Add option to use source phase imports for wasm module loading

Now that node support has been landed we can test this, at least
against the latest node canary builds.

See nodejs/node#56919

Fixes: emscripten-core#23047

Author: sbc100

ChangeLog.md

@@ -20,6 +20,9 @@ See docs/process.md for more on how version tagging works.
 
 4.0.5 (in development)
 ----------------------
+- Added initial support for wasm source phase imports via
+  `-sSOURCE_PHASE_IMPORTS`.  This is currently experimental and not yet
+  implemented in browsers. (#23175)
 
 4.0.4 - 02/25/25
 ----------------

site/source/docs/tools_reference/settings_reference.rst

@@ -3347,3 +3347,15 @@ Use _ for non-pointer arguments, p for pointer/i53 arguments, and P for optional
 Example use -sSIGNATURE_CONVERSIONS=someFunction:_p,anotherFunction:p
 
 Default value: []
+
+.. _source_phase_imports:
+
+SOURCE_PHASE_IMPORTS
+====================
+
+Experimental support for wasm source phase imports.
+This is only currenty implement in the pre-release/nightly version of node,
+and not yet supported by browsers.
+Requires EXPORT_ES6
+
+Default value: false

src/preamble.js

@@ -558,11 +558,40 @@ function instrumentWasmTableWithAbort() {
 }
 #endif
 
+#if LOAD_SOURCE_MAP
+var wasmSourceMap;
+#include "source_map_support.js"
+
+function receiveSourceMapJSON(sourceMap) {
+  wasmSourceMap = new WasmSourceMap(sourceMap);
+  {{{ runIfMainThread("removeRunDependency('source-map');") }}}
+}
+#endif
+
+#if (PTHREADS || WASM_WORKERS) && (LOAD_SOURCE_MAP || USE_OFFSET_CONVERTER)
+// When using postMessage to send an object, it is processed by the structured
+// clone algorithm.  The prototype, and hence methods, on that object is then
+// lost. This function adds back the lost prototype.  This does not work with
+// nested objects that has prototypes, but it suffices for WasmSourceMap and
+// WasmOffsetConverter.
+function resetPrototype(constructor, attrs) {
+  var object = Object.create(constructor.prototype);
+  return Object.assign(object, attrs);
+}
+#endif
+
+#if USE_OFFSET_CONVERTER
+var wasmOffsetConverter;
+#include "wasm_offset_converter.js"
+#endif
+
+#if !SOURCE_PHASE_IMPORTS
 #if SINGLE_FILE
 // In SINGLE_FILE mode the wasm binary is encoded inline here as a data: URL.
 var wasmBinaryFile = '{{{ WASM_BINARY_FILE }}}';
 #else
 var wasmBinaryFile;
+
 function findWasmBinary() {
 #if EXPORT_ES6 && !AUDIO_WORKLET
   if (Module['locateFile']) {
@@ -647,13 +676,6 @@ var splitModuleProxyHandler = {
 };
 #endif
 
-#if LOAD_SOURCE_MAP
-function receiveSourceMapJSON(sourceMap) {
-  wasmSourceMap = new WasmSourceMap(sourceMap);
-  {{{ runIfMainThread("removeRunDependency('source-map');") }}}
-}
-#endif
-
 #if SPLIT_MODULE || !WASM_ASYNC_COMPILATION
 function instantiateSync(file, info) {
   var module;
@@ -701,18 +723,6 @@ function instantiateSync(file, info) {
 }
 #endif
 
-#if (PTHREADS || WASM_WORKERS) && (LOAD_SOURCE_MAP || USE_OFFSET_CONVERTER)
-// When using postMessage to send an object, it is processed by the structured
-// clone algorithm.  The prototype, and hence methods, on that object is then
-// lost. This function adds back the lost prototype.  This does not work with
-// nested objects that has prototypes, but it suffices for WasmSourceMap and
-// WasmOffsetConverter.
-function resetPrototype(constructor, attrs) {
-  var object = Object.create(constructor.prototype);
-  return Object.assign(object, attrs);
-}
-#endif
-
 #if WASM_ASYNC_COMPILATION
 async function instantiateArrayBuffer(binaryFile, imports) {
   try {
@@ -815,6 +825,7 @@ async function instantiateAsync(binary, binaryFile, imports) {
   return instantiateArrayBuffer(binaryFile, imports);
 }
 #endif // WASM_ASYNC_COMPILATION
+#endif // SOURCE_PHASE_IMPORTS
 
 function getWasmImports() {
 #if PTHREADS
@@ -1016,10 +1027,14 @@ function getWasmImports() {
   }
 #endif
 
+#if SOURCE_PHASE_IMPORTS
+  var instance = await WebAssembly.instantiate(wasmModule, info);
+  var exports = receiveInstantiationResult({instance, 'module':wasmModule});
+  return exports;
+#else
 #if !SINGLE_FILE
   wasmBinaryFile ??= findWasmBinary();
 #endif
-
 #if WASM_ASYNC_COMPILATION
 #if RUNTIME_DEBUG
   dbg('asynchronously preparing wasm');
@@ -1051,6 +1066,7 @@ function getWasmImports() {
   return receiveInstance(result[0]);
 #endif
 #endif // WASM_ASYNC_COMPILATION
+#endif // SOURCE_PHASE_IMPORTS
 }
 
 #if !WASM_BIGINT

src/settings.js

@@ -2182,6 +2182,13 @@ var LEGACY_RUNTIME = false;
 // [link]
 var SIGNATURE_CONVERSIONS = [];
 
+// Experimental support for wasm source phase imports.
+// This is only currenty implement in the pre-release/nightly version of node,
+// and not yet supported by browsers.
+// Requires EXPORT_ES6
+// [link]
+var SOURCE_PHASE_IMPORTS = false;
+
 // For renamed settings the format is:
 // [OLD_NAME, NEW_NAME]
 // For removed settings (which now effectively have a fixed value and can no

test/test_other.py

@@ -358,13 +358,23 @@ def test_emcc_generate_config(self, compiler):
   @parameterized({
     '': ([],),
     'node': (['-sENVIRONMENT=node'],),
+    # load a worker before startup to check ES6 modules there as well
+    'pthreads': (['-pthread', '-sPTHREAD_POOL_SIZE=1'],),
   })
   def test_esm(self, args):
     self.run_process([EMCC, '-o', 'hello_world.mjs',
                       '--extern-post-js', test_file('modularize_post_js.js'),
                       test_file('hello_world.c')] + args)
-    src = read_file('hello_world.mjs')
-    self.assertContained('export default Module;', src)
+    self.assertContained('export default Module;', read_file('hello_world.mjs'))
+    self.assertContained('hello, world!', self.run_js('hello_world.mjs'))
+
+  @requires_node_canary
+  def test_esm_source_phase_imports(self):
+    self.node_args += ['--experimental-wasm-modules']
+    self.run_process([EMCC, '-o', 'hello_world.mjs', '-sSOURCE_PHASE_IMPORTS',
+                      '--extern-post-js', test_file('modularize_post_js.js'),
+                      test_file('hello_world.c')])
+    self.assertContained('import source wasmModule from', read_file('hello_world.mjs'))
     self.assertContained('hello, world!', self.run_js('hello_world.mjs'))
 
   @parameterized({

tools/link.py

@@ -1757,6 +1757,9 @@ def get_full_import_name(name):
     if settings.ASYNCIFY == 2:
       diagnostics.warning('experimental', '-sASYNCIFY=2 (JSPI) is still experimental')
 
+  if settings.SOURCE_PHASE_IMPORTS:
+    diagnostics.warning('experimental', '-sSOURCE_PHASE_IMPORTS is still experimental and not yet supported in browsers')
+
   if settings.WASM2JS:
     if settings.GENERATE_SOURCE_MAP:
       exit_with_error('wasm2js does not support source maps yet (debug in wasm for now)')
@@ -2478,6 +2481,9 @@ def modularize():
 })();
 ''' % {'EXPORT_NAME': settings.EXPORT_NAME}
 
+  if settings.SOURCE_PHASE_IMPORTS:
+    src = f"import source wasmModule from './{settings.WASM_BINARY_FILE}';\n\n" + src
+
   # Given the async nature of how the Module function and Module object
   # come into existence in AudioWorkletGlobalScope, store the Module
   # function under a different variable name so that AudioWorkletGlobalScope