1: <?php
  2: /**
  3:  * Text Helper
  4:  *
  5:  * Text manipulations: Highlight, excerpt, truncate, strip of links, convert email addresses to mailto: links...
  6:  *
  7:  * PHP 5
  8:  *
  9:  * CakePHP(tm) : Rapid Development Framework (http://cakephp.org)
 10:  * Copyright 2005-2011, Cake Software Foundation, Inc. (http://cakefoundation.org)
 11:  *
 12:  * Licensed under The MIT License
 13:  * Redistributions of files must retain the above copyright notice.
 14:  *
 15:  * @copyright     Copyright 2005-2011, Cake Software Foundation, Inc. (http://cakefoundation.org)
 16:  * @link          http://cakephp.org CakePHP(tm) Project
 17:  * @package       Cake.View.Helper
 18:  * @since         CakePHP(tm) v 0.10.0.1076
 19:  * @license       MIT License (http://www.opensource.org/licenses/mit-license.php)
 20:  */
 21: 
 22: App::uses('AppHelper', 'View/Helper');
 23: App::uses('HtmlHelper', 'Helper');
 24: App::uses('Multibyte', 'I18n');
 25: 
 26: /**
 27:  * Text helper library.
 28:  *
 29:  * Text manipulations: Highlight, excerpt, truncate, strip of links, convert email addresses to mailto: links...
 30:  *
 31:  * @package       Cake.View.Helper
 32:  * @property      HtmlHelper $Html
 33:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html
 34:  */
 35: class TextHelper extends AppHelper {
 36: 
 37: /**
 38:  * helpers
 39:  *
 40:  * @var array
 41:  */
 42:     public $helpers = array('Html');
 43: 
 44: /**
 45:  * Highlights a given phrase in a text. You can specify any expression in highlighter that
 46:  * may include the \1 expression to include the $phrase found.
 47:  *
 48:  * ### Options:
 49:  *
 50:  * - `format` The piece of html with that the phrase will be highlighted
 51:  * - `html` If true, will ignore any HTML tags, ensuring that only the correct text is highlighted
 52:  *
 53:  * @param string $text Text to search the phrase in
 54:  * @param string $phrase The phrase that will be searched
 55:  * @param array $options An array of html attributes and options.
 56:  * @return string The highlighted text
 57:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::highlight
 58:  */
 59:     public function highlight($text, $phrase, $options = array()) {
 60:         if (empty($phrase)) {
 61:             return $text;
 62:         }
 63: 
 64:         $default = array(
 65:             'format' => '<span class="highlight">\1</span>',
 66:             'html' => false
 67:         );
 68:         $options = array_merge($default, $options);
 69:         extract($options);
 70: 
 71:         if (is_array($phrase)) {
 72:             $replace = array();
 73:             $with = array();
 74: 
 75:             foreach ($phrase as $key => $segment) {
 76:                 $segment = '(' . preg_quote($segment, '|') . ')';
 77:                 if ($html) {
 78:                     $segment = "(?![^<]+>)$segment(?![^<]+>)";
 79:                 }
 80: 
 81:                 $with[] = (is_array($format)) ? $format[$key] : $format;
 82:                 $replace[] = "|$segment|iu";
 83:             }
 84: 
 85:             return preg_replace($replace, $with, $text);
 86:         } else {
 87:             $phrase = '(' . preg_quote($phrase, '|') . ')';
 88:             if ($html) {
 89:                 $phrase = "(?![^<]+>)$phrase(?![^<]+>)";
 90:             }
 91: 
 92:             return preg_replace("|$phrase|iu", $format, $text);
 93:         }
 94:     }
 95: 
 96: /**
 97:  * Strips given text of all links (<a href=....)
 98:  *
 99:  * @param string $text Text
100:  * @return string The text without links
101:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::stripLinks
102:  */
103:     public function stripLinks($text) {
104:         return preg_replace('|<a\s+[^>]+>|im', '', preg_replace('|<\/a>|im', '', $text));
105:     }
106: 
107: /**
108:  * Adds links (<a href=....) to a given text, by finding text that begins with
109:  * strings like http:// and ftp://.
110:  *
111:  * @param string $text Text to add links to
112:  * @param array $htmlOptions Array of HTML options.
113:  * @return string The text with links
114:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::autoLinkUrls
115:  */
116:     public function autoLinkUrls($text, $htmlOptions = array()) {
117:         $this->_linkOptions = $htmlOptions;
118:         $text = preg_replace_callback(
119:             '#(?<!href="|src="|">)((?:https?|ftp|nntp)://[^\s<>()]+)#i',
120:             array(&$this, '_linkBareUrl'),
121:             $text
122:         );
123:         return preg_replace_callback(
124:             '#(?<!href="|">)(?<!http://|https://|ftp://|nntp://)(www\.[^\n\%\ <]+[^<\n\%\,\.\ <])(?<!\))#i',
125:             array(&$this, '_linkUrls'),
126:             $text
127:         );
128:     }
129: 
130: /**
131:  * Links urls that include http://
132:  *
133:  * @param array $matches
134:  * @return string
135:  * @see TextHelper::autoLinkUrls()
136:  */
137:     protected function _linkBareUrl($matches) {
138:         return $this->Html->link($matches[0], $matches[0], $this->_linkOptions);
139:     }
140: 
141: /**
142:  * Links urls missing http://
143:  *
144:  * @param array $matches
145:  * @return string
146:  * @see TextHelper::autoLinkUrls()
147:  */
148:     protected function _linkUrls($matches) {
149:         return $this->Html->link($matches[0], 'http://' . $matches[0], $this->_linkOptions);
150:     }
151: 
152: /**
153:  * Links email addresses
154:  *
155:  * @param array $matches
156:  * @return string
157:  * @see TextHelper::autoLinkUrls()
158:  */
159:     protected function _linkEmails($matches) {
160:         return $this->Html->link($matches[0], 'mailto:' . $matches[0], $this->_linkOptions);
161:     }
162: 
163: /**
164:  * Adds email links (<a href="mailto:....) to a given text.
165:  *
166:  * @param string $text Text
167:  * @param array $options Array of HTML options.
168:  * @return string The text with links
169:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::autoLinkEmails
170:  */
171:     public function autoLinkEmails($text, $options = array()) {
172:         $this->_linkOptions = $options;
173:         $atom = '[a-z0-9!#$%&\'*+\/=?^_`{|}~-]';
174:         return preg_replace_callback(
175:             '/(' . $atom . '+(?:\.' . $atom . '+)*@[a-z0-9-]+(?:\.[a-z0-9-]+)+)/i',
176:             array(&$this, '_linkEmails'),
177:             $text
178:         );
179:     }
180: 
181: /**
182:  * Convert all links and email addresses to HTML links.
183:  *
184:  * @param string $text Text
185:  * @param array $options Array of HTML options.
186:  * @return string The text with links
187:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::autoLink
188:  */
189:     public function autoLink($text, $options = array()) {
190:         return $this->autoLinkEmails($this->autoLinkUrls($text, $options), $options);
191:     }
192: 
193: /**
194:  * Truncates text.
195:  *
196:  * Cuts a string to the length of $length and replaces the last characters
197:  * with the ending if the text is longer than length.
198:  *
199:  * ### Options:
200:  *
201:  * - `ending` Will be used as Ending and appended to the trimmed string
202:  * - `exact` If false, $text will not be cut mid-word
203:  * - `html` If true, HTML tags would be handled correctly
204:  *
205:  * @param string $text String to truncate.
206:  * @param integer $length Length of returned string, including ellipsis.
207:  * @param array $options An array of html attributes and options.
208:  * @return string Trimmed string.
209:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::truncate
210:  */
211:     public function truncate($text, $length = 100, $options = array()) {
212:         $default = array(
213:             'ending' => '...', 'exact' => true, 'html' => false
214:         );
215:         $options = array_merge($default, $options);
216:         extract($options);
217: 
218:         if (!function_exists('mb_strlen')) {
219:             class_exists('Multibyte');
220:         }
221: 
222:         if ($html) {
223:             if (mb_strlen(preg_replace('/<.*?>/', '', $text)) <= $length) {
224:                 return $text;
225:             }
226:             $totalLength = mb_strlen(strip_tags($ending));
227:             $openTags = array();
228:             $truncate = '';
229: 
230:             preg_match_all('/(<\/?([\w+]+)[^>]*>)?([^<>]*)/', $text, $tags, PREG_SET_ORDER);
231:             foreach ($tags as $tag) {
232:                 if (!preg_match('/img|br|input|hr|area|base|basefont|col|frame|isindex|link|meta|param/s', $tag[2])) {
233:                     if (preg_match('/<[\w]+[^>]*>/s', $tag[0])) {
234:                         array_unshift($openTags, $tag[2]);
235:                     } else if (preg_match('/<\/([\w]+)[^>]*>/s', $tag[0], $closeTag)) {
236:                         $pos = array_search($closeTag[1], $openTags);
237:                         if ($pos !== false) {
238:                             array_splice($openTags, $pos, 1);
239:                         }
240:                     }
241:                 }
242:                 $truncate .= $tag[1];
243: 
244:                 $contentLength = mb_strlen(preg_replace('/&[0-9a-z]{2,8};|&#[0-9]{1,7};|&#x[0-9a-f]{1,6};/i', ' ', $tag[3]));
245:                 if ($contentLength + $totalLength > $length) {
246:                     $left = $length - $totalLength;
247:                     $entitiesLength = 0;
248:                     if (preg_match_all('/&[0-9a-z]{2,8};|&#[0-9]{1,7};|&#x[0-9a-f]{1,6};/i', $tag[3], $entities, PREG_OFFSET_CAPTURE)) {
249:                         foreach ($entities[0] as $entity) {
250:                             if ($entity[1] + 1 - $entitiesLength <= $left) {
251:                                 $left--;
252:                                 $entitiesLength += mb_strlen($entity[0]);
253:                             } else {
254:                                 break;
255:                             }
256:                         }
257:                     }
258: 
259:                     $truncate .= mb_substr($tag[3], 0 , $left + $entitiesLength);
260:                     break;
261:                 } else {
262:                     $truncate .= $tag[3];
263:                     $totalLength += $contentLength;
264:                 }
265:                 if ($totalLength >= $length) {
266:                     break;
267:                 }
268:             }
269:         } else {
270:             if (mb_strlen($text) <= $length) {
271:                 return $text;
272:             } else {
273:                 $truncate = mb_substr($text, 0, $length - mb_strlen($ending));
274:             }
275:         }
276:         if (!$exact) {
277:             $spacepos = mb_strrpos($truncate, ' ');
278:             if ($html) {
279:                 $truncateCheck = mb_substr($truncate, 0, $spacepos);
280:                 $lastOpenTag = mb_strrpos($truncateCheck, '<');
281:                 $lastCloseTag = mb_strrpos($truncateCheck, '>');
282:                 if ($lastOpenTag > $lastCloseTag) {
283:                     preg_match_all('/<[\w]+[^>]*>/s', $truncate, $lastTagMatches);
284:                     $lastTag = array_pop($lastTagMatches[0]);
285:                     $spacepos = mb_strrpos($truncate, $lastTag) + mb_strlen($lastTag);
286:                 }
287:                 $bits = mb_substr($truncate, $spacepos);
288:                 preg_match_all('/<\/([a-z]+)>/', $bits, $droppedTags, PREG_SET_ORDER);
289:                 if (!empty($droppedTags)) {
290:                     if (!empty($openTags)) {
291:                         foreach ($droppedTags as $closingTag) {
292:                             if (!in_array($closingTag[1], $openTags)) {
293:                                 array_unshift($openTags, $closingTag[1]);
294:                             }
295:                         }
296:                     } else {
297:                         foreach ($droppedTags as $closingTag) {
298:                             array_push($openTags, $closingTag[1]);
299:                         }
300:                     }
301:                 }
302:             }
303:             $truncate = mb_substr($truncate, 0, $spacepos);
304:         }
305:         $truncate .= $ending;
306: 
307:         if ($html) {
308:             foreach ($openTags as $tag) {
309:                 $truncate .= '</' . $tag . '>';
310:             }
311:         }
312: 
313:         return $truncate;
314:     }
315: 
316: /**
317:  * Extracts an excerpt from the text surrounding the phrase with a number of characters on each side
318:  * determined by radius.
319:  *
320:  * @param string $text String to search the phrase in
321:  * @param string $phrase Phrase that will be searched for
322:  * @param integer $radius The amount of characters that will be returned on each side of the founded phrase
323:  * @param string $ending Ending that will be appended
324:  * @return string Modified string
325:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::excerpt
326:  */
327:     public function excerpt($text, $phrase, $radius = 100, $ending = '...') {
328:         if (empty($text) or empty($phrase)) {
329:             return $this->truncate($text, $radius * 2, array('ending' => $ending));
330:         }
331: 
332:         $append = $prepend = $ending;
333: 
334:         $phraseLen = mb_strlen($phrase);
335:         $textLen = mb_strlen($text);
336: 
337:         $pos = mb_strpos(mb_strtolower($text), mb_strtolower($phrase));
338:         if ($pos === false) {
339:             return mb_substr($text, 0, $radius) . $ending;
340:         }
341: 
342:         $startPos = $pos - $radius;
343:         if ($startPos <= 0) {
344:             $startPos = 0;
345:             $prepend = '';
346:         }
347: 
348:         $endPos = $pos + $phraseLen + $radius;
349:         if ($endPos >= $textLen) {
350:             $endPos = $textLen;
351:             $append = '';
352:         }
353: 
354:         $excerpt = mb_substr($text, $startPos, $endPos - $startPos);
355:         $excerpt = $prepend . $excerpt . $append;
356:         
357:         return $excerpt;
358:     }
359: 
360: /**
361:  * Creates a comma separated list where the last two items are joined with 'and', forming natural English
362:  *
363:  * @param array $list The list to be joined
364:  * @param string $and The word used to join the last and second last items together with. Defaults to 'and'
365:  * @param string $separator The separator used to join all the other items together. Defaults to ', '
366:  * @return string The glued together string.
367:  * @link http://book.cakephp.org/2.0/en/core-libraries/helpers/text.html#TextHelper::toList
368:  */
369:     public function toList($list, $and = 'and', $separator = ', ') {
370:         if (count($list) > 1) {
371:             return implode($separator, array_slice($list, null, -1)) . ' ' . $and . ' ' . array_pop($list);
372:         } else {
373:             return array_pop($list);
374:         }
375:     }
376: }
377: