1//===- Lexer.h - C Language Family Lexer ------------------------*- C++ -*-===//
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 defines the Lexer interface.
10//
11//===----------------------------------------------------------------------===//
12
13#ifndef LLVM_CLANG_LEX_LEXER_H
14#define LLVM_CLANG_LEX_LEXER_H
15
16#include "clang/Basic/LangOptions.h"
17#include "clang/Basic/SourceLocation.h"
18#include "clang/Basic/TokenKinds.h"
19#include "clang/Lex/PreprocessorLexer.h"
20#include "clang/Lex/Token.h"
21#include "llvm/ADT/Optional.h"
22#include "llvm/ADT/SmallVector.h"
23#include "llvm/ADT/StringRef.h"
24#include <cassert>
25#include <cstdint>
26#include <string>
27
28namespace llvm {
29
30class MemoryBuffer;
31
32} // namespace llvm
33
34namespace clang {
35
36class DiagnosticBuilder;
37class Preprocessor;
38class SourceManager;
39
40/// ConflictMarkerKind - Kinds of conflict marker which the lexer might be
41/// recovering from.
42enum ConflictMarkerKind {
43 /// Not within a conflict marker.
44 CMK_None,
45
46 /// A normal or diff3 conflict marker, initiated by at least 7 "<"s,
47 /// separated by at least 7 "="s or "|"s, and terminated by at least 7 ">"s.
48 CMK_Normal,
49
50 /// A Perforce-style conflict marker, initiated by 4 ">"s,
51 /// separated by 4 "="s, and terminated by 4 "<"s.
52 CMK_Perforce
53};
54
55/// Describes the bounds (start, size) of the preamble and a flag required by
56/// PreprocessorOptions::PrecompiledPreambleBytes.
57/// The preamble includes the BOM, if any.
58struct PreambleBounds {
59 /// Size of the preamble in bytes.
60 unsigned Size;
61
62 /// Whether the preamble ends at the start of a new line.
63 ///
64 /// Used to inform the lexer as to whether it's starting at the beginning of
65 /// a line after skipping the preamble.
66 bool PreambleEndsAtStartOfLine;
67
68 PreambleBounds(unsigned Size, bool PreambleEndsAtStartOfLine)
69 : Size(Size), PreambleEndsAtStartOfLine(PreambleEndsAtStartOfLine) {}
70};
71
72/// Lexer - This provides a simple interface that turns a text buffer into a
73/// stream of tokens. This provides no support for file reading or buffering,
74/// or buffering/seeking of tokens, only forward lexing is supported. It relies
75/// on the specified Preprocessor object to handle preprocessor directives, etc.
76class Lexer : public PreprocessorLexer {
77 friend class Preprocessor;
78
79 void anchor() override;
80
81 //===--------------------------------------------------------------------===//
82 // Constant configuration values for this lexer.
83
84 // Start of the buffer.
85 const char *BufferStart;
86
87 // End of the buffer.
88 const char *BufferEnd;
89
90 // Location for start of file.
91 SourceLocation FileLoc;
92
93 // LangOpts enabled by this language (cache).
94 LangOptions LangOpts;
95
96 // True if lexer for _Pragma handling.
97 bool Is_PragmaLexer;
98
99 //===--------------------------------------------------------------------===//
100 // Context-specific lexing flags set by the preprocessor.
101 //
102
103 /// ExtendedTokenMode - The lexer can optionally keep comments and whitespace
104 /// and return them as tokens. This is used for -C and -CC modes, and
105 /// whitespace preservation can be useful for some clients that want to lex
106 /// the file in raw mode and get every character from the file.
107 ///
108 /// When this is set to 2 it returns comments and whitespace. When set to 1
109 /// it returns comments, when it is set to 0 it returns normal tokens only.
110 unsigned char ExtendedTokenMode;
111
112 //===--------------------------------------------------------------------===//
113 // Context that changes as the file is lexed.
114 // NOTE: any state that mutates when in raw mode must have save/restore code
115 // in Lexer::isNextPPTokenLParen.
116
117 // BufferPtr - Current pointer into the buffer. This is the next character
118 // to be lexed.
119 const char *BufferPtr;
120
121 // IsAtStartOfLine - True if the next lexed token should get the "start of
122 // line" flag set on it.
123 bool IsAtStartOfLine;
124
125 bool IsAtPhysicalStartOfLine;
126
127 bool HasLeadingSpace;
128
129 bool HasLeadingEmptyMacro;
130
131 // CurrentConflictMarkerState - The kind of conflict marker we are handling.
132 ConflictMarkerKind CurrentConflictMarkerState;
133
134 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd);
135
136public:
137 /// Lexer constructor - Create a new lexer object for the specified buffer
138 /// with the specified preprocessor managing the lexing process. This lexer
139 /// assumes that the associated file buffer and Preprocessor objects will
140 /// outlive it, so it doesn't take ownership of either of them.
141 Lexer(FileID FID, const llvm::MemoryBuffer *InputFile, Preprocessor &PP);
142
143 /// Lexer constructor - Create a new raw lexer object. This object is only
144 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
145 /// text range will outlive it, so it doesn't take ownership of it.
146 Lexer(SourceLocation FileLoc, const LangOptions &LangOpts,
147 const char *BufStart, const char *BufPtr, const char *BufEnd);
148
149 /// Lexer constructor - Create a new raw lexer object. This object is only
150 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
151 /// text range will outlive it, so it doesn't take ownership of it.
152 Lexer(FileID FID, const llvm::MemoryBuffer *FromFile,
153 const SourceManager &SM, const LangOptions &LangOpts);
154
155 Lexer(const Lexer &) = delete;
156 Lexer &operator=(const Lexer &) = delete;
157
158 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for
159 /// _Pragma expansion. This has a variety of magic semantics that this method
160 /// sets up. It returns a new'd Lexer that must be delete'd when done.
161 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc,
162 SourceLocation ExpansionLocStart,
163 SourceLocation ExpansionLocEnd,
164 unsigned TokLen, Preprocessor &PP);
165
166 /// getLangOpts - Return the language features currently enabled.
167 /// NOTE: this lexer modifies features as a file is parsed!
168 const LangOptions &getLangOpts() const { return LangOpts; }
169
170 /// getFileLoc - Return the File Location for the file we are lexing out of.
171 /// The physical location encodes the location where the characters come from,
172 /// the virtual location encodes where we should *claim* the characters came
173 /// from. Currently this is only used by _Pragma handling.
174 SourceLocation getFileLoc() const { return FileLoc; }
175
176private:
177 /// Lex - Return the next token in the file. If this is the end of file, it
178 /// return the tok::eof token. This implicitly involves the preprocessor.
179 bool Lex(Token &Result);
180
181public:
182 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma.
183 bool isPragmaLexer() const { return Is_PragmaLexer; }
184
185private:
186 /// IndirectLex - An indirect call to 'Lex' that can be invoked via
187 /// the PreprocessorLexer interface.
188 void IndirectLex(Token &Result) override { Lex(Result); }
189
190public:
191 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no
192 /// associated preprocessor object. Return true if the 'next character to
193 /// read' pointer points at the end of the lexer buffer, false otherwise.
194 bool LexFromRawLexer(Token &Result) {
195 assert(LexingRawMode && "Not already in raw mode!");
196 Lex(Result);
197 // Note that lexing to the end of the buffer doesn't implicitly delete the
198 // lexer when in raw mode.
199 return BufferPtr == BufferEnd;
200 }
201
202 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for
203 /// every character in the file, including whitespace and comments. This
204 /// should only be used in raw mode, as the preprocessor is not prepared to
205 /// deal with the excess tokens.
206 bool isKeepWhitespaceMode() const {
207 return ExtendedTokenMode > 1;
208 }
209
210 /// SetKeepWhitespaceMode - This method lets clients enable or disable
211 /// whitespace retention mode.
212 void SetKeepWhitespaceMode(bool Val) {
213 assert((!Val || LexingRawMode || LangOpts.TraditionalCPP) &&
214 "Can only retain whitespace in raw mode or -traditional-cpp");
215 ExtendedTokenMode = Val ? 2 : 0;
216 }
217
218 /// inKeepCommentMode - Return true if the lexer should return comments as
219 /// tokens.
220 bool inKeepCommentMode() const {
221 return ExtendedTokenMode > 0;
222 }
223
224 /// SetCommentRetentionMode - Change the comment retention mode of the lexer
225 /// to the specified mode. This is really only useful when lexing in raw
226 /// mode, because otherwise the lexer needs to manage this.
227 void SetCommentRetentionState(bool Mode) {
228 assert(!isKeepWhitespaceMode() &&
229 "Can't play with comment retention state when retaining whitespace");
230 ExtendedTokenMode = Mode ? 1 : 0;
231 }
232
233 /// Sets the extended token mode back to its initial value, according to the
234 /// language options and preprocessor. This controls whether the lexer
235 /// produces comment and whitespace tokens.
236 ///
237 /// This requires the lexer to have an associated preprocessor. A standalone
238 /// lexer has nothing to reset to.
239 void resetExtendedTokenMode();
240
241 /// Gets source code buffer.
242 StringRef getBuffer() const {
243 return StringRef(BufferStart, BufferEnd - BufferStart);
244 }
245
246 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an
247 /// uninterpreted string. This switches the lexer out of directive mode.
248 void ReadToEndOfLine(SmallVectorImpl<char> *Result = nullptr);
249
250
251 /// Diag - Forwarding function for diagnostics. This translate a source
252 /// position in the current buffer into a SourceLocation object for rendering.
253 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const;
254
255 /// getSourceLocation - Return a source location identifier for the specified
256 /// offset in the current file.
257 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const;
258
259 /// getSourceLocation - Return a source location for the next character in
260 /// the current file.
261 SourceLocation getSourceLocation() override {
262 return getSourceLocation(BufferPtr);
263 }
264
265 /// Return the current location in the buffer.
266 const char *getBufferLocation() const { return BufferPtr; }
267
268 /// Stringify - Convert the specified string into a C string by i) escaping
269 /// '\\' and " characters and ii) replacing newline character(s) with "\\n".
270 /// If Charify is true, this escapes the ' character instead of ".
271 static std::string Stringify(StringRef Str, bool Charify = false);
272
273 /// Stringify - Convert the specified string into a C string by i) escaping
274 /// '\\' and " characters and ii) replacing newline character(s) with "\\n".
275 static void Stringify(SmallVectorImpl<char> &Str);
276
277 /// getSpelling - This method is used to get the spelling of a token into a
278 /// preallocated buffer, instead of as an std::string. The caller is required
279 /// to allocate enough space for the token, which is guaranteed to be at least
280 /// Tok.getLength() bytes long. The length of the actual result is returned.
281 ///
282 /// Note that this method may do two possible things: it may either fill in
283 /// the buffer specified with characters, or it may *change the input pointer*
284 /// to point to a constant buffer with the data already in it (avoiding a
285 /// copy). The caller is not allowed to modify the returned buffer pointer
286 /// if an internal buffer is returned.
287 static unsigned getSpelling(const Token &Tok, const char *&Buffer,
288 const SourceManager &SourceMgr,
289 const LangOptions &LangOpts,
290 bool *Invalid = nullptr);
291
292 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a
293 /// token is the characters used to represent the token in the source file
294 /// after trigraph expansion and escaped-newline folding. In particular, this
295 /// wants to get the true, uncanonicalized, spelling of things like digraphs
296 /// UCNs, etc.
297 static std::string getSpelling(const Token &Tok,
298 const SourceManager &SourceMgr,
299 const LangOptions &LangOpts,
300 bool *Invalid = nullptr);
301
302 /// getSpelling - This method is used to get the spelling of the
303 /// token at the given source location. If, as is usually true, it
304 /// is not necessary to copy any data, then the returned string may
305 /// not point into the provided buffer.
306 ///
307 /// This method lexes at the expansion depth of the given
308 /// location and does not jump to the expansion or spelling
309 /// location.
310 static StringRef getSpelling(SourceLocation loc,
311 SmallVectorImpl<char> &buffer,
312 const SourceManager &SM,
313 const LangOptions &options,
314 bool *invalid = nullptr);
315
316 /// MeasureTokenLength - Relex the token at the specified location and return
317 /// its length in bytes in the input file. If the token needs cleaning (e.g.
318 /// includes a trigraph or an escaped newline) then this count includes bytes
319 /// that are part of that.
320 static unsigned MeasureTokenLength(SourceLocation Loc,
321 const SourceManager &SM,
322 const LangOptions &LangOpts);
323
324 /// Relex the token at the specified location.
325 /// \returns true if there was a failure, false on success.
326 static bool getRawToken(SourceLocation Loc, Token &Result,
327 const SourceManager &SM,
328 const LangOptions &LangOpts,
329 bool IgnoreWhiteSpace = false);
330
331 /// Given a location any where in a source buffer, find the location
332 /// that corresponds to the beginning of the token in which the original
333 /// source location lands.
334 static SourceLocation GetBeginningOfToken(SourceLocation Loc,
335 const SourceManager &SM,
336 const LangOptions &LangOpts);
337
338 /// Get the physical length (including trigraphs and escaped newlines) of the
339 /// first \p Characters characters of the token starting at TokStart.
340 static unsigned getTokenPrefixLength(SourceLocation TokStart,
341 unsigned CharNo,
342 const SourceManager &SM,
343 const LangOptions &LangOpts);
344
345 /// AdvanceToTokenCharacter - If the current SourceLocation specifies a
346 /// location at the start of a token, return a new location that specifies a
347 /// character within the token. This handles trigraphs and escaped newlines.
348 static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
349 unsigned Characters,
350 const SourceManager &SM,
351 const LangOptions &LangOpts) {
352 return TokStart.getLocWithOffset(
353 getTokenPrefixLength(TokStart, Characters, SM, LangOpts));
354 }
355
356 /// Computes the source location just past the end of the
357 /// token at this source location.
358 ///
359 /// This routine can be used to produce a source location that
360 /// points just past the end of the token referenced by \p Loc, and
361 /// is generally used when a diagnostic needs to point just after a
362 /// token where it expected something different that it received. If
363 /// the returned source location would not be meaningful (e.g., if
364 /// it points into a macro), this routine returns an invalid
365 /// source location.
366 ///
367 /// \param Offset an offset from the end of the token, where the source
368 /// location should refer to. The default offset (0) produces a source
369 /// location pointing just past the end of the token; an offset of 1 produces
370 /// a source location pointing to the last character in the token, etc.
371 static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset,
372 const SourceManager &SM,
373 const LangOptions &LangOpts);
374
375 /// Given a token range, produce a corresponding CharSourceRange that
376 /// is not a token range. This allows the source range to be used by
377 /// components that don't have access to the lexer and thus can't find the
378 /// end of the range for themselves.
379 static CharSourceRange getAsCharRange(SourceRange Range,
380 const SourceManager &SM,
381 const LangOptions &LangOpts) {
382 SourceLocation End = getLocForEndOfToken(Range.getEnd(), 0, SM, LangOpts);
383 return End.isInvalid() ? CharSourceRange()
384 : CharSourceRange::getCharRange(
385 Range.getBegin(), End.getLocWithOffset(-1));
386 }
387 static CharSourceRange getAsCharRange(CharSourceRange Range,
388 const SourceManager &SM,
389 const LangOptions &LangOpts) {
390 return Range.isTokenRange()
391 ? getAsCharRange(Range.getAsRange(), SM, LangOpts)
392 : Range;
393 }
394
395 /// Returns true if the given MacroID location points at the first
396 /// token of the macro expansion.
397 ///
398 /// \param MacroBegin If non-null and function returns true, it is set to
399 /// begin location of the macro.
400 static bool isAtStartOfMacroExpansion(SourceLocation loc,
401 const SourceManager &SM,
402 const LangOptions &LangOpts,
403 SourceLocation *MacroBegin = nullptr);
404
405 /// Returns true if the given MacroID location points at the last
406 /// token of the macro expansion.
407 ///
408 /// \param MacroEnd If non-null and function returns true, it is set to
409 /// end location of the macro.
410 static bool isAtEndOfMacroExpansion(SourceLocation loc,
411 const SourceManager &SM,
412 const LangOptions &LangOpts,
413 SourceLocation *MacroEnd = nullptr);
414
415 /// Accepts a range and returns a character range with file locations.
416 ///
417 /// Returns a null range if a part of the range resides inside a macro
418 /// expansion or the range does not reside on the same FileID.
419 ///
420 /// This function is trying to deal with macros and return a range based on
421 /// file locations. The cases where it can successfully handle macros are:
422 ///
423 /// -begin or end range lies at the start or end of a macro expansion, in
424 /// which case the location will be set to the expansion point, e.g:
425 /// \#define M 1 2
426 /// a M
427 /// If you have a range [a, 2] (where 2 came from the macro), the function
428 /// will return a range for "a M"
429 /// if you have range [a, 1], the function will fail because the range
430 /// overlaps with only a part of the macro
431 ///
432 /// -The macro is a function macro and the range can be mapped to the macro
433 /// arguments, e.g:
434 /// \#define M 1 2
435 /// \#define FM(x) x
436 /// FM(a b M)
437 /// if you have range [b, 2], the function will return the file range "b M"
438 /// inside the macro arguments.
439 /// if you have range [a, 2], the function will return the file range
440 /// "FM(a b M)" since the range includes all of the macro expansion.
441 static CharSourceRange makeFileCharRange(CharSourceRange Range,
442 const SourceManager &SM,
443 const LangOptions &LangOpts);
444
445 /// Returns a string for the source that the range encompasses.
446 static StringRef getSourceText(CharSourceRange Range,
447 const SourceManager &SM,
448 const LangOptions &LangOpts,
449 bool *Invalid = nullptr);
450
451 /// Retrieve the name of the immediate macro expansion.
452 ///
453 /// This routine starts from a source location, and finds the name of the macro
454 /// responsible for its immediate expansion. It looks through any intervening
455 /// macro argument expansions to compute this. It returns a StringRef which
456 /// refers to the SourceManager-owned buffer of the source where that macro
457 /// name is spelled. Thus, the result shouldn't out-live that SourceManager.
458 static StringRef getImmediateMacroName(SourceLocation Loc,
459 const SourceManager &SM,
460 const LangOptions &LangOpts);
461
462 /// Retrieve the name of the immediate macro expansion.
463 ///
464 /// This routine starts from a source location, and finds the name of the
465 /// macro responsible for its immediate expansion. It looks through any
466 /// intervening macro argument expansions to compute this. It returns a
467 /// StringRef which refers to the SourceManager-owned buffer of the source
468 /// where that macro name is spelled. Thus, the result shouldn't out-live
469 /// that SourceManager.
470 ///
471 /// This differs from Lexer::getImmediateMacroName in that any macro argument
472 /// location will result in the topmost function macro that accepted it.
473 /// e.g.
474 /// \code
475 /// MAC1( MAC2(foo) )
476 /// \endcode
477 /// for location of 'foo' token, this function will return "MAC1" while
478 /// Lexer::getImmediateMacroName will return "MAC2".
479 static StringRef getImmediateMacroNameForDiagnostics(
480 SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts);
481
482 /// Compute the preamble of the given file.
483 ///
484 /// The preamble of a file contains the initial comments, include directives,
485 /// and other preprocessor directives that occur before the code in this
486 /// particular file actually begins. The preamble of the main source file is
487 /// a potential prefix header.
488 ///
489 /// \param Buffer The memory buffer containing the file's contents.
490 ///
491 /// \param MaxLines If non-zero, restrict the length of the preamble
492 /// to fewer than this number of lines.
493 ///
494 /// \returns The offset into the file where the preamble ends and the rest
495 /// of the file begins along with a boolean value indicating whether
496 /// the preamble ends at the beginning of a new line.
497 static PreambleBounds ComputePreamble(StringRef Buffer,
498 const LangOptions &LangOpts,
499 unsigned MaxLines = 0);
500
501 /// Finds the token that comes right after the given location.
502 ///
503 /// Returns the next token, or none if the location is inside a macro.
504 static Optional<Token> findNextToken(SourceLocation Loc,
505 const SourceManager &SM,
506 const LangOptions &LangOpts);
507
508 /// Checks that the given token is the first token that occurs after
509 /// the given location (this excludes comments and whitespace). Returns the
510 /// location immediately after the specified token. If the token is not found
511 /// or the location is inside a macro, the returned source location will be
512 /// invalid.
513 static SourceLocation findLocationAfterToken(SourceLocation loc,
514 tok::TokenKind TKind,
515 const SourceManager &SM,
516 const LangOptions &LangOpts,
517 bool SkipTrailingWhitespaceAndNewLine);
518
519 /// Returns true if the given character could appear in an identifier.
520 static bool isIdentifierBodyChar(char c, const LangOptions &LangOpts);
521
522 /// Checks whether new line pointed by Str is preceded by escape
523 /// sequence.
524 static bool isNewLineEscaped(const char *BufferStart, const char *Str);
525
526 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever
527 /// emit a warning.
528 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size,
529 const LangOptions &LangOpts) {
530 // If this is not a trigraph and not a UCN or escaped newline, return
531 // quickly.
532 if (isObviouslySimpleCharacter(Ptr[0])) {
533 Size = 1;
534 return *Ptr;
535 }
536
537 Size = 0;
538 return getCharAndSizeSlowNoWarn(Ptr, Size, LangOpts);
539 }
540
541 /// Returns the leading whitespace for line that corresponds to the given
542 /// location \p Loc.
543 static StringRef getIndentationForLine(SourceLocation Loc,
544 const SourceManager &SM);
545
546private:
547 //===--------------------------------------------------------------------===//
548 // Internal implementation interfaces.
549
550 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called
551 /// by Lex.
552 ///
553 bool LexTokenInternal(Token &Result, bool TokAtPhysicalStartOfLine);
554
555 bool CheckUnicodeWhitespace(Token &Result, uint32_t C, const char *CurPtr);
556
557 /// Given that a token begins with the Unicode character \p C, figure out
558 /// what kind of token it is and dispatch to the appropriate lexing helper
559 /// function.
560 bool LexUnicode(Token &Result, uint32_t C, const char *CurPtr);
561
562 /// FormTokenWithChars - When we lex a token, we have identified a span
563 /// starting at BufferPtr, going to TokEnd that forms the token. This method
564 /// takes that range and assigns it to the token as its location and size. In
565 /// addition, since tokens cannot overlap, this also updates BufferPtr to be
566 /// TokEnd.
567 void FormTokenWithChars(Token &Result, const char *TokEnd,
568 tok::TokenKind Kind) {
569 unsigned TokLen = TokEnd-BufferPtr;
570 Result.setLength(TokLen);
571 Result.setLocation(getSourceLocation(BufferPtr, TokLen));
572 Result.setKind(Kind);
573 BufferPtr = TokEnd;
574 }
575
576 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a
577 /// tok::l_paren token, 0 if it is something else and 2 if there are no more
578 /// tokens in the buffer controlled by this lexer.
579 unsigned isNextPPTokenLParen();
580
581 //===--------------------------------------------------------------------===//
582 // Lexer character reading interfaces.
583
584 // This lexer is built on two interfaces for reading characters, both of which
585 // automatically provide phase 1/2 translation. getAndAdvanceChar is used
586 // when we know that we will be reading a character from the input buffer and
587 // that this character will be part of the result token. This occurs in (f.e.)
588 // string processing, because we know we need to read until we find the
589 // closing '"' character.
590 //
591 // The second interface is the combination of getCharAndSize with
592 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character,
593 // returning it and its size. If the lexer decides that this character is
594 // part of the current token, it calls ConsumeChar on it. This two stage
595 // approach allows us to emit diagnostics for characters (e.g. warnings about
596 // trigraphs), knowing that they only are emitted if the character is
597 // consumed.
598
599 /// isObviouslySimpleCharacter - Return true if the specified character is
600 /// obviously the same in translation phase 1 and translation phase 3. This
601 /// can return false for characters that end up being the same, but it will
602 /// never return true for something that needs to be mapped.
603 static bool isObviouslySimpleCharacter(char C) {
604 return C != '?' && C != '\\';
605 }
606
607 /// getAndAdvanceChar - Read a single 'character' from the specified buffer,
608 /// advance over it, and return it. This is tricky in several cases. Here we
609 /// just handle the trivial case and fall-back to the non-inlined
610 /// getCharAndSizeSlow method to handle the hard case.
611 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) {
612 // If this is not a trigraph and not a UCN or escaped newline, return
613 // quickly.
614 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++;
615
616 unsigned Size = 0;
617 char C = getCharAndSizeSlow(Ptr, Size, &Tok);
618 Ptr += Size;
619 return C;
620 }
621
622 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed
623 /// and added to a given token, check to see if there are diagnostics that
624 /// need to be emitted or flags that need to be set on the token. If so, do
625 /// it.
626 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) {
627 // Normal case, we consumed exactly one token. Just return it.
628 if (Size == 1)
629 return Ptr+Size;
630
631 // Otherwise, re-lex the character with a current token, allowing
632 // diagnostics to be emitted and flags to be set.
633 Size = 0;
634 getCharAndSizeSlow(Ptr, Size, &Tok);
635 return Ptr+Size;
636 }
637
638 /// getCharAndSize - Peek a single 'character' from the specified buffer,
639 /// get its size, and return it. This is tricky in several cases. Here we
640 /// just handle the trivial case and fall-back to the non-inlined
641 /// getCharAndSizeSlow method to handle the hard case.
642 inline char getCharAndSize(const char *Ptr, unsigned &Size) {
643 // If this is not a trigraph and not a UCN or escaped newline, return
644 // quickly.
645 if (isObviouslySimpleCharacter(Ptr[0])) {
646 Size = 1;
647 return *Ptr;
648 }
649
650 Size = 0;
651 return getCharAndSizeSlow(Ptr, Size);
652 }
653
654 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize
655 /// method.
656 char getCharAndSizeSlow(const char *Ptr, unsigned &Size,
657 Token *Tok = nullptr);
658
659 /// getEscapedNewLineSize - Return the size of the specified escaped newline,
660 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry
661 /// to this function.
662 static unsigned getEscapedNewLineSize(const char *P);
663
664 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of
665 /// them), skip over them and return the first non-escaped-newline found,
666 /// otherwise return P.
667 static const char *SkipEscapedNewLines(const char *P);
668
669 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a
670 /// diagnostic.
671 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size,
672 const LangOptions &LangOpts);
673
674 //===--------------------------------------------------------------------===//
675 // Other lexer functions.
676
677 void SetByteOffset(unsigned Offset, bool StartOfLine);
678
679 void PropagateLineStartLeadingSpaceInfo(Token &Result);
680
681 const char *LexUDSuffix(Token &Result, const char *CurPtr,
682 bool IsStringLiteral);
683
684 // Helper functions to lex the remainder of a token of the specific type.
685 bool LexIdentifier (Token &Result, const char *CurPtr);
686 bool LexNumericConstant (Token &Result, const char *CurPtr);
687 bool LexStringLiteral (Token &Result, const char *CurPtr,
688 tok::TokenKind Kind);
689 bool LexRawStringLiteral (Token &Result, const char *CurPtr,
690 tok::TokenKind Kind);
691 bool LexAngledStringLiteral(Token &Result, const char *CurPtr);
692 bool LexCharConstant (Token &Result, const char *CurPtr,
693 tok::TokenKind Kind);
694 bool LexEndOfFile (Token &Result, const char *CurPtr);
695 bool SkipWhitespace (Token &Result, const char *CurPtr,
696 bool &TokAtPhysicalStartOfLine);
697 bool SkipLineComment (Token &Result, const char *CurPtr,
698 bool &TokAtPhysicalStartOfLine);
699 bool SkipBlockComment (Token &Result, const char *CurPtr,
700 bool &TokAtPhysicalStartOfLine);
701 bool SaveLineComment (Token &Result, const char *CurPtr);
702
703 bool IsStartOfConflictMarker(const char *CurPtr);
704 bool HandleEndOfConflictMarker(const char *CurPtr);
705
706 bool lexEditorPlaceholder(Token &Result, const char *CurPtr);
707
708 bool isCodeCompletionPoint(const char *CurPtr) const;
709 void cutOffLexing() { BufferPtr = BufferEnd; }
710
711 bool isHexaLiteral(const char *Start, const LangOptions &LangOpts);
712
713 void codeCompleteIncludedFile(const char *PathStart,
714 const char *CompletionPoint, bool IsAngled);
715
716 /// Read a universal character name.
717 ///
718 /// \param StartPtr The position in the source buffer after the initial '\'.
719 /// If the UCN is syntactically well-formed (but not
720 /// necessarily valid), this parameter will be updated to
721 /// point to the character after the UCN.
722 /// \param SlashLoc The position in the source buffer of the '\'.
723 /// \param Result The token being formed. Pass \c nullptr to suppress
724 /// diagnostics and handle token formation in the caller.
725 ///
726 /// \return The Unicode codepoint specified by the UCN, or 0 if the UCN is
727 /// invalid.
728 uint32_t tryReadUCN(const char *&StartPtr, const char *SlashLoc, Token *Result);
729
730 /// Try to consume a UCN as part of an identifier at the current
731 /// location.
732 /// \param CurPtr Initially points to the range of characters in the source
733 /// buffer containing the '\'. Updated to point past the end of
734 /// the UCN on success.
735 /// \param Size The number of characters occupied by the '\' (including
736 /// trigraphs and escaped newlines).
737 /// \param Result The token being produced. Marked as containing a UCN on
738 /// success.
739 /// \return \c true if a UCN was lexed and it produced an acceptable
740 /// identifier character, \c false otherwise.
741 bool tryConsumeIdentifierUCN(const char *&CurPtr, unsigned Size,
742 Token &Result);
743
744 /// Try to consume an identifier character encoded in UTF-8.
745 /// \param CurPtr Points to the start of the (potential) UTF-8 code unit
746 /// sequence. On success, updated to point past the end of it.
747 /// \return \c true if a UTF-8 sequence mapping to an acceptable identifier
748 /// character was lexed, \c false otherwise.
749 bool tryConsumeIdentifierUTF8Char(const char *&CurPtr);
750};
751
752} // namespace clang
753
754#endif // LLVM_CLANG_LEX_LEXER_H
755