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 546125949 : 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 545061376 : : bit_field_(CodePointField::encode(0) | ValueField::encode(0)) {}
40 : inline CacheEntry(uchar code_point, bool value)
41 : : bit_field_(CodePointField::encode(code_point) |
42 3200659 : 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 42298839 : 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 42243840 : 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 169164 : return IsLeadSurrogate(lead) && IsTrailSurrogate(trail);
98 : }
99 : static inline bool IsLeadSurrogate(int code) {
100 1012831102 : if (code == kNoPreviousCharacter) return false;
101 1010798289 : return (code & 0xfc00) == 0xd800;
102 : }
103 : static inline bool IsTrailSurrogate(int code) {
104 23588611 : if (code == kNoPreviousCharacter) return false;
105 23589214 : return (code & 0xfc00) == 0xdc00;
106 : }
107 :
108 : static inline int CombineSurrogatePair(uchar lead, uchar trail) {
109 2410 : 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 10032050 : return 0xd800 + (((char_code - 0x10000) >> 10) & 0x3ff);
124 : }
125 : static inline uint16_t TrailSurrogate(uint32_t char_code) {
126 10032050 : return 0xdc00 + (char_code & 0x3ff);
127 : }
128 : };
129 :
130 : class V8_EXPORT_PRIVATE Utf8 {
131 : public:
132 : static inline uchar Length(uchar chr, int previous);
133 : static inline unsigned EncodeOneByte(char* out, uint8_t c);
134 : static inline unsigned Encode(char* out,
135 : uchar c,
136 : int previous,
137 : bool replace_invalid = false);
138 : static uchar CalculateValue(const byte* str, size_t length, size_t* cursor);
139 :
140 : // The unicode replacement character, used to signal invalid unicode
141 : // sequences (e.g. an orphan surrogate) when converting to a UTF-8 encoding.
142 : static const uchar kBadChar = 0xFFFD;
143 : static const uchar kBufferEmpty = 0x0;
144 : static const uchar kIncomplete = 0xFFFFFFFC; // any non-valid code point.
145 : static const unsigned kMaxEncodedSize = 4;
146 : static const unsigned kMaxOneByteChar = 0x7f;
147 : static const unsigned kMaxTwoByteChar = 0x7ff;
148 : static const unsigned kMaxThreeByteChar = 0xffff;
149 : static const unsigned kMaxFourByteChar = 0x1fffff;
150 :
151 : // A single surrogate is coded as a 3 byte UTF-8 sequence, but two together
152 : // that match are coded as a 4 byte UTF-8 sequence.
153 : static const unsigned kBytesSavedByCombiningSurrogates = 2;
154 : static const unsigned kSizeOfUnmatchedSurrogate = 3;
155 : // The maximum size a single UTF-16 code unit may take up when encoded as
156 : // UTF-8.
157 : static const unsigned kMax16BitCodeUnitSize = 3;
158 : static inline uchar ValueOf(const byte* str, size_t length, size_t* cursor);
159 :
160 : typedef uint32_t Utf8IncrementalBuffer;
161 : static uchar ValueOfIncremental(byte next_byte,
162 : Utf8IncrementalBuffer* buffer);
163 : static uchar ValueOfIncrementalFinish(Utf8IncrementalBuffer* buffer);
164 :
165 : // Excludes non-characters from the set of valid code points.
166 : static inline bool IsValidCharacter(uchar c);
167 :
168 : // Validate if the input has a valid utf-8 encoding. Unlike JS source code
169 : // this validation function will accept any unicode code point, including
170 : // kBadChar and BOMs.
171 : //
172 : // This method checks for:
173 : // - valid utf-8 endcoding (e.g. no over-long encodings),
174 : // - absence of surrogates,
175 : // - valid code point range.
176 : static bool ValidateEncoding(const byte* str, size_t length);
177 : };
178 :
179 : struct Uppercase {
180 : static bool Is(uchar c);
181 : };
182 : struct Letter {
183 : static bool Is(uchar c);
184 : };
185 : #ifndef V8_INTL_SUPPORT
186 : struct V8_EXPORT_PRIVATE ID_Start {
187 : static bool Is(uchar c);
188 : };
189 : struct V8_EXPORT_PRIVATE ID_Continue {
190 : static bool Is(uchar c);
191 : };
192 : struct V8_EXPORT_PRIVATE WhiteSpace {
193 : static bool Is(uchar c);
194 : };
195 : #endif // !V8_INTL_SUPPORT
196 :
197 : // LineTerminator: 'JS_Line_Terminator' in point.properties
198 : // ES#sec-line-terminators lists exactly 4 code points:
199 : // LF (U+000A), CR (U+000D), LS(U+2028), PS(U+2029)
200 : V8_INLINE bool IsLineTerminator(uchar c) {
201 216006189 : return c == 0x000A || c == 0x000D || c == 0x2028 || c == 0x2029;
202 : }
203 :
204 : #ifndef V8_INTL_SUPPORT
205 : struct ToLowercase {
206 : static const int kMaxWidth = 3;
207 : static const bool kIsToLower = true;
208 : static int Convert(uchar c,
209 : uchar n,
210 : uchar* result,
211 : bool* allow_caching_ptr);
212 : };
213 : struct ToUppercase {
214 : static const int kMaxWidth = 3;
215 : static const bool kIsToLower = false;
216 : static int Convert(uchar c,
217 : uchar n,
218 : uchar* result,
219 : bool* allow_caching_ptr);
220 : };
221 : #endif
222 : struct Ecma262Canonicalize {
223 : static const int kMaxWidth = 1;
224 : static int Convert(uchar c,
225 : uchar n,
226 : uchar* result,
227 : bool* allow_caching_ptr);
228 : };
229 : struct Ecma262UnCanonicalize {
230 : static const int kMaxWidth = 4;
231 : static int Convert(uchar c,
232 : uchar n,
233 : uchar* result,
234 : bool* allow_caching_ptr);
235 : };
236 : struct CanonicalizationRange {
237 : static const int kMaxWidth = 1;
238 : static int Convert(uchar c,
239 : uchar n,
240 : uchar* result,
241 : bool* allow_caching_ptr);
242 : };
243 :
244 : } // namespace unibrow
245 :
246 : #endif // V8_UNICODE_H_
|
