|
347 | { |
---|
348 | return (false !== getFormData('spelling_suggestions', false)) || (false !== getFormData('spelling_corrections', false)); |
---|
349 | } |
---|
350 | |
---|
351 | /** |
---|
352 | * Replace the misspelled words in the text of a specified form with the corrections. |
---|
353 | * |
---|
354 | * @access public |
---|
355 | * @param string $form_name Name of form to apply corrections to. |
---|
356 | * @return string Corrected form text. |
---|
357 | * @author Quinn Comendant <quinn@strangecode.com> |
---|
358 | * @version 1.0 |
---|
359 | * @since 09 Jun 2005 23:18:51 |
---|
360 | */ |
---|
361 | function applyFormCorrections($form_name) |
---|
362 | { |
---|
363 | // Split words on punctuation except apostrophes (this regex is used in several places in this class). |
---|
364 | // http://stackoverflow.com/questions/790596/split-a-text-into-single-words |
---|
365 | $form_words = preg_split("/((?:^\p{P}+)|(?:\p{P}*\s+\p{P}*)|[\p{Pd}ââ-]+|(?:\p{P}+$))/", getFormData($form_name), -1, PREG_SPLIT_DELIM_CAPTURE); |
---|
366 | $suggestions = getFormData('spelling_suggestions'); |
---|
367 | $corrections = getFormData('spelling_corrections'); |
---|
368 | |
---|
369 | $form_words = array_diff($corrections[$form_name], array('')) + $suggestions[$form_name] + $form_words; |
---|
370 | ksort($form_words); |
---|
371 | |
---|
372 | if ($this->_use_personal_wordlist) { |
---|
373 | $save_to_personal_wordlist = getFormData('save_to_personal_wordlist'); |
---|
374 | if (is_array($save_to_personal_wordlist) && !empty($save_to_personal_wordlist)) { |
---|
375 | foreach ($save_to_personal_wordlist as $cust) { |
---|
376 | $this->add($form_words[$cust]); |
---|
377 | } |
---|
378 | } |
---|
379 | $this->save(); |
---|
380 | } |
---|
381 | |
---|
382 | if (is_array($form_words) && !empty($form_words)) { |
---|
383 | return join('', $form_words); |
---|
384 | } else { |
---|
385 | return getFormData($form_name); |
---|
386 | } |
---|
387 | } |
---|
388 | |
---|
389 | } // End. |
---|
390 | |
---|
391 | ?> |
---|