feat:CJK 修改为中文

ricoa · Apr 24, 2017 · bbe2748 · bbe2748
1 parent c0930db
commit bbe2748
Show file tree

Hide file tree

Showing 4 changed files with 34 additions and 43 deletions.
diff --git a/src/Correctors/CharacterCorrector.php b/src/Correctors/CharacterCorrector.php
@@ -26,7 +26,7 @@ public function handle($text)
  $text= str_replace(getDBC(),getSBC(),$text);
 
  //中文后使用全角中文标点
- $text = preg_replace_callback("/([".getCJK().'])([!?\.,\(\):;\'\"])/iu', function($m){
+ $text = preg_replace_callback("/([".getCN().'])([!?\.,\(\):;\'\"])/iu', function($m){
  $replace=[
  '!'=>'！',
  '?'=>'？',

diff --git a/src/Correctors/SpaceCorrector.php b/src/Correctors/SpaceCorrector.php
@@ -4,7 +4,7 @@
 
 /**
  *
- * 在 CJK 字符与英文字符间添加空格
+ * 在中文字符与英文字符间添加空格
  *
  *update base on https://github.com/Rakume/pangu.php/blob/fc7d1c54ada1c85bb0e2725d41ce41b449eb3737/pangu.php
  *
@@ -28,15 +28,15 @@ class SpaceCorrector extends Corrector{
  */
  protected function __construct()
  {
- $cjk = getCJK();
+ $cn = getCN();
  $this->patterns = array(
- 'cjk_quote' => array(
- '([' . $cjk . '])(["\'])',
+ 'cn_quote' => array(
+ '([' . $cn . '])(["\'])',
  '$1 $2'
  ),
 
- 'quote_cjk' => array(
- '(["\'])([' . $cjk . '])',
+ 'quote_cn' => array(
+ '(["\'])([' . $cn . '])',
  '$1 $2'
  ),
 
@@ -45,40 +45,40 @@ protected function __construct()
  '$1$3$5'
  ),
 
- 'cjk_hash' => array(
- '([' . $cjk . '])(#(\S+))',
+ 'cn_hash' => array(
+ '([' . $cn . '])(#(\S+))',
  '$1 $2'
  ),
 
- 'hash_cjk' => array(
- '((\S+)#)([' . $cjk . '])',
+ 'hash_cn' => array(
+ '((\S+)#)([' . $cn . '])',
  '$1 $3'
  ),
 
- 'cjk_operator_ans' => array(
- '([' . $cjk . '])([A-Za-z0-9])([\+\-\*\/=&\\|<>])',
+ 'cn_operator_ans' => array(
+ '([' . $cn . '])([A-Za-z0-9])([\+\-\*\/=&\\|<>])',
  '$1 $2 $3'
  ),
 
- 'ans_operator_cjk' => array(
- '([\+\-\*\/=&\\|<>])([A-Za-z0-9])([' . $cjk . '])',
+ 'ans_operator_cn' => array(
+ '([\+\-\*\/=&\\|<>])([A-Za-z0-9])([' . $cn . '])',
  '$1 $2 $3'
  ),
 
  'bracket' => array(
  array(
- '([' . $cjk . '])([<\[\{\(]+(.*?)[>\]\}\)]+)([' . $cjk . '])',
+ '([' . $cn . '])([<\[\{\(]+(.*?)[>\]\}\)]+)([' . $cn . '])',
  '$1 $2 $4'
  ),
 
  array(
- 'cjk_bracket' => array(
- '([' . $cjk . '])([<>\[\]\{\}\(\)])',
+ 'cn_bracket' => array(
+ '([' . $cn . '])([<>\[\]\{\}\(\)])',
  '$1 $2'
  ),
 
- 'bracket_cjk' => array(
- '([<>\[\]\{\}\(\)])([' . $cjk . '])',
+ 'bracket_cn' => array(
+ '([<>\[\]\{\}\(\)])([' . $cn . '])',
  '$1 $2'
  )
  )
@@ -89,13 +89,13 @@ protected function __construct()
  '$1$3$5'
  ),
 
- 'cjk_ans' => array(
- '([' . $cjk . '])([A-Za-z0-9`@&%\=\$\^\*\-\+\\/|\\\])',
+ 'cn_ans' => array(
+ '([' . $cn . '])([A-Za-z0-9`@&%\=\$\^\*\-\+\\/|\\\])',
  '$1 $2'
  ),
 
- 'ans_cjk' => array(
- '([A-Za-z0-9`~!%&=;\|\,\.\:\?\$\^\*\-\+\/\\\])([' . $cjk . '])',
+ 'ans_cn' => array(
+ '([A-Za-z0-9`~!%&=;\|\,\.\:\?\$\^\*\-\+\/\\\])([' . $cn . '])',
  '$1 $2'
  ),
  'number_letters'=>array(
@@ -106,12 +106,12 @@ protected function __construct()
  "([ ]*)([\u{FF00}-\u{FFFF}。])([ ]*)",
  '$2'
  ),
- 'cjk_greek'=>array(
- '([' . $cjk . '0-9A-Za-z])([\p{Greek}])',
+ 'cn_greek'=>array(
+ '([' . $cn . '0-9A-Za-z])([\p{Greek}])',
  '$1 $2'
  ),
- 'greek_cjk'=>array(
- '([\p{Greek}])([' . $cjk . '0-9A-Za-z])',
+ 'greek_cn'=>array(
+ '([\p{Greek}])([' . $cn . '0-9A-Za-z])',
  '$1 $2'
  ),
  );

diff --git a/src/helpers.php b/src/helpers.php
@@ -1,16 +1,7 @@
 <?php
 
-function getCJK(){
- return '' .
- '\x{2e80}-\x{2eff}' .
- '\x{2f00}-\x{2fdf}' .
- '\x{3040}-\x{309f}' .
- '\x{30a0}-\x{30ff}' .
- '\x{3100}-\x{312f}' .
- '\x{3200}-\x{32ff}' .
- '\x{3400}-\x{4dbf}' .
- '\x{4e00}-\x{9fff}' .
- '\x{f900}-\x{faff}';
+function getCN(){
+ return '\p{Han}';
 }
 
 //获得全角字符

diff --git a/tests/unit/SpaceTest.php b/tests/unit/SpaceTest.php
@@ -3,7 +3,7 @@
 use Ricoa\CopyWritingCorrect\CopyWritingCorrectService;
 
 /**
- * 检测 CJK和半角字符之间增加空格
+ * 检测 中文和半角字符之间增加空格
  *
  * Class SpaceBetweenCjkAndEngTest
  */
@@ -22,13 +22,13 @@ public function testMe()
  {
  $service=new CopyWritingCorrectService();
 
- //CJK 与英文
+ //中文与英文
  $this->assertEquals(
  '在 LeanCloud 上，数据存储是围绕 AVObject 进行的。每个 AVObject 都包含了与 JSON 兼容的 key-value 对应的数据。数据是 schema-free 的，你不需要在每个 AVObject 上提前指定存在哪些键，只要直接设定对应的 key-value 即可。',
  $service->correct('在LeanCloud上，数据存储是围绕AVObject进行的。每个AVObject都包含了与JSON兼容的key-value对应的数据。数据是schema-free的，你不需要在每个AVObject上提前指定存在哪些键，只要直接设定对应的key-value即可。')
  );
 
- //CJK 与数字
+ //中文与数字
  $this->assertEquals(
  '今天出去买菜花了 5000 元。',
  $service->correct('今天出去买菜花了5000元。')
@@ -46,7 +46,7 @@ public function testMe()
  $service->correct('我家的光纤入户宽带有10Gbps ，SSD一共有20TB。 ')
  );
 
- //希腊字母与 CJK 字符以及数字英文字符之间添加空格
+ //希腊字母与中文字符以及数字英文字符之间添加空格
  $this->assertEquals(
  '电阻为 1 Ω',
  $service->correct('电阻为1Ω')