First pass on JS built-ins support

gershnik · gershnik · commit 00180e570a4a · 2025-09-25T16:42:41.000-07:00
diff --git a/.github/workflows/test.yml b/.github/workflows/test.yml
@@ -20,6 +20,7 @@ jobs:
     outputs:
       hosts: ${{ steps.matrix.outputs.hosts }}
       containers: ${{ steps.matrix.outputs.containers }}
+      emscriptens: ${{ steps.matrix.outputs.emscriptens }}
 
     steps:
       - name: Define Matrix
@@ -53,9 +54,15 @@ jobs:
             20: 'ubuntu-latest',
             21: 'ubuntu-latest'
           }
+          emscripten_versions = [
+            '3.1.74',
+            '4.0.15',
+          ]
           
           hosts = []
           containers=[]
+          emscriptens=[]
+          
           
           #macOS
           for runon, xcode in macos_map.items():
@@ -81,9 +88,14 @@ jobs:
           for clang, runon in clang_map.items():
             hosts.append({'os': runon, 'compiler': 'clang', 'version': clang, 'jobname': f'Linux - Clang{clang}'})
 
+          #emscripten
+          for emscripten in emscripten_versions:
+            emscriptens.append({'version': emscripten, 'jobname': f'Emscripten - {emscripten}'})
+
           with open(os.environ['GITHUB_OUTPUT'], 'w') as env:
             print('hosts=' + json.dumps(hosts), file=env) 
             print('containers=' + json.dumps(containers), file=env)
+            print('emscriptens=' + json.dumps(emscriptens), file=env)
 
   desktop:
     needs: define-matrix
@@ -206,7 +218,7 @@ jobs:
     - name: Configure
       shell: bash
       run: |
-        cmake -S . -B out -DCMAKE_BUILD_TYPE=MinSizeRel
+        cmake -GNinja -S . -B out -DCMAKE_BUILD_TYPE=MinSizeRel
 
     - name: Build and Test
       shell: bash
@@ -297,8 +309,14 @@ jobs:
             echo "::endgroup::"
 
   emscripten:
+    needs: define-matrix
+    name: ${{ matrix.jobname }}
     runs-on: ubuntu-latest
-    container: emscripten/emsdk:3.1.70
+    container: emscripten/emsdk:${{ matrix.version }}
+    strategy:
+      fail-fast: false
+      matrix: 
+        include: ${{ fromJSON(needs.define-matrix.outputs.emscriptens) }}
 
     steps:
     - name: Checkout
@@ -319,7 +337,7 @@ jobs:
 
     - name: Configure
       shell: bash
-      run: cmake -S . -B out -DCMAKE_TOOLCHAIN_FILE=$EMSDK/upstream/emscripten/cmake/Modules/Platform/Emscripten.cmake -DCMAKE_BUILD_TYPE=MinSizeRel
+      run: cmake -GNinja -S . -B out -DCMAKE_TOOLCHAIN_FILE=$EMSDK/upstream/emscripten/cmake/Modules/Platform/Emscripten.cmake -DCMAKE_BUILD_TYPE=MinSizeRel
 
 
     - name: Build and Test
diff --git a/lib/inc/sys_string/impl/platforms/emscripten_javascript.h b/lib/inc/sys_string/impl/platforms/emscripten_javascript.h
@@ -12,6 +12,19 @@
 
 #include <sys_string/impl/util/generic_impl.h>
 
+//Like EM_JS but with inline and extern "C" only on data
+#define SYSSTR_EM_JS(ret, c_name, js_name, params, ...)                        \
+  ret c_name params EM_IMPORT(js_name);                                        \
+  extern "C"                                                                   \
+  __attribute__((visibility("hidden")))                                        \
+  inline void * __em_js_ref_##c_name = (void*)&c_name;                         \
+  extern "C"                                                                   \
+  EMSCRIPTEN_KEEPALIVE                                                         \
+  __attribute__((section("em_js"), aligned(1)))                                \
+  inline char __em_js__##js_name[] =                                           \
+    #params "<::>" #__VA_ARGS__;
+
+
 namespace sysstr::util
 {
     struct emscripten_traits
@@ -33,6 +46,40 @@ namespace sysstr::util
 
     using emscripten_char_access       = generic::char_access<emscripten_traits::storage_type, emscripten_traits::size_type>;
 
+#ifdef __wasm_reference_types__
+
+    #if SYS_STRING_USE_WASM_JS_STRING
+        uint32_t js_length(__externref_t str) __attribute__((import_module("wasm:js-string"), import_name("length")));
+        uint32_t js_charCodeAt(__externref_t str, uint32_t idx) __attribute__((import_module("wasm:js-string"), import_name("charCodeAt")));
+    #else
+
+        SYSSTR_EM_JS(uint32_t, js_length, sysstr_length, (__externref_t str), {
+            return str.length;
+        });
+
+        SYSSTR_EM_JS(uint32_t, js_charCodeAt, sysstr_charCodeAt, (__externref_t str, uint32_t idx), {
+            return str.charCodeAt(idx);
+        });
+    #endif
+
+    SYSSTR_EM_JS(__externref_t, js_UTF16ToString, sysstr_UTF16ToString, (const void * ptr, size_t length), {
+        let start = (ptr>>1);
+        if (length > 16 && UTF16Decoder)
+            return UTF16Decoder.decode(HEAPU16.subarray(start, start + length));
+
+        let ret = "";
+
+        for (let i = 0; i < length; ++i) {
+            const codeUnit = HEAPU16[start];
+            ret += String.fromCharCode(codeUnit);
+            ++start;
+        }
+        
+        return ret;
+    });
+    
+#endif
+
 }
 
 namespace sysstr
@@ -59,6 +106,12 @@ namespace sysstr
             super(create_buffer(js_str))
         {}
 
+#ifdef __wasm_reference_types__
+        emscripten_storage(__externref_t js_str):
+            super(create_buffer(js_str))
+        {}
+#endif
+
 #ifdef _MSC_VER
         template<has_utf_encoding Char>
         emscripten_storage(const Char * str, size_t len):
@@ -78,26 +131,12 @@ namespace sysstr
         using super::data;
         using super::copy_data;
 
-        auto make_js_string() const noexcept -> emscripten::EM_VAL 
-        {
+        auto make_js_string() const noexcept -> emscripten::EM_VAL {
             #pragma clang diagnostic push
             #pragma clang diagnostic ignored "-Wdollar-in-identifier-extension"
             
             auto ret = (emscripten::EM_VAL)EM_ASM_PTR({
-                let ptr = $0;
-                let length = $1;
-                if (length > 16 && UTF16Decoder)
-                    return Emval.toHandle(UTF16Decoder.decode(HEAPU8.subarray(ptr, ptr + length  * 2)));
-                
-                let ret = "";
-  
-                for (let i = 0; i < length; ++i) {
-                    const codeUnit = HEAP16[((ptr)>>1)];
-                    ret += String.fromCharCode(codeUnit);
-                    ptr += 2;
-                }
-                
-                return Emval.toHandle(ret);
+                return Emval.toHandle(sysstr_UTF16ToString($0, $1));
             }, data(), size());
 
             return ret;
@@ -106,6 +145,12 @@ namespace sysstr
             
         }
 
+#ifdef __wasm_reference_types__
+        auto make_js_string_ref() const noexcept -> __externref_t {
+            return util::js_UTF16ToString(data(), size());
+        }
+#endif
+
     protected:
         using super::size;
         using super::operator[];
@@ -133,14 +178,29 @@ namespace sysstr
                 let ptr = $1;
                 let length = str.length;
                 for (let i = 0; i < length; ++i) {
-                    HEAP16[((ptr)>>1)] = str.charCodeAt(i);
+                    HEAPU16[((ptr)>>1)] = str.charCodeAt(i);
                     ptr += 2;
                 }
             }, js_str, ret.data());
             return ret;
 
             #pragma clang diagnostic pop
         }
+
+#ifdef __wasm_reference_types__
+        static auto create_buffer(__externref_t js_str) -> buffer {
+            uint32_t len = util::js_length(js_str);
+
+            if (len == 0)
+                return buffer();
+
+            buffer ret(len);
+            auto ptr = ret.data();
+            for(uint32_t i = 0; i < len; ++i)
+                ptr[i] = util::js_charCodeAt(js_str, i);
+            return ret;
+        }
+#endif
     };
 }
 
diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
@@ -19,6 +19,24 @@ FetchContent_MakeAvailable(doctest)
 
 find_package(ICU 67 COMPONENTS uc)
 
+if (EMSCRIPTEN)
+    message(CHECK_START "Checking if ${NODE_JS_EXECUTABLE} supports wasm:js-string")
+
+    execute_process(
+        COMMAND ${NODE_JS_EXECUTABLE} ${CMAKE_CURRENT_LIST_DIR}/check-wasm-js-string.js 
+        RESULT_VARIABLE NODE_JS_HAS_WASM_JS_STRING_RESULT
+    )
+
+    if (${NODE_JS_HAS_WASM_JS_STRING_RESULT} EQUAL 0)
+        set(NODE_JS_HAS_WASM_JS_STRING TRUE)
+        message(CHECK_PASS "Success")
+    else()
+        set(NODE_JS_HAS_WASM_JS_STRING FALSE)
+        message(CHECK_FAIL "Failure")
+    endif()
+
+endif()
+
 if(${Python3_Development_FOUND} AND ${Python3_Interpreter_FOUND})
     include_directories(
     SYSTEM
@@ -162,6 +180,15 @@ foreach(LIB_SUFFIX ${LIB_TYPES})
         $<$<CXX_COMPILER_ID:GNU>:-Wall;-Wextra;-pedantic>
     )
 
+    if (EMSCRIPTEN AND STORAGE_SUFFIX STREQUAL "emscr")
+        target_compile_definitions(test-${TEST_SUFFIX} PRIVATE
+            SYS_STRING_USE_WASM_JS_STRING=$<IF:$<BOOL:${NODE_JS_HAS_WASM_JS_STRING}>,1,0>
+        )
+        target_compile_options(test-${TEST_SUFFIX} PRIVATE
+            -sMAIN_MODULE=2 -sSTRICT
+        )
+    endif()
+
     if ("${CMAKE_CXX_COMPILER_FRONTEND_VARIANT}" STREQUAL "MSVC")
         target_compile_options(test-${TEST_SUFFIX}
         PRIVATE
@@ -192,10 +219,21 @@ foreach(LIB_SUFFIX ${LIB_TYPES})
     endif()
 
     target_link_options(test-${TEST_SUFFIX} PRIVATE
-    
-        "$<$<PLATFORM_ID:Android>:-Wl,--export-dynamic>"
+        $<$<PLATFORM_ID:Android>:-Wl,--export-dynamic>
     )
 
+    if (EMSCRIPTEN AND STORAGE_SUFFIX STREQUAL "emscr")
+        target_link_options(test-${TEST_SUFFIX} PRIVATE
+            -sMAIN_MODULE=2
+            -sSTRICT
+            -sERROR_ON_UNDEFINED_SYMBOLS=0
+            -sWARN_ON_UNDEFINED_SYMBOLS=0
+            --pre-js ${CMAKE_CURRENT_LIST_DIR}/prefix.js
+            -sINCOMING_MODULE_JS_API=arguments,canvas,monitorRunDependencies,print,setStatus>
+        )
+        set_target_properties(test-${TEST_SUFFIX} PROPERTIES LINK_DEPENDS ${CMAKE_CURRENT_LIST_DIR}/prefix.js)
+    endif()
+
     if (SYS_STRING_COLLECT_COVERAGE)
 
         target_compile_options(test-${TEST_SUFFIX} PRIVATE
diff --git a/test/check-wasm-js-string.js b/test/check-wasm-js-string.js
@@ -0,0 +1,7 @@
+let bytes = new Uint8Array([
+    0, 97, 115, 109, 1, 0, 0, 0, 1, 4, 1, 96, 0, 0, 2, 23, 1, 14, 119, 97, 115,
+    109, 58, 106, 115, 45, 115, 116, 114, 105, 110, 103, 4, 99, 97, 115, 116, 0,
+    0,
+  ]);
+
+process.exit(!WebAssembly.validate(bytes, { builtins: ["js-string"] }) ? 0 : 1);  
diff --git a/test/prefix.js b/test/prefix.js
@@ -0,0 +1,14 @@
+
+
+WebAssembly.instantiate = function() {
+    let originalInstantiate = WebAssembly.instantiate;
+    return function(binary, imports) {
+        return originalInstantiate(binary, imports, {
+            builtins: ["js-string"]
+        });
+    }
+}();
+
+
+
+
diff --git a/test/test_javascript.cpp b/test/test_javascript.cpp
@@ -20,6 +20,10 @@ using namespace emscripten;
 
 #if !SYS_STRING_USE_GENERIC && !SYS_STRING_USE_PYTHON
 
+    EM_JS(__externref_t, make_test_string, (const char * str), {
+        return UTF8ToString(str);
+    });
+
     TEST_SUITE("javascript") {
 
     TEST_CASE( "Javascript Conversions" ) {
@@ -52,6 +56,18 @@ using namespace emscripten;
         }, handle);
         CHECK(result);
     }
+    
+#if defined(__wasm_reference_types__)
+    TEST_CASE( "Javascript Conversions with builtins" ) {
+
+        auto js_str = make_test_string("абвгдеёжзийклмнопрстуфхцчшщьыъэюя");
+
+        CHECK(sys_string(js_str) == S("абвгдеёжзийклмнопрстуфхцчшщьыъэюя"));
+
+        auto js_str2 = S("абвгдеёжзийклмнопрстуфхцчшщьыъэюя").make_js_string_ref();
+        CHECK(sys_string(js_str2) == S("абвгдеёжзийклмнопрстуфхцчшщьыъэюя"));
+    }
+#endif
 
     }
 
