1 // Copyright 2023 The PDFium Authors
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 #ifndef CORE_FXCRT_UTF16_H_
6 #define CORE_FXCRT_UTF16_H_
7
8 #include "third_party/base/check.h"
9
10 namespace pdfium {
11
12 // The number of suffix bits in a UTF-16 surrogate.
13 inline constexpr int kSurrogateBits = 10;
14
15 // A bitmask for the suffix of a UTF-16 surrogate.
16 inline constexpr char16_t kSurrogateMask = (1 << kSurrogateBits) - 1;
17
18 // The first supplementary code point, `U+10000`.
19 inline constexpr char32_t kMinimumSupplementaryCodePoint = 0x10000;
20
21 // The last supplementary code point, `U+10FFFF`.
22 inline constexpr char32_t kMaximumSupplementaryCodePoint =
23 kMinimumSupplementaryCodePoint +
24 (kSurrogateMask << kSurrogateBits | kSurrogateMask);
25
26 // The first UTF-16 high surrogate code unit, `U+D800`.
27 inline constexpr char16_t kMinimumHighSurrogateCodeUnit = 0xd800;
28
29 // The last UTF-16 high surrogate code unit, `U+DBFF`.
30 inline constexpr char16_t kMaximumHighSurrogateCodeUnit =
31 kMinimumHighSurrogateCodeUnit | kSurrogateMask;
32
33 // The first UTF-16 low surrogate code unit, `U+DC00`.
34 inline constexpr char16_t kMinimumLowSurrogateCodeUnit =
35 kMaximumHighSurrogateCodeUnit + 1;
36
37 // The last UTF-16 low surrogate code unit, `U+DFFF`.
38 inline constexpr char16_t kMaximumLowSurrogateCodeUnit =
39 kMinimumLowSurrogateCodeUnit | kSurrogateMask;
40
41 // Returns `true` if `code_point` is in a supplementary plane, and therefore
42 // requires encoding as a UTF-16 surrogate pair.
IsSupplementary(char32_t code_point)43 constexpr bool IsSupplementary(char32_t code_point) {
44 return code_point >= kMinimumSupplementaryCodePoint &&
45 code_point <= kMaximumSupplementaryCodePoint;
46 }
47
48 // Returns `true` if `code_point` is a UTF-16 high surrogate.
IsHighSurrogate(char32_t code_point)49 constexpr bool IsHighSurrogate(char32_t code_point) {
50 return code_point >= kMinimumHighSurrogateCodeUnit &&
51 code_point <= kMaximumHighSurrogateCodeUnit;
52 }
53
54 // Returns `true` if `code_point` is a UTF-16 low surrogate.
IsLowSurrogate(char32_t code_point)55 constexpr bool IsLowSurrogate(char32_t code_point) {
56 return code_point >= kMinimumLowSurrogateCodeUnit &&
57 code_point <= kMaximumLowSurrogateCodeUnit;
58 }
59
60 // A UTF-16 surrogate pair.
61 class SurrogatePair final {
62 public:
63 // Constructs a surrogate pair from a high and a low surrogate.
SurrogatePair(char16_t high,char16_t low)64 constexpr SurrogatePair(char16_t high, char16_t low)
65 : high_(high), low_(low) {
66 DCHECK(IsHighSurrogate(high_));
67 DCHECK(IsLowSurrogate(low_));
68 }
69
70 // Constructs a surrogate pair from a code point.
SurrogatePair(char32_t code_point)71 explicit constexpr SurrogatePair(char32_t code_point)
72 : high_(GetHighSurrogate(code_point)), low_(GetLowSurrogate(code_point)) {
73 // This constructor initializes `high_` and `low_` using helper functions
74 // because C++17 requires it for `constexpr` constructors.
75 DCHECK(IsSupplementary(code_point));
76 }
77
high()78 constexpr char16_t high() const { return high_; }
low()79 constexpr char16_t low() const { return low_; }
80
81 // Decodes this surrogate pair to a code point.
ToCodePoint()82 constexpr char32_t ToCodePoint() const {
83 char32_t code_point = low_ & kSurrogateMask;
84 code_point |= (high_ & kSurrogateMask) << kSurrogateBits;
85 return kMinimumSupplementaryCodePoint + code_point;
86 }
87
88 private:
GetHighSurrogate(char32_t code_point)89 static constexpr char16_t GetHighSurrogate(char32_t code_point) {
90 code_point -= kMinimumSupplementaryCodePoint;
91 char16_t code_unit = (code_point >> kSurrogateBits) & kSurrogateMask;
92 return kMinimumHighSurrogateCodeUnit | code_unit;
93 }
94
GetLowSurrogate(char32_t code_point)95 static constexpr char16_t GetLowSurrogate(char32_t code_point) {
96 code_point -= kMinimumSupplementaryCodePoint;
97 char16_t code_unit = code_point & kSurrogateMask;
98 return kMinimumLowSurrogateCodeUnit | code_unit;
99 }
100
101 char16_t high_;
102 char16_t low_;
103 };
104
105 } // namespace pdfium
106
107 #endif // CORE_FXCRT_UTF16_H_
108