1 | /* |
2 | * Copyright (C) 2003, 2006, 2007, 2008, 2009, 2012 Apple Inc. All rights reserved. |
3 | * |
4 | * This library is free software; you can redistribute it and/or |
5 | * modify it under the terms of the GNU Library General Public |
6 | * License as published by the Free Software Foundation; either |
7 | * version 2 of the License, or (at your option) any later version. |
8 | * |
9 | * This library is distributed in the hope that it will be useful, |
10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
12 | * Library General Public License for more details. |
13 | * |
14 | * You should have received a copy of the GNU Library General Public License |
15 | * along with this library; see the file COPYING.LIB. If not, write to |
16 | * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
17 | * Boston, MA 02110-1301, USA. |
18 | * |
19 | */ |
20 | |
21 | #pragma once |
22 | |
23 | #include "PrivateName.h" |
24 | #include "VM.h" |
25 | #include <wtf/Optional.h> |
26 | #include <wtf/text/CString.h> |
27 | #include <wtf/text/UniquedStringImpl.h> |
28 | #include <wtf/text/WTFString.h> |
29 | |
30 | namespace JSC { |
31 | |
32 | class ExecState; |
33 | |
34 | ALWAYS_INLINE bool isIndex(uint32_t index) |
35 | { |
36 | return index != 0xFFFFFFFFU; |
37 | } |
38 | |
39 | template <typename CharType> |
40 | ALWAYS_INLINE Optional<uint32_t> parseIndex(const CharType* characters, unsigned length) |
41 | { |
42 | // An empty string is not a number. |
43 | if (!length) |
44 | return WTF::nullopt; |
45 | |
46 | // Get the first character, turning it into a digit. |
47 | uint32_t value = characters[0] - '0'; |
48 | if (value > 9) |
49 | return WTF::nullopt; |
50 | |
51 | // Check for leading zeros. If the first characher is 0, then the |
52 | // length of the string must be one - e.g. "042" is not equal to "42". |
53 | if (!value && length > 1) |
54 | return WTF::nullopt; |
55 | |
56 | while (--length) { |
57 | // Multiply value by 10, checking for overflow out of 32 bits. |
58 | if (value > 0xFFFFFFFFU / 10) |
59 | return WTF::nullopt; |
60 | value *= 10; |
61 | |
62 | // Get the next character, turning it into a digit. |
63 | uint32_t newValue = *(++characters) - '0'; |
64 | if (newValue > 9) |
65 | return WTF::nullopt; |
66 | |
67 | // Add in the old value, checking for overflow out of 32 bits. |
68 | newValue += value; |
69 | if (newValue < value) |
70 | return WTF::nullopt; |
71 | value = newValue; |
72 | } |
73 | |
74 | if (!isIndex(value)) |
75 | return WTF::nullopt; |
76 | return value; |
77 | } |
78 | |
79 | ALWAYS_INLINE Optional<uint32_t> parseIndex(StringImpl& impl) |
80 | { |
81 | if (impl.is8Bit()) |
82 | return parseIndex(impl.characters8(), impl.length()); |
83 | return parseIndex(impl.characters16(), impl.length()); |
84 | } |
85 | |
86 | class Identifier { |
87 | friend class Structure; |
88 | public: |
89 | Identifier() { } |
90 | enum EmptyIdentifierFlag { EmptyIdentifier }; |
91 | Identifier(EmptyIdentifierFlag) : m_string(StringImpl::empty()) { ASSERT(m_string.impl()->isAtom()); } |
92 | |
93 | const String& string() const { return m_string; } |
94 | UniquedStringImpl* impl() const { return static_cast<UniquedStringImpl*>(m_string.impl()); } |
95 | |
96 | int length() const { return m_string.length(); } |
97 | |
98 | CString ascii() const { return m_string.ascii(); } |
99 | CString utf8() const { return m_string.utf8(); } |
100 | |
101 | // There's 2 functions to construct Identifier from string, (1) fromString and (2) fromUid. |
102 | // They have different meanings in keeping or discarding symbol-ness of strings. |
103 | // (1): fromString |
104 | // Just construct Identifier from string. String held by Identifier is always atomized. |
105 | // Symbol-ness of StringImpl*, which represents that the string is inteded to be used for ES6 Symbols, is discarded. |
106 | // So a constructed Identifier never represents a symbol. |
107 | // (2): fromUid |
108 | // `StringImpl* uid` represents ether String or Symbol property. |
109 | // fromUid keeps symbol-ness of provided StringImpl* while fromString discards it. |
110 | // Use fromUid when constructing Identifier from StringImpl* which may represent symbols. |
111 | |
112 | // Only to be used with string literals. |
113 | template<unsigned charactersCount> |
114 | static Identifier fromString(VM*, const char (&characters)[charactersCount]); |
115 | template<unsigned charactersCount> |
116 | static Identifier fromString(ExecState*, const char (&characters)[charactersCount]); |
117 | static Identifier fromString(VM*, const LChar*, int length); |
118 | static Identifier fromString(VM*, const UChar*, int length); |
119 | static Identifier fromString(VM*, const String&); |
120 | static Identifier fromString(ExecState*, AtomStringImpl*); |
121 | static Identifier fromString(ExecState*, const AtomString&); |
122 | static Identifier fromString(ExecState*, const String&); |
123 | static Identifier fromString(ExecState*, const char*); |
124 | static Identifier fromString(VM* vm, const Vector<LChar>& characters) { return fromString(vm, characters.data(), characters.size()); } |
125 | |
126 | static Identifier fromUid(VM*, UniquedStringImpl* uid); |
127 | static Identifier fromUid(ExecState*, UniquedStringImpl* uid); |
128 | static Identifier fromUid(const PrivateName&); |
129 | static Identifier fromUid(SymbolImpl&); |
130 | |
131 | static Identifier createLCharFromUChar(VM* vm, const UChar* s, int length) { return Identifier(vm, add8(vm, s, length)); } |
132 | |
133 | JS_EXPORT_PRIVATE static Identifier from(ExecState*, unsigned y); |
134 | JS_EXPORT_PRIVATE static Identifier from(ExecState*, int y); |
135 | static Identifier from(ExecState*, double y); |
136 | static Identifier from(VM*, unsigned y); |
137 | static Identifier from(VM*, int y); |
138 | static Identifier from(VM*, double y); |
139 | |
140 | bool isNull() const { return m_string.isNull(); } |
141 | bool isEmpty() const { return m_string.isEmpty(); } |
142 | bool isSymbol() const { return !isNull() && impl()->isSymbol(); } |
143 | bool isPrivateName() const { return isSymbol() && static_cast<const SymbolImpl*>(impl())->isPrivate(); } |
144 | |
145 | friend bool operator==(const Identifier&, const Identifier&); |
146 | friend bool operator!=(const Identifier&, const Identifier&); |
147 | |
148 | friend bool operator==(const Identifier&, const LChar*); |
149 | friend bool operator==(const Identifier&, const char*); |
150 | friend bool operator!=(const Identifier&, const LChar*); |
151 | friend bool operator!=(const Identifier&, const char*); |
152 | |
153 | static bool equal(const StringImpl*, const LChar*); |
154 | static inline bool equal(const StringImpl*a, const char*b) { return Identifier::equal(a, reinterpret_cast<const LChar*>(b)); }; |
155 | static bool equal(const StringImpl*, const LChar*, unsigned length); |
156 | static bool equal(const StringImpl*, const UChar*, unsigned length); |
157 | static bool equal(const StringImpl* a, const StringImpl* b) { return ::equal(a, b); } |
158 | |
159 | // Only to be used with string literals. |
160 | JS_EXPORT_PRIVATE static Ref<StringImpl> add(VM*, const char*); |
161 | JS_EXPORT_PRIVATE static Ref<StringImpl> add(ExecState*, const char*); |
162 | |
163 | void dump(PrintStream&) const; |
164 | |
165 | private: |
166 | String m_string; |
167 | |
168 | // Only to be used with string literals. |
169 | template<unsigned charactersCount> |
170 | Identifier(VM* vm, const char (&characters)[charactersCount]) : m_string(add(vm, characters)) { ASSERT(m_string.impl()->isAtom()); } |
171 | |
172 | Identifier(VM* vm, const LChar* s, int length) : m_string(add(vm, s, length)) { ASSERT(m_string.impl()->isAtom()); } |
173 | Identifier(VM* vm, const UChar* s, int length) : m_string(add(vm, s, length)) { ASSERT(m_string.impl()->isAtom()); } |
174 | Identifier(ExecState*, AtomStringImpl*); |
175 | Identifier(ExecState*, const AtomString&); |
176 | Identifier(VM* vm, const String& string) : m_string(add(vm, string.impl())) { ASSERT(m_string.impl()->isAtom()); } |
177 | Identifier(VM* vm, StringImpl* rep) : m_string(add(vm, rep)) { ASSERT(m_string.impl()->isAtom()); } |
178 | |
179 | Identifier(SymbolImpl& uid) |
180 | : m_string(&uid) |
181 | { |
182 | } |
183 | |
184 | template <typename CharType> |
185 | ALWAYS_INLINE static uint32_t toUInt32FromCharacters(const CharType* characters, unsigned length, bool& ok); |
186 | |
187 | static bool equal(const Identifier& a, const Identifier& b) { return a.m_string.impl() == b.m_string.impl(); } |
188 | static bool equal(const Identifier& a, const LChar* b) { return equal(a.m_string.impl(), b); } |
189 | |
190 | template <typename T> static Ref<StringImpl> add(VM*, const T*, int length); |
191 | static Ref<StringImpl> add8(VM*, const UChar*, int length); |
192 | template <typename T> ALWAYS_INLINE static bool canUseSingleCharacterString(T); |
193 | |
194 | static Ref<StringImpl> add(ExecState*, StringImpl*); |
195 | static Ref<StringImpl> add(VM*, StringImpl*); |
196 | |
197 | #ifndef NDEBUG |
198 | JS_EXPORT_PRIVATE static void checkCurrentAtomStringTable(ExecState*); |
199 | JS_EXPORT_PRIVATE static void checkCurrentAtomStringTable(VM*); |
200 | #else |
201 | JS_EXPORT_PRIVATE NO_RETURN_DUE_TO_CRASH static void checkCurrentAtomStringTable(ExecState*); |
202 | JS_EXPORT_PRIVATE NO_RETURN_DUE_TO_CRASH static void checkCurrentAtomStringTable(VM*); |
203 | #endif |
204 | }; |
205 | |
206 | template <> ALWAYS_INLINE bool Identifier::canUseSingleCharacterString(LChar) |
207 | { |
208 | ASSERT(maxSingleCharacterString == 0xff); |
209 | return true; |
210 | } |
211 | |
212 | template <> ALWAYS_INLINE bool Identifier::canUseSingleCharacterString(UChar c) |
213 | { |
214 | return (c <= maxSingleCharacterString); |
215 | } |
216 | |
217 | template <typename T> |
218 | Ref<StringImpl> Identifier::add(VM* vm, const T* s, int length) |
219 | { |
220 | if (length == 1) { |
221 | T c = s[0]; |
222 | if (canUseSingleCharacterString(c)) |
223 | return vm->smallStrings.singleCharacterStringRep(c); |
224 | } |
225 | if (!length) |
226 | return *StringImpl::empty(); |
227 | |
228 | return *AtomStringImpl::add(s, length); |
229 | } |
230 | |
231 | inline bool operator==(const Identifier& a, const Identifier& b) |
232 | { |
233 | return Identifier::equal(a, b); |
234 | } |
235 | |
236 | inline bool operator!=(const Identifier& a, const Identifier& b) |
237 | { |
238 | return !Identifier::equal(a, b); |
239 | } |
240 | |
241 | inline bool operator==(const Identifier& a, const LChar* b) |
242 | { |
243 | return Identifier::equal(a, b); |
244 | } |
245 | |
246 | inline bool operator==(const Identifier& a, const char* b) |
247 | { |
248 | return Identifier::equal(a, reinterpret_cast<const LChar*>(b)); |
249 | } |
250 | |
251 | inline bool operator!=(const Identifier& a, const LChar* b) |
252 | { |
253 | return !Identifier::equal(a, b); |
254 | } |
255 | |
256 | inline bool operator!=(const Identifier& a, const char* b) |
257 | { |
258 | return !Identifier::equal(a, reinterpret_cast<const LChar*>(b)); |
259 | } |
260 | |
261 | inline bool Identifier::equal(const StringImpl* r, const LChar* s) |
262 | { |
263 | return WTF::equal(r, s); |
264 | } |
265 | |
266 | inline bool Identifier::equal(const StringImpl* r, const LChar* s, unsigned length) |
267 | { |
268 | return WTF::equal(r, s, length); |
269 | } |
270 | |
271 | inline bool Identifier::equal(const StringImpl* r, const UChar* s, unsigned length) |
272 | { |
273 | return WTF::equal(r, s, length); |
274 | } |
275 | |
276 | ALWAYS_INLINE Optional<uint32_t> parseIndex(const Identifier& identifier) |
277 | { |
278 | auto uid = identifier.impl(); |
279 | if (!uid) |
280 | return WTF::nullopt; |
281 | if (uid->isSymbol()) |
282 | return WTF::nullopt; |
283 | return parseIndex(*uid); |
284 | } |
285 | |
286 | JSValue identifierToJSValue(VM&, const Identifier&); |
287 | // This will stringify private symbols. When leaking JSValues to |
288 | // non-internal code, make sure to use this function and not the above one. |
289 | JSValue identifierToSafePublicJSValue(VM&, const Identifier&); |
290 | |
291 | // FIXME: It may be better for this to just be a typedef for PtrHash, since PtrHash may be cheaper to |
292 | // compute than loading the StringImpl's hash from memory. That change would also reduce the likelihood of |
293 | // crashes in code that somehow dangled a StringImpl. |
294 | // https://bugs.webkit.org/show_bug.cgi?id=150137 |
295 | struct IdentifierRepHash : PtrHash<RefPtr<UniquedStringImpl>> { |
296 | static unsigned hash(const RefPtr<UniquedStringImpl>& key) { return key->existingSymbolAwareHash(); } |
297 | static unsigned hash(UniquedStringImpl* key) { return key->existingSymbolAwareHash(); } |
298 | }; |
299 | |
300 | struct IdentifierMapIndexHashTraits : HashTraits<int> { |
301 | static int emptyValue() { return std::numeric_limits<int>::max(); } |
302 | static const bool emptyValueIsZero = false; |
303 | }; |
304 | |
305 | typedef HashSet<RefPtr<UniquedStringImpl>, IdentifierRepHash> IdentifierSet; |
306 | typedef HashMap<RefPtr<UniquedStringImpl>, int, IdentifierRepHash, HashTraits<RefPtr<UniquedStringImpl>>, IdentifierMapIndexHashTraits> IdentifierMap; |
307 | typedef HashMap<UniquedStringImpl*, int, IdentifierRepHash, HashTraits<UniquedStringImpl*>, IdentifierMapIndexHashTraits> BorrowedIdentifierMap; |
308 | |
309 | } // namespace JSC |
310 | |
311 | namespace WTF { |
312 | |
313 | template <> struct VectorTraits<JSC::Identifier> : SimpleClassVectorTraits { }; |
314 | |
315 | } // namespace WTF |
316 | |