githubmirror/node

src/string_bytes.cc
801 строка27 KB
Mert Can Altin
src: add Latin1 fast path in StringBytes::Encode utf8
18 июн 2026, 11:09
18 июн 2026, 11:091682264
// Copyright Joyent, Inc. and other Node contributors.
//
// Permission is hereby granted, free of charge, to any person obtaining a
// copy of this software and associated documentation files (the
// "Software"), to deal in the Software without restriction, including
// without limitation the rights to use, copy, modify, merge, publish,
// distribute, sublicense, and/or sell copies of the Software, and to permit
// persons to whom the Software is furnished to do so, subject to the
// following conditions:
//
// The above copyright notice and this permission notice shall be included
// in all copies or substantial portions of the Software.
//
// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
// OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
// MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN
// NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
// DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
// OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
// USE OR OTHER DEALINGS IN THE SOFTWARE.

#include "string_bytes.h"

#include "env-inl.h"
#include "nbytes.h"
#include "node_buffer.h"
#include "node_errors.h"
#include "simdutf.h"
#include "util.h"
#include "v8-external-memory-accounter.h"

#include <climits>
#include <cstring>  // memcpy

#include <algorithm>

// When creating strings >= this length v8's gc spins up and consumes
// most of the execution time. For these cases it's more performant to
// use external string resources.
#define EXTERN_APEX 0xFBEE9

namespace node {

using v8::ExternalMemoryAccounter;
using v8::HandleScope;
using v8::Isolate;
using v8::Just;
using v8::Local;
using v8::Maybe;
using v8::MaybeLocal;
using v8::Nothing;
using v8::String;
using v8::Value;

namespace {

template <typename ResourceType, typename TypeName>
class ExternString: public ResourceType {
 public:
  ~ExternString() override {
    free(const_cast<TypeName*>(data_));
    external_memory_accounter_->Decrease(isolate(), byte_length());
    delete external_memory_accounter_;
  }

  const TypeName* data() const override {
    return data_;
  }

  size_t length() const override {
    return length_;
  }

  size_t byte_length() const { return length() * sizeof(*data()); }

  static MaybeLocal<Value> NewFromCopy(Isolate* isolate,
                                       const TypeName* data,
                                       size_t length) {
    if (length == 0) {
      return String::Empty(isolate);
    }

    if (length < EXTERN_APEX) {
      return NewSimpleFromCopy(isolate, data, length);
    }

    TypeName* new_data = node::UncheckedMalloc<TypeName>(length);
    if (new_data == nullptr) {
      isolate->ThrowException(node::ERR_MEMORY_ALLOCATION_FAILED(isolate));
      return MaybeLocal<Value>();
    }
    memcpy(new_data, data, length * sizeof(*new_data));

    return ExternString<ResourceType, TypeName>::New(isolate, new_data, length);
  }

  // uses "data" for external resource, and will be free'd on gc
  static MaybeLocal<Value> New(Isolate* isolate,
                               TypeName* data,
                               size_t length) {
    if (length == 0)
      return String::Empty(isolate);

    if (length < EXTERN_APEX) {
      MaybeLocal<Value> str = NewSimpleFromCopy(isolate, data, length);
      free(data);
      return str;
    }

    ExternString* h_str = new ExternString<ResourceType, TypeName>(isolate,
                                                                   data,
                                                                   length);
    Local<Value> str;

    if (!NewExternal(isolate, h_str).ToLocal(&str)) {
      delete h_str;
      isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
      return MaybeLocal<Value>();
    }

    return str;
  }

  inline Isolate* isolate() const { return isolate_; }

 private:
  ExternString(Isolate* isolate, const TypeName* data, size_t length)
      : isolate_(isolate),
        external_memory_accounter_(new ExternalMemoryAccounter()),
        data_(data),
        length_(length) {
    external_memory_accounter_->Increase(isolate, byte_length());
  }
  static MaybeLocal<Value> NewExternal(Isolate* isolate,
                                       ExternString* h_str);

  // This method does not actually create ExternString instances.
  static MaybeLocal<Value> NewSimpleFromCopy(Isolate* isolate,
                                             const TypeName* data,
                                             size_t length);

  Isolate* isolate_;
  ExternalMemoryAccounter* external_memory_accounter_;
  const TypeName* data_;
  size_t length_;
};

typedef ExternString<String::ExternalOneByteStringResource, char>
    ExternOneByteString;
typedef ExternString<String::ExternalStringResource, uint16_t>
    ExternTwoByteString;

template <typename EncodeFn>
static MaybeLocal<Value> EncodeOneByteString(Isolate* isolate,
                                             size_t length,
                                             EncodeFn encode) {
  // 512B stack threshold: covers common small outputs (hex SHA-256/512, UUIDs).
  // Larger thresholds were benchmarked
  MaybeStackBuffer<char, 512> buf(length);
  encode(buf.out());
  // Copy stack-backed data, but release heap-backed storage to V8.
  if (buf.IsAllocated()) {
    char* data = buf.out();
    buf.Release();
    return ExternOneByteString::New(isolate, data, length);
  }
  return String::NewFromOneByte(isolate,
                                reinterpret_cast<const uint8_t*>(buf.out()),
                                v8::NewStringType::kNormal,
                                static_cast<int>(length));
}

template <typename EncodeFn>
static MaybeLocal<Value> EncodeTwoByteString(Isolate* isolate,
                                             size_t char_length,
                                             EncodeFn encode) {
  // 256 uint16_t = 512 bytes on the stack, matching the one-byte
  MaybeStackBuffer<uint16_t, 256> buf(char_length);
  encode(buf.out());
  // Copy stack-backed data, but release heap-backed storage to V8.
  if (buf.IsAllocated()) {
    uint16_t* data = buf.out();
    buf.Release();
    return ExternTwoByteString::New(isolate, data, char_length);
  }
  return String::NewFromTwoByte(isolate,
                                buf.out(),
                                v8::NewStringType::kNormal,
                                static_cast<int>(char_length));
}

template <>
MaybeLocal<Value> ExternOneByteString::NewExternal(
    Isolate* isolate, ExternOneByteString* h_str) {
  return String::NewExternalOneByte(isolate, h_str).FromMaybe(Local<Value>());
}


template <>
MaybeLocal<Value> ExternTwoByteString::NewExternal(
    Isolate* isolate, ExternTwoByteString* h_str) {
  return String::NewExternalTwoByte(isolate, h_str).FromMaybe(Local<Value>());
}

template <>
MaybeLocal<Value> ExternOneByteString::NewSimpleFromCopy(Isolate* isolate,
                                                         const char* data,
                                                         size_t length) {
  Local<String> str;
  if (!String::NewFromOneByte(isolate,
                              reinterpret_cast<const uint8_t*>(data),
                              v8::NewStringType::kNormal,
                              length)
           .ToLocal(&str)) {
    isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
    return MaybeLocal<Value>();
  }
  return str;
}

template <>
MaybeLocal<Value> ExternTwoByteString::NewSimpleFromCopy(Isolate* isolate,
                                                         const uint16_t* data,
                                                         size_t length) {
  Local<String> str;
  if (!String::NewFromTwoByte(isolate, data, v8::NewStringType::kNormal, length)
           .ToLocal(&str)) {
    isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
    return MaybeLocal<Value>();
  }
  return str;
}

}  // anonymous namespace

static size_t keep_buflen_in_range(size_t len) {
  if (len > static_cast<size_t>(std::numeric_limits<int>::max())) {
    return static_cast<size_t>(std::numeric_limits<int>::max());
  }
  return len;
}

size_t StringBytes::WriteUCS2(Isolate* isolate,
                              char* buf,
                              size_t buflen,
                              Local<String> str) {
  uint16_t* const dst = reinterpret_cast<uint16_t*>(buf);

  const size_t max_chars = buflen / sizeof(*dst);
  const size_t nchars = std::min(max_chars, static_cast<size_t>(str->Length()));
  if (nchars == 0) {
    return 0;
  }

  uint16_t* const aligned_dst = nbytes::AlignUp(dst, sizeof(*dst));
  CHECK_EQ(reinterpret_cast<uintptr_t>(aligned_dst) % sizeof(*dst), 0);
  if (aligned_dst == dst) {
    str->WriteV2(isolate, 0, nchars, dst);
  } else {
    // Write all but the last char.
    str->WriteV2(isolate, 0, nchars - 1, aligned_dst);

    // Shift everything to unaligned-left.
    memmove(dst, aligned_dst, (nchars - 1) * sizeof(*dst));

    // One more char to be written.
    uint16_t last;
    str->WriteV2(isolate, nchars - 1, 1, &last);
    memcpy(dst + nchars - 1, &last, sizeof(last));
  }

  return nchars * sizeof(*dst);
}

size_t StringBytes::Write(Isolate* isolate,
                          char* buf,
                          size_t buflen,
                          Local<Value> val,
                          enum encoding encoding) {
  HandleScope scope(isolate);
  size_t nbytes;
  buflen = keep_buflen_in_range(buflen);
  CHECK(val->IsString() == true);
  Local<String> str = val.As<String>();
  String::ValueView input_view(isolate, str);

  switch (encoding) {
    case ASCII:
    case LATIN1:
      if (input_view.is_one_byte()) {
        nbytes = std::min(buflen, static_cast<size_t>(input_view.length()));
        memcpy(buf, input_view.data8(), nbytes);
      } else {
        nbytes = std::min(buflen, static_cast<size_t>(input_view.length()));
        // Do not use v8::String::WriteOneByteV2 as it asserts the string to be
        // a one byte string. For compatibility, convert the uint16_t to uint8_t
        // even though this may loose accuracy.
        for (size_t i = 0; i < nbytes; i++) {
          buf[i] = static_cast<uint8_t>(input_view.data16()[i]);
        }
      }
      break;

    case BUFFER:
    case UTF8:
      if (input_view.is_one_byte()) {
        // Use simdutf for one-byte strings instead of V8's WriteUtf8V2.
        nbytes = simdutf::convert_latin1_to_utf8_safe(
            reinterpret_cast<const char*>(input_view.data8()),
            input_view.length(),
            buf,
            buflen);
      } else {
        nbytes = str->WriteUtf8V2(
            isolate, buf, buflen, String::WriteFlags::kReplaceInvalidUtf8);
      }
      break;

    case UCS2: {
      nbytes = WriteUCS2(isolate, buf, buflen, str);

      // Node's "ucs2" encoding wants LE character data stored in
      // the Buffer, so we need to reorder on BE platforms.  See
      // https://nodejs.org/api/buffer.html regarding Node's "ucs2"
      // encoding specification
      if constexpr (IsBigEndian()) CHECK(nbytes::SwapBytes16(buf, nbytes));

      break;
    }

    case BASE64URL:
      if (input_view.is_one_byte()) {  // 8-bit case
        size_t written_len = buflen;
        auto result = simdutf::base64_to_binary_safe(
            reinterpret_cast<const char*>(input_view.data8()),
            input_view.length(),
            buf,
            written_len,
            simdutf::base64_url);
        if (result.error == simdutf::error_code::SUCCESS) {
          nbytes = written_len;
        } else {
          // The input does not follow the WHATWG forgiving-base64 specification
          // (adapted for base64url with + and / replaced by - and _).
          // https://infra.spec.whatwg.org/#forgiving-base64-decode
          nbytes = nbytes::Base64Decode(
              buf,
              buflen,
              reinterpret_cast<const char*>(input_view.data8()),
              input_view.length());
        }
      } else {
        TwoByteValue value(isolate, str);
        size_t written_len = buflen;
        auto result = simdutf::base64_to_binary_safe(
            reinterpret_cast<const char16_t*>(value.out()),
            value.length(),
            buf,
            written_len,
            simdutf::base64_url);
        if (result.error == simdutf::error_code::SUCCESS) {
          nbytes = written_len;
        } else {
          // The input does not follow the WHATWG forgiving-base64 specification
          // (adapted for base64url with + and / replaced by - and _).
          // https://infra.spec.whatwg.org/#forgiving-base64-decode
          nbytes = nbytes::Base64Decode(buf, buflen, *value, value.length());
        }
      }
      break;

    case BASE64: {
      if (input_view.is_one_byte()) {  // 8-bit case
        size_t written_len = buflen;
        auto result = simdutf::base64_to_binary_safe(
            reinterpret_cast<const char*>(input_view.data8()),
            input_view.length(),
            buf,
            written_len);
        if (result.error == simdutf::error_code::SUCCESS) {
          nbytes = written_len;
        } else {
          // The input does not follow the WHATWG forgiving-base64 specification
          // https://infra.spec.whatwg.org/#forgiving-base64-decode
          nbytes = nbytes::Base64Decode(
              buf,
              buflen,
              reinterpret_cast<const char*>(input_view.data8()),
              input_view.length());
        }
      } else {
        TwoByteValue value(isolate, str);
        size_t written_len = buflen;
        auto result = simdutf::base64_to_binary_safe(
            reinterpret_cast<const char16_t*>(value.out()),
            value.length(),
            buf,
            written_len);
        if (result.error == simdutf::error_code::SUCCESS) {
          nbytes = written_len;
        } else {
          // The input does not follow the WHATWG base64 specification
          // https://infra.spec.whatwg.org/#forgiving-base64-decode
          nbytes = nbytes::Base64Decode(buf, buflen, *value, value.length());
        }
      }
      break;
    }
    case HEX:
      if (input_view.is_one_byte()) {
        nbytes =
            nbytes::HexDecode(buf,
                              buflen,
                              reinterpret_cast<const char*>(input_view.data8()),
                              input_view.length());
      } else {
        TwoByteValue value(isolate, str);
        nbytes = nbytes::HexDecode(buf, buflen, value.out(), value.length());
      }
      break;

    default:
      UNREACHABLE("unknown encoding");
  }

  return nbytes;
}

// Quick and dirty size calculation
// Will always be at least big enough, but may have some extra
// UTF8 can be as much as 3x the size, Base64 can have 1-2 extra bytes
Maybe<size_t> StringBytes::StorageSize(Isolate* isolate,
                                       Local<Value> val,
                                       enum encoding encoding) {
  HandleScope scope(isolate);

  if (Buffer::HasInstance(val) && (encoding == BUFFER || encoding == LATIN1)) {
    return Just(Buffer::Length(val));
  }

  Local<String> str;
  if (!val->ToString(isolate->GetCurrentContext()).ToLocal(&str))
    return Nothing<size_t>();
  String::ValueView view(isolate, str);
  size_t data_size = 0;

  switch (encoding) {
    case ASCII:
    case LATIN1:
      data_size = view.length();
      break;

    case BUFFER:
    case UTF8:
      // A single UCS2 codepoint never takes up more than 3 utf8 bytes.
      // It is an exercise for the caller to decide when a string is
      // long enough to justify calling Size() instead of StorageSize()
      data_size = 3 * view.length();
      break;

    case UCS2:
      data_size = view.length() * sizeof(uint16_t);
      break;

    case BASE64URL:
      data_size = simdutf::base64_length_from_binary(view.length(),
                                                     simdutf::base64_url);
      break;

    case BASE64:
      data_size = simdutf::base64_length_from_binary(view.length());
      break;

    case HEX:
      CHECK(view.length() % 2 == 0 && "invalid hex string length");
      data_size = view.length() / 2;
      break;

    default:
      UNREACHABLE("unknown encoding");
  }

  return Just(data_size);
}

Maybe<size_t> StringBytes::Size(Isolate* isolate,
                                Local<Value> val,
                                enum encoding encoding) {
  HandleScope scope(isolate);

  if (Buffer::HasInstance(val) && (encoding == BUFFER || encoding == LATIN1))
    return Just(Buffer::Length(val));

  Local<String> str;
  if (!val->ToString(isolate->GetCurrentContext()).ToLocal(&str))
    return Nothing<size_t>();
  String::ValueView view(isolate, str);

  switch (encoding) {
    case ASCII:
    case LATIN1:
      return Just<size_t>(view.length());

    case BUFFER:
    case UTF8:
      if (view.is_one_byte()) {
        return Just<size_t>(simdutf::utf8_length_from_latin1(
            reinterpret_cast<const char*>(view.data8()), view.length()));
      }
      return Just<size_t>(simdutf::utf8_length_from_utf16(
          reinterpret_cast<const char16_t*>(view.data16()), view.length()));

    case UCS2:
      return Just(view.length() * sizeof(uint16_t));

    case BASE64URL: {
      return Just(simdutf::base64_length_from_binary(view.length(),
                                                     simdutf::base64_url));
    }

    case BASE64: {
      return Just(simdutf::base64_length_from_binary(view.length()));
    }

    case HEX:
      return Just<size_t>(view.length() / 2);
  }

  UNREACHABLE();
}

#define CHECK_BUFLEN_IN_RANGE(len)                                             \
  do {                                                                         \
    if ((len) > Buffer::kMaxLength) {                                          \
      isolate->ThrowException(node::ERR_BUFFER_TOO_LARGE(isolate));            \
      return MaybeLocal<Value>();                                              \
    }                                                                          \
  } while (0)

MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
                                      const char* buf,
                                      size_t buflen,
                                      enum encoding encoding) {
  CHECK_BUFLEN_IN_RANGE(buflen);

  if (!buflen && encoding != BUFFER) {
    return String::Empty(isolate);
  }

  MaybeLocal<String> val;

  switch (encoding) {
    case BUFFER:
      {
        auto maybe_buf = Buffer::Copy(isolate, buf, buflen);
        Local<v8::Object> buf;
        if (!maybe_buf.ToLocal(&buf)) {
          isolate->ThrowException(node::ERR_MEMORY_ALLOCATION_FAILED(isolate));
        }
        return buf;
      }

    case ASCII:
      buflen = keep_buflen_in_range(buflen);
      if (simdutf::validate_ascii_with_errors(buf, buflen).error) {
        // The input contains non-ASCII bytes.

        return EncodeOneByteString(isolate, buflen, [buf, buflen](char* dst) {
          nbytes::ForceAscii(buf, dst, buflen);
        });
      } else {
        return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
      }

    case UTF8: {
      buflen = keep_buflen_in_range(buflen);

      // ASCII fast path
      // TODO(chalker): remove when String::NewFromUtf8 is fast enough itself
      // This is cheap compared to the benefits though
      if (!simdutf::validate_ascii_with_errors(buf, buflen).error) {
        return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
      }

      // Latin1-fits fast path: one-byte V8 string, half the heap of UTF-16.
      // Capped at 1 MiB (above that the prescan cost erases the win).
      constexpr size_t kLatin1Max = 1u << 20;
      if (buflen >= 256 && buflen <= kLatin1Max) {
        // Skip the allocation when any byte >= 0xC4 (UTF-8 lead for a
        // codepoint > U+FF). Inner loop has no early exit so clang
        // vectorizes it.
        constexpr size_t kChunk = 64;
        bool maybe_latin1 = true;
        size_t i = 0;
        for (; i + kChunk <= buflen; i += kChunk) {
          uint8_t acc = 0;
          for (size_t j = 0; j < kChunk; j++) {
            acc |= static_cast<uint8_t>(buf[i + j]) >= 0xC4 ? 1 : 0;
          }
          if (acc) {
            maybe_latin1 = false;
            break;
          }
        }
        if (maybe_latin1) {
          for (; i < buflen; i++) {
            if (static_cast<uint8_t>(buf[i]) >= 0xC4) {
              maybe_latin1 = false;
              break;
            }
          }
        }
        if (maybe_latin1) {
          MaybeStackBuffer<char, 4096> latin1;
          latin1.AllocateSufficientStorage(buflen);
          simdutf::result l1 = simdutf::convert_utf8_to_latin1_with_errors(
              buf, buflen, latin1.out());
          if (l1.error == simdutf::error_code::SUCCESS) {
            return ExternOneByteString::NewFromCopy(
                isolate, latin1.out(), l1.count);
          }
        }
      }

      if (buflen >= 32) {
        // Single-pass UTF-16: over-allocate (1 char16_t per byte), then
        // shrink. Above 1 MiB the exact-size 3-pass below is cheaper.
        constexpr size_t kSinglePassMax = 1u << 20;
        if (buflen <= kSinglePassMax) {
          MaybeStackBuffer<uint16_t, 256> u16;
          u16.AllocateSufficientStorage(buflen);
          simdutf::result r = simdutf::convert_utf8_to_utf16_with_errors(
              buf, buflen, reinterpret_cast<char16_t*>(u16.out()));
          if (r.error == simdutf::error_code::SUCCESS) {
            if (r.count > static_cast<size_t>(v8::String::kMaxLength)) {
              isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
              return MaybeLocal<Value>();
            }
            if (u16.IsAllocated()) {
              uint16_t* data = u16.out();
              u16.Release();
              uint16_t* shrunk = static_cast<uint16_t*>(
                  realloc(data, r.count * sizeof(uint16_t)));
              if (shrunk == nullptr) shrunk = data;
              return ExternTwoByteString::New(isolate, shrunk, r.count);
            }
            return String::NewFromTwoByte(isolate,
                                          u16.out(),
                                          v8::NewStringType::kNormal,
                                          static_cast<int>(r.count));
          }
        } else if (simdutf::validate_utf8(buf, buflen)) {
          size_t u16size = simdutf::utf16_length_from_utf8(buf, buflen);
          if (u16size > static_cast<size_t>(v8::String::kMaxLength)) {
            isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
            return MaybeLocal<Value>();
          }
          return EncodeTwoByteString(
              isolate, u16size, [buf, buflen, u16size](uint16_t* dst) {
                size_t written = simdutf::convert_valid_utf8_to_utf16(
                    buf, buflen, reinterpret_cast<char16_t*>(dst));
                CHECK_EQ(written, u16size);
              });
        }
      }

      val =
          String::NewFromUtf8(isolate, buf, v8::NewStringType::kNormal, buflen);
      Local<String> str;
      if (!val.ToLocal(&str)) {
        isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
      }
      return str;
    }

    case LATIN1:
      buflen = keep_buflen_in_range(buflen);
      return ExternOneByteString::NewFromCopy(isolate, buf, buflen);

    case BASE64: {
      buflen = keep_buflen_in_range(buflen);
      size_t dlen = simdutf::base64_length_from_binary(buflen);
      return EncodeOneByteString(isolate, dlen, [buf, buflen, dlen](char* dst) {
        size_t written = simdutf::binary_to_base64(buf, buflen, dst);
        CHECK_EQ(written, dlen);
      });
    }

    case BASE64URL: {
      buflen = keep_buflen_in_range(buflen);
      size_t dlen =
          simdutf::base64_length_from_binary(buflen, simdutf::base64_url);
      return EncodeOneByteString(isolate, dlen, [buf, buflen, dlen](char* dst) {
        size_t written =
            simdutf::binary_to_base64(buf, buflen, dst, simdutf::base64_url);
        CHECK_EQ(written, dlen);
      });
    }

    case HEX: {
      buflen = keep_buflen_in_range(buflen);
      size_t dlen = buflen * 2;
      return EncodeOneByteString(isolate, dlen, [buf, buflen, dlen](char* dst) {
        size_t written = nbytes::HexEncode(buf, buflen, dst, dlen);
        CHECK_EQ(written, dlen);
      });
    }

    case UCS2: {
      buflen = keep_buflen_in_range(buflen);
      size_t str_len = buflen / 2;
      if constexpr (IsBigEndian()) {
        return EncodeTwoByteString(
            isolate, str_len, [buf, str_len](uint16_t* dst) {
              for (size_t i = 0, k = 0; k < str_len; i += 2, k += 1) {
                // The input is in *little endian*, because that's what Node.js
                // expects, so the high byte comes after the low byte.
                const uint8_t hi = static_cast<uint8_t>(buf[i + 1]);
                const uint8_t lo = static_cast<uint8_t>(buf[i + 0]);
                dst[k] = static_cast<uint16_t>(hi) << 8 | lo;
              }
            });
      }
      if (reinterpret_cast<uintptr_t>(buf) % 2 != 0) {
        return EncodeTwoByteString(
            isolate, str_len, [buf, buflen](uint16_t* dst) {
              memcpy(dst, buf, buflen);
            });
      }
      return ExternTwoByteString::NewFromCopy(
          isolate, reinterpret_cast<const uint16_t*>(buf), str_len);
    }

    default:
      UNREACHABLE("unknown encoding");
  }
}

MaybeLocal<Value> StringBytes::EncodeValidUtf8(Isolate* isolate,
                                               const char* buf,
                                               size_t buflen) {
  CHECK_BUFLEN_IN_RANGE(buflen);
  if (!buflen) return String::Empty(isolate);
  buflen = keep_buflen_in_range(buflen);

  // ASCII fast path
  if (!simdutf::validate_ascii_with_errors(buf, buflen).error) {
    return ExternOneByteString::NewFromCopy(isolate, buf, buflen);
  }

  if (buflen >= 32) {
    size_t u16size = simdutf::utf16_length_from_utf8(buf, buflen);
    if (u16size > static_cast<size_t>(v8::String::kMaxLength)) {
      isolate->ThrowException(ERR_STRING_TOO_LONG(isolate));
      return MaybeLocal<Value>();
    }
    return EncodeTwoByteString(
        isolate, u16size, [buf, buflen, u16size](uint16_t* dst) {
          size_t written = simdutf::convert_valid_utf8_to_utf16(
              buf, buflen, reinterpret_cast<char16_t*>(dst));
          CHECK_EQ(written, u16size);
        });
  }

  Local<String> str;
  if (!String::NewFromUtf8(isolate, buf, v8::NewStringType::kNormal, buflen)
           .ToLocal(&str)) {
    isolate->ThrowException(node::ERR_STRING_TOO_LONG(isolate));
  }
  return str;
}

MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
                                      const uint16_t* buf,
                                      size_t buflen) {
  if (buflen == 0) return String::Empty(isolate);
  CHECK_BUFLEN_IN_RANGE(buflen);

  // Node's "ucs2" encoding expects LE character data inside a
  // Buffer, so we need to reorder on BE platforms.  See
  // https://nodejs.org/api/buffer.html regarding Node's "ucs2"
  // encoding specification
  if constexpr (IsBigEndian()) {
    return EncodeTwoByteString(isolate, buflen, [buf, buflen](uint16_t* dst) {
      size_t nbytes = buflen * sizeof(uint16_t);
      memcpy(dst, buf, nbytes);
      CHECK(nbytes::SwapBytes16(reinterpret_cast<char*>(dst), nbytes));
    });
  } else {
    return ExternTwoByteString::NewFromCopy(isolate, buf, buflen);
  }
}

MaybeLocal<Value> StringBytes::Encode(Isolate* isolate,
                                      const char* buf,
                                      enum encoding encoding) {
  const size_t len = strlen(buf);
  return Encode(isolate, buf, len, encoding);
}

}  // namespace node