Fix -sPROXY_TO_WORKER + -sSINGLE_FILE + node (#22610)

The proxyClient needs to jump through a few more hoops here to load the
embedded generated JS file which is encoded as a data URL.

This change also fixes the mime type used when we embed JS as a data
URL.

Author: sbc100

src/proxyClient.js

@@ -18,7 +18,23 @@
 #if ENVIRONMENT_MAY_BE_NODE
 var ENVIRONMENT_IS_NODE = typeof process == 'object' && typeof process.versions == 'object' && typeof process.versions.node == 'string';
 if (ENVIRONMENT_IS_NODE) {
-  global.Worker = require('worker_threads').Worker;
+  var NodeWorker = require('worker_threads').Worker;
+  global.Worker = function(url, options) {
+    // Special handling for `data:` URL argument, to match the behaviour
+    // of the Web API.
+    if (typeof url == 'string' && url.startsWith('data:')) {
+#if EXPORT_ES6
+      // worker_threads always assume data URLs are ES6 modules
+      url = new URL(url);
+#else
+      // For class modules we decode the data URL and use `eval: true`.
+      url = Buffer.from(url.split(",")[1], 'base64').toString();
+      options ||= {}
+      options.eval = true;
+#endif
+    }
+    return new NodeWorker(url, options);
+  }
   var Module = Module || {}
 } else
 #endif

src/shell.js

@@ -136,8 +136,12 @@ if (ENVIRONMENT_IS_NODE) {
   // TODO: Swap all `require()`'s with `import()`'s?
 #if EXPORT_ES6 && ENVIRONMENT_MAY_BE_WEB
   const { createRequire } = await import('module');
+  let dirname = import.meta.url;
+  if (dirname.startsWith("data:")) {
+    dirname = '/';
+  }
   /** @suppress{duplicate} */
-  var require = createRequire(import.meta.url);
+  var require = createRequire(dirname);
 #endif
 
 #if PTHREADS || WASM_WORKERS
@@ -235,7 +239,9 @@ if (ENVIRONMENT_IS_NODE) {
   // EXPORT_ES6 + ENVIRONMENT_IS_NODE always requires use of import.meta.url,
   // since there's no way getting the current absolute path of the module when
   // support for that is not available.
-  scriptDirectory = nodePath.dirname(require('url').fileURLToPath(import.meta.url)) + '/';
+  if (!import.meta.url.startsWith('data:')) {
+    scriptDirectory = nodePath.dirname(require('url').fileURLToPath(import.meta.url)) + '/';
+  }
 #else
   scriptDirectory = __dirname + '/';
 #endif

test/test_other.py

@@ -14722,8 +14722,13 @@ def test_standalone_whole_archive(self):
     self.emcc_args += ['-sSTANDALONE_WASM', '-pthread', '-Wl,--whole-archive', '-lbulkmemory', '-lstandalonewasm', '-Wl,--no-whole-archive']
     self.do_runf('hello_world.c')
 
-  def test_proxy_to_worker(self):
-    self.do_runf('hello_world.c', emcc_args=['--proxy-to-worker'])
+  @parameterized({
+    '':   ([],),
+    '_single_file': (['-sSINGLE_FILE'],),
+    '_single_file_es6': (['-sSINGLE_FILE', '-sEXPORT_ES6', '--extern-post-js', test_file('modularize_post_js.js')],),
+  })
+  def test_proxy_to_worker(self, args):
+    self.do_runf('hello_world.c', emcc_args=['--proxy-to-worker'] + args)
 
   @also_with_standalone_wasm()
   def test_console_out(self):

tools/link.py

@@ -2501,7 +2501,7 @@ def generate_traditional_runtime_html(target, options, js_target, target_basenam
   var filename = '%s';
   if ((',' + window.location.search.substr(1) + ',').indexOf(',noProxy,') < 0) {
     console.log('running code in a web worker');
-''' % get_subresource_location(proxy_worker_filename)) + worker_js + '''
+''' % get_subresource_location_js(proxy_worker_filename)) + worker_js + '''
   } else {
     console.log('running code on the main thread');
     var fileBytes = tryParseAsDataURI(filename);
@@ -2569,7 +2569,7 @@ def generate_traditional_runtime_html(target, options, js_target, target_basenam
             // Current browser supports Wasm, proceed with loading the main JS runtime.
             loadMainJs();
           }
-''' % (script.inline, get_subresource_location(wasm_target) + '.js')
+''' % (script.inline, get_subresource_location_js(wasm_target + '.js'))
 
   shell = do_replace(shell, '{{{ SCRIPT }}}', script.replacement())
   shell = shell.replace('{{{ SHELL_CSS }}}', utils.read_file(utils.path_from_root('src/shell.css')))
@@ -2645,7 +2645,7 @@ def generate_html(target, options, js_target, target_basename, wasm_target):
 def generate_worker_js(target, js_target, target_basename):
   if settings.SINGLE_FILE:
     # compiler output is embedded as base64 data URL
-    proxy_worker_filename = get_subresource_location(js_target)
+    proxy_worker_filename = get_subresource_location_js(js_target)
   else:
     # compiler output goes in .worker.js file
     move_file(js_target, shared.replace_suffix(js_target, get_worker_js_suffix()))
@@ -2958,13 +2958,17 @@ def move_file(src, dst):
 
 
 # Returns the subresource location for run-time access
-def get_subresource_location(path):
+def get_subresource_location(path, mimetype='application/octet-stream'):
   if settings.SINGLE_FILE:
-    return 'data:application/octet-stream;base64,' + base64_encode(path)
+    return f'data:{mimetype};base64,{base64_encode(path)}'
   else:
     return os.path.basename(path)
 
 
+def get_subresource_location_js(path):
+  return get_subresource_location(path, 'text/javascript')
+
+
 @ToolchainProfiler.profile()
 def package_files(options, target):
   rtn = []