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
#include "encoding-legacy.h"
6
 
7
#include <kj-rs/convert.h>
8
#include <rust/cxx.h>
9
 
10
#include <kj/common.h>
11
 
12
namespace workerd::api {
13
 
14
namespace {
15
 
16
// Map workerd::api::Encoding to the Rust-side RustEncoding enum.
17
::workerd::rust::encoding::Encoding toRustEncoding(Encoding encoding) {
18
  using RE = ::workerd::rust::encoding::Encoding;
19
  switch (encoding) {
20
    case Encoding::Big5:
21
      return RE::Big5;
22
    case Encoding::Euc_Jp:
23
      return RE::EucJp;
24
    case Encoding::Euc_Kr:
25
      return RE::EucKr;
26
    case Encoding::Gb18030:
27
      return RE::Gb18030;
28
    case Encoding::Gbk:
29
      return RE::Gbk;
30
    case Encoding::Iso2022_Jp:
31
      return RE::Iso2022Jp;
32
    case Encoding::Shift_Jis:
33
      return RE::ShiftJis;
34
    case Encoding::Windows_1252:
35
      return RE::Windows1252;
36
    case Encoding::X_User_Defined:
37
      return RE::XUserDefined;
38
    default:
39
      KJ_UNREACHABLE;
40
  }
41
}
42
 
43
}  // namespace
44
 
45
LegacyDecoder::LegacyDecoder(Encoding encoding, DecoderFatal fatal)
46
    : encoding(encoding),
47
      fatal(fatal),
48
      state(::workerd::rust::encoding::new_decoder(toRustEncoding(encoding))) {}
49
 
50
void LegacyDecoder::reset() {
51
  ::workerd::rust::encoding::reset(*state);
52
}
53
 
54
kj::Maybe<jsg::JsString> LegacyDecoder::decode(
55
    jsg::Lock& js, kj::ArrayPtr<const kj::byte> buffer, bool flush) {
56
  if (!flush && buffer.size() == 0) {
57
    // Avoid encoding_rs empty-chunk streaming bug:
58
    // https://github.com/hsivonen/encoding_rs/issues/126#issuecomment-3677642122
59
    return js.str();
60
  }
61
 
62
  ::workerd::rust::encoding::DecodeOptions options{.flush = flush, .fatal = fatal.toBool()};
63
  // Decode into the Rust-side reusable buffer. The Rust decoder handles
64
  // lazy reset internally when a previous call used flush=true.
65
  auto result =
66
      ::workerd::rust::encoding::decode(*state, buffer.as<kj_rs::RustMutable>(), kj::mv(options));
67
 
68
  if (fatal.toBool() && result.had_error) {
69
    return kj::none;
70
  }
71
 
72
  // Zero-copy view of the UTF-16 output slice from the Rust-owned buffer.
73
  return js.str(kj::from<kj_rs::Rust>(result.output));
74
}
75
 
76
}  // namespace workerd::api