/firstrend/src/core/class/chinese.class.php

http://ownerpress.googlecode.com/ · PHP · 255 lines · 221 code · 14 blank · 20 comment · 62 complexity · 89049bc1f697832e34117fe2576a96cc MD5 · raw file 

    

  1. <?php

    2. 

  2. define('CODETABLE_DIR', FANWE_ROOT.'./public/table/');

    3. 

  3. 

    4. 

  4. /**  

    5. 

  5.  * chinese.class.php

    6. 

  6.  *

    7. 

  7.  * ???????

    8. 

  8.  *

    9. 

  9.  * @package class

    10. 

  10.  * @author awfigq <awfigq@qq.com>

    11. 

  11.  */

    12. 

  12. class Chinese

    13. 

  13. {

    14. 

  14. 	var $table = '';

    15. 

  15. 	var $iconv_enabled = false;

    16. 

  16. 	var $convertbig5 = false;

    17. 

  17. 	var $unicode_table = array();

    18. 

  18. 	var $config  =  array (

    19. 

  19. 		'SourceLang'		=> '',

    20. 

  20. 		'TargetLang'		=> '',

    21. 

  21. 		'GBtoUnicode_table'	=> 'gb-unicode.table',

    22. 

  22. 		'BIG5toUnicode_table'	=> 'big5-unicode.table',

    23. 

  23. 		'GBtoBIG5_table'   	=> 'gb-big5.table',

    24. 

  24. 	);

    25. 

  25. 	

    26. 

  26. 	/**  

    27. 

  27. 	 * ?????

    28. 

  28. 	 * @param string $SourceLang ???

    29. 

  29. 	 * @param string $TargetLang ????

    30. 

  30. 	 * @param bool   $ForceTable ??????????? ????

    31. 

  31. 	 * @return void

    32. 

  32. 	 */

    33. 

  33. 	function Chinese($SourceLang, $TargetLang, $ForceTable = FALSE) {

    34. 

  34. 		$this->config['SourceLang'] = $this->_lang($SourceLang);

    35. 

  35. 		$this->config['TargetLang'] = $this->_lang($TargetLang);

    36. 

  36. 

    37. 

  37. 		if(ICONV_ENABLE && $this->config['TargetLang'] != 'BIG5' && !$ForceTable) {

    38. 

  38. 			$this->iconv_enabled = true;

    39. 

  39. 		} else {

    40. 

  40. 			$this->iconv_enabled = false;

    41. 

  41. 			$this->OpenTable();

    42. 

  42. 		}

    43. 

  43. 	}

    44. 

  44. 	

    45. 

  45. 	/**  

    46. 

  46. 	 * ??????????

    47. 

  47. 	 * @param string $SourceText ???????

    48. 

  48. 	 * @return string

    49. 

  49. 	 */

    50. 

  50. 	function convert($SourceText) {

    51. 

  51. 		if($this->config['SourceLang'] == $this->config['TargetLang']) {

    52. 

  52. 			return $SourceText;

    53. 

  53. 		} elseif($this->iconv_enabled) {

    54. 

  54. 			if($this->config['TargetLang'] <> 'UNICODE') {

    55. 

  55. 				return iconv($this->config['SourceLang'], $this->config['TargetLang'], $SourceText);

    56. 

  56. 			} else {

    57. 

  57. 				$return = '';

    58. 

  58. 				while($SourceText != '') {

    59. 

  59. 					if(ord(substr($SourceText, 0, 1)) > 127) {

    60. 

  60. 						$return .= "&#x".dechex($this->Utf8_Unicode(iconv($this->config['SourceLang'],"UTF-8", substr($SourceText, 0, 2)))).";";

    61. 

  61. 						$SourceText = substr($SourceText, 2, strlen($SourceText));

    62. 

  62. 					} else {

    63. 

  63. 						$return .= substr($SourceText, 0, 1);

    64. 

  64. 						$SourceText = substr($SourceText, 1, strlen($SourceText));

    65. 

  65. 					}

    66. 

  66. 				}

    67. 

  67. 				return $return;

    68. 

  68. 			}

    69. 

  69. 

    70. 

  70. 		} elseif($this->config['TargetLang'] == 'UNICODE') {

    71. 

  71. 			$utf = '';

    72. 

  72. 			while($SourceText != '') {

    73. 

  73. 				if(ord(substr($SourceText, 0, 1)) > 127) {

    74. 

  74. 					if($this->config['SourceLang'] == 'GBK') {

    75. 

  75. 						$utf .= '&#x'.$this->unicode_table[hexdec(bin2hex(substr($SourceText, 0, 2))) - 0x8080].';';

    76. 

  76. 					} elseif($this->config['SourceLang'] == 'BIG5') {

    77. 

  77. 						$utf .= '&#x'.$this->unicode_table[hexdec(bin2hex(substr($SourceText, 0, 2)))].';';

    78. 

  78. 					}

    79. 

  79. 					$SourceText = substr($SourceText, 2, strlen($SourceText));

    80. 

  80. 				} else {

    81. 

  81. 					$utf .= substr($SourceText, 0, 1);

    82. 

  82. 					$SourceText = substr($SourceText, 1, strlen($SourceText));

    83. 

  83. 				}

    84. 

  84. 			}

    85. 

  85. 			return $utf;

    86. 

  86. 		} else {

    87. 

  87. 			$ret = '';

    88. 

  88. 			if($this->config['SourceLang'] == 'UTF-8') {

    89. 

  89. 				$out = '';

    90. 

  90. 				$len = strlen($SourceText);

    91. 

  91. 				$i = 0;

    92. 

  92. 				while($i < $len) {

    93. 

  93. 					$c = ord(substr($SourceText, $i++, 1));

    94. 

  94. 					switch($c >> 4) {

    95. 

  95. 						case 0: case 1: case 2: case 3: case 4: case 5: case 6: case 7:

    96. 

  96. 							$out .= substr($SourceText, $i - 1, 1);

    97. 

  97. 							break;

    98. 

  98. 						case 12: case 13:

    99. 

  99. 							$char2 = ord(substr($SourceText, $i++, 1));

    100. 

  100. 							$char3 = $this->unicode_table[(($c & 0x1F) << 6) | ($char2 & 0x3F)];

    101. 

  101. 							if($this->config['TargetLang'] == 'GBK') {

    102. 

  102. 								$out .= $this->_hex2bin(dechex($char3 + 0x8080));

    103. 

  103. 							} elseif($this->config['TargetLang'] == 'BIG5') {

    104. 

  104. 								$out .= $this->_hex2bin($char3);

    105. 

  105. 							}

    106. 

  106. 							break;

    107. 

  107. 						case 14:

    108. 

  108. 							$char2 = ord(substr($SourceText, $i++, 1));

    109. 

  109. 							$char3 = ord(substr($SourceText, $i++, 1));

    110. 

  110. 							$char4 = $this->unicode_table[(($c & 0x0F) << 12) | (($char2 & 0x3F) << 6) | (($char3 & 0x3F) << 0)];

    111. 

  111. 							if($this->config['TargetLang'] == 'GBK') {

    112. 

  112. 								$out .= $this->_hex2bin(dechex($char4 + 0x8080));

    113. 

  113. 							} elseif($this->config['TargetLang'] == 'BIG5') {

    114. 

  114. 								$out .= $this->_hex2bin($char4);

    115. 

  115. 							}

    116. 

  116. 							break;

    117. 

  117. 					}

    118. 

  118. 				}

    119. 

  119. 				return !$this->convertbig5 ? $out : $this->GB2312toBIG5($out);

    120. 

  120. 			} else {

    121. 

  121. 				while($SourceText != '') {

    122. 

  122. 					if(ord(substr($SourceText, 0, 1)) > 127) {

    123. 

  123. 						if($this->config['SourceLang'] == 'BIG5') {

    124. 

  124. 							$utf8 = $this->CHSUtoUTF8(hexdec($this->unicode_table[hexdec(bin2hex(substr($SourceText, 0, 2)))]));

    125. 

  125. 						} elseif($this->config['SourceLang'] == 'GBK') {

    126. 

  126. 							$utf8=$this->CHSUtoUTF8(hexdec($this->unicode_table[hexdec(bin2hex(substr($SourceText, 0, 2))) - 0x8080]));

    127. 

  127. 						}

    128. 

  128. 						for($i = 0; $i < strlen($utf8); $i += 3) {

    129. 

  129. 							$ret .= chr(substr($utf8, $i, 3));

    130. 

  130. 						}

    131. 

  131. 						$SourceText = substr($SourceText, 2, strlen($SourceText));

    132. 

  132. 					} else {

    133. 

  133. 						$ret .= substr($SourceText, 0, 1);

    134. 

  134. 						$SourceText = substr($SourceText, 1, strlen($SourceText));

    135. 

  135. 					}

    136. 

  136. 				}

    137. 

  137. 				$SourceText = '';

    138. 

  138. 				return $ret;

    139. 

  139. 			}

    140. 

  140. 		}

    141. 

  141. 	}

    142. 

  142. 

    143. 

  143. 	function _lang($LangCode) {

    144. 

  144. 		$LangCode = strtoupper($LangCode);

    145. 

  145. 

    146. 

  146. 		if(substr($LangCode, 0, 2) == 'GB') {

    147. 

  147. 			return 'GBK';

    148. 

  148. 		} elseif(substr($LangCode, 0, 3) == 'BIG') {

    149. 

  149. 			return 'BIG5';

    150. 

  150. 		} elseif(substr($LangCode, 0, 3) == 'UTF') {

    151. 

  151. 			return 'UTF-8';

    152. 

  152. 		} elseif(substr($LangCode, 0, 3) == 'UNI') {

    153. 

  153. 			return 'UNICODE';

    154. 

  154. 		}

    155. 

  155. 	}

    156. 

  156. 

    157. 

  157. 	function _hex2bin($hexdata) {

    158. 

  158. 		for($i=0; $i < strlen($hexdata); $i += 2) {

    159. 

  159. 			$bindata .= chr(hexdec(substr($hexdata, $i, 2)));

    160. 

  160. 		}

    161. 

  161. 		return $bindata;

    162. 

  162. 	}

    163. 

  163. 

    164. 

  164. 	function OpenTable() {

    165. 

  165. 		$this->unicode_table = array();

    166. 

  166. 		if(!$this->iconv_enabled && $this->config['TargetLang'] == 'BIG5') {

    167. 

  167. 			$this->config['TargetLang'] = 'GBK';

    168. 

  168. 			$this->convertbig5 = TRUE;

    169. 

  169. 		}

    170. 

  170. 		if($this->config['SourceLang'] == 'GBK' || $this->config['TargetLang'] == 'GBK') {

    171. 

  171. 			$this->table = CODETABLE_DIR.$this->config['GBtoUnicode_table'];

    172. 

  172. 		} elseif($this->config['SourceLang'] == 'BIG5' || $this->config['TargetLang'] == 'BIG5') {

    173. 

  173. 			$this->table = CODETABLE_DIR.$this->config['BIG5toUnicode_table'];

    174. 

  174. 		}

    175. 

  175. 		$fp = fopen($this->table, 'rb');

    176. 

  176. 		$tabletmp = fread($fp, filesize($this->table));

    177. 

  177. 		for($i = 0; $i < strlen($tabletmp); $i += 4) {

    178. 

  178. 			$tmp = unpack('nkey/nvalue', substr($tabletmp, $i, 4));

    179. 

  179. 			if($this->config['TargetLang'] == 'UTF-8') {

    180. 

  180. 				$this->unicode_table[$tmp['key']] = '0x'.dechex($tmp['value']);

    181. 

  181. 			} elseif($this->config['SourceLang'] == 'UTF-8') {

    182. 

  182. 				$this->unicode_table[$tmp['value']] = '0x'.dechex($tmp['key']);

    183. 

  183. 			} elseif($this->config['TargetLang'] == 'UNICODE') {

    184. 

  184. 				$this->unicode_table[$tmp['key']] = dechex($tmp['value']);

    185. 

  185. 			}

    186. 

  186. 		}

    187. 

  187. 	}

    188. 

  188. 

    189. 

  189. 	function CHSUtoUTF8($c) {

    190. 

  190. 		$str = '';

    191. 

  191. 		if($c < 0x80) {

    192. 

  192. 			$str .= $c;

    193. 

  193. 		} elseif($c < 0x800) {

    194. 

  194. 			$str .= (0xC0 | $c >> 6);

    195. 

  195. 			$str .= (0x80 | $c & 0x3F);

    196. 

  196. 		} elseif($c < 0x10000) {

    197. 

  197. 			$str .= (0xE0 | $c >> 12);

    198. 

  198. 			$str .= (0x80 | $c >> 6 & 0x3F);

    199. 

  199. 			$str .=( 0x80 | $c & 0x3F);

    200. 

  200. 		} elseif($c < 0x200000) {

    201. 

  201. 			$str .= (0xF0 | $c >> 18);

    202. 

  202. 			$str .= (0x80 | $c >> 12 & 0x3F);

    203. 

  203. 			$str .= (0x80 | $c >> 6 & 0x3F);

    204. 

  204. 			$str .= (0x80 | $c & 0x3F);

    205. 

  205. 		}

    206. 

  206. 		return $str;

    207. 

  207. 	}

    208. 

  208. 

    209. 

  209. 	function GB2312toBIG5($c) {

    210. 

  210. 		$f = fopen(CODETABLE_DIR.$this->config['GBtoBIG5_table'], 'r');

    211. 

  211. 		$max=strlen($c)-1;

    212. 

  212. 		for($i = 0;$i < $max;$i++){

    213. 

  213. 			$h=ord($c[$i]);

    214. 

  214. 			if($h>=160) {

    215. 

  215. 				$l=ord($c[$i+1]);

    216. 

  216. 				if($h==161 && $l==64){

    217. 

  217. 					$gb="  ";

    218. 

  218. 				} else{

    219. 

  219. 					fseek($f,($h-160)*510+($l-1)*2);

    220. 

  220. 					$gb=fread($f,2);

    221. 

  221. 				}

    222. 

  222. 				$c[$i]=$gb[0];

    223. 

  223. 				$c[$i+1]=$gb[1];

    224. 

  224. 				$i++;

    225. 

  225. 			}

    226. 

  226. 		}

    227. 

  227. 		$result = $c;

    228. 

  228. 		return $result;

    229. 

  229. 	}

    230. 

  230. 

    231. 

  231. 	function Utf8_Unicode($char) {

    232. 

  232. 		switch(strlen($char)) {

    233. 

  233. 			case 1:

    234. 

  234. 				return ord($char);

    235. 

  235. 			case 2:

    236. 

  236. 				$n = (ord($char[0]) & 0x3f) << 6;

    237. 

  237. 				$n += ord($char[1]) & 0x3f;

    238. 

  238. 				return $n;

    239. 

  239. 			case 3:

    240. 

  240. 				$n = (ord($char[0]) & 0x1f) << 12;

    241. 

  241. 				$n += (ord($char[1]) & 0x3f) << 6;

    242. 

  242. 				$n += ord($char[2]) & 0x3f;

    243. 

  243. 				return $n;

    244. 

  244. 			case 4:

    245. 

  245. 				$n = (ord($char[0]) & 0x0f) << 18;

    246. 

  246. 				$n += (ord($char[1]) & 0x3f) << 12;

    247. 

  247. 				$n += (ord($char[2]) & 0x3f) << 6;

    248. 

  248. 				$n += ord($char[3]) & 0x3f;

    249. 

  249. 				return $n;

    250. 

  250. 		}

    251. 

  251. 	}

    252. 

  252. 

    253. 

  253. }

    254. 

  254. 

    255. 

  255. ?>
    256.