src: provide faster string character counts

anonrig · anonrig · commit 1a90cbbd3053 · 2023-02-02T16:42:15.000-05:00
diff --git a/src/inspector/node_string.cc b/src/inspector/node_string.cc
@@ -1,5 +1,6 @@
 #include "node_string.h"
 #include "node/inspector/protocol/Protocol.h"
+#include "simdutf.h"
 
 #include <unicode/unistr.h>
 
@@ -118,9 +119,8 @@ const uint8_t* CharactersUTF8(const String& s){
 }
 
 size_t CharacterCount(const String& s){
- icu::UnicodeString utf16 =
- icu::UnicodeString::fromUTF8(icu::StringPiece(s.data(), s.length()));
- return utf16.countChar32();
+ // TODO(@anonrig): Test to make sure CharacterCount returns correctly.
+ return simdutf::utf32_length_from_utf8(s.data(), s.length());
 }
 
 } // namespace StringUtil

Original file line number	Diff line number	Diff line change
`@@ -1,5 +1,6 @@`
`1`	`1`	`#include"node_string.h"`
`2`	`2`	`#include"node/inspector/protocol/Protocol.h"`
	`3`	`+#include"simdutf.h"`
`3`	`4`
`4`	`5`	`#include<unicode/unistr.h>`
`5`	`6`
`@@ -118,9 +119,8 @@ const uint8_t* CharactersUTF8(const String& s){`
`118`	`119`	`}`
`119`	`120`
`120`	`121`	`size_tCharacterCount(const String& s){`
`121`		`- icu::UnicodeString utf16 =`
`122`		`-icu::UnicodeString::fromUTF8(icu::StringPiece(s.data(), s.length()));`
`123`		`-return utf16.countChar32();`
	`122`	`+// TODO(@anonrig): Test to make sure CharacterCount returns correctly.`
	`123`	`+returnsimdutf::utf32_length_from_utf8(s.data(), s.length());`
`124`	`124`	`}`
`125`	`125`
`126`	`126`	`} // namespace StringUtil`