1
// Copyright (c) 2017-2022 Cloudflare, Inc.
2
// Licensed under the Apache 2.0 license found in the LICENSE file or at:
3
//     https://opensource.org/licenses/Apache-2.0
4
 
5
// WHATWG-compliant legacy decoders (CJK multi-byte, windows-1252,
6
// x-user-defined) implemented via the encoding_rs Rust crate through
7
// a CXX bridge. A single LegacyDecoder class wraps an opaque Rust-side
8
// decoder that handles all the encoding-specific state machines.
9
 
10
#pragma once
11
 
12
#include "encoding-shared.h"
13
 
14
#include <workerd/rust/encoding/lib.rs.h>
15
 
16
#include <rust/cxx.h>
17
 
18
#include <kj/common.h>
19
 
20
namespace workerd::api {
21
 
22
// Unified legacy decoder using encoding_rs via Rust CXX bridge.
23
// encoding_rs implements the full WHATWG decoder algorithms for all
24
// legacy encodings, including streaming, error recovery, and ASCII
25
// byte pushback.
26
//
27
// According to WHATWG spec, any encoding except UTF-8 and UTF-16 is considered legacy.
28
class LegacyDecoder final: public Decoder {
29
 public:
30
  LegacyDecoder(Encoding encoding, DecoderFatal fatal);
31
  ~LegacyDecoder() noexcept = default;
32
  LegacyDecoder(LegacyDecoder&&) noexcept = default;
33
  LegacyDecoder& operator=(LegacyDecoder&&) noexcept = default;
34
  KJ_DISALLOW_COPY(LegacyDecoder);
35
 
36
  Encoding getEncoding() override {
37
    return encoding;
38
  }
39
 
40
  kj::Maybe<jsg::JsString> decode(
41
      jsg::Lock& js, kj::ArrayPtr<const kj::byte> buffer, bool flush = false) override;
42
 
43
  void reset() override;
44
 
45
 private:
46
  Encoding encoding;
47
  DecoderFatal fatal;
48
  ::rust::Box<::workerd::rust::encoding::Decoder> state;
49
};
50
 
51
}  // namespace workerd::api