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