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 MemoryBufferRef;
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 // NewLinePtr - A pointer to new line character '\n' being lexed. For '\r\n',
132 // it also points to '\n.'
133 const char *NewLinePtr;
134
135 // CurrentConflictMarkerState - The kind of conflict marker we are handling.
136 ConflictMarkerKind CurrentConflictMarkerState;
137
138 void InitLexer(const char *BufStart, const char *BufPtr, const char *BufEnd);
139
140public:
141 /// Lexer constructor - Create a new lexer object for the specified buffer
142 /// with the specified preprocessor managing the lexing process. This lexer
143 /// assumes that the associated file buffer and Preprocessor objects will
144 /// outlive it, so it doesn't take ownership of either of them.
145 Lexer(FileID FID, const llvm::MemoryBufferRef &InputFile, Preprocessor &PP);
146
147 /// Lexer constructor - Create a new raw lexer object. This object is only
148 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
149 /// text range will outlive it, so it doesn't take ownership of it.
150 Lexer(SourceLocation FileLoc, const LangOptions &LangOpts,
151 const char *BufStart, const char *BufPtr, const char *BufEnd);
152
153 /// Lexer constructor - Create a new raw lexer object. This object is only
154 /// suitable for calls to 'LexFromRawLexer'. This lexer assumes that the
155 /// text range will outlive it, so it doesn't take ownership of it.
156 Lexer(FileID FID, const llvm::MemoryBufferRef &FromFile,
157 const SourceManager &SM, const LangOptions &LangOpts);
158
159 Lexer(const Lexer &) = delete;
160 Lexer &operator=(const Lexer &) = delete;
161
162 /// Create_PragmaLexer: Lexer constructor - Create a new lexer object for
163 /// _Pragma expansion. This has a variety of magic semantics that this method
164 /// sets up. It returns a new'd Lexer that must be delete'd when done.
165 static Lexer *Create_PragmaLexer(SourceLocation SpellingLoc,
166 SourceLocation ExpansionLocStart,
167 SourceLocation ExpansionLocEnd,
168 unsigned TokLen, Preprocessor &PP);
169
170 /// getLangOpts - Return the language features currently enabled.
171 /// NOTE: this lexer modifies features as a file is parsed!
172 const LangOptions &getLangOpts() const { return LangOpts; }
173
174 /// getFileLoc - Return the File Location for the file we are lexing out of.
175 /// The physical location encodes the location where the characters come from,
176 /// the virtual location encodes where we should *claim* the characters came
177 /// from. Currently this is only used by _Pragma handling.
178 SourceLocation getFileLoc() const { return FileLoc; }
179
180private:
181 /// Lex - Return the next token in the file. If this is the end of file, it
182 /// return the tok::eof token. This implicitly involves the preprocessor.
183 bool Lex(Token &Result);
184
185public:
186 /// isPragmaLexer - Returns true if this Lexer is being used to lex a pragma.
187 bool isPragmaLexer() const { return Is_PragmaLexer; }
188
189private:
190 /// IndirectLex - An indirect call to 'Lex' that can be invoked via
191 /// the PreprocessorLexer interface.
192 void IndirectLex(Token &Result) override { Lex(Result); }
193
194public:
195 /// LexFromRawLexer - Lex a token from a designated raw lexer (one with no
196 /// associated preprocessor object. Return true if the 'next character to
197 /// read' pointer points at the end of the lexer buffer, false otherwise.
198 bool LexFromRawLexer(Token &Result) {
199 assert(LexingRawMode && "Not already in raw mode!");
200 Lex(Result);
201 // Note that lexing to the end of the buffer doesn't implicitly delete the
202 // lexer when in raw mode.
203 return BufferPtr == BufferEnd;
204 }
205
206 /// isKeepWhitespaceMode - Return true if the lexer should return tokens for
207 /// every character in the file, including whitespace and comments. This
208 /// should only be used in raw mode, as the preprocessor is not prepared to
209 /// deal with the excess tokens.
210 bool isKeepWhitespaceMode() const {
211 return ExtendedTokenMode > 1;
212 }
213
214 /// SetKeepWhitespaceMode - This method lets clients enable or disable
215 /// whitespace retention mode.
216 void SetKeepWhitespaceMode(bool Val) {
217 assert((!Val || LexingRawMode || LangOpts.TraditionalCPP) &&
218 "Can only retain whitespace in raw mode or -traditional-cpp");
219 ExtendedTokenMode = Val ? 2 : 0;
220 }
221
222 /// inKeepCommentMode - Return true if the lexer should return comments as
223 /// tokens.
224 bool inKeepCommentMode() const {
225 return ExtendedTokenMode > 0;
226 }
227
228 /// SetCommentRetentionMode - Change the comment retention mode of the lexer
229 /// to the specified mode. This is really only useful when lexing in raw
230 /// mode, because otherwise the lexer needs to manage this.
231 void SetCommentRetentionState(bool Mode) {
232 assert(!isKeepWhitespaceMode() &&
233 "Can't play with comment retention state when retaining whitespace");
234 ExtendedTokenMode = Mode ? 1 : 0;
235 }
236
237 /// Sets the extended token mode back to its initial value, according to the
238 /// language options and preprocessor. This controls whether the lexer
239 /// produces comment and whitespace tokens.
240 ///
241 /// This requires the lexer to have an associated preprocessor. A standalone
242 /// lexer has nothing to reset to.
243 void resetExtendedTokenMode();
244
245 /// Gets source code buffer.
246 StringRef getBuffer() const {
247 return StringRef(BufferStart, BufferEnd - BufferStart);
248 }
249
250 /// ReadToEndOfLine - Read the rest of the current preprocessor line as an
251 /// uninterpreted string. This switches the lexer out of directive mode.
252 void ReadToEndOfLine(SmallVectorImpl<char> *Result = nullptr);
253
254
255 /// Diag - Forwarding function for diagnostics. This translate a source
256 /// position in the current buffer into a SourceLocation object for rendering.
257 DiagnosticBuilder Diag(const char *Loc, unsigned DiagID) const;
258
259 /// getSourceLocation - Return a source location identifier for the specified
260 /// offset in the current file.
261 SourceLocation getSourceLocation(const char *Loc, unsigned TokLen = 1) const;
262
263 /// getSourceLocation - Return a source location for the next character in
264 /// the current file.
265 SourceLocation getSourceLocation() override {
266 return getSourceLocation(BufferPtr);
267 }
268
269 /// Return the current location in the buffer.
270 const char *getBufferLocation() const { return BufferPtr; }
271
272 /// Returns the current lexing offset.
273 unsigned getCurrentBufferOffset() {
274 assert(BufferPtr >= BufferStart && "Invalid buffer state");
275 return BufferPtr - BufferStart;
276 }
277
278 /// Skip over \p NumBytes bytes.
279 ///
280 /// If the skip is successful, the next token will be lexed from the new
281 /// offset. The lexer also assumes that we skipped to the start of the line.
282 ///
283 /// \returns true if the skip failed (new offset would have been past the
284 /// end of the buffer), false otherwise.
285 bool skipOver(unsigned NumBytes);
286
287 /// Stringify - Convert the specified string into a C string by i) escaping
288 /// '\\' and " characters and ii) replacing newline character(s) with "\\n".
289 /// If Charify is true, this escapes the ' character instead of ".
290 static std::string Stringify(StringRef Str, bool Charify = false);
291
292 /// Stringify - Convert the specified string into a C string by i) escaping
293 /// '\\' and " characters and ii) replacing newline character(s) with "\\n".
294 static void Stringify(SmallVectorImpl<char> &Str);
295
296 /// getSpelling - This method is used to get the spelling of a token into a
297 /// preallocated buffer, instead of as an std::string. The caller is required
298 /// to allocate enough space for the token, which is guaranteed to be at least
299 /// Tok.getLength() bytes long. The length of the actual result is returned.
300 ///
301 /// Note that this method may do two possible things: it may either fill in
302 /// the buffer specified with characters, or it may *change the input pointer*
303 /// to point to a constant buffer with the data already in it (avoiding a
304 /// copy). The caller is not allowed to modify the returned buffer pointer
305 /// if an internal buffer is returned.
306 static unsigned getSpelling(const Token &Tok, const char *&Buffer,
307 const SourceManager &SourceMgr,
308 const LangOptions &LangOpts,
309 bool *Invalid = nullptr);
310
311 /// getSpelling() - Return the 'spelling' of the Tok token. The spelling of a
312 /// token is the characters used to represent the token in the source file
313 /// after trigraph expansion and escaped-newline folding. In particular, this
314 /// wants to get the true, uncanonicalized, spelling of things like digraphs
315 /// UCNs, etc.
316 static std::string getSpelling(const Token &Tok,
317 const SourceManager &SourceMgr,
318 const LangOptions &LangOpts,
319 bool *Invalid = nullptr);
320
321 /// getSpelling - This method is used to get the spelling of the
322 /// token at the given source location. If, as is usually true, it
323 /// is not necessary to copy any data, then the returned string may
324 /// not point into the provided buffer.
325 ///
326 /// This method lexes at the expansion depth of the given
327 /// location and does not jump to the expansion or spelling
328 /// location.
329 static StringRef getSpelling(SourceLocation loc,
330 SmallVectorImpl<char> &buffer,
331 const SourceManager &SM,
332 const LangOptions &options,
333 bool *invalid = nullptr);
334
335 /// MeasureTokenLength - Relex the token at the specified location and return
336 /// its length in bytes in the input file. If the token needs cleaning (e.g.
337 /// includes a trigraph or an escaped newline) then this count includes bytes
338 /// that are part of that.
339 static unsigned MeasureTokenLength(SourceLocation Loc,
340 const SourceManager &SM,
341 const LangOptions &LangOpts);
342
343 /// Relex the token at the specified location.
344 /// \returns true if there was a failure, false on success.
345 static bool getRawToken(SourceLocation Loc, Token &Result,
346 const SourceManager &SM,
347 const LangOptions &LangOpts,
348 bool IgnoreWhiteSpace = false);
349
350 /// Given a location any where in a source buffer, find the location
351 /// that corresponds to the beginning of the token in which the original
352 /// source location lands.
353 static SourceLocation GetBeginningOfToken(SourceLocation Loc,
354 const SourceManager &SM,
355 const LangOptions &LangOpts);
356
357 /// Get the physical length (including trigraphs and escaped newlines) of the
358 /// first \p Characters characters of the token starting at TokStart.
359 static unsigned getTokenPrefixLength(SourceLocation TokStart,
360 unsigned CharNo,
361 const SourceManager &SM,
362 const LangOptions &LangOpts);
363
364 /// AdvanceToTokenCharacter - If the current SourceLocation specifies a
365 /// location at the start of a token, return a new location that specifies a
366 /// character within the token. This handles trigraphs and escaped newlines.
367 static SourceLocation AdvanceToTokenCharacter(SourceLocation TokStart,
368 unsigned Characters,
369 const SourceManager &SM,
370 const LangOptions &LangOpts) {
371 return TokStart.getLocWithOffset(
372 getTokenPrefixLength(TokStart, Characters, SM, LangOpts));
373 }
374
375 /// Computes the source location just past the end of the
376 /// token at this source location.
377 ///
378 /// This routine can be used to produce a source location that
379 /// points just past the end of the token referenced by \p Loc, and
380 /// is generally used when a diagnostic needs to point just after a
381 /// token where it expected something different that it received. If
382 /// the returned source location would not be meaningful (e.g., if
383 /// it points into a macro), this routine returns an invalid
384 /// source location.
385 ///
386 /// \param Offset an offset from the end of the token, where the source
387 /// location should refer to. The default offset (0) produces a source
388 /// location pointing just past the end of the token; an offset of 1 produces
389 /// a source location pointing to the last character in the token, etc.
390 static SourceLocation getLocForEndOfToken(SourceLocation Loc, unsigned Offset,
391 const SourceManager &SM,
392 const LangOptions &LangOpts);
393
394 /// Given a token range, produce a corresponding CharSourceRange that
395 /// is not a token range. This allows the source range to be used by
396 /// components that don't have access to the lexer and thus can't find the
397 /// end of the range for themselves.
398 static CharSourceRange getAsCharRange(SourceRange Range,
399 const SourceManager &SM,
400 const LangOptions &LangOpts) {
401 SourceLocation End = getLocForEndOfToken(Range.getEnd(), 0, SM, LangOpts);
402 return End.isInvalid() ? CharSourceRange()
403 : CharSourceRange::getCharRange(
404 Range.getBegin(), End);
405 }
406 static CharSourceRange getAsCharRange(CharSourceRange Range,
407 const SourceManager &SM,
408 const LangOptions &LangOpts) {
409 return Range.isTokenRange()
410 ? getAsCharRange(Range.getAsRange(), SM, LangOpts)
411 : Range;
412 }
413
414 /// Returns true if the given MacroID location points at the first
415 /// token of the macro expansion.
416 ///
417 /// \param MacroBegin If non-null and function returns true, it is set to
418 /// begin location of the macro.
419 static bool isAtStartOfMacroExpansion(SourceLocation loc,
420 const SourceManager &SM,
421 const LangOptions &LangOpts,
422 SourceLocation *MacroBegin = nullptr);
423
424 /// Returns true if the given MacroID location points at the last
425 /// token of the macro expansion.
426 ///
427 /// \param MacroEnd If non-null and function returns true, it is set to
428 /// end location of the macro.
429 static bool isAtEndOfMacroExpansion(SourceLocation loc,
430 const SourceManager &SM,
431 const LangOptions &LangOpts,
432 SourceLocation *MacroEnd = nullptr);
433
434 /// Accepts a range and returns a character range with file locations.
435 ///
436 /// Returns a null range if a part of the range resides inside a macro
437 /// expansion or the range does not reside on the same FileID.
438 ///
439 /// This function is trying to deal with macros and return a range based on
440 /// file locations. The cases where it can successfully handle macros are:
441 ///
442 /// -begin or end range lies at the start or end of a macro expansion, in
443 /// which case the location will be set to the expansion point, e.g:
444 /// \#define M 1 2
445 /// a M
446 /// If you have a range [a, 2] (where 2 came from the macro), the function
447 /// will return a range for "a M"
448 /// if you have range [a, 1], the function will fail because the range
449 /// overlaps with only a part of the macro
450 ///
451 /// -The macro is a function macro and the range can be mapped to the macro
452 /// arguments, e.g:
453 /// \#define M 1 2
454 /// \#define FM(x) x
455 /// FM(a b M)
456 /// if you have range [b, 2], the function will return the file range "b M"
457 /// inside the macro arguments.
458 /// if you have range [a, 2], the function will return the file range
459 /// "FM(a b M)" since the range includes all of the macro expansion.
460 static CharSourceRange makeFileCharRange(CharSourceRange Range,
461 const SourceManager &SM,
462 const LangOptions &LangOpts);
463
464 /// Returns a string for the source that the range encompasses.
465 static StringRef getSourceText(CharSourceRange Range,
466 const SourceManager &SM,
467 const LangOptions &LangOpts,
468 bool *Invalid = nullptr);
469
470 /// Retrieve the name of the immediate macro expansion.
471 ///
472 /// This routine starts from a source location, and finds the name of the macro
473 /// responsible for its immediate expansion. It looks through any intervening
474 /// macro argument expansions to compute this. It returns a StringRef which
475 /// refers to the SourceManager-owned buffer of the source where that macro
476 /// name is spelled. Thus, the result shouldn't out-live that SourceManager.
477 static StringRef getImmediateMacroName(SourceLocation Loc,
478 const SourceManager &SM,
479 const LangOptions &LangOpts);
480
481 /// Retrieve the name of the immediate macro expansion.
482 ///
483 /// This routine starts from a source location, and finds the name of the
484 /// macro responsible for its immediate expansion. It looks through any
485 /// intervening macro argument expansions to compute this. It returns a
486 /// StringRef which refers to the SourceManager-owned buffer of the source
487 /// where that macro name is spelled. Thus, the result shouldn't out-live
488 /// that SourceManager.
489 ///
490 /// This differs from Lexer::getImmediateMacroName in that any macro argument
491 /// location will result in the topmost function macro that accepted it.
492 /// e.g.
493 /// \code
494 /// MAC1( MAC2(foo) )
495 /// \endcode
496 /// for location of 'foo' token, this function will return "MAC1" while
497 /// Lexer::getImmediateMacroName will return "MAC2".
498 static StringRef getImmediateMacroNameForDiagnostics(
499 SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts);
500
501 /// Compute the preamble of the given file.
502 ///
503 /// The preamble of a file contains the initial comments, include directives,
504 /// and other preprocessor directives that occur before the code in this
505 /// particular file actually begins. The preamble of the main source file is
506 /// a potential prefix header.
507 ///
508 /// \param Buffer The memory buffer containing the file's contents.
509 ///
510 /// \param MaxLines If non-zero, restrict the length of the preamble
511 /// to fewer than this number of lines.
512 ///
513 /// \returns The offset into the file where the preamble ends and the rest
514 /// of the file begins along with a boolean value indicating whether
515 /// the preamble ends at the beginning of a new line.
516 static PreambleBounds ComputePreamble(StringRef Buffer,
517 const LangOptions &LangOpts,
518 unsigned MaxLines = 0);
519
520 /// Finds the token that comes right after the given location.
521 ///
522 /// Returns the next token, or none if the location is inside a macro.
523 static Optional<Token> findNextToken(SourceLocation Loc,
524 const SourceManager &SM,
525 const LangOptions &LangOpts);
526
527 /// Checks that the given token is the first token that occurs after
528 /// the given location (this excludes comments and whitespace). Returns the
529 /// location immediately after the specified token. If the token is not found
530 /// or the location is inside a macro, the returned source location will be
531 /// invalid.
532 static SourceLocation findLocationAfterToken(SourceLocation loc,
533 tok::TokenKind TKind,
534 const SourceManager &SM,
535 const LangOptions &LangOpts,
536 bool SkipTrailingWhitespaceAndNewLine);
537
538 /// Returns true if the given character could appear in an identifier.
539 static bool isIdentifierBodyChar(char c, const LangOptions &LangOpts);
540
541 /// Checks whether new line pointed by Str is preceded by escape
542 /// sequence.
543 static bool isNewLineEscaped(const char *BufferStart, const char *Str);
544
545 /// getCharAndSizeNoWarn - Like the getCharAndSize method, but does not ever
546 /// emit a warning.
547 static inline char getCharAndSizeNoWarn(const char *Ptr, unsigned &Size,
548 const LangOptions &LangOpts) {
549 // If this is not a trigraph and not a UCN or escaped newline, return
550 // quickly.
551 if (isObviouslySimpleCharacter(Ptr[0])) {
552 Size = 1;
553 return *Ptr;
554 }
555
556 Size = 0;
557 return getCharAndSizeSlowNoWarn(Ptr, Size, LangOpts);
558 }
559
560 /// Returns the leading whitespace for line that corresponds to the given
561 /// location \p Loc.
562 static StringRef getIndentationForLine(SourceLocation Loc,
563 const SourceManager &SM);
564
565private:
566 //===--------------------------------------------------------------------===//
567 // Internal implementation interfaces.
568
569 /// LexTokenInternal - Internal interface to lex a preprocessing token. Called
570 /// by Lex.
571 ///
572 bool LexTokenInternal(Token &Result, bool TokAtPhysicalStartOfLine);
573
574 bool CheckUnicodeWhitespace(Token &Result, uint32_t C, const char *CurPtr);
575
576 /// Given that a token begins with the Unicode character \p C, figure out
577 /// what kind of token it is and dispatch to the appropriate lexing helper
578 /// function.
579 bool LexUnicode(Token &Result, uint32_t C, const char *CurPtr);
580
581 /// FormTokenWithChars - When we lex a token, we have identified a span
582 /// starting at BufferPtr, going to TokEnd that forms the token. This method
583 /// takes that range and assigns it to the token as its location and size. In
584 /// addition, since tokens cannot overlap, this also updates BufferPtr to be
585 /// TokEnd.
586 void FormTokenWithChars(Token &Result, const char *TokEnd,
587 tok::TokenKind Kind) {
588 unsigned TokLen = TokEnd-BufferPtr;
589 Result.setLength(TokLen);
590 Result.setLocation(getSourceLocation(BufferPtr, TokLen));
591 Result.setKind(Kind);
592 BufferPtr = TokEnd;
593 }
594
595 /// isNextPPTokenLParen - Return 1 if the next unexpanded token will return a
596 /// tok::l_paren token, 0 if it is something else and 2 if there are no more
597 /// tokens in the buffer controlled by this lexer.
598 unsigned isNextPPTokenLParen();
599
600 //===--------------------------------------------------------------------===//
601 // Lexer character reading interfaces.
602
603 // This lexer is built on two interfaces for reading characters, both of which
604 // automatically provide phase 1/2 translation. getAndAdvanceChar is used
605 // when we know that we will be reading a character from the input buffer and
606 // that this character will be part of the result token. This occurs in (f.e.)
607 // string processing, because we know we need to read until we find the
608 // closing '"' character.
609 //
610 // The second interface is the combination of getCharAndSize with
611 // ConsumeChar. getCharAndSize reads a phase 1/2 translated character,
612 // returning it and its size. If the lexer decides that this character is
613 // part of the current token, it calls ConsumeChar on it. This two stage
614 // approach allows us to emit diagnostics for characters (e.g. warnings about
615 // trigraphs), knowing that they only are emitted if the character is
616 // consumed.
617
618 /// isObviouslySimpleCharacter - Return true if the specified character is
619 /// obviously the same in translation phase 1 and translation phase 3. This
620 /// can return false for characters that end up being the same, but it will
621 /// never return true for something that needs to be mapped.
622 static bool isObviouslySimpleCharacter(char C) {
623 return C != '?' && C != '\\';
624 }
625
626 /// getAndAdvanceChar - Read a single 'character' from the specified buffer,
627 /// advance over it, and return it. This is tricky in several cases. Here we
628 /// just handle the trivial case and fall-back to the non-inlined
629 /// getCharAndSizeSlow method to handle the hard case.
630 inline char getAndAdvanceChar(const char *&Ptr, Token &Tok) {
631 // If this is not a trigraph and not a UCN or escaped newline, return
632 // quickly.
633 if (isObviouslySimpleCharacter(Ptr[0])) return *Ptr++;
634
635 unsigned Size = 0;
636 char C = getCharAndSizeSlow(Ptr, Size, &Tok);
637 Ptr += Size;
638 return C;
639 }
640
641 /// ConsumeChar - When a character (identified by getCharAndSize) is consumed
642 /// and added to a given token, check to see if there are diagnostics that
643 /// need to be emitted or flags that need to be set on the token. If so, do
644 /// it.
645 const char *ConsumeChar(const char *Ptr, unsigned Size, Token &Tok) {
646 // Normal case, we consumed exactly one token. Just return it.
647 if (Size == 1)
648 return Ptr+Size;
649
650 // Otherwise, re-lex the character with a current token, allowing
651 // diagnostics to be emitted and flags to be set.
652 Size = 0;
653 getCharAndSizeSlow(Ptr, Size, &Tok);
654 return Ptr+Size;
655 }
656
657 /// getCharAndSize - Peek a single 'character' from the specified buffer,
658 /// get its size, and return it. This is tricky in several cases. Here we
659 /// just handle the trivial case and fall-back to the non-inlined
660 /// getCharAndSizeSlow method to handle the hard case.
661 inline char getCharAndSize(const char *Ptr, unsigned &Size) {
662 // If this is not a trigraph and not a UCN or escaped newline, return
663 // quickly.
664 if (isObviouslySimpleCharacter(Ptr[0])) {
665 Size = 1;
666 return *Ptr;
667 }
668
669 Size = 0;
670 return getCharAndSizeSlow(Ptr, Size);
671 }
672
673 /// getCharAndSizeSlow - Handle the slow/uncommon case of the getCharAndSize
674 /// method.
675 char getCharAndSizeSlow(const char *Ptr, unsigned &Size,
676 Token *Tok = nullptr);
677
678 /// getEscapedNewLineSize - Return the size of the specified escaped newline,
679 /// or 0 if it is not an escaped newline. P[-1] is known to be a "\" on entry
680 /// to this function.
681 static unsigned getEscapedNewLineSize(const char *P);
682
683 /// SkipEscapedNewLines - If P points to an escaped newline (or a series of
684 /// them), skip over them and return the first non-escaped-newline found,
685 /// otherwise return P.
686 static const char *SkipEscapedNewLines(const char *P);
687
688 /// getCharAndSizeSlowNoWarn - Same as getCharAndSizeSlow, but never emits a
689 /// diagnostic.
690 static char getCharAndSizeSlowNoWarn(const char *Ptr, unsigned &Size,
691 const LangOptions &LangOpts);
692
693 //===--------------------------------------------------------------------===//
694 // Other lexer functions.
695
696 void SetByteOffset(unsigned Offset, bool StartOfLine);
697
698 void PropagateLineStartLeadingSpaceInfo(Token &Result);
699
700 const char *LexUDSuffix(Token &Result, const char *CurPtr,
701 bool IsStringLiteral);
702
703 // Helper functions to lex the remainder of a token of the specific type.
704 bool LexIdentifier (Token &Result, const char *CurPtr);
705 bool LexNumericConstant (Token &Result, const char *CurPtr);
706 bool LexStringLiteral (Token &Result, const char *CurPtr,
707 tok::TokenKind Kind);
708 bool LexRawStringLiteral (Token &Result, const char *CurPtr,
709 tok::TokenKind Kind);
710 bool LexAngledStringLiteral(Token &Result, const char *CurPtr);
711 bool LexCharConstant (Token &Result, const char *CurPtr,
712 tok::TokenKind Kind);
713 bool LexEndOfFile (Token &Result, const char *CurPtr);
714 bool SkipWhitespace (Token &Result, const char *CurPtr,
715 bool &TokAtPhysicalStartOfLine);
716 bool SkipLineComment (Token &Result, const char *CurPtr,
717 bool &TokAtPhysicalStartOfLine);
718 bool SkipBlockComment (Token &Result, const char *CurPtr,
719 bool &TokAtPhysicalStartOfLine);
720 bool SaveLineComment (Token &Result, const char *CurPtr);
721
722 bool IsStartOfConflictMarker(const char *CurPtr);
723 bool HandleEndOfConflictMarker(const char *CurPtr);
724
725 bool lexEditorPlaceholder(Token &Result, const char *CurPtr);
726
727 bool isCodeCompletionPoint(const char *CurPtr) const;
728 void cutOffLexing() { BufferPtr = BufferEnd; }
729
730 bool isHexaLiteral(const char *Start, const LangOptions &LangOpts);
731
732 void codeCompleteIncludedFile(const char *PathStart,
733 const char *CompletionPoint, bool IsAngled);
734
735 /// Read a universal character name.
736 ///
737 /// \param StartPtr The position in the source buffer after the initial '\'.
738 /// If the UCN is syntactically well-formed (but not
739 /// necessarily valid), this parameter will be updated to
740 /// point to the character after the UCN.
741 /// \param SlashLoc The position in the source buffer of the '\'.
742 /// \param Result The token being formed. Pass \c nullptr to suppress
743 /// diagnostics and handle token formation in the caller.
744 ///
745 /// \return The Unicode codepoint specified by the UCN, or 0 if the UCN is
746 /// invalid.
747 uint32_t tryReadUCN(const char *&StartPtr, const char *SlashLoc, Token *Result);
748
749 /// Try to consume a UCN as part of an identifier at the current
750 /// location.
751 /// \param CurPtr Initially points to the range of characters in the source
752 /// buffer containing the '\'. Updated to point past the end of
753 /// the UCN on success.
754 /// \param Size The number of characters occupied by the '\' (including
755 /// trigraphs and escaped newlines).
756 /// \param Result The token being produced. Marked as containing a UCN on
757 /// success.
758 /// \return \c true if a UCN was lexed and it produced an acceptable
759 /// identifier character, \c false otherwise.
760 bool tryConsumeIdentifierUCN(const char *&CurPtr, unsigned Size,
761 Token &Result);
762
763 /// Try to consume an identifier character encoded in UTF-8.
764 /// \param CurPtr Points to the start of the (potential) UTF-8 code unit
765 /// sequence. On success, updated to point past the end of it.
766 /// \return \c true if a UTF-8 sequence mapping to an acceptable identifier
767 /// character was lexed, \c false otherwise.
768 bool tryConsumeIdentifierUTF8Char(const char *&CurPtr);
769};
770
771} // namespace clang
772
773#endif // LLVM_CLANG_LEX_LEXER_H
774