use builtin emscripten UTF8 functions

Author: emnh

web/vim_lib.js

@@ -72,38 +72,6 @@ var LibraryVIM = {
 
     dropbox: null,
 
-    // from https://github.com/s-macke/jor1k/blob/master/js/lib/utf8.js
-    // LICENSE: https://github.com/s-macke/jor1k/blob/master/LICENSE.md
-    UnicodeToUTF8Stream: function(key) {
-      key = key|0;
-      if (key < 0x80) {
-          return [key];
-      }
-      if (key <= 0x7FF) {
-          return [
-              (key >> 6) | 0xC0,
-              (key & 0x3F) | 0x80
-              ];
-      }
-      if (key <= 0xFFFF) {
-          return [
-              (key >> 12) | 0xE0,
-              ((key >> 6) & 0x3F) | 0x80,
-              (key & 0x3F) | 0x80
-              ];
-      }
-      if (key <= 0x10FFFF) {
-          return [
-              (key >> 18) | 0xF0,
-              ((key >> 12) & 0x3F) | 0x80,
-              ((key >> 6) & 0x3F) | 0x80,
-              (key & 0x3F) | 0x80
-              ];
-      }
-      //message.Debug("Error in utf-8 encoding: Invalid key");
-      return [];
-    },
-
     // functions that are not exposed to C
     handle_key: function(charCode, keyCode, e) {//VIMJS_FOLD_START
       // macros defined in keymap.h
@@ -124,12 +92,14 @@ var LibraryVIM = {
       }
 
       if(!handled) {
-        var chars = vimjs.UnicodeToUTF8Stream(charCode || keyCode);
-        if (chars.length == 1) {
+        var MAX_UTF8_BYTES = 6;
+        var chars = new Uint8Array(MAX_UTF8_BYTES + 1); // null-terminated
+        var charLen = stringToUTF8Array(String.fromCharCode(charCode), chars, 0, MAX_UTF8_BYTES);
+        if (charLen == 1) {
           vimjs.gui_web_handle_key(chars[0], modifiers, 0, 0);
         } else {
           // no modifers for UTF-8, should be handled in chars already
-          for (var i = 0; i < chars.length; i++) {
+          for (var i = 0; i < charLen; i++) {
             vimjs.gui_web_handle_key(chars[i], 0, 0, 0);
           }
         }
@@ -820,81 +790,13 @@ var LibraryVIM = {
 
   vimjs_draw_string__deps: ['vimjs_clear_block'],
   vimjs_draw_string: function(row, col, s_ptr, len, flags) {
-    // from https://github.com/s-macke/jor1k/blob/master/js/lib/utf8.js
-    // LICENSE: https://github.com/s-macke/jor1k/blob/master/LICENSE.md
-    var UTF8StreamToUnicode = function() {
-      this.stream = new Uint8Array(5);
-      this.ofs = 0;
-
-      this.Put = function(key) {
-          this.stream[this.ofs] = key;
-          this.ofs++;
-          switch(this.ofs) {
-              case 1:
-                  if (this.stream[0] < 0x80) {
-                      this.ofs = 0;
-                      return this.stream[0];
-                  }
-                  break;
-
-              case 2:
-                  if ((this.stream[0]&0xE0) == 0xC0)
-                  if ((this.stream[1]&0xC0) == 0x80) {
-                      this.ofs = 0;
-                      return ((this.stream[0]&0x1F)<<6) |
-                          ((this.stream[1]&0x3F)<<0);
-                  }
-                  break;
-
-              case 3:
-                  if ((this.stream[0]&0xF0) == 0xE0)
-                  if ((this.stream[1]&0xC0) == 0x80)
-                  if ((this.stream[2]&0xC0) == 0x80) {
-                      this.ofs = 0;
-                      return ((this.stream[0]&0xF ) << 12) |
-                          ((this.stream[1]&0x3F) << 6)  |
-                          ((this.stream[2]&0x3F) << 0);
-                  }
-                  break;
-
-              case 4:
-                  if ((this.stream[0]&0xF8) == 0xF0)
-                  if ((this.stream[1]&0xC0) == 0x80)
-                  if ((this.stream[2]&0xC0) == 0x80)
-                  if ((this.stream[3]&0xC0) == 0x80) {
-                      this.ofs = 0;
-                      return ((this.stream[0]&0x7 ) << 18) |
-                          ((this.stream[1]&0x3F) << 12) |
-                          ((this.stream[2]&0x3F) << 6)  |
-                          ((this.stream[3]&0x3F) << 0);
-                  }
-                  this.ofs = 0;
-                  return -1; //obviously illegal character, so reset
-                  break;
-
-              default:
-                  this.ofs = 0;
-                  return -1;
-                  break;
-          }
-          return -1;
-      }
-    };
-
-    // Pointer_stringify includes UTF-8 decoding,
-    // but is buggy somehow, so we decode here instead.
-    // TODO: Make a test case for Pointer_stringify and report bug to Emscripten.
-    // var s = Pointer_stringify(s_ptr, len);
-    var utf8converter = new UTF8StreamToUnicode();
-    var news = "";
+    var byteArray = [];
     for (var i = 0; i < len; i++) {
       c = getValue(s_ptr + i, 'i8', true);
-      c = utf8converter.Put(c);
-      if (c != -1) {
-        news += String.fromCharCode(c);
-      }
+      byteArray.push(c);
     }
-    var s = news;
+    byteArray.push(0);
+    var s = UTF8ArrayToString(byteArray, 0);
     len = s.length;
 
     // TODO: use macros for flag constants