utf parsing in writeStringToMemory as well

src/library.js

@@ -373,7 +373,7 @@ LibraryManager.library = {
           output.printer(output.buffer.join(''));
           output.buffer = [];
         } else {
-          output.buffer.push(utf8.feed(val));
+          output.buffer.push(utf8.processCChar(val));
         }
       }
       if (!output) {

src/preamble.js

@@ -521,7 +521,7 @@ function Pointer_stringify(ptr, /* optional */ length) {
   while (1) {
     t = {{{ makeGetValue('ptr', 'i', 'i8', 0, 1) }}};
     if (nullTerminated && t == 0) break;
-    ret += utf8.feed(t);
+    ret += utf8.processCChar(t);
     i += 1;
     if (!nullTerminated && i == length) break;
   }
@@ -734,22 +734,9 @@ Module['String_len'] = String_len;
 // This processes a JS string into a C-line array of numbers, 0-terminated.
 // For LLVM-originating strings, see parser.js:parseLLVMString function
 function intArrayFromString(stringy, dontAddNull, length /* optional */) {
-  var ret = [];
-  var t;
-  var i = 0;
-  if (length === undefined) {
-    length = stringy.length;
-  }
-  while (i < length) {
-    var chr = stringy.charCodeAt(i);
-    if (chr > 0xFF) {
-#if ASSERTIONS
-        assert(false, 'Character code ' + chr + ' (' + stringy[i] + ')  at offset ' + i + ' not in 0x00-0xFF.');
-#endif
-      chr &= 0xFF;
-    }
-    ret.push(chr);
-    i = i + 1;
+  var ret = (new Runtime.UTF8Processor()).processJSString(stringy);
+  if (length) {
+    ret.length = length;
   }
   if (!dontAddNull) {
     ret.push(0);
@@ -776,21 +763,13 @@ Module['intArrayToString'] = intArrayToString;
 
 // Write a Javascript array to somewhere in the heap
 function writeStringToMemory(string, buffer, dontAddNull) {
+  var array = intArrayFromString(string, dontAddNull);
   var i = 0;
-  while (i < string.length) {
-    var chr = string.charCodeAt(i);
-    if (chr > 0xFF) {
-#if ASSERTIONS
-        assert(false, 'Character code ' + chr + ' (' + string[i] + ')  at offset ' + i + ' not in 0x00-0xFF.');
-#endif
-      chr &= 0xFF;
-    }
+  while (i < array.length) {
+    var chr = array[i];
     {{{ makeSetValue('buffer', 'i', 'chr', 'i8') }}}
     i = i + 1;
   }
-  if (!dontAddNull) {
-    {{{ makeSetValue('buffer', 'i', '0', 'i8') }}}
-  }
 }
 Module['writeStringToMemory'] = writeStringToMemory;
 

src/runtime.js

@@ -334,10 +334,13 @@ var Runtime = {
     return Runtime.funcWrappers[func];
   },
 
+  // Returns a processor of UTF.
+  // processCChar() receives characters from a C-like UTF representation and returns JS string fragments.
+  // processJSString() receives a JS string and returns a C-like UTF representation in an array
   UTF8Processor: function() {
     var buffer = [];
     var needed = 0;
-    this.feed = function (code) {
+    this.processCChar = function (code) {
       code = code & 0xff;
       if (needed) {
         buffer.push(code);
@@ -366,6 +369,14 @@ var Runtime = {
       buffer.length = 0;
       return ret;
     }
+    this.processJSString = function(string) {
+      string = unescape(encodeURIComponent(string));
+      var ret = [];
+      for (var i = 0; i < string.length; i++) {
+        ret.push(string.charCodeAt(i));
+      }
+      return ret;
+    }
   },
 
 #if RUNTIME_DEBUG

tests/runner.py

@@ -4485,14 +4485,15 @@ def process(filename):
         #include <stdio.h>
         #include <emscripten.h>
 
-        char *c = "μ†ℱ ╋ℯ╳╋";
-
         int main() {
+          char *c = "μ†ℱ ╋ℯ╳╋";
           printf("%d %d %d %d %s\n", c[0]&0xff, c[1]&0xff, c[2]&0xff, c[3]&0xff, c);
-          //emscripten_run_script("Module.print(Pointer_stringify(Module.getValue(_c, '*')))");
+          emscripten_run_script("cheez = Module._malloc(100);"
+                                "Module.writeStringToMemory(\"μ†ℱ ╋ℯ╳╋\", cheez);"
+                                "Module.print([Pointer_stringify(cheez), Module.getValue(cheez, 'i8')&0xff, Module.getValue(cheez+1, 'i8')&0xff, Module.getValue(cheez+2, 'i8')&0xff, Module.getValue(cheez+3, 'i8')&0xff, ]);");
         }
       '''
-      self.do_run(src, '206 188 226 128 μ†ℱ ╋ℯ╳╋\n')#μ†ℱ ╋ℯ╳╋\n');
+      self.do_run(src, '206 188 226 128 μ†ℱ ╋ℯ╳╋\nμ†ℱ ╋ℯ╳╋,206,188,226,128\n');
 
     def test_direct_string_constant_usage(self):
       if self.emcc_args is None: return self.skip('requires libcxx')