1
// Copyright (c) 2023 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
#pragma once
5
 
6
#include <kj/string.h>
7
 
8
#include <cstdint>
9
 
10
namespace workerd {
11
 
12
enum CharAttributeFlag : uint8_t {
13
  NONE = 0,
14
  ALPHA = 1 << 0,
15
  DIGIT = 1 << 1,
16
  HEX = 1 << 2,
17
  ASCII = 1 << 3,
18
  ASCII_WHITESPACE = 1 << 4,
19
  UPPER_CASE = 1 << 5,
20
  LOWER_CASE = 1 << 6,
21
  SEPARATOR = 1 << 7,
22
};
23
 
24
// Construct a lookup table for various interesting character properties.
25
constexpr kj::FixedArray<uint8_t, 256> kCharLookupTable = []() consteval {
26
  kj::FixedArray<uint8_t, 256> result{};
27
  for (uint8_t c = 'A'; c <= 'Z'; c++) {
28
    if (c <= 'F') {
29
      result[c] |= CharAttributeFlag::HEX;
30
      result[c + 0x20] |= CharAttributeFlag::HEX;
31
    }
32
    result[c] |= CharAttributeFlag::ALPHA | CharAttributeFlag::UPPER_CASE;
33
    result[c + 0x20] |= CharAttributeFlag::ALPHA | CharAttributeFlag::LOWER_CASE;
34
  }
35
  for (uint8_t c = '0'; c <= '9'; c++) {
36
    result[c] |= CharAttributeFlag::DIGIT | CharAttributeFlag::HEX;
37
  }
38
  for (uint8_t c = 0; c <= 0x7f; c++) {
39
    result[c] |= CharAttributeFlag::ASCII;
40
  }
41
  for (uint8_t c: {0x09, 0x0a, 0x0c, 0x0d, 0x20}) {
42
    result[c] |= CharAttributeFlag::ASCII_WHITESPACE;
43
  }
44
  result['+'] |= CharAttributeFlag::SEPARATOR;
45
  result['-'] |= CharAttributeFlag::SEPARATOR;
46
  result['_'] |= CharAttributeFlag::SEPARATOR;
47
  return result;
48
}();
49
 
50
constexpr bool isAlpha(const kj::byte c) noexcept {
51
  return kCharLookupTable[c] & CharAttributeFlag::ALPHA;
52
}
53
 
54
constexpr bool isDigit(const kj::byte c) noexcept {
55
  return kCharLookupTable[c] & CharAttributeFlag::DIGIT;
56
}
57
 
58
// Check if `c` is the ASCII code of a hexadecimal digit.
59
constexpr bool isHexDigit(const kj::byte c) noexcept {
60
  return kCharLookupTable[c] & CharAttributeFlag::HEX;
61
}
62
 
63
constexpr bool isAscii(const kj::byte c) noexcept {
64
  return kCharLookupTable[c] & CharAttributeFlag::ASCII;
65
}
66
 
67
constexpr bool isAsciiWhitespace(const kj::byte c) noexcept {
68
  return kCharLookupTable[c] & CharAttributeFlag::ASCII_WHITESPACE;
69
}
70
 
71
constexpr bool isAlphaUpper(const kj::byte c) noexcept {
72
  return kCharLookupTable[c] & CharAttributeFlag::UPPER_CASE;
73
}
74
 
75
constexpr bool isAlphaLower(const kj::byte c) noexcept {
76
  return kCharLookupTable[c] & CharAttributeFlag::LOWER_CASE;
77
}
78
 
79
// TODO(later): If kj::ArrayPtr ever supports a constexpr [] operator,
80
// make this function constexpr.
81
bool strcaseeq(kj::ArrayPtr<const char> a, kj::ArrayPtr<const char> b);
82
 
83
// Convert ASCII alpha characters in the given string to lowercase in place.
84
kj::String toLower(kj::String&& str);
85
 
86
// Convert ASCII alpha characters in the given string to uppercase in place.
87
kj::String toUpper(kj::String&& str);
88
 
89
// Copy the input and convert ASCII alpha characters in the given string to lowercase.
90
kj::String toLower(kj::ArrayPtr<const char> ptr);
91
kj::String toUpper(kj::ArrayPtr<const char> ptr);
92
 
93
kj::ArrayPtr<const char> trimLeadingAndTrailingWhitespace(kj::ArrayPtr<const char> ptr);
94
kj::ArrayPtr<const char> trimTailingWhitespace(kj::ArrayPtr<const char> ptr);
95
kj::Array<kj::byte> stripInnerWhitespace(kj::ArrayPtr<kj::byte> input);
96
 
97
}  // namespace workerd