Make the utf8_range implementation just in C

PiperOrigin-RevId: 590961088

php/ext/google/protobuf/config.m4

@@ -4,7 +4,7 @@ if test "$PHP_PROTOBUF" != "no"; then
 
   PHP_NEW_EXTENSION(
     protobuf,
-    arena.c array.c convert.c def.c map.c message.c names.c php-upb.c protobuf.c third_party/utf8_range/naive.c third_party/utf8_range/range2-neon.c third_party/utf8_range/range2-sse.c,
+    arena.c array.c convert.c def.c map.c message.c names.c php-upb.c protobuf.c third_party/utf8_range/utf8_range.c,
     $ext_shared, , -std=gnu99 -I@ext_srcdir@/third_party/utf8_range)
   PHP_ADD_BUILD_DIR($ext_builddir/third_party/utf8_range)
 

python/convert.c

@@ -241,7 +241,7 @@ bool PyUpb_PyToUpb(PyObject* obj, const upb_FieldDef* f, upb_MessageValue* val,
         // Use the object's bytes if they are valid UTF-8.
         char* ptr;
         if (PyBytes_AsStringAndSize(obj, &ptr, &size) < 0) return false;
-        if (utf8_range2((const unsigned char*)ptr, size) != 0) {
+        if (!utf8_range_IsValid(ptr, size)) {
           // Invalid UTF-8.  Try to convert the message to a Python Unicode
           // object, even though we know this will fail, just to get the
           // idiomatic Python error message.

ruby/.gitignore

@@ -8,8 +8,6 @@ pkg/
 tmp/
 tests/google/
 ext/google/protobuf_c/third_party/utf8_range/utf8_range.h
-ext/google/protobuf_c/third_party/utf8_range/range2-sse.c
-ext/google/protobuf_c/third_party/utf8_range/range2-neon.c
-ext/google/protobuf_c/third_party/utf8_range/naive.c
+ext/google/protobuf_c/third_party/utf8_range/utf8_range.c
 ext/google/protobuf_c/third_party/utf8_range/LICENSE
 lib/google/protobuf/*_pb.rb

ruby/Rakefile

@@ -75,7 +75,7 @@ task :copy_third_party do
     # We need utf8_range in-tree.
     utf8_root = '../third_party/utf8_range'
     %w[
-      utf8_range.h naive.c range2-neon.c range2-neon.c range2-sse.c LICENSE
+      utf8_range.h utf8_range.c LICENSE
     ].each do |file|
       FileUtils.cp File.join(utf8_root, file),
                    "ext/google/protobuf_c/third_party/utf8_range"

ruby/ext/google/protobuf_c/extconf.rb

@@ -22,7 +22,7 @@
 
 $srcs = ["protobuf.c", "convert.c", "defs.c", "message.c",
          "repeated_field.c", "map.c", "ruby-upb.c", "wrap_memcpy.c",
-         "naive.c", "range2-neon.c", "range2-sse.c", "shared_convert.c",
+         "utf8_range.c", "shared_convert.c",
          "shared_message.c"]
 
 create_makefile(ext_name)

ruby/lib/google/tasks/ffi.rake

@@ -74,9 +74,7 @@ begin
       FFI::Compiler::CompileTask.new 'protobuf_c_ffi' do |c|
         configure_common_compile_task c
         # Ruby UPB was already compiled with different flags.
-        c.exclude << "/range2-neon.c"
-        c.exclude << "/range2-sse.c"
-        c.exclude << "/naive.c"
+        c.exclude << "/utf8_range.c"
         c.exclude << "/ruby-upb.c"
       end
 

third_party/utf8_range/BUILD.bazel

@@ -23,9 +23,7 @@ exports_files([
 filegroup(
     name = "utf8_range_srcs",
     srcs = [
-        "naive.c",
-        "range2-neon.c",
-        "range2-sse.c",
+        "utf8_range.c",
         "utf8_range.h",
     ],
     visibility = ["//:__subpackages__"],
@@ -34,9 +32,7 @@ filegroup(
 cc_library(
     name = "utf8_range",
     srcs = [
-        "naive.c",
-        "range2-neon.c",
-        "range2-sse.c",
+        "utf8_range.c",
     ],
     hdrs = ["utf8_range.h"],
     strip_include_prefix = "/third_party/utf8_range",
@@ -48,14 +44,19 @@ cc_library(
     hdrs = ["utf8_validity.h"],
     strip_include_prefix = "/third_party/utf8_range",
     deps = [
+        ":utf8_range",
         "@com_google_absl//absl/strings",
     ],
 )
 
 cc_test(
     name = "utf8_validity_test",
-    srcs = ["utf8_validity_test.cc"],
+    srcs = [
+        "utf8_range.c",
+        "utf8_validity_test.cc",
+    ],
     deps = [
+        ":utf8_range",
         ":utf8_validity",
         "@com_google_absl//absl/strings",
         "@com_google_googletest//:gtest_main",

third_party/utf8_range/CMakeLists.txt

@@ -12,14 +12,12 @@ option (utf8_range_ENABLE_INSTALL "Configure installation" ON)
 ##
 # Create the lightweight C library
 add_library (utf8_range STATIC
-  naive.c
-  range2-neon.c
-  range2-sse.c
+  utf8_range.c
 )
 
 ##
 # A heavier-weight C++ wrapper that supports Abseil.
-add_library (utf8_validity STATIC utf8_validity.cc)
+add_library (utf8_validity STATIC utf8_validity.cc utf8_range.c)
 
 # Load Abseil dependency.
 if (NOT TARGET absl::strings)