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/utils.h"
11 : /**
12 : * \file
13 : * Definitions and convenience functions for working with unicode.
14 : */
15 :
16 : namespace unibrow {
17 :
18 : typedef unsigned int uchar;
19 : typedef unsigned char byte;
20 :
21 : /**
22 : * The max length of the result of converting the case of a single
23 : * character.
24 : */
25 : const int kMaxMappingSize = 4;
26 :
27 : template <class T, int size = 256>
28 : class Predicate {
29 : public:
30 93733430 : inline Predicate() { }
31 : inline bool get(uchar c);
32 :
33 : private:
34 : friend class Test;
35 : bool CalculateValue(uchar c);
36 : class CacheEntry {
37 : public:
38 : inline CacheEntry()
39 93587200 : : bit_field_(CodePointField::encode(0) | ValueField::encode(0)) {}
40 : inline CacheEntry(uchar code_point, bool value)
41 : : bit_field_(CodePointField::encode(code_point) |
42 5388768 : ValueField::encode(value)) {}
43 :
44 : uchar code_point() const { return CodePointField::decode(bit_field_); }
45 : bool value() const { return ValueField::decode(bit_field_); }
46 :
47 : private:
48 : class CodePointField : public v8::internal::BitField<uchar, 0, 21> {};
49 : class ValueField : public v8::internal::BitField<bool, 21, 1> {};
50 :
51 : uint32_t bit_field_;
52 : };
53 : static const int kSize = size;
54 : static const int kMask = kSize - 1;
55 : CacheEntry entries_[kSize];
56 : };
57 :
58 :
59 : // A cache used in case conversion. It caches the value for characters
60 : // that either have no mapping or map to a single character independent
61 : // of context. Characters that map to more than one character or that
62 : // map differently depending on context are always looked up.
63 : template <class T, int size = 256>
64 : class Mapping {
65 : public:
66 62292188 : inline Mapping() { }
67 : inline int get(uchar c, uchar n, uchar* result);
68 : private:
69 : friend class Test;
70 : int CalculateValue(uchar c, uchar n, uchar* result);
71 : struct CacheEntry {
72 62170624 : inline CacheEntry() : code_point_(kNoChar), offset_(0) { }
73 : inline CacheEntry(uchar code_point, signed offset)
74 : : code_point_(code_point),
75 : offset_(offset) { }
76 : uchar code_point_;
77 : signed offset_;
78 : static const int kNoChar = (1 << 21) - 1;
79 : };
80 : static const int kSize = size;
81 : static const int kMask = kSize - 1;
82 : CacheEntry entries_[kSize];
83 : };
84 :
85 :
86 : class UnicodeData {
87 : private:
88 : friend class Test;
89 : static int GetByteCount();
90 : static const uchar kMaxCodePoint;
91 : };
92 :
93 :
94 : class Utf16 {
95 : public:
96 : static inline bool IsSurrogatePair(int lead, int trail) {
97 199187 : return IsLeadSurrogate(lead) && IsTrailSurrogate(trail);
98 : }
99 : static inline bool IsLeadSurrogate(int code) {
100 1218881334 : if (code == kNoPreviousCharacter) return false;
101 1215733974 : return (code & 0xfc00) == 0xd800;
102 : }
103 : static inline bool IsTrailSurrogate(int code) {
104 34735258 : if (code == kNoPreviousCharacter) return false;
105 34736049 : return (code & 0xfc00) == 0xdc00;
106 : }
107 :
108 : static inline int CombineSurrogatePair(uchar lead, uchar trail) {
109 3234 : return 0x10000 + ((lead & 0x3ff) << 10) + (trail & 0x3ff);
110 : }
111 : static const int kNoPreviousCharacter = -1;
112 : static const uchar kMaxNonSurrogateCharCode = 0xffff;
113 : // Encoding a single UTF-16 code unit will produce 1, 2 or 3 bytes
114 : // of UTF-8 data. The special case where the unit is a surrogate
115 : // trail produces 1 byte net, because the encoding of the pair is
116 : // 4 bytes and the 3 bytes that were used to encode the lead surrogate
117 : // can be reclaimed.
118 : static const int kMaxExtraUtf8BytesForOneUtf16CodeUnit = 3;
119 : // One UTF-16 surrogate is endoded (illegally) as 3 UTF-8 bytes.
120 : // The illegality stems from the surrogate not being part of a pair.
121 : static const int kUtf8BytesToCodeASurrogate = 3;
122 : static inline uint16_t LeadSurrogate(uint32_t char_code) {
123 15543986 : return 0xd800 + (((char_code - 0x10000) >> 10) & 0x3ff);
124 : }
125 : static inline uint16_t TrailSurrogate(uint32_t char_code) {
126 15543993 : return 0xdc00 + (char_code & 0x3ff);
127 : }
128 : };
129 :
130 :
131 : class Utf8 {
132 : public:
133 : static inline uchar Length(uchar chr, int previous);
134 : static inline unsigned EncodeOneByte(char* out, uint8_t c);
135 : static inline unsigned Encode(char* out,
136 : uchar c,
137 : int previous,
138 : bool replace_invalid = false);
139 : static uchar CalculateValue(const byte* str, size_t length, size_t* cursor);
140 :
141 : // The unicode replacement character, used to signal invalid unicode
142 : // sequences (e.g. an orphan surrogate) when converting to a UTF-8 encoding.
143 : static const uchar kBadChar = 0xFFFD;
144 : static const uchar kBufferEmpty = 0x0;
145 : static const uchar kIncomplete = 0xFFFFFFFC; // any non-valid code point.
146 : static const unsigned kMaxEncodedSize = 4;
147 : static const unsigned kMaxOneByteChar = 0x7f;
148 : static const unsigned kMaxTwoByteChar = 0x7ff;
149 : static const unsigned kMaxThreeByteChar = 0xffff;
150 : static const unsigned kMaxFourByteChar = 0x1fffff;
151 :
152 : // A single surrogate is coded as a 3 byte UTF-8 sequence, but two together
153 : // that match are coded as a 4 byte UTF-8 sequence.
154 : static const unsigned kBytesSavedByCombiningSurrogates = 2;
155 : static const unsigned kSizeOfUnmatchedSurrogate = 3;
156 : // The maximum size a single UTF-16 code unit may take up when encoded as
157 : // UTF-8.
158 : static const unsigned kMax16BitCodeUnitSize = 3;
159 : static inline uchar ValueOf(const byte* str, size_t length, size_t* cursor);
160 :
161 : typedef uint32_t Utf8IncrementalBuffer;
162 : static uchar ValueOfIncremental(byte next_byte,
163 : Utf8IncrementalBuffer* buffer);
164 : static uchar ValueOfIncrementalFinish(Utf8IncrementalBuffer* buffer);
165 :
166 : // Excludes non-characters from the set of valid code points.
167 : static inline bool IsValidCharacter(uchar c);
168 :
169 : static bool Validate(const byte* str, size_t length);
170 : };
171 :
172 : struct Uppercase {
173 : static bool Is(uchar c);
174 : };
175 : struct Lowercase {
176 : static bool Is(uchar c);
177 : };
178 : struct Letter {
179 : static bool Is(uchar c);
180 : };
181 : struct V8_EXPORT_PRIVATE ID_Start {
182 : static bool Is(uchar c);
183 : };
184 : struct V8_EXPORT_PRIVATE ID_Continue {
185 : static bool Is(uchar c);
186 : };
187 : struct V8_EXPORT_PRIVATE WhiteSpace {
188 : static bool Is(uchar c);
189 : };
190 : struct V8_EXPORT_PRIVATE LineTerminator {
191 : static bool Is(uchar c);
192 : };
193 : struct ToLowercase {
194 : static const int kMaxWidth = 3;
195 : static const bool kIsToLower = true;
196 : static int Convert(uchar c,
197 : uchar n,
198 : uchar* result,
199 : bool* allow_caching_ptr);
200 : };
201 : struct ToUppercase {
202 : static const int kMaxWidth = 3;
203 : static const bool kIsToLower = false;
204 : static int Convert(uchar c,
205 : uchar n,
206 : uchar* result,
207 : bool* allow_caching_ptr);
208 : };
209 : struct Ecma262Canonicalize {
210 : static const int kMaxWidth = 1;
211 : static int Convert(uchar c,
212 : uchar n,
213 : uchar* result,
214 : bool* allow_caching_ptr);
215 : };
216 : struct Ecma262UnCanonicalize {
217 : static const int kMaxWidth = 4;
218 : static int Convert(uchar c,
219 : uchar n,
220 : uchar* result,
221 : bool* allow_caching_ptr);
222 : };
223 : struct CanonicalizationRange {
224 : static const int kMaxWidth = 1;
225 : static int Convert(uchar c,
226 : uchar n,
227 : uchar* result,
228 : bool* allow_caching_ptr);
229 : };
230 :
231 : } // namespace unibrow
232 :
233 : #endif // V8_UNICODE_H_
|