| 1 | <?php |
| 2 | /** |
| 3 | * PHP pspell spellcheck class functions |
| 4 | * @copyright 2006-2012 The SquirrelMail Project Team |
| 5 | * @license http://opensource.org/licenses/gpl-license.php GNU Public License |
| 6 | * @version $Id$ |
| 7 | * @package plugins |
| 8 | * @subpackage squirrelspell |
| 9 | */ |
| 10 | |
| 11 | /** |
| 12 | * PHP Pspell class |
| 13 | * @package plugins |
| 14 | * @subpackage squirrelspell |
| 15 | */ |
| 16 | class php_pspell extends squirrelspell { |
| 17 | // |
| 18 | var $dict = 'en'; |
| 19 | var $subdict = ''; |
| 20 | var $jargon = ''; |
| 21 | var $charset = 'utf-8'; |
| 22 | var $mode = null; |
| 23 | var $userdic = array(); |
| 24 | |
| 25 | /** |
| 26 | */ |
| 27 | var $missed_words = array(); |
| 28 | |
| 29 | /** |
| 30 | * Error buffer |
| 31 | * @var string |
| 32 | */ |
| 33 | var $error = ''; |
| 34 | /** |
| 35 | */ |
| 36 | var $dictionary_link = null; |
| 37 | |
| 38 | /** |
| 39 | * Constructor function |
| 40 | * @param array $aParams |
| 41 | */ |
| 42 | function php_pspell($aParams=array()) { |
| 43 | if (! extension_loaded('pspell')) { |
| 44 | return $this->set_error('Pspell extension is not available'); |
| 45 | } |
| 46 | // |
| 47 | if (isset($aParams['dictionary'])) { |
| 48 | $aDict = explode(',',$aParams['dictionary']); |
| 49 | if (isset($aDict[0])) $this->dict = trim($aDict[0]); |
| 50 | if (isset($aDict[1])) $this->subdict = trim($aDict[1]); |
| 51 | if (isset($aDict[2])) $this->jargon = trim($aDict[2]); |
| 52 | } |
| 53 | if (isset($aParams['charset'])) { |
| 54 | $this->charset = $aParams['charset']; |
| 55 | } |
| 56 | if (isset($aParams['userdic'])) { |
| 57 | $this->userdic = $aParams['userdic']; |
| 58 | } |
| 59 | if (isset($aParams['mode'])) { |
| 60 | $this->mode = $aParams['mode']; |
| 61 | } else { |
| 62 | $this->mode = PSPELL_FAST; |
| 63 | } |
| 64 | // dict, subdict, jargon, charset, spellcheck_type |
| 65 | $this->dictionary_link = pspell_new($this->dict,$this->subdict,$this->jargon,$this->charset,$this->mode); |
| 66 | } |
| 67 | |
| 68 | // private functions |
| 69 | function check_word($sWord) { |
| 70 | return pspell_check($this->dictionary_link,$sWord); |
| 71 | } |
| 72 | |
| 73 | function suggest($sWord) { |
| 74 | return pspell_suggest($this->dictionary_link,$sWord); |
| 75 | } |
| 76 | |
| 77 | // public function |
| 78 | |
| 79 | /** |
| 80 | * Check block of text |
| 81 | * @return array |
| 82 | */ |
| 83 | function check_text($sText) { |
| 84 | // resets missed words array |
| 85 | $this->missed_words = array(); |
| 86 | |
| 87 | $line = 0; |
| 88 | $start = 0; |
| 89 | $position = 0; |
| 90 | $word = ''; |
| 91 | // parse text. sq_* functions are used in order to work with characters and not with bytes |
| 92 | for ($i = 0; $i <= sq_strlen($sText,$this->charset); $i++) { |
| 93 | if ($i == sq_strlen($sText,$this->charset)) { |
| 94 | // add space in order to check last $word. |
| 95 | $char = ' '; |
| 96 | } else { |
| 97 | $char = sq_substr($sText,$i,1,$this->charset); |
| 98 | } |
| 99 | // Current |
| 100 | switch($char) { |
| 101 | case ' ': |
| 102 | case '.': |
| 103 | case ';': |
| 104 | case "\t": |
| 105 | case "\r": |
| 106 | case "\n": |
| 107 | if (!empty($word)) { |
| 108 | if (isset($this->missed_words[$word]) || !$this->check_word($word)) { |
| 109 | if (! isset($this->missed_words[$word]['suggestions'])) { |
| 110 | $this->missed_words[$word]['suggestions'] = $this->suggest($word); |
| 111 | } |
| 112 | $this->missed_words[$word]['locations'][] = "$line:$start"; |
| 113 | } |
| 114 | $word = ''; |
| 115 | } |
| 116 | if ($char == "\n") { |
| 117 | $position = 0; |
| 118 | $line++; |
| 119 | } else { |
| 120 | $position++; |
| 121 | } |
| 122 | break; |
| 123 | default: |
| 124 | // a-zA-Z0-9' + 8bit chars (nbspace and other spaces excluded, depends on charset) |
| 125 | // add char to word |
| 126 | if(empty($word)) { |
| 127 | $start = $position; // squirrelspell adds one space to checked text |
| 128 | } |
| 129 | $position++; |
| 130 | $word.=$char; |
| 131 | } |
| 132 | } |
| 133 | return $this->missed_words; |
| 134 | } |
| 135 | } |