Merge pull request #11520 from keymanapp/feat/core/9467-devolve-norm-…

…to-js2-epic-ldml feat(core): change normalize_nfd() to use JS native call instead of ICU on wasm 🙀
keymanapp · May 27, 2024 · 7d9b1a1 · 7d9b1a1
2 parents ab7fd5e + 948f979
commit 7d9b1a1
Show file tree

Hide file tree

Showing 4 changed files with 36 additions and 3 deletions.
diff --git a/core/meson.build b/core/meson.build
@@ -35,6 +35,11 @@ if get_option('buildtype') == 'debug'
   add_global_arguments('-DDEBUG', language : 'cpp')
 endif
 
+# shared with a number of subdirs
+if cpp_compiler.get_id() == 'emscripten'
+  wasm_exported_runtime_methods = '-sEXPORTED_RUNTIME_METHODS=[\'UTF8ToString\',\'stringToNewUTF8\']'
+endif
+
 subdir('doc')
 subdir('include')
 subdir('src')

diff --git a/core/src/util_normalize.cpp b/core/src/util_normalize.cpp
@@ -10,11 +10,24 @@
 #include "core_icu.h"
 #include "kmx/kmx_xstring.h"
 
+#ifdef __EMSCRIPTEN__
+#include <emscripten.h>
+#include "utfcodec.hpp"
+// JS implementations
+
+EM_JS(char*, NormalizeNFD, (const char* input), {
+  if (!input) return input; // pass through null
+  const instr = Module.UTF8ToString(input);
+  const nfd = instr.normalize("NFD");
+  return stringToNewUTF8(nfd);
+});
+#endif
 
 namespace km {
 namespace core {
 namespace util {
 
+#ifndef __EMSCRIPTEN__
 
 /**
  * Internal function to normalize with a specified mode.
@@ -41,6 +54,7 @@ static bool normalize(const icu::Normalizer2 *n, std::u16string &str, UErrorCode
   }
   return U_SUCCESS(status);
 }
+#endif
 
 bool normalize_nfd(std::u32string &str) {
   std::u16string rstr = km::core::kmx::u32string_to_u16string(str);
@@ -53,10 +67,24 @@ bool normalize_nfd(std::u32string &str) {
 }
 
 bool normalize_nfd(std::u16string &str) {
+#ifdef __EMSCRIPTEN__
+  std::string instr = convert<char16_t,char>(str);
+  const char *in = instr.c_str();
+  char *out = NormalizeNFD(in);
+  if (out == nullptr) {
+    assert(out != nullptr);
+    return false;
+  }
+  std::string outstr(out);
+  str = convert<char,char16_t>(outstr);
+  free(out);
+  return true;
+#else
   UErrorCode status = U_ZERO_ERROR;
   const icu::Normalizer2 *nfd = icu::Normalizer2::getNFDInstance(status);
   UASSERT_SUCCESS(status);
   return normalize(nfd, str, status);
+#endif
 }
 
 /**

diff --git a/core/tests/unit/kmx/meson.build b/core/tests/unit/kmx/meson.build
@@ -23,7 +23,7 @@ common_test_keyboards_baseline = source_path / '../../../../common/test/keyboard
 if cpp_compiler.get_id() == 'emscripten'
   # TODO: we will move the tests to ES6 in the future
   # '-sMODULARIZE', '-sEXPORT_ES6',
-  tests_flags += ['-lnodefs.js', '-sEXPORTED_RUNTIME_METHODS=[\'UTF8ToString\']']
+  tests_flags += ['-lnodefs.js', wasm_exported_runtime_methods]
 endif
 
 kmx = executable('kmx',

diff --git a/core/tests/unit/ldml/meson.build b/core/tests/unit/ldml/meson.build
@@ -50,7 +50,7 @@ if cpp_compiler.get_id() == 'emscripten'
   invalid_test_path = '/'
   tests_flags += ['-lnodefs.js',
                   '-sNO_DISABLE_EXCEPTION_CATCHING', # for test exceptions
-                  '-sEXPORTED_RUNTIME_METHODS=[\'UTF8ToString\']']
+                  wasm_exported_runtime_methods]
 else
   tests_flags = []
   test_path         = meson.current_build_dir() / 'keyboards'
@@ -117,7 +117,7 @@ test('test_transforms', t, suite: 'ldml')
 normalization_tests_flags = tests_flags
 
 if cpp_compiler.get_id() == 'emscripten'
-  normalization_tests_flags += ['-lnodefs.js', '-sEXPORTED_RUNTIME_METHODS=[\'UTF8ToString\']']
+  normalization_tests_flags += ['-lnodefs.js', wasm_exported_runtime_methods]
 endif
 
 test_context_normalization = executable('test_context_normalization',