1 | //===- IdentifierTable.cpp - Hash table for identifier lookup -------------===// |
2 | // |
3 | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | // See https://llvm.org/LICENSE.txt for license information. |
5 | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | // |
7 | //===----------------------------------------------------------------------===// |
8 | // |
9 | // This file implements the IdentifierInfo, IdentifierVisitor, and |
10 | // IdentifierTable interfaces. |
11 | // |
12 | //===----------------------------------------------------------------------===// |
13 | |
14 | #include "clang/Basic/IdentifierTable.h" |
15 | #include "clang/Basic/CharInfo.h" |
16 | #include "clang/Basic/DiagnosticLex.h" |
17 | #include "clang/Basic/LangOptions.h" |
18 | #include "clang/Basic/OperatorKinds.h" |
19 | #include "clang/Basic/Specifiers.h" |
20 | #include "clang/Basic/TargetBuiltins.h" |
21 | #include "clang/Basic/TokenKinds.h" |
22 | #include "llvm/ADT/DenseMapInfo.h" |
23 | #include "llvm/ADT/FoldingSet.h" |
24 | #include "llvm/ADT/StringMap.h" |
25 | #include "llvm/ADT/StringRef.h" |
26 | #include "llvm/Support/Allocator.h" |
27 | #include "llvm/Support/raw_ostream.h" |
28 | #include <cassert> |
29 | #include <cstdio> |
30 | #include <cstring> |
31 | #include <string> |
32 | |
33 | using namespace clang; |
34 | |
35 | // A check to make sure the ObjCOrBuiltinID has sufficient room to store the |
36 | // largest possible target/aux-target combination. If we exceed this, we likely |
37 | // need to just change the ObjCOrBuiltinIDBits value in IdentifierTable.h. |
38 | static_assert(2 * LargestBuiltinID < (2 << (InterestingIdentifierBits - 1)), |
39 | "Insufficient ObjCOrBuiltinID Bits" ); |
40 | |
41 | //===----------------------------------------------------------------------===// |
42 | // IdentifierTable Implementation |
43 | //===----------------------------------------------------------------------===// |
44 | |
45 | IdentifierIterator::~IdentifierIterator() = default; |
46 | |
47 | IdentifierInfoLookup::~IdentifierInfoLookup() = default; |
48 | |
49 | namespace { |
50 | |
51 | /// A simple identifier lookup iterator that represents an |
52 | /// empty sequence of identifiers. |
53 | class EmptyLookupIterator : public IdentifierIterator { |
54 | public: |
55 | StringRef Next() override { return StringRef(); } |
56 | }; |
57 | |
58 | } // namespace |
59 | |
60 | IdentifierIterator *IdentifierInfoLookup::getIdentifiers() { |
61 | return new EmptyLookupIterator(); |
62 | } |
63 | |
64 | IdentifierTable::IdentifierTable(IdentifierInfoLookup *ExternalLookup) |
65 | : HashTable(8192), // Start with space for 8K identifiers. |
66 | ExternalLookup(ExternalLookup) {} |
67 | |
68 | IdentifierTable::IdentifierTable(const LangOptions &LangOpts, |
69 | IdentifierInfoLookup *ExternalLookup) |
70 | : IdentifierTable(ExternalLookup) { |
71 | // Populate the identifier table with info about keywords for the current |
72 | // language. |
73 | AddKeywords(LangOpts); |
74 | } |
75 | |
76 | //===----------------------------------------------------------------------===// |
77 | // Language Keyword Implementation |
78 | //===----------------------------------------------------------------------===// |
79 | |
80 | // Constants for TokenKinds.def |
81 | namespace { |
82 | |
83 | enum TokenKey : unsigned { |
84 | KEYC99 = 0x1, |
85 | KEYCXX = 0x2, |
86 | KEYCXX11 = 0x4, |
87 | KEYGNU = 0x8, |
88 | KEYMS = 0x10, |
89 | BOOLSUPPORT = 0x20, |
90 | KEYALTIVEC = 0x40, |
91 | KEYNOCXX = 0x80, |
92 | KEYBORLAND = 0x100, |
93 | KEYOPENCLC = 0x200, |
94 | KEYC23 = 0x400, |
95 | KEYNOMS18 = 0x800, |
96 | KEYNOOPENCL = 0x1000, |
97 | WCHARSUPPORT = 0x2000, |
98 | HALFSUPPORT = 0x4000, |
99 | CHAR8SUPPORT = 0x8000, |
100 | KEYOBJC = 0x10000, |
101 | KEYZVECTOR = 0x20000, |
102 | KEYCOROUTINES = 0x40000, |
103 | KEYMODULES = 0x80000, |
104 | KEYCXX20 = 0x100000, |
105 | KEYOPENCLCXX = 0x200000, |
106 | KEYMSCOMPAT = 0x400000, |
107 | KEYSYCL = 0x800000, |
108 | KEYCUDA = 0x1000000, |
109 | KEYZOS = 0x2000000, |
110 | KEYNOZOS = 0x4000000, |
111 | KEYHLSL = 0x8000000, |
112 | KEYFIXEDPOINT = 0x10000000, |
113 | KEYMAX = KEYFIXEDPOINT, // The maximum key |
114 | KEYALLCXX = KEYCXX | KEYCXX11 | KEYCXX20, |
115 | KEYALL = (KEYMAX | (KEYMAX - 1)) & ~KEYNOMS18 & ~KEYNOOPENCL & |
116 | ~KEYNOZOS // KEYNOMS18, KEYNOOPENCL, KEYNOZOS are excluded. |
117 | }; |
118 | |
119 | /// How a keyword is treated in the selected standard. This enum is ordered |
120 | /// intentionally so that the value that 'wins' is the most 'permissive'. |
121 | enum KeywordStatus { |
122 | KS_Unknown, // Not yet calculated. Used when figuring out the status. |
123 | KS_Disabled, // Disabled |
124 | KS_Future, // Is a keyword in future standard |
125 | KS_Extension, // Is an extension |
126 | KS_Enabled, // Enabled |
127 | }; |
128 | |
129 | } // namespace |
130 | |
131 | // This works on a single TokenKey flag and checks the LangOpts to get the |
132 | // KeywordStatus based exclusively on this flag, so that it can be merged in |
133 | // getKeywordStatus. Most should be enabled/disabled, but some might imply |
134 | // 'future' versions, or extensions. Returns 'unknown' unless this is KNOWN to |
135 | // be disabled, and the calling function makes it 'disabled' if no other flag |
136 | // changes it. This is necessary for the KEYNOCXX and KEYNOOPENCL flags. |
137 | static KeywordStatus getKeywordStatusHelper(const LangOptions &LangOpts, |
138 | TokenKey Flag) { |
139 | // Flag is a single bit version of TokenKey (that is, not |
140 | // KEYALL/KEYALLCXX/etc), so we can check with == throughout this function. |
141 | assert((Flag & ~(Flag - 1)) == Flag && "Multiple bits set?" ); |
142 | |
143 | switch (Flag) { |
144 | case KEYC99: |
145 | if (LangOpts.C99) |
146 | return KS_Enabled; |
147 | return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown; |
148 | case KEYC23: |
149 | if (LangOpts.C23) |
150 | return KS_Enabled; |
151 | return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown; |
152 | case KEYCXX: |
153 | return LangOpts.CPlusPlus ? KS_Enabled : KS_Unknown; |
154 | case KEYCXX11: |
155 | if (LangOpts.CPlusPlus11) |
156 | return KS_Enabled; |
157 | return LangOpts.CPlusPlus ? KS_Future : KS_Unknown; |
158 | case KEYCXX20: |
159 | if (LangOpts.CPlusPlus20) |
160 | return KS_Enabled; |
161 | return LangOpts.CPlusPlus ? KS_Future : KS_Unknown; |
162 | case KEYGNU: |
163 | return LangOpts.GNUKeywords ? KS_Extension : KS_Unknown; |
164 | case KEYMS: |
165 | return LangOpts.MicrosoftExt ? KS_Extension : KS_Unknown; |
166 | case BOOLSUPPORT: |
167 | if (LangOpts.Bool) return KS_Enabled; |
168 | return !LangOpts.CPlusPlus ? KS_Future : KS_Unknown; |
169 | case KEYALTIVEC: |
170 | return LangOpts.AltiVec ? KS_Enabled : KS_Unknown; |
171 | case KEYBORLAND: |
172 | return LangOpts.Borland ? KS_Extension : KS_Unknown; |
173 | case KEYOPENCLC: |
174 | return LangOpts.OpenCL && !LangOpts.OpenCLCPlusPlus ? KS_Enabled |
175 | : KS_Unknown; |
176 | case WCHARSUPPORT: |
177 | return LangOpts.WChar ? KS_Enabled : KS_Unknown; |
178 | case HALFSUPPORT: |
179 | return LangOpts.Half ? KS_Enabled : KS_Unknown; |
180 | case CHAR8SUPPORT: |
181 | if (LangOpts.Char8) return KS_Enabled; |
182 | if (LangOpts.CPlusPlus20) return KS_Unknown; |
183 | if (LangOpts.CPlusPlus) return KS_Future; |
184 | return KS_Unknown; |
185 | case KEYOBJC: |
186 | // We treat bridge casts as objective-C keywords so we can warn on them |
187 | // in non-arc mode. |
188 | return LangOpts.ObjC ? KS_Enabled : KS_Unknown; |
189 | case KEYZVECTOR: |
190 | return LangOpts.ZVector ? KS_Enabled : KS_Unknown; |
191 | case KEYCOROUTINES: |
192 | return LangOpts.Coroutines ? KS_Enabled : KS_Unknown; |
193 | case KEYMODULES: |
194 | return KS_Unknown; |
195 | case KEYOPENCLCXX: |
196 | return LangOpts.OpenCLCPlusPlus ? KS_Enabled : KS_Unknown; |
197 | case KEYMSCOMPAT: |
198 | return LangOpts.MSVCCompat ? KS_Enabled : KS_Unknown; |
199 | case KEYSYCL: |
200 | return LangOpts.isSYCL() ? KS_Enabled : KS_Unknown; |
201 | case KEYCUDA: |
202 | return LangOpts.CUDA ? KS_Enabled : KS_Unknown; |
203 | case KEYZOS: |
204 | return LangOpts.ZOSExt ? KS_Enabled : KS_Unknown; |
205 | case KEYHLSL: |
206 | return LangOpts.HLSL ? KS_Enabled : KS_Unknown; |
207 | case KEYNOCXX: |
208 | // This is enabled in all non-C++ modes, but might be enabled for other |
209 | // reasons as well. |
210 | return LangOpts.CPlusPlus ? KS_Unknown : KS_Enabled; |
211 | case KEYNOOPENCL: |
212 | case KEYNOMS18: |
213 | case KEYNOZOS: |
214 | // The disable behavior for this is handled in getKeywordStatus. |
215 | return KS_Unknown; |
216 | case KEYFIXEDPOINT: |
217 | return LangOpts.FixedPoint ? KS_Enabled : KS_Disabled; |
218 | default: |
219 | llvm_unreachable("Unknown KeywordStatus flag" ); |
220 | } |
221 | } |
222 | |
223 | /// Translates flags as specified in TokenKinds.def into keyword status |
224 | /// in the given language standard. |
225 | static KeywordStatus getKeywordStatus(const LangOptions &LangOpts, |
226 | unsigned Flags) { |
227 | // KEYALL means always enabled, so special case this one. |
228 | if (Flags == KEYALL) return KS_Enabled; |
229 | // These are tests that need to 'always win', as they are special in that they |
230 | // disable based on certain conditions. |
231 | if (LangOpts.OpenCL && (Flags & KEYNOOPENCL)) return KS_Disabled; |
232 | if (LangOpts.MSVCCompat && (Flags & KEYNOMS18) && |
233 | !LangOpts.isCompatibleWithMSVC(MajorVersion: LangOptions::MSVC2015)) |
234 | return KS_Disabled; |
235 | if (LangOpts.ZOSExt && (Flags & KEYNOZOS)) |
236 | return KS_Disabled; |
237 | KeywordStatus CurStatus = KS_Unknown; |
238 | |
239 | while (Flags != 0) { |
240 | unsigned CurFlag = Flags & ~(Flags - 1); |
241 | Flags = Flags & ~CurFlag; |
242 | CurStatus = std::max( |
243 | a: CurStatus, |
244 | b: getKeywordStatusHelper(LangOpts, Flag: static_cast<TokenKey>(CurFlag))); |
245 | } |
246 | |
247 | if (CurStatus == KS_Unknown) |
248 | return KS_Disabled; |
249 | return CurStatus; |
250 | } |
251 | |
252 | static bool IsKeywordInCpp(unsigned Flags) { |
253 | return (Flags & (KEYCXX | KEYCXX11 | KEYCXX20 | BOOLSUPPORT | WCHARSUPPORT | |
254 | CHAR8SUPPORT)) != 0; |
255 | } |
256 | |
257 | static void MarkIdentifierAsKeywordInCpp(IdentifierTable &Table, |
258 | StringRef Name) { |
259 | IdentifierInfo &II = Table.get(Name, TokenCode: tok::identifier); |
260 | II.setIsKeywordInCPlusPlus(); |
261 | II.setHandleIdentifierCase(); |
262 | } |
263 | |
264 | /// AddKeyword - This method is used to associate a token ID with specific |
265 | /// identifiers because they are language keywords. This causes the lexer to |
266 | /// automatically map matching identifiers to specialized token codes. |
267 | static void AddKeyword(StringRef Keyword, |
268 | tok::TokenKind TokenCode, unsigned Flags, |
269 | const LangOptions &LangOpts, IdentifierTable &Table) { |
270 | KeywordStatus AddResult = getKeywordStatus(LangOpts, Flags); |
271 | |
272 | // Don't add this keyword if disabled in this language and isn't otherwise |
273 | // special. |
274 | if (AddResult == KS_Disabled) { |
275 | // We do not consider any identifiers to be C++ keywords when in |
276 | // Objective-C because @ effectively introduces a custom grammar where C++ |
277 | // keywords can be used (and similar for selectors). We could enable this |
278 | // for Objective-C, but it would require more logic to ensure we do not |
279 | // issue compatibility diagnostics in these cases. |
280 | if (!LangOpts.ObjC && IsKeywordInCpp(Flags)) |
281 | MarkIdentifierAsKeywordInCpp(Table, Name: Keyword); |
282 | return; |
283 | } |
284 | |
285 | IdentifierInfo &Info = |
286 | Table.get(Name: Keyword, TokenCode: AddResult == KS_Future ? tok::identifier : TokenCode); |
287 | Info.setIsExtensionToken(AddResult == KS_Extension); |
288 | Info.setIsFutureCompatKeyword(AddResult == KS_Future); |
289 | } |
290 | |
291 | /// AddCXXOperatorKeyword - Register a C++ operator keyword alternative |
292 | /// representations. |
293 | static void AddCXXOperatorKeyword(StringRef Keyword, |
294 | tok::TokenKind TokenCode, |
295 | IdentifierTable &Table) { |
296 | IdentifierInfo &Info = Table.get(Name: Keyword, TokenCode); |
297 | Info.setIsCPlusPlusOperatorKeyword(); |
298 | } |
299 | |
300 | /// AddObjCKeyword - Register an Objective-C \@keyword like "class" "selector" |
301 | /// or "property". |
302 | static void AddObjCKeyword(StringRef Name, |
303 | tok::ObjCKeywordKind ObjCID, |
304 | IdentifierTable &Table) { |
305 | Table.get(Name).setObjCKeywordID(ObjCID); |
306 | } |
307 | |
308 | static void AddNotableIdentifier(StringRef Name, |
309 | tok::NotableIdentifierKind BTID, |
310 | IdentifierTable &Table) { |
311 | // Don't add 'not_notable' identifier. |
312 | if (BTID != tok::not_notable) { |
313 | IdentifierInfo &Info = Table.get(Name, TokenCode: tok::identifier); |
314 | Info.setNotableIdentifierID(BTID); |
315 | } |
316 | } |
317 | |
318 | /// AddKeywords - Add all keywords to the symbol table. |
319 | /// |
320 | void IdentifierTable::AddKeywords(const LangOptions &LangOpts) { |
321 | // Add keywords and tokens for the current language. |
322 | #define KEYWORD(NAME, FLAGS) \ |
323 | AddKeyword(StringRef(#NAME), tok::kw_ ## NAME, \ |
324 | FLAGS, LangOpts, *this); |
325 | #define ALIAS(NAME, TOK, FLAGS) \ |
326 | AddKeyword(StringRef(NAME), tok::kw_ ## TOK, \ |
327 | FLAGS, LangOpts, *this); |
328 | #define CXX_KEYWORD_OPERATOR(NAME, ALIAS) \ |
329 | if (LangOpts.CXXOperatorNames) \ |
330 | AddCXXOperatorKeyword(StringRef(#NAME), tok::ALIAS, *this); \ |
331 | else \ |
332 | MarkIdentifierAsKeywordInCpp(*this, StringRef(#NAME)); |
333 | #define OBJC_AT_KEYWORD(NAME) \ |
334 | if (LangOpts.ObjC) \ |
335 | AddObjCKeyword(StringRef(#NAME), tok::objc_##NAME, *this); |
336 | #define NOTABLE_IDENTIFIER(NAME) \ |
337 | AddNotableIdentifier(StringRef(#NAME), tok::NAME, *this); |
338 | |
339 | #define TESTING_KEYWORD(NAME, FLAGS) |
340 | #include "clang/Basic/TokenKinds.def" |
341 | |
342 | if (LangOpts.ParseUnknownAnytype) |
343 | AddKeyword(Keyword: "__unknown_anytype" , TokenCode: tok::kw___unknown_anytype, Flags: KEYALL, |
344 | LangOpts, Table&: *this); |
345 | |
346 | if (LangOpts.DeclSpecKeyword) |
347 | AddKeyword(Keyword: "__declspec" , TokenCode: tok::kw___declspec, Flags: KEYALL, LangOpts, Table&: *this); |
348 | |
349 | if (LangOpts.IEEE128) |
350 | AddKeyword(Keyword: "__ieee128" , TokenCode: tok::kw___float128, Flags: KEYALL, LangOpts, Table&: *this); |
351 | |
352 | // Add the 'import' contextual keyword. |
353 | get(Name: "import" ).setModulesImport(true); |
354 | } |
355 | |
356 | /// Checks if the specified token kind represents a keyword in the |
357 | /// specified language. |
358 | /// \returns Status of the keyword in the language. |
359 | static KeywordStatus getTokenKwStatus(const LangOptions &LangOpts, |
360 | tok::TokenKind K) { |
361 | switch (K) { |
362 | #define KEYWORD(NAME, FLAGS) \ |
363 | case tok::kw_##NAME: return getKeywordStatus(LangOpts, FLAGS); |
364 | #include "clang/Basic/TokenKinds.def" |
365 | default: return KS_Disabled; |
366 | } |
367 | } |
368 | |
369 | /// Returns true if the identifier represents a keyword in the |
370 | /// specified language. |
371 | bool IdentifierInfo::isKeyword(const LangOptions &LangOpts) const { |
372 | switch (getTokenKwStatus(LangOpts, K: getTokenID())) { |
373 | case KS_Enabled: |
374 | case KS_Extension: |
375 | return true; |
376 | default: |
377 | return false; |
378 | } |
379 | } |
380 | |
381 | /// Returns true if the identifier represents a C++ keyword in the |
382 | /// specified language. |
383 | bool IdentifierInfo::isCPlusPlusKeyword(const LangOptions &LangOpts) const { |
384 | if (!LangOpts.CPlusPlus || !isKeyword(LangOpts)) |
385 | return false; |
386 | // This is a C++ keyword if this identifier is not a keyword when checked |
387 | // using LangOptions without C++ support. |
388 | LangOptions LangOptsNoCPP = LangOpts; |
389 | LangOptsNoCPP.CPlusPlus = false; |
390 | LangOptsNoCPP.CPlusPlus11 = false; |
391 | LangOptsNoCPP.CPlusPlus20 = false; |
392 | return !isKeyword(LangOpts: LangOptsNoCPP); |
393 | } |
394 | |
395 | ReservedIdentifierStatus |
396 | IdentifierInfo::isReserved(const LangOptions &LangOpts) const { |
397 | StringRef Name = getName(); |
398 | |
399 | // '_' is a reserved identifier, but its use is so common (e.g. to store |
400 | // ignored values) that we don't warn on it. |
401 | if (Name.size() <= 1) |
402 | return ReservedIdentifierStatus::NotReserved; |
403 | |
404 | // [lex.name] p3 |
405 | if (Name[0] == '_') { |
406 | |
407 | // Each name that begins with an underscore followed by an uppercase letter |
408 | // or another underscore is reserved. |
409 | if (Name[1] == '_') |
410 | return ReservedIdentifierStatus::StartsWithDoubleUnderscore; |
411 | |
412 | if ('A' <= Name[1] && Name[1] <= 'Z') |
413 | return ReservedIdentifierStatus:: |
414 | StartsWithUnderscoreFollowedByCapitalLetter; |
415 | |
416 | // This is a bit misleading: it actually means it's only reserved if we're |
417 | // at global scope because it starts with an underscore. |
418 | return ReservedIdentifierStatus::StartsWithUnderscoreAtGlobalScope; |
419 | } |
420 | |
421 | // Each name that contains a double underscore (__) is reserved. |
422 | if (LangOpts.CPlusPlus && Name.contains(Other: "__" )) |
423 | return ReservedIdentifierStatus::ContainsDoubleUnderscore; |
424 | |
425 | return ReservedIdentifierStatus::NotReserved; |
426 | } |
427 | |
428 | ReservedLiteralSuffixIdStatus |
429 | IdentifierInfo::isReservedLiteralSuffixId() const { |
430 | StringRef Name = getName(); |
431 | |
432 | // Note: the diag::warn_deprecated_literal_operator_id diagnostic depends on |
433 | // this being the first check we do, so if this order changes, we have to fix |
434 | // that as well. |
435 | if (Name[0] != '_') |
436 | return ReservedLiteralSuffixIdStatus::NotStartsWithUnderscore; |
437 | |
438 | if (Name.contains(Other: "__" )) |
439 | return ReservedLiteralSuffixIdStatus::ContainsDoubleUnderscore; |
440 | |
441 | return ReservedLiteralSuffixIdStatus::NotReserved; |
442 | } |
443 | |
444 | StringRef IdentifierInfo::deuglifiedName() const { |
445 | StringRef Name = getName(); |
446 | if (Name.size() >= 2 && Name.front() == '_' && |
447 | (Name[1] == '_' || (Name[1] >= 'A' && Name[1] <= 'Z'))) |
448 | return Name.ltrim(Char: '_'); |
449 | return Name; |
450 | } |
451 | |
452 | tok::PPKeywordKind IdentifierInfo::getPPKeywordID() const { |
453 | // We use a perfect hash function here involving the length of the keyword, |
454 | // the first and third character. For preprocessor ID's there are no |
455 | // collisions (if there were, the switch below would complain about duplicate |
456 | // case values). Note that this depends on 'if' being null terminated. |
457 | |
458 | #define HASH(LEN, FIRST, THIRD) \ |
459 | (LEN << 6) + (((FIRST - 'a') - (THIRD - 'a')) & 63) |
460 | #define CASE(LEN, FIRST, THIRD, NAME) \ |
461 | case HASH(LEN, FIRST, THIRD): \ |
462 | return memcmp(Name, #NAME, LEN) ? tok::pp_not_keyword : tok::pp_ ## NAME |
463 | |
464 | unsigned Len = getLength(); |
465 | if (Len < 2) return tok::pp_not_keyword; |
466 | const char *Name = getNameStart(); |
467 | switch (HASH(Len, Name[0], Name[2])) { |
468 | default: return tok::pp_not_keyword; |
469 | CASE( 2, 'i', '\0', if); |
470 | CASE( 4, 'e', 'i', elif); |
471 | CASE( 4, 'e', 's', else); |
472 | CASE( 4, 'l', 'n', line); |
473 | CASE( 4, 's', 'c', sccs); |
474 | CASE( 5, 'e', 'b', embed); |
475 | CASE( 5, 'e', 'd', endif); |
476 | CASE( 5, 'e', 'r', error); |
477 | CASE( 5, 'i', 'e', ident); |
478 | CASE( 5, 'i', 'd', ifdef); |
479 | CASE( 5, 'u', 'd', undef); |
480 | |
481 | CASE( 6, 'a', 's', assert); |
482 | CASE( 6, 'd', 'f', define); |
483 | CASE( 6, 'i', 'n', ifndef); |
484 | CASE( 6, 'i', 'p', import); |
485 | CASE( 6, 'p', 'a', pragma); |
486 | |
487 | CASE( 7, 'd', 'f', defined); |
488 | CASE( 7, 'e', 'i', elifdef); |
489 | CASE( 7, 'i', 'c', include); |
490 | CASE( 7, 'w', 'r', warning); |
491 | |
492 | CASE( 8, 'e', 'i', elifndef); |
493 | CASE( 8, 'u', 'a', unassert); |
494 | CASE(12, 'i', 'c', include_next); |
495 | |
496 | CASE(14, '_', 'p', __public_macro); |
497 | |
498 | CASE(15, '_', 'p', __private_macro); |
499 | |
500 | CASE(16, '_', 'i', __include_macros); |
501 | #undef CASE |
502 | #undef HASH |
503 | } |
504 | } |
505 | |
506 | //===----------------------------------------------------------------------===// |
507 | // Stats Implementation |
508 | //===----------------------------------------------------------------------===// |
509 | |
510 | /// PrintStats - Print statistics about how well the identifier table is doing |
511 | /// at hashing identifiers. |
512 | void IdentifierTable::PrintStats() const { |
513 | unsigned NumBuckets = HashTable.getNumBuckets(); |
514 | unsigned NumIdentifiers = HashTable.getNumItems(); |
515 | unsigned NumEmptyBuckets = NumBuckets-NumIdentifiers; |
516 | unsigned AverageIdentifierSize = 0; |
517 | unsigned MaxIdentifierLength = 0; |
518 | |
519 | // TODO: Figure out maximum times an identifier had to probe for -stats. |
520 | for (llvm::StringMap<IdentifierInfo*, llvm::BumpPtrAllocator>::const_iterator |
521 | I = HashTable.begin(), E = HashTable.end(); I != E; ++I) { |
522 | unsigned IdLen = I->getKeyLength(); |
523 | AverageIdentifierSize += IdLen; |
524 | if (MaxIdentifierLength < IdLen) |
525 | MaxIdentifierLength = IdLen; |
526 | } |
527 | |
528 | fprintf(stderr, format: "\n*** Identifier Table Stats:\n" ); |
529 | fprintf(stderr, format: "# Identifiers: %d\n" , NumIdentifiers); |
530 | fprintf(stderr, format: "# Empty Buckets: %d\n" , NumEmptyBuckets); |
531 | fprintf(stderr, format: "Hash density (#identifiers per bucket): %f\n" , |
532 | NumIdentifiers/(double)NumBuckets); |
533 | fprintf(stderr, format: "Ave identifier length: %f\n" , |
534 | (AverageIdentifierSize/(double)NumIdentifiers)); |
535 | fprintf(stderr, format: "Max identifier length: %d\n" , MaxIdentifierLength); |
536 | |
537 | // Compute statistics about the memory allocated for identifiers. |
538 | HashTable.getAllocator().PrintStats(); |
539 | } |
540 | |
541 | //===----------------------------------------------------------------------===// |
542 | // SelectorTable Implementation |
543 | //===----------------------------------------------------------------------===// |
544 | |
545 | unsigned llvm::DenseMapInfo<clang::Selector>::getHashValue(clang::Selector S) { |
546 | return DenseMapInfo<void*>::getHashValue(PtrVal: S.getAsOpaquePtr()); |
547 | } |
548 | |
549 | bool Selector::isKeywordSelector(ArrayRef<StringRef> Names) const { |
550 | assert(!Names.empty() && "must have >= 1 selector slots" ); |
551 | if (getNumArgs() != Names.size()) |
552 | return false; |
553 | for (unsigned I = 0, E = Names.size(); I != E; ++I) { |
554 | if (getNameForSlot(argIndex: I) != Names[I]) |
555 | return false; |
556 | } |
557 | return true; |
558 | } |
559 | |
560 | bool Selector::isUnarySelector(StringRef Name) const { |
561 | return isUnarySelector() && getNameForSlot(argIndex: 0) == Name; |
562 | } |
563 | |
564 | unsigned Selector::getNumArgs() const { |
565 | unsigned IIF = getIdentifierInfoFlag(); |
566 | if (IIF <= ZeroArg) |
567 | return 0; |
568 | if (IIF == OneArg) |
569 | return 1; |
570 | // We point to a MultiKeywordSelector. |
571 | MultiKeywordSelector *SI = getMultiKeywordSelector(); |
572 | return SI->getNumArgs(); |
573 | } |
574 | |
575 | const IdentifierInfo * |
576 | Selector::getIdentifierInfoForSlot(unsigned argIndex) const { |
577 | if (getIdentifierInfoFlag() < MultiArg) { |
578 | assert(argIndex == 0 && "illegal keyword index" ); |
579 | return getAsIdentifierInfo(); |
580 | } |
581 | |
582 | // We point to a MultiKeywordSelector. |
583 | MultiKeywordSelector *SI = getMultiKeywordSelector(); |
584 | return SI->getIdentifierInfoForSlot(i: argIndex); |
585 | } |
586 | |
587 | StringRef Selector::getNameForSlot(unsigned int argIndex) const { |
588 | const IdentifierInfo *II = getIdentifierInfoForSlot(argIndex); |
589 | return II ? II->getName() : StringRef(); |
590 | } |
591 | |
592 | std::string MultiKeywordSelector::getName() const { |
593 | SmallString<256> Str; |
594 | llvm::raw_svector_ostream OS(Str); |
595 | for (keyword_iterator I = keyword_begin(), E = keyword_end(); I != E; ++I) { |
596 | if (*I) |
597 | OS << (*I)->getName(); |
598 | OS << ':'; |
599 | } |
600 | |
601 | return std::string(OS.str()); |
602 | } |
603 | |
604 | std::string Selector::getAsString() const { |
605 | if (isNull()) |
606 | return "<null selector>" ; |
607 | |
608 | if (getIdentifierInfoFlag() < MultiArg) { |
609 | const IdentifierInfo *II = getAsIdentifierInfo(); |
610 | |
611 | if (getNumArgs() == 0) { |
612 | assert(II && "If the number of arguments is 0 then II is guaranteed to " |
613 | "not be null." ); |
614 | return std::string(II->getName()); |
615 | } |
616 | |
617 | if (!II) |
618 | return ":" ; |
619 | |
620 | return II->getName().str() + ":" ; |
621 | } |
622 | |
623 | // We have a multiple keyword selector. |
624 | return getMultiKeywordSelector()->getName(); |
625 | } |
626 | |
627 | void Selector::print(llvm::raw_ostream &OS) const { |
628 | OS << getAsString(); |
629 | } |
630 | |
631 | LLVM_DUMP_METHOD void Selector::dump() const { print(OS&: llvm::errs()); } |
632 | |
633 | /// Interpreting the given string using the normal CamelCase |
634 | /// conventions, determine whether the given string starts with the |
635 | /// given "word", which is assumed to end in a lowercase letter. |
636 | static bool startsWithWord(StringRef name, StringRef word) { |
637 | if (name.size() < word.size()) return false; |
638 | return ((name.size() == word.size() || !isLowercase(c: name[word.size()])) && |
639 | name.starts_with(Prefix: word)); |
640 | } |
641 | |
642 | ObjCMethodFamily Selector::getMethodFamilyImpl(Selector sel) { |
643 | const IdentifierInfo *first = sel.getIdentifierInfoForSlot(argIndex: 0); |
644 | if (!first) return OMF_None; |
645 | |
646 | StringRef name = first->getName(); |
647 | if (sel.isUnarySelector()) { |
648 | if (name == "autorelease" ) return OMF_autorelease; |
649 | if (name == "dealloc" ) return OMF_dealloc; |
650 | if (name == "finalize" ) return OMF_finalize; |
651 | if (name == "release" ) return OMF_release; |
652 | if (name == "retain" ) return OMF_retain; |
653 | if (name == "retainCount" ) return OMF_retainCount; |
654 | if (name == "self" ) return OMF_self; |
655 | if (name == "initialize" ) return OMF_initialize; |
656 | } |
657 | |
658 | if (name == "performSelector" || name == "performSelectorInBackground" || |
659 | name == "performSelectorOnMainThread" ) |
660 | return OMF_performSelector; |
661 | |
662 | // The other method families may begin with a prefix of underscores. |
663 | name = name.ltrim(Char: '_'); |
664 | |
665 | if (name.empty()) return OMF_None; |
666 | switch (name.front()) { |
667 | case 'a': |
668 | if (startsWithWord(name, word: "alloc" )) return OMF_alloc; |
669 | break; |
670 | case 'c': |
671 | if (startsWithWord(name, word: "copy" )) return OMF_copy; |
672 | break; |
673 | case 'i': |
674 | if (startsWithWord(name, word: "init" )) return OMF_init; |
675 | break; |
676 | case 'm': |
677 | if (startsWithWord(name, word: "mutableCopy" )) return OMF_mutableCopy; |
678 | break; |
679 | case 'n': |
680 | if (startsWithWord(name, word: "new" )) return OMF_new; |
681 | break; |
682 | default: |
683 | break; |
684 | } |
685 | |
686 | return OMF_None; |
687 | } |
688 | |
689 | ObjCInstanceTypeFamily Selector::getInstTypeMethodFamily(Selector sel) { |
690 | const IdentifierInfo *first = sel.getIdentifierInfoForSlot(argIndex: 0); |
691 | if (!first) return OIT_None; |
692 | |
693 | StringRef name = first->getName(); |
694 | |
695 | if (name.empty()) return OIT_None; |
696 | switch (name.front()) { |
697 | case 'a': |
698 | if (startsWithWord(name, word: "array" )) return OIT_Array; |
699 | break; |
700 | case 'd': |
701 | if (startsWithWord(name, word: "default" )) return OIT_ReturnsSelf; |
702 | if (startsWithWord(name, word: "dictionary" )) return OIT_Dictionary; |
703 | break; |
704 | case 's': |
705 | if (startsWithWord(name, word: "shared" )) return OIT_ReturnsSelf; |
706 | if (startsWithWord(name, word: "standard" )) return OIT_Singleton; |
707 | break; |
708 | case 'i': |
709 | if (startsWithWord(name, word: "init" )) return OIT_Init; |
710 | break; |
711 | default: |
712 | break; |
713 | } |
714 | return OIT_None; |
715 | } |
716 | |
717 | ObjCStringFormatFamily Selector::getStringFormatFamilyImpl(Selector sel) { |
718 | const IdentifierInfo *first = sel.getIdentifierInfoForSlot(argIndex: 0); |
719 | if (!first) return SFF_None; |
720 | |
721 | StringRef name = first->getName(); |
722 | |
723 | switch (name.front()) { |
724 | case 'a': |
725 | if (name == "appendFormat" ) return SFF_NSString; |
726 | break; |
727 | |
728 | case 'i': |
729 | if (name == "initWithFormat" ) return SFF_NSString; |
730 | break; |
731 | |
732 | case 'l': |
733 | if (name == "localizedStringWithFormat" ) return SFF_NSString; |
734 | break; |
735 | |
736 | case 's': |
737 | if (name == "stringByAppendingFormat" || |
738 | name == "stringWithFormat" ) return SFF_NSString; |
739 | break; |
740 | } |
741 | return SFF_None; |
742 | } |
743 | |
744 | namespace { |
745 | |
746 | struct SelectorTableImpl { |
747 | llvm::FoldingSet<MultiKeywordSelector> Table; |
748 | llvm::BumpPtrAllocator Allocator; |
749 | }; |
750 | |
751 | } // namespace |
752 | |
753 | static SelectorTableImpl &getSelectorTableImpl(void *P) { |
754 | return *static_cast<SelectorTableImpl*>(P); |
755 | } |
756 | |
757 | SmallString<64> |
758 | SelectorTable::constructSetterName(StringRef Name) { |
759 | SmallString<64> SetterName("set" ); |
760 | SetterName += Name; |
761 | SetterName[3] = toUppercase(c: SetterName[3]); |
762 | return SetterName; |
763 | } |
764 | |
765 | Selector |
766 | SelectorTable::constructSetterSelector(IdentifierTable &Idents, |
767 | SelectorTable &SelTable, |
768 | const IdentifierInfo *Name) { |
769 | IdentifierInfo *SetterName = |
770 | &Idents.get(Name: constructSetterName(Name: Name->getName())); |
771 | return SelTable.getUnarySelector(ID: SetterName); |
772 | } |
773 | |
774 | std::string SelectorTable::getPropertyNameFromSetterSelector(Selector Sel) { |
775 | StringRef Name = Sel.getNameForSlot(argIndex: 0); |
776 | assert(Name.starts_with("set" ) && "invalid setter name" ); |
777 | return (Twine(toLowercase(c: Name[3])) + Name.drop_front(N: 4)).str(); |
778 | } |
779 | |
780 | size_t SelectorTable::getTotalMemory() const { |
781 | SelectorTableImpl &SelTabImpl = getSelectorTableImpl(P: Impl); |
782 | return SelTabImpl.Allocator.getTotalMemory(); |
783 | } |
784 | |
785 | Selector SelectorTable::getSelector(unsigned nKeys, |
786 | const IdentifierInfo **IIV) { |
787 | if (nKeys < 2) |
788 | return Selector(IIV[0], nKeys); |
789 | |
790 | SelectorTableImpl &SelTabImpl = getSelectorTableImpl(P: Impl); |
791 | |
792 | // Unique selector, to guarantee there is one per name. |
793 | llvm::FoldingSetNodeID ID; |
794 | MultiKeywordSelector::Profile(ID, ArgTys: IIV, NumArgs: nKeys); |
795 | |
796 | void *InsertPos = nullptr; |
797 | if (MultiKeywordSelector *SI = |
798 | SelTabImpl.Table.FindNodeOrInsertPos(ID, InsertPos)) |
799 | return Selector(SI); |
800 | |
801 | // MultiKeywordSelector objects are not allocated with new because they have a |
802 | // variable size array (for parameter types) at the end of them. |
803 | unsigned Size = sizeof(MultiKeywordSelector) + nKeys*sizeof(IdentifierInfo *); |
804 | MultiKeywordSelector *SI = |
805 | (MultiKeywordSelector *)SelTabImpl.Allocator.Allocate( |
806 | Size, Alignment: alignof(MultiKeywordSelector)); |
807 | new (SI) MultiKeywordSelector(nKeys, IIV); |
808 | SelTabImpl.Table.InsertNode(N: SI, InsertPos); |
809 | return Selector(SI); |
810 | } |
811 | |
812 | SelectorTable::SelectorTable() { |
813 | Impl = new SelectorTableImpl(); |
814 | } |
815 | |
816 | SelectorTable::~SelectorTable() { |
817 | delete &getSelectorTableImpl(P: Impl); |
818 | } |
819 | |
820 | const char *clang::getOperatorSpelling(OverloadedOperatorKind Operator) { |
821 | switch (Operator) { |
822 | case OO_None: |
823 | case NUM_OVERLOADED_OPERATORS: |
824 | return nullptr; |
825 | |
826 | #define OVERLOADED_OPERATOR(Name,Spelling,Token,Unary,Binary,MemberOnly) \ |
827 | case OO_##Name: return Spelling; |
828 | #include "clang/Basic/OperatorKinds.def" |
829 | } |
830 | |
831 | llvm_unreachable("Invalid OverloadedOperatorKind!" ); |
832 | } |
833 | |
834 | StringRef clang::getNullabilitySpelling(NullabilityKind kind, |
835 | bool isContextSensitive) { |
836 | switch (kind) { |
837 | case NullabilityKind::NonNull: |
838 | return isContextSensitive ? "nonnull" : "_Nonnull" ; |
839 | |
840 | case NullabilityKind::Nullable: |
841 | return isContextSensitive ? "nullable" : "_Nullable" ; |
842 | |
843 | case NullabilityKind::NullableResult: |
844 | assert(!isContextSensitive && |
845 | "_Nullable_result isn't supported as context-sensitive keyword" ); |
846 | return "_Nullable_result" ; |
847 | |
848 | case NullabilityKind::Unspecified: |
849 | return isContextSensitive ? "null_unspecified" : "_Null_unspecified" ; |
850 | } |
851 | llvm_unreachable("Unknown nullability kind." ); |
852 | } |
853 | |
854 | llvm::raw_ostream &clang::operator<<(llvm::raw_ostream &OS, |
855 | NullabilityKind NK) { |
856 | switch (NK) { |
857 | case NullabilityKind::NonNull: |
858 | return OS << "NonNull" ; |
859 | case NullabilityKind::Nullable: |
860 | return OS << "Nullable" ; |
861 | case NullabilityKind::NullableResult: |
862 | return OS << "NullableResult" ; |
863 | case NullabilityKind::Unspecified: |
864 | return OS << "Unspecified" ; |
865 | } |
866 | llvm_unreachable("Unknown nullability kind." ); |
867 | } |
868 | |
869 | diag::kind |
870 | IdentifierTable::getFutureCompatDiagKind(const IdentifierInfo &II, |
871 | const LangOptions &LangOpts) { |
872 | assert(II.isFutureCompatKeyword() && "diagnostic should not be needed" ); |
873 | |
874 | unsigned Flags = llvm::StringSwitch<unsigned>(II.getName()) |
875 | #define KEYWORD(NAME, FLAGS) .Case(#NAME, FLAGS) |
876 | #include "clang/Basic/TokenKinds.def" |
877 | #undef KEYWORD |
878 | ; |
879 | |
880 | if (LangOpts.CPlusPlus) { |
881 | if ((Flags & KEYCXX11) == KEYCXX11) |
882 | return diag::warn_cxx11_keyword; |
883 | |
884 | // char8_t is not modeled as a CXX20_KEYWORD because it's not |
885 | // unconditionally enabled in C++20 mode. (It can be disabled |
886 | // by -fno-char8_t.) |
887 | if (((Flags & KEYCXX20) == KEYCXX20) || |
888 | ((Flags & CHAR8SUPPORT) == CHAR8SUPPORT)) |
889 | return diag::warn_cxx20_keyword; |
890 | } else { |
891 | if ((Flags & KEYC99) == KEYC99) |
892 | return diag::warn_c99_keyword; |
893 | if ((Flags & KEYC23) == KEYC23) |
894 | return diag::warn_c23_keyword; |
895 | } |
896 | |
897 | llvm_unreachable( |
898 | "Keyword not known to come from a newer Standard or proposed Standard" ); |
899 | } |
900 | |