js解析支持unicode fix #44

army8735 · Feb 4, 2015 · 5ee86cb · 5ee86cb
1 parent 334da49
commit 5ee86cb
Show file tree

Hide file tree

Showing 7 changed files with 50,718 additions and 14 deletions.
diff --git a/lib/unicode.js b/lib/unicode.js
@@ -6,21 +6,29 @@ var Lt = require('unicode-7.0.0/categories/Lt/code-points');
 var Lm = require('unicode-7.0.0/categories/Lm/code-points');
 var Lo = require('unicode-7.0.0/categories/Lo/code-points');
 var Nl = require('unicode-7.0.0/categories/Nl/code-points');
+var Mn = require('unicode-7.0.0/categories/Mn/code-points');
+var Mc = require('unicode-7.0.0/categories/Mc/code-points');
+var Nd = require('unicode-7.0.0/categories/Nd/code-points');
+var Pc = require('unicode-7.0.0/categories/Pc/code-points');
 
 var sort = require('../src/util/sort');
 
-var 编码集 = Lu.concat(Ll).concat(Lt).concat(Lm).concat(Lo).concat(Nl);
+var 编码集 = Lu.concat(Ll, Lt, Lm, Lo, Nl);
 sort(编码集);
+编码集 = 过滤(编码集);
 
-编码集 = 编码集.filter(function(编码) {
-  if(编码 >= 0x010000 && 编码 <= 0x10FFFF) {
-    return false;
-  }
-  if(编码 >= 0x0 && 编码 <= 0x7F) {
-    return false;
-  }
-  return true;
-});
+function 过滤(编码集) {
+  编码集 = 编码集.filter(function(编码) {
+    if(编码 >= 0x010000 && 编码 <= 0x10FFFF) {
+      return false;
+    }
+    if(编码 >= 0x0 && 编码 <= 0x7F) {
+      return false;
+    }
+    return true;
+  });
+  return 编码集;
+}
 
 fs.writeFileSync('./编码集.txt', 编码集.join('\n'), { encoding: 'utf-8' });
 
@@ -67,4 +75,34 @@ function 补前缀(编码) {
   }
 }
 
-fs.writeFileSync('./结果集.txt', 结果集.join('\n'), { encoding: 'utf-8' });
+fs.writeFileSync('./结果集.txt', 结果集.join('\n'), { encoding: 'utf-8' });
+
+
+
+
+var 后缀编码集 = [0x200C, 0x200D].concat(Lu, Ll, Lt, Lm, Lo, Nl, Mn, Mc, Nd, Pc);
+sort(后缀编码集);
+后缀编码集 = 过滤(后缀编码集);
+
+fs.writeFileSync('./后缀编码集.txt', 后缀编码集.join('\n'), { encoding: 'utf-8' });
+
+var 后缀结果集 = [];
+var 上个索引 = 0;
+for(var i = 1, len = 后缀编码集.length; i < len; i++) {
+  var 当前 = 后缀编码集[i];
+  var 前个 = 后缀编码集[i-1];
+  if(当前 == 前个 + 1) {
+    //
+  }
+  else {
+    if(上个索引 == i - 1) {
+      后缀结果集.push(转为Unicode(前个));
+    }
+    else {
+      后缀结果集.push(转为Unicode(后缀编码集[上个索引]) + '-' + 转为Unicode(前个));
+    }
+    上个索引 = i;
+  }
+}
+
+fs.writeFileSync('./后缀结果集.txt', 后缀结果集.join('\n'), { encoding: 'utf-8' });