Use higher-level string helpers where possible. NFC (#24486)

I guess one functional change is in implementation of
`emscripten_fetch_get_response_headers_length`.

The docs say that it just returns the length, and you must use `length +
1` for buffer size passed to `emscripten_fetch_get_response_headers`,
and that's what tests do, but we were actually returning `length + 1`
already. This seems like a bugfix, but happy to revert if deemed too
risky.

Author: RReverser

ChangeLog.md

@@ -46,6 +46,9 @@ See docs/process.md for more on how version tagging works.
 - The field `responseUrl` is added to `emscripten_fetch_t`. This is notably
   usable for obtaining resolved URL, in line with JS `XMLHttpRequest.responseURL`
   field. (#24414)
+- `emscripten_fetch_get_response_headers_length` now excludes the trailing
+  null character from the length calculation to match the documented behaviour.
+  (#24486)
 
 4.0.9 - 05/19/25
 ----------------

src/Fetch.js

@@ -584,14 +584,12 @@ function startFetch(fetch, successcb, errorcb, progresscb, readystatechangecb) {
 }
 
 function fetchGetResponseHeadersLength(id) {
-  return lengthBytesUTF8(Fetch.xhrs.get(id).getAllResponseHeaders()) + 1;
+  return lengthBytesUTF8(Fetch.xhrs.get(id).getAllResponseHeaders());
 }
 
 function fetchGetResponseHeaders(id, dst, dstSizeBytes) {
   var responseHeaders = Fetch.xhrs.get(id).getAllResponseHeaders();
-  var lengthBytes = lengthBytesUTF8(responseHeaders) + 1;
-  stringToUTF8(responseHeaders, dst, dstSizeBytes);
-  return Math.min(lengthBytes, dstSizeBytes);
+  return stringToUTF8(responseHeaders, dst, dstSizeBytes) + 1;
 }
 
 //Delete the xhr JS object, allowing it to be garbage collected.

src/lib/libfs.js

@@ -10,8 +10,6 @@ var LibraryFS = {
     '$FS_modeStringToFlags',
     '$FS_getMode',
     '$intArrayFromString',
-    '$stringToUTF8Array',
-    '$lengthBytesUTF8',
 #if LibraryManager.has('libidbfs.js')
     '$IDBFS',
 #endif
@@ -1347,28 +1345,24 @@ FS.staticInit();`;
       if (opts.encoding !== 'utf8' && opts.encoding !== 'binary') {
         throw new Error(`Invalid encoding type "${opts.encoding}"`);
       }
-      var ret;
       var stream = FS.open(path, opts.flags);
       var stat = FS.stat(path);
       var length = stat.size;
       var buf = new Uint8Array(length);
       FS.read(stream, buf, 0, length, 0);
       if (opts.encoding === 'utf8') {
-        ret = UTF8ArrayToString(buf);
-      } else if (opts.encoding === 'binary') {
-        ret = buf;
+        buf = UTF8ArrayToString(buf);
       }
       FS.close(stream);
-      return ret;
+      return buf;
     },
     writeFile(path, data, opts = {}) {
       opts.flags = opts.flags || {{{ cDefs.O_TRUNC | cDefs.O_CREAT | cDefs.O_WRONLY }}};
       var stream = FS.open(path, opts.flags, opts.mode);
       if (typeof data == 'string') {
-        var buf = new Uint8Array(lengthBytesUTF8(data)+1);
-        var actualNumBytes = stringToUTF8Array(data, buf, 0, buf.length);
-        FS.write(stream, buf, 0, actualNumBytes, undefined, opts.canOwn);
-      } else if (ArrayBuffer.isView(data)) {
+        data = new Uint8Array(intArrayFromString(data, true));
+      }
+      if (ArrayBuffer.isView(data)) {
         FS.write(stream, data, 0, data.byteLength, undefined, opts.canOwn);
       } else {
         throw new Error('Unsupported data type');

src/lib/libsdl.js

@@ -20,7 +20,7 @@ var LibrarySDL = {
     '$PATH', '$Browser', 'SDL_GetTicks', 'SDL_LockSurface',
     '$MainLoop',
     // For makeCEvent().
-    '$intArrayFromString',
+    '$stringToUTF8',
     // Many SDL functions depend on malloc/free
     'malloc', 'free',
     'memcpy',
@@ -964,10 +964,7 @@ var LibrarySDL = {
         case 'keypress': {
           {{{ makeSetValue('ptr', C_STRUCTS.SDL_TextInputEvent.type, 'SDL.DOMEventToSDLEvent[event.type]', 'i32') }}};
           // Not filling in windowID for now
-          var cStr = intArrayFromString(String.fromCharCode(event.charCode));
-          for (var i = 0; i < cStr.length; ++i) {
-            {{{ makeSetValue('ptr', C_STRUCTS.SDL_TextInputEvent.text + ' + i', 'cStr[i]', 'i8') }}};
-          }
+          stringToUTF8(String.fromCharCode(event.charCode), ptr + {{{ C_STRUCTS.SDL_TextInputEvent.text }}}, 4);
           break;
         }
         case 'mousedown': case 'mouseup': case 'mousemove': {
@@ -1782,7 +1779,7 @@ var LibrarySDL = {
   SDL_GetKeyState: () => _SDL_GetKeyboardState(0),
 
   SDL_GetKeyName__proxy: 'sync',
-  SDL_GetKeyName__deps: ['$stringToUTF8', 'realloc'],
+  SDL_GetKeyName__deps: ['$lengthBytesUTF8', '$stringToUTF8', 'realloc'],
   SDL_GetKeyName: (key) => {
     var name = '';
     /* ASCII A-Z or 0-9 */
@@ -2904,8 +2901,8 @@ var LibrarySDL = {
       audio.frequency = info.audio.frequency;
     }
     audio['onended'] = function() { // TODO: cache these
-      if (channelInfo.audio === this || channelInfo.audio.webAudioNode === this) { 
-        channelInfo.audio.paused = true; channelInfo.audio = null; 
+      if (channelInfo.audio === this || channelInfo.audio.webAudioNode === this) {
+        channelInfo.audio.paused = true; channelInfo.audio = null;
       }
       if (SDL.channelFinished) {{{ makeDynCall('vi', 'SDL.channelFinished') }}}(channel);
     }

src/lib/libtime.js

@@ -257,7 +257,7 @@ addToLibrary({
   },
 
   strptime__deps: ['$isLeapYear', '$arraySum', '$addDays', '$MONTH_DAYS_REGULAR', '$MONTH_DAYS_LEAP',
-                   '$intArrayFromString' ],
+                   '$lengthBytesUTF8'],
   strptime: (buf, format, tm) => {
     // char *strptime(const char *restrict buf, const char *restrict format, struct tm *restrict tm);
     // http://pubs.opengroup.org/onlinepubs/009695399/functions/strptime.html
@@ -474,7 +474,7 @@ addToLibrary({
         // GMT offset as either 'Z' or +-HH:MM or +-HH or +-HHMM
         if (value.toLowerCase() === 'z'){
           date.gmtoff = 0;
-        } else {          
+        } else {
           var match = value.match(/^((?:\-|\+)\d\d):?(\d\d)?/);
           date.gmtoff = match[1] * 3600;
           if (match[2]) {
@@ -507,10 +507,10 @@ addToLibrary({
       {{{ makeSetValue('tm', C_STRUCTS.tm.tm_yday, 'arraySum(isLeapYear(fullDate.getFullYear()) ? MONTH_DAYS_LEAP : MONTH_DAYS_REGULAR, fullDate.getMonth()-1)+fullDate.getDate()-1', 'i32') }}};
       {{{ makeSetValue('tm', C_STRUCTS.tm.tm_isdst, '0', 'i32') }}};
       {{{ makeSetValue('tm', C_STRUCTS.tm.tm_gmtoff, 'date.gmtoff', LONG_TYPE) }}};
- 
+
       // we need to convert the matched sequence into an integer array to take care of UTF-8 characters > 0x7F
       // TODO: not sure that intArrayFromString handles all unicode characters correctly
-      return buf+intArrayFromString(matches[0]).length-1;
+      return buf+lengthBytesUTF8(matches[0]);
     }
 
     return 0;

src/lib/libwasi.js

@@ -442,7 +442,7 @@ var WasiLibrary = {
       return {{{ cDefs.EBADF }}};
     }
     var preopen_path = preopens[fd];
-    stringToUTF8Array(preopen_path, HEAP8, path, path_len)
+    stringToUTF8(preopen_path, path, path_len)
 #if SYSCALL_DEBUG
     dbg(`fd_prestat_dir_name -> "${preopen_path}"`);
 #endif

src/lib/libwasmfs.js

@@ -141,8 +141,7 @@ addToLibrary({
 
       // Default return type is binary.
       // The buffer contents exist 8 bytes after the returned pointer.
-      var ret = new Uint8Array(HEAPU8.subarray(buf, buf + length));
-      return opts.encoding === 'utf8' ? UTF8ArrayToString(ret) : ret;
+      return opts.encoding === 'utf8' ? UTF8ToString(buf, length) : HEAPU8.slice(buf, buf + length);
     },
 #endif
 
@@ -260,7 +259,7 @@ addToLibrary({
           ino: {{{ makeGetValue('statBuf', C_STRUCTS.stat.st_ino, "u53") }}}
       }
     },
-    stat(path) { 
+    stat(path) {
       return withStackSave(() => {
         var statBuf = stackAlloc({{{ C_STRUCTS.stat.__size__ }}});
         FS.handleError(__wasmfs_stat(stringToUTF8OnStack(path), statBuf));
@@ -507,19 +506,17 @@ addToLibrary({
   $FS_writeFile: (path, data) => {
     var sp = stackSave();
     var pathBuffer = stringToUTF8OnStack(path);
-    if (typeof data == 'string') {
-      var buf = new Uint8Array(lengthBytesUTF8(data) + 1);
-      var actualNumBytes = stringToUTF8Array(data, buf, 0, buf.length);
-      data = buf.slice(0, actualNumBytes);
-    }
-    var dataBuffer = _malloc(data.length);
+    var len = typeof data == 'string' ? lengthBytesUTF8(data) + 1 : data.length;
+    var dataBuffer = _malloc(len);
 #if ASSERTIONS
     assert(dataBuffer);
 #endif
-    for (var i = 0; i < data.length; i++) {
-      {{{ makeSetValue('dataBuffer', 'i', 'data[i]', 'i8') }}};
+    if (typeof data == 'string') {
+      len = stringToUTF8(data, dataBuffer, len);
+    } else {
+      HEAPU8.set(data, dataBuffer);
     }
-    var ret = __wasmfs_write_file(pathBuffer, dataBuffer, data.length);
+    var ret = __wasmfs_write_file(pathBuffer, dataBuffer, len);
     _free(dataBuffer);
     stackRestore(sp);
     return ret;

test/other/codesize/test_codesize_hello_dylink.gzsize

@@ -1 +1 @@
-11676
+11657

test/other/codesize/test_codesize_hello_dylink.jssize

@@ -1 +1 @@
-27564
+27541

test/wasmfs/wasmfs_fetch.js

@@ -2,10 +2,7 @@ addToLibrary({
   getUrlOrigin__sig: 'vpp',
   getUrlOrigin: function (ptr, len) {
     try {
-      var orig = self.location.origin;
-      var nb = lengthBytesUTF8(orig) + 1;
-      if (nb <= len)
-        stringToUTF8(orig, ptr, len);
+      stringToUTF8(location.origin, ptr, len);
     } catch (e) {
       console.warn(e);
     }