Line data Source code
1 : // Copyright 2011 the V8 project authors. All rights reserved.
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 V8_UNICODE_H_
6 : #define V8_UNICODE_H_
7 :
8 : #include <sys/types.h>
9 : #include "src/globals.h"
10 : #include "src/third_party/utf8-decoder/utf8-decoder.h"
11 : #include "src/utils.h"
12 : /**
13 : * \file
14 : * Definitions and convenience functions for working with unicode.
15 : */
16 :
17 : namespace unibrow {
18 :
19 : typedef unsigned int uchar;
20 : typedef unsigned char byte;
21 :
22 : /**
23 : * The max length of the result of converting the case of a single
24 : * character.
25 : */
26 : const int kMaxMappingSize = 4;
27 :
28 : template <class T, int size = 256>
29 : class Predicate {
30 : public:
31 : inline Predicate() = default;
32 : inline bool get(uchar c);
33 :
34 : private:
35 : friend class Test;
36 : bool CalculateValue(uchar c);
37 : class CacheEntry {
38 : public:
39 : inline CacheEntry()
40 : : bit_field_(CodePointField::encode(0) | ValueField::encode(0)) {}
41 : inline CacheEntry(uchar code_point, bool value)
42 : : bit_field_(
43 : CodePointField::encode(CodePointField::kMask & code_point) |
44 : ValueField::encode(value)) {
45 : DCHECK_IMPLIES((CodePointField::kMask & code_point) != code_point,
46 : code_point == static_cast<uchar>(-1));
47 : }
48 :
49 : uchar code_point() const { return CodePointField::decode(bit_field_); }
50 : bool value() const { return ValueField::decode(bit_field_); }
51 :
52 : private:
53 : class CodePointField : public v8::internal::BitField<uchar, 0, 21> {};
54 : class ValueField : public v8::internal::BitField<bool, 21, 1> {};
55 :
56 : uint32_t bit_field_;
57 : };
58 : static const int kSize = size;
59 : static const int kMask = kSize - 1;
60 : CacheEntry entries_[kSize];
61 : };
62 :
63 :
64 : // A cache used in case conversion. It caches the value for characters
65 : // that either have no mapping or map to a single character independent
66 : // of context. Characters that map to more than one character or that
67 : // map differently depending on context are always looked up.
68 : template <class T, int size = 256>
69 : class Mapping {
70 : public:
71 47264621 : inline Mapping() = default;
72 : inline int get(uchar c, uchar n, uchar* result);
73 : private:
74 : friend class Test;
75 : int CalculateValue(uchar c, uchar n, uchar* result);
76 : struct CacheEntry {
77 47199232 : inline CacheEntry() : code_point_(kNoChar), offset_(0) { }
78 : inline CacheEntry(uchar code_point, signed offset)
79 : : code_point_(code_point),
80 : offset_(offset) { }
81 : uchar code_point_;
82 : signed offset_;
83 : static const int kNoChar = (1 << 21) - 1;
84 : };
85 : static const int kSize = size;
86 : static const int kMask = kSize - 1;
87 : CacheEntry entries_[kSize];
88 : };
89 :
90 :
91 : class UnicodeData {
92 : private:
93 : friend class Test;
94 : static int GetByteCount();
95 : static const uchar kMaxCodePoint;
96 : };
97 :
98 :
99 : class Utf16 {
100 : public:
101 : static const int kNoPreviousCharacter = -1;
102 : static inline bool IsSurrogatePair(int lead, int trail) {
103 105541475 : return IsLeadSurrogate(lead) && IsTrailSurrogate(trail);
104 : }
105 : static inline bool IsLeadSurrogate(int code) {
106 96229454 : return (code & 0xfc00) == 0xd800;
107 : }
108 : static inline bool IsTrailSurrogate(int code) {
109 15064760 : return (code & 0xfc00) == 0xdc00;
110 : }
111 :
112 : static inline int CombineSurrogatePair(uchar lead, uchar trail) {
113 15075059 : return 0x10000 + ((lead & 0x3ff) << 10) + (trail & 0x3ff);
114 : }
115 : static const uchar kMaxNonSurrogateCharCode = 0xffff;
116 : // Encoding a single UTF-16 code unit will produce 1, 2 or 3 bytes
117 : // of UTF-8 data. The special case where the unit is a surrogate
118 : // trail produces 1 byte net, because the encoding of the pair is
119 : // 4 bytes and the 3 bytes that were used to encode the lead surrogate
120 : // can be reclaimed.
121 : static const int kMaxExtraUtf8BytesForOneUtf16CodeUnit = 3;
122 : // One UTF-16 surrogate is endoded (illegally) as 3 UTF-8 bytes.
123 : // The illegality stems from the surrogate not being part of a pair.
124 : static const int kUtf8BytesToCodeASurrogate = 3;
125 : static inline uint16_t LeadSurrogate(uint32_t char_code) {
126 3240356 : return 0xd800 + (((char_code - 0x10000) >> 10) & 0x3ff);
127 : }
128 : static inline uint16_t TrailSurrogate(uint32_t char_code) {
129 3240356 : return 0xdc00 + (char_code & 0x3ff);
130 : }
131 : };
132 :
133 : class V8_EXPORT_PRIVATE Utf8 {
134 : public:
135 : using State = Utf8DfaDecoder::State;
136 :
137 : static inline uchar Length(uchar chr, int previous);
138 : static inline unsigned EncodeOneByte(char* out, uint8_t c);
139 : static inline unsigned Encode(char* out,
140 : uchar c,
141 : int previous,
142 : bool replace_invalid = false);
143 : static uchar CalculateValue(const byte* str, size_t length, size_t* cursor);
144 :
145 : // The unicode replacement character, used to signal invalid unicode
146 : // sequences (e.g. an orphan surrogate) when converting to a UTF-8 encoding.
147 : static const uchar kBadChar = 0xFFFD;
148 : static const uchar kBufferEmpty = 0x0;
149 : static const uchar kIncomplete = 0xFFFFFFFC; // any non-valid code point.
150 : static const unsigned kMaxEncodedSize = 4;
151 : static const unsigned kMaxOneByteChar = 0x7f;
152 : static const unsigned kMaxTwoByteChar = 0x7ff;
153 : static const unsigned kMaxThreeByteChar = 0xffff;
154 : static const unsigned kMaxFourByteChar = 0x1fffff;
155 :
156 : // A single surrogate is coded as a 3 byte UTF-8 sequence, but two together
157 : // that match are coded as a 4 byte UTF-8 sequence.
158 : static const unsigned kBytesSavedByCombiningSurrogates = 2;
159 : static const unsigned kSizeOfUnmatchedSurrogate = 3;
160 : // The maximum size a single UTF-16 code unit may take up when encoded as
161 : // UTF-8.
162 : static const unsigned kMax16BitCodeUnitSize = 3;
163 : static inline uchar ValueOf(const byte* str, size_t length, size_t* cursor);
164 :
165 : typedef uint32_t Utf8IncrementalBuffer;
166 : static inline uchar ValueOfIncremental(const byte** cursor, State* state,
167 : Utf8IncrementalBuffer* buffer);
168 : static uchar ValueOfIncrementalFinish(State* state);
169 :
170 : // Excludes non-characters from the set of valid code points.
171 : static inline bool IsValidCharacter(uchar c);
172 :
173 : // Validate if the input has a valid utf-8 encoding. Unlike JS source code
174 : // this validation function will accept any unicode code point, including
175 : // kBadChar and BOMs.
176 : //
177 : // This method checks for:
178 : // - valid utf-8 endcoding (e.g. no over-long encodings),
179 : // - absence of surrogates,
180 : // - valid code point range.
181 : static bool ValidateEncoding(const byte* str, size_t length);
182 : };
183 :
184 : struct Uppercase {
185 : static bool Is(uchar c);
186 : };
187 : struct Letter {
188 : static bool Is(uchar c);
189 : };
190 : #ifndef V8_INTL_SUPPORT
191 : struct V8_EXPORT_PRIVATE ID_Start {
192 : static bool Is(uchar c);
193 : };
194 : struct V8_EXPORT_PRIVATE ID_Continue {
195 : static bool Is(uchar c);
196 : };
197 : struct V8_EXPORT_PRIVATE WhiteSpace {
198 : static bool Is(uchar c);
199 : };
200 : #endif // !V8_INTL_SUPPORT
201 :
202 : // LineTerminator: 'JS_Line_Terminator' in point.properties
203 : // ES#sec-line-terminators lists exactly 4 code points:
204 : // LF (U+000A), CR (U+000D), LS(U+2028), PS(U+2029)
205 : V8_INLINE bool IsLineTerminator(uchar c) {
206 744523429 : return c == 0x000A || c == 0x000D || c == 0x2028 || c == 0x2029;
207 : }
208 :
209 : V8_INLINE bool IsStringLiteralLineTerminator(uchar c) {
210 9168290 : return c == 0x000A || c == 0x000D;
211 : }
212 :
213 : #ifndef V8_INTL_SUPPORT
214 : struct ToLowercase {
215 : static const int kMaxWidth = 3;
216 : static const bool kIsToLower = true;
217 : static int Convert(uchar c,
218 : uchar n,
219 : uchar* result,
220 : bool* allow_caching_ptr);
221 : };
222 : struct ToUppercase {
223 : static const int kMaxWidth = 3;
224 : static const bool kIsToLower = false;
225 : static int Convert(uchar c,
226 : uchar n,
227 : uchar* result,
228 : bool* allow_caching_ptr);
229 : };
230 : #endif
231 : struct Ecma262Canonicalize {
232 : static const int kMaxWidth = 1;
233 : static int Convert(uchar c,
234 : uchar n,
235 : uchar* result,
236 : bool* allow_caching_ptr);
237 : };
238 : struct Ecma262UnCanonicalize {
239 : static const int kMaxWidth = 4;
240 : static int Convert(uchar c,
241 : uchar n,
242 : uchar* result,
243 : bool* allow_caching_ptr);
244 : };
245 : struct CanonicalizationRange {
246 : static const int kMaxWidth = 1;
247 : static int Convert(uchar c,
248 : uchar n,
249 : uchar* result,
250 : bool* allow_caching_ptr);
251 : };
252 :
253 : } // namespace unibrow
254 :
255 : #endif // V8_UNICODE_H_
|