1
#include "data-url.h"
2
 
3
#include <workerd/api/encoding.h>
4
#include <workerd/util/strings.h>
5
 
6
#include <kj/encoding.h>
7
 
8
namespace workerd::api {
9
 
10
kj::Maybe<DataUrl> DataUrl::tryParse(kj::StringPtr url) {
11
  KJ_IF_SOME(url, jsg::Url::tryParse(url)) {
12
    return from(url);
13
  }
14
  return kj::none;
15
}
16
 
17
kj::Maybe<DataUrl> DataUrl::from(const jsg::Url& url) {
18
  if (url.getProtocol() != "data:"_kj) return kj::none;
19
  auto clone = url.clone(jsg::Url::EquivalenceOption::IGNORE_FRAGMENTS);
20
 
21
  // Remove the "data:" prefix.
22
  auto href = clone.getHref().slice(5);
23
 
24
  // We scan for the first comma, which separates the MIME type from the data.
25
  // Per the fetch spec, it doesn't matter if the comma is within a quoted
26
  // string value in the MIME type... which is fun.
27
 
28
  static const auto isBase64 = [](kj::ArrayPtr<const char> label) -> bool {
29
    KJ_IF_SOME(pos, label.findLast(';')) {
30
      auto res = trimLeadingAndTrailingWhitespace(label.slice(pos + 1));
31
      return res.size() == 6 && (res[0] | 0x20) == 'b' && (res[1] | 0x20) == 'a' &&
32
          (res[2] | 0x20) == 's' && (res[3] | 0x20) == 'e' && (res[4] == '6') && (res[5] == '4');
33
    }
34
    return false;
35
  };
36
 
37
  static const auto create = [](auto input, auto decoded) {
38
    KJ_IF_SOME(parsed, MimeType::tryParse(input)) {
39
      return DataUrl(kj::mv(parsed), kj::mv(decoded));
40
    } else {
41
      return DataUrl(MimeType::PLAINTEXT_ASCII.clone(), kj::mv(decoded));
42
    }
43
  };
44
 
45
  KJ_IF_SOME(pos, href.findFirst(',')) {
46
    auto unparsed = href.first(pos);
47
    auto data = href.slice(pos + 1);
48
 
49
    // We need to trim leading and trailing whitespace from the mimetype
50
    unparsed = trimLeadingAndTrailingWhitespace(unparsed);
51
 
52
    // Determine if the data is base64 encoded
53
    kj::Array<kj::byte> decoded = nullptr;
54
    if (isBase64(unparsed)) {
55
      unparsed = unparsed.first(KJ_ASSERT_NONNULL(unparsed.findLast(';')));
56
      decoded =
57
          kj::decodeBase64(stripInnerWhitespace(jsg::Url::percentDecode(data.asBytes())).asChars());
58
    } else {
59
      decoded = jsg::Url::percentDecode(data.asBytes());
60
    }
61
 
62
    if (unparsed.startsWith(";"_kj)) {
63
      // If the mime type starts with ;, then the spec tells us to
64
      // prepend "text/plain" to the mime type.
65
      auto fixed = kj::str("text/plain", unparsed);
66
      return create(fixed.asPtr(), kj::mv(decoded));
67
    }
68
 
69
    return create(unparsed, kj::mv(decoded));
70
  }
71
 
72
  // If there are no commas, the data: url is invalid.
73
  return kj::none;
74
}
75
 
76
}  // namespace workerd::api