Implement V8::String::{Utf8Length, IsOneByte, `ContainsOnlyOneB…

…yte`, `IsExternal`, `IsExternalTwoByte`, `IsExternalOneByte`} (#13417)

src/bun.js/bindings/v8/V8String.cpp

@@ -1,6 +1,7 @@
 #include "V8String.h"
 
 #include "V8HandleScope.h"
+#include "wtf/SIMDUTF.h"
 
 using JSC::JSString;
 
@@ -30,6 +31,75 @@ MaybeLocal<String> String::NewFromUtf8(Isolate* isolate, char const* data, NewSt
     return MaybeLocal<String>(isolate->globalInternals()->currentHandleScope()->createLocal<String>(jsString));
 }
 
+int String::Utf8Length(Isolate* isolate) const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return 0;
+    }
+
+    auto str = jsString->view(isolate->globalObject());
+    if (str->is8Bit()) {
+        const auto span = str->span8();
+        size_t len = simdutf::utf8_length_from_latin1(reinterpret_cast<const char*>(span.data()), span.size());
+        return static_cast<int>(len);
+    } else {
+        const auto span = str->span16();
+        size_t len = simdutf::utf8_length_from_utf16(span.data(), span.size());
+        return static_cast<int>(len);
+    }
+}
+
+bool String::IsOneByte() const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return true;
+    }
+    auto impl = jsString->tryGetValue();
+    return impl->is8Bit();
+}
+
+bool String::ContainsOnlyOneByte() const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return true;
+    }
+    auto impl = jsString->tryGetValue();
+    return impl->containsOnlyLatin1();
+}
+
+bool String::IsExternal() const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return false;
+    }
+    auto impl = jsString->tryGetValue();
+    return !impl->isNull() && impl->impl()->isExternal();
+}
+
+bool String::IsExternalTwoByte() const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return false;
+    }
+    auto impl = jsString->tryGetValue();
+    return !impl->isNull() && impl->impl()->isExternal() && !impl->is8Bit();
+}
+
+bool String::IsExternalOneByte() const
+{
+    auto jsString = localToObjectPointer<JSString>();
+    if (jsString->length() == 0) {
+        return false;
+    }
+    auto impl = jsString->tryGetValue();
+    return !impl->isNull() && impl->impl()->isExternal() && impl->is8Bit();
+}
+
 extern "C" size_t TextEncoder__encodeInto8(const LChar* stringPtr, size_t stringLen, void* ptr, size_t len);
 extern "C" size_t TextEncoder__encodeInto16(const UChar* stringPtr, size_t stringLen, void* ptr, size_t len);
 
@@ -69,9 +139,7 @@ int String::WriteUtf8(Isolate* isolate, char* buffer, int length, int* nchars_re
 int String::Length() const
 {
     auto jsString = localToObjectPointer<JSString>();
-    RELEASE_ASSERT(jsString->isString());
-    WTF::String s = jsString->getString(Isolate::GetCurrent()->globalObject());
-    return s.length();
+    return static_cast<int>(jsString->length());
 }
 
 }

src/bun.js/bindings/v8/V8String.h

@@ -33,6 +33,42 @@ class String : Primitive {
     BUN_EXPORT int WriteUtf8(Isolate* isolate, char* buffer, int length = -1, int* nchars_ref = nullptr, int options = NO_OPTIONS) const;
     BUN_EXPORT int Length() const;
 
+    /**
+     * Returns the number of bytes in the UTF-8 encoded
+     * representation of this string.
+     */
+    BUN_EXPORT int Utf8Length(Isolate* isolate) const;
+
+    /**
+     * Returns whether this string is known to contain only one byte data,
+     * i.e. ISO-8859-1 code points.
+     * Does not read the string.
+     * False negatives are possible.
+     */
+    BUN_EXPORT bool IsOneByte() const;
+
+    /**
+     * Returns whether this string contain only one byte data,
+     * i.e. ISO-8859-1 code points.
+     * Will read the entire string in some cases.
+     */
+    BUN_EXPORT bool ContainsOnlyOneByte() const;
+
+    /**
+     * Returns true if the string is external.
+     */
+    BUN_EXPORT bool IsExternal() const;
+
+    /**
+     * Returns true if the string is both external and two-byte.
+     */
+    BUN_EXPORT bool IsExternalTwoByte() const;
+
+    /**
+     * Returns true if the string is both external and one-byte.
+     */
+    BUN_EXPORT bool IsExternalOneByte() const;
+
     JSC::JSString* localToJSString()
     {
         return localToObjectPointer<JSC::JSString>();

src/napi/napi.zig

@@ -1744,6 +1744,12 @@ const V8API = if (!bun.Environment.isWindows) struct {
     pub extern fn _ZNK2v85Value8IsStringEv() *anyopaque;
     pub extern fn _ZN2v87Boolean3NewEPNS_7IsolateEb() *anyopaque;
     pub extern fn _ZN2v86Object16GetInternalFieldEi() *anyopaque;
+    pub extern fn _ZNK2v86String10Utf8LengthEPNS_7IsolateE() *anyopaque;
+    pub extern fn _ZNK2v86String10IsExternalEv() *anyopaque;
+    pub extern fn _ZNK2v86String17IsExternalOneByteEv() *anyopaque;
+    pub extern fn _ZNK2v86String17IsExternalTwoByteEv() *anyopaque;
+    pub extern fn _ZNK2v86String9IsOneByteEv() *anyopaque;
+    pub extern fn _ZNK2v86String19ContainsOnlyOneByteEv() *anyopaque;
 } else struct {
     // MSVC name mangling is different than it is on unix.
     // To make this easier to deal with, I have provided a script to generate the list of functions.
@@ -1800,6 +1806,12 @@ const V8API = if (!bun.Environment.isWindows) struct {
     pub extern fn @"?IsString@Value@v8@@QEBA_NXZ"() *anyopaque;
     pub extern fn @"?New@Boolean@v8@@SA?AV?$Local@VBoolean@v8@@@2@PEAVIsolate@2@_N@Z"() *anyopaque;
     pub extern fn @"?GetInternalField@Object@v8@@QEAA?AV?$Local@VData@v8@@@2@H@Z"() *anyopaque;
+    pub extern fn @"?IsExternal@String@v8@@QEBA_NXZ"() *anyopaque;
+    pub extern fn @"?IsExternalOneByte@String@v8@@QEBA_NXZ"() *anyopaque;
+    pub extern fn @"?IsExternalTwoByte@String@v8@@QEBA_NXZ"() *anyopaque;
+    pub extern fn @"?IsOneByte@String@v8@@QEBA_NXZ"() *anyopaque;
+    pub extern fn @"?Utf8Length@String@v8@@QEBAHPEAVIsolate@2@@Z"() *anyopaque;
+    pub extern fn @"?ContainsOnlyOneByte@String@v8@@QEBA_NXZ"() *anyopaque;
 };
 
 pub fn fixDeadCodeElimination() void {

src/symbols.def

@@ -616,3 +616,9 @@ EXPORTS
   ?IsString@Value@v8@@QEBA_NXZ
   ?New@Boolean@v8@@SA?AV?$Local@VBoolean@v8@@@2@PEAVIsolate@2@_N@Z
   ?GetInternalField@Object@v8@@QEAA?AV?$Local@VData@v8@@@2@H@Z
+  ?IsExternal@String@v8@@QEBA_NXZ
+  ?IsExternalOneByte@String@v8@@QEBA_NXZ
+  ?IsExternalTwoByte@String@v8@@QEBA_NXZ
+  ?IsOneByte@String@v8@@QEBA_NXZ
+  ?Utf8Length@String@v8@@QEBAHPEAVIsolate@2@@Z
+  ?ContainsOnlyOneByte@String@v8@@QEBA_NXZ

src/symbols.dyn

@@ -203,4 +203,10 @@
     __ZNK2v85Value8IsStringEv;
     __ZN2v87Boolean3NewEPNS_7IsolateEb;
     __ZN2v86Object16GetInternalFieldEi;
+    __ZNK2v86String10Utf8LengthEPNS_7IsolateE;
+    __ZNK2v86String10IsExternalEv;
+    __ZNK2v86String17IsExternalOneByteEv;
+    __ZNK2v86String17IsExternalTwoByteEv;
+    __ZNK2v86String9IsOneByteEv;
+    __ZNK2v86String19ContainsOnlyOneByteEv;
 };

src/symbols.txt

@@ -202,3 +202,9 @@ __ZNK2v85Value7IsFalseEv
 __ZNK2v85Value8IsStringEv
 __ZN2v87Boolean3NewEPNS_7IsolateEb
 __ZN2v86Object16GetInternalFieldEi
+__ZNK2v86String10Utf8LengthEPNS_7IsolateE
+__ZNK2v86String10IsExternalEv
+__ZNK2v86String17IsExternalOneByteEv
+__ZNK2v86String17IsExternalTwoByteEv
+__ZNK2v86String9IsOneByteEv
+__ZNK2v86String19ContainsOnlyOneByteEv

test/v8/v8-module/main.cpp

@@ -238,6 +238,17 @@ static bool perform_string_test(const FunctionCallbackInfo<Value> &info,
     return false;
   }
 
+  if (v8_string->Utf8Length(isolate) != encoded_utf_8_length) {
+    fail(info, "String::Utf8Length returned wrong length: expected %d got %d",
+         encoded_utf_8_length, v8_string->Utf8Length(isolate));
+    return false;
+  }
+
+  if (v8_string->IsExternal()) {
+    fail(info, "String::IsExternal returned true");
+    return false;
+  }
+
   ok(info);
   return true;
 }