1 //===- Tokens.h - collect tokens from preprocessing --------------*- 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 // Record tokens that a preprocessor emits and define operations to map between
9 // the tokens written in a file and tokens produced by the preprocessor.
10 //
11 // When running the compiler, there are two token streams we are interested in:
12 //   - "spelled" tokens directly correspond to a substring written in some
13 //     source file.
14 //   - "expanded" tokens represent the result of preprocessing, parses consumes
15 //     this token stream to produce the AST.
16 //
17 // Expanded tokens correspond directly to locations found in the AST, allowing
18 // to find subranges of the token stream covered by various AST nodes. Spelled
19 // tokens correspond directly to the source code written by the user.
20 //
21 // To allow composing these two use-cases, we also define operations that map
22 // between expanded and spelled tokens that produced them (macro calls,
23 // directives, etc).
24 //
25 //===----------------------------------------------------------------------===//
26 
27 #ifndef LLVM_CLANG_TOOLING_SYNTAX_TOKENS_H
28 #define LLVM_CLANG_TOOLING_SYNTAX_TOKENS_H
29 
30 #include "clang/Basic/FileManager.h"
31 #include "clang/Basic/LangOptions.h"
32 #include "clang/Basic/SourceLocation.h"
33 #include "clang/Basic/SourceManager.h"
34 #include "clang/Basic/TokenKinds.h"
35 #include "clang/Lex/Token.h"
36 #include "llvm/ADT/ArrayRef.h"
37 #include "llvm/ADT/DenseMap.h"
38 #include "llvm/ADT/Optional.h"
39 #include "llvm/ADT/StringRef.h"
40 #include "llvm/Support/Compiler.h"
41 #include "llvm/Support/raw_ostream.h"
42 #include <cstdint>
43 #include <tuple>
44 
45 namespace clang {
46 class Preprocessor;
47 
48 namespace syntax {
49 
50 /// A half-open character range inside a particular file, the start offset is
51 /// included and the end offset is excluded from the range.
52 struct FileRange {
53   /// EXPECTS: File.isValid() && Begin <= End.
54   FileRange(FileID File, unsigned BeginOffset, unsigned EndOffset);
55   /// EXPECTS: BeginLoc.isValid() && BeginLoc.isFileID().
56   FileRange(const SourceManager &SM, SourceLocation BeginLoc, unsigned Length);
57   /// EXPECTS: BeginLoc.isValid() && BeginLoc.isFileID(), Begin <= End and files
58   ///          are the same.
59   FileRange(const SourceManager &SM, SourceLocation BeginLoc,
60             SourceLocation EndLoc);
61 
62   FileID file() const { return File; }
63   /// Start is a start offset (inclusive) in the corresponding file.
64   unsigned beginOffset() const { return Begin; }
65   /// End offset (exclusive) in the corresponding file.
66   unsigned endOffset() const { return End; }
67 
68   unsigned length() const { return End - Begin; }
69 
70   /// Check if \p Offset is inside the range.
71   bool contains(unsigned Offset) const {
72     return Begin <= Offset && Offset < End;
73   }
74   /// Check \p Offset is inside the range or equal to its endpoint.
75   bool touches(unsigned Offset) const {
76     return Begin <= Offset && Offset <= End;
77   }
78 
79   /// Gets the substring that this FileRange refers to.
80   llvm::StringRef text(const SourceManager &SM) const;
81 
82   /// Convert to the clang range. The returned range is always a char range,
83   /// never a token range.
84   CharSourceRange toCharRange(const SourceManager &SM) const;
85 
86   friend bool operator==(const FileRange &L, const FileRange &R) {
87     return std::tie(L.File, L.Begin, L.End) == std::tie(R.File, R.Begin, R.End);
88   }
89   friend bool operator!=(const FileRange &L, const FileRange &R) {
90     return !(L == R);
91   }
92 
93 private:
94   FileID File;
95   unsigned Begin;
96   unsigned End;
97 };
98 
99 /// For debugging purposes.
100 llvm::raw_ostream &operator<<(llvm::raw_ostream &OS, const FileRange &R);
101 
102 /// A token coming directly from a file or from a macro invocation. Has just
103 /// enough information to locate the token in the source code.
104 /// Can represent both expanded and spelled tokens.
105 class Token {
106 public:
107   Token(SourceLocation Location, unsigned Length, tok::TokenKind Kind);
108   /// EXPECTS: clang::Token is not an annotation token.
109   explicit Token(const clang::Token &T);
110 
111   tok::TokenKind kind() const { return Kind; }
112   /// Location of the first character of a token.
113   SourceLocation location() const { return Location; }
114   /// Location right after the last character of a token.
115   SourceLocation endLocation() const {
116     return Location.getLocWithOffset(Length);
117   }
118   unsigned length() const { return Length; }
119 
120   /// Get the substring covered by the token. Note that will include all
121   /// digraphs, newline continuations, etc. E.g. tokens for 'int' and
122   ///    in\
123   ///    t
124   /// both have the same kind tok::kw_int, but results of text() are different.
125   llvm::StringRef text(const SourceManager &SM) const;
126 
127   /// Gets a range of this token.
128   /// EXPECTS: token comes from a file, not from a macro expansion.
129   FileRange range(const SourceManager &SM) const;
130 
131   /// Given two tokens inside the same file, returns a file range that starts at
132   /// \p First and ends at \p Last.
133   /// EXPECTS: First and Last are file tokens from the same file, Last starts
134   ///          after First.
135   static FileRange range(const SourceManager &SM, const syntax::Token &First,
136                          const syntax::Token &Last);
137 
138   std::string dumpForTests(const SourceManager &SM) const;
139   /// For debugging purposes.
140   std::string str() const;
141 
142 private:
143   SourceLocation Location;
144   unsigned Length;
145   tok::TokenKind Kind;
146 };
147 /// For debugging purposes. Equivalent to a call to Token::str().
148 llvm::raw_ostream &operator<<(llvm::raw_ostream &OS, const Token &T);
149 
150 /// A list of tokens obtained by preprocessing a text buffer and operations to
151 /// map between the expanded and spelled tokens, i.e. TokenBuffer has
152 /// information about two token streams:
153 ///    1. Expanded tokens: tokens produced by the preprocessor after all macro
154 ///       replacements,
155 ///    2. Spelled tokens: corresponding directly to the source code of a file
156 ///       before any macro replacements occurred.
157 /// Here's an example to illustrate a difference between those two:
158 ///     #define FOO 10
159 ///     int a = FOO;
160 ///
161 /// Spelled tokens are {'#','define','FOO','10','int','a','=','FOO',';'}.
162 /// Expanded tokens are {'int','a','=','10',';','eof'}.
163 ///
164 /// Note that the expanded token stream has a tok::eof token at the end, the
165 /// spelled tokens never store a 'eof' token.
166 ///
167 /// The full list expanded tokens can be obtained with expandedTokens(). Spelled
168 /// tokens for each of the files can be obtained via spelledTokens(FileID).
169 ///
170 /// To map between the expanded and spelled tokens use findSpelledByExpanded().
171 ///
172 /// To build a token buffer use the TokenCollector class. You can also compute
173 /// the spelled tokens of a file using the tokenize() helper.
174 ///
175 /// FIXME: allow mappings into macro arguments.
176 class TokenBuffer {
177 public:
178   TokenBuffer(const SourceManager &SourceMgr) : SourceMgr(&SourceMgr) {}
179 
180   TokenBuffer(TokenBuffer &&) = default;
181   TokenBuffer(const TokenBuffer &) = delete;
182   TokenBuffer &operator=(TokenBuffer &&) = default;
183   TokenBuffer &operator=(const TokenBuffer &) = delete;
184 
185   /// All tokens produced by the preprocessor after all macro replacements,
186   /// directives, etc. Source locations found in the clang AST will always
187   /// point to one of these tokens.
188   /// Tokens are in TU order (per SourceManager::isBeforeInTranslationUnit()).
189   /// FIXME: figure out how to handle token splitting, e.g. '>>' can be split
190   ///        into two '>' tokens by the parser. However, TokenBuffer currently
191   ///        keeps it as a single '>>' token.
192   llvm::ArrayRef<syntax::Token> expandedTokens() const {
193     return ExpandedTokens;
194   }
195 
196   /// Builds a cache to make future calls to expandedToken(SourceRange) faster.
197   /// Creates an index only once. Further calls to it will be no-op.
198   void indexExpandedTokens();
199 
200   /// Returns the subrange of expandedTokens() corresponding to the closed
201   /// token range R.
202   /// Consider calling indexExpandedTokens() before for faster lookups.
203   llvm::ArrayRef<syntax::Token> expandedTokens(SourceRange R) const;
204 
205   /// Returns the subrange of spelled tokens corresponding to AST node spanning
206   /// \p Expanded. This is the text that should be replaced if a refactoring
207   /// were to rewrite the node. If \p Expanded is empty, the returned value is
208   /// llvm::None.
209   ///
210   /// Will fail if the expanded tokens do not correspond to a sequence of
211   /// spelled tokens. E.g. for the following example:
212   ///
213   ///   #define FIRST f1 f2 f3
214   ///   #define SECOND s1 s2 s3
215   ///   #define ID2(X, Y) X Y
216   ///
217   ///   a FIRST b SECOND c // expanded tokens are: a f1 f2 f3 b s1 s2 s3 c
218   ///   d ID2(e f g, h) i  // expanded tokens are: d e f g h i
219   ///
220   /// the results would be:
221   ///   expanded   => spelled
222   ///   ------------------------
223   ///            a => a
224   ///     s1 s2 s3 => SECOND
225   ///   a f1 f2 f3 => a FIRST
226   ///         a f1 => can't map
227   ///        s1 s2 => can't map
228   ///         e f  => e f
229   ///         g h  => can't map
230   ///
231   /// EXPECTS: \p Expanded is a subrange of expandedTokens().
232   /// Complexity is logarithmic.
233   llvm::Optional<llvm::ArrayRef<syntax::Token>>
234   spelledForExpanded(llvm::ArrayRef<syntax::Token> Expanded) const;
235 
236   /// Find the subranges of expanded tokens, corresponding to \p Spelled.
237   ///
238   /// Some spelled tokens may not be present in the expanded token stream, so
239   /// this function can return an empty vector, e.g. for tokens of macro
240   /// directives or disabled preprocessor branches.
241   ///
242   /// Some spelled tokens can be duplicated in the expanded token stream
243   /// multiple times and this function will return multiple results in those
244   /// cases. This happens when \p Spelled is inside a macro argument.
245   ///
246   /// FIXME: return correct results on macro arguments. For now, we return an
247   ///        empty list.
248   ///
249   /// (!) will return empty vector on tokens from #define body:
250   /// E.g. for the following example:
251   ///
252   ///   #define FIRST(A) f1 A = A f2
253   ///   #define SECOND s
254   ///
255   ///   a FIRST(arg) b SECOND c // expanded tokens are: a f1 arg = arg f2 b s
256   /// The results would be
257   ///   spelled           => expanded
258   ///   ------------------------
259   ///   #define FIRST     => {}
260   ///   a FIRST(arg)      => {a f1 arg = arg f2}
261   ///   arg               => {arg, arg} // arg #1 is before `=` and arg #2 is
262   ///                                   // after `=` in the expanded tokens.
263   llvm::SmallVector<llvm::ArrayRef<syntax::Token>, 1>
264   expandedForSpelled(llvm::ArrayRef<syntax::Token> Spelled) const;
265 
266   /// An expansion produced by the preprocessor, includes macro expansions and
267   /// preprocessor directives. Preprocessor always maps a non-empty range of
268   /// spelled tokens to a (possibly empty) range of expanded tokens. Here is a
269   /// few examples of expansions:
270   ///    #pragma once      // Expands to an empty range.
271   ///    #define FOO 1 2 3 // Expands an empty range.
272   ///    FOO               // Expands to "1 2 3".
273   /// FIXME(ibiryukov): implement this, currently #include expansions are empty.
274   ///    #include <vector> // Expands to tokens produced by the include.
275   struct Expansion {
276     llvm::ArrayRef<syntax::Token> Spelled;
277     llvm::ArrayRef<syntax::Token> Expanded;
278   };
279   /// If \p Spelled starts a mapping (e.g. if it's a macro name or '#' starting
280   /// a preprocessor directive) return the subrange of expanded tokens that the
281   /// macro expands to.
282   llvm::Optional<Expansion>
283   expansionStartingAt(const syntax::Token *Spelled) const;
284   /// Returns all expansions (partially) expanded from the specified tokens.
285   /// This is the expansions whose Spelled range intersects \p Spelled.
286   std::vector<Expansion>
287   expansionsOverlapping(llvm::ArrayRef<syntax::Token> Spelled) const;
288 
289   /// Lexed tokens of a file before preprocessing. E.g. for the following input
290   ///     #define DECL(name) int name = 10
291   ///     DECL(a);
292   /// spelledTokens() returns
293   ///    {"#", "define", "DECL", "(", "name", ")", "int", "name", "=", "10",
294   ///     "DECL", "(", "a", ")", ";"}
295   llvm::ArrayRef<syntax::Token> spelledTokens(FileID FID) const;
296 
297   /// Returns the spelled Token starting at Loc, if there are no such tokens
298   /// returns nullptr.
299   const syntax::Token *spelledTokenAt(SourceLocation Loc) const;
300 
301   /// Get all tokens that expand a macro in \p FID. For the following input
302   ///     #define FOO B
303   ///     #define FOO2(X) int X
304   ///     FOO2(XY)
305   ///     int B;
306   ///     FOO;
307   /// macroExpansions() returns {"FOO2", "FOO"} (from line 3 and 5
308   /// respecitvely).
309   std::vector<const syntax::Token *> macroExpansions(FileID FID) const;
310 
311   const SourceManager &sourceManager() const { return *SourceMgr; }
312 
313   std::string dumpForTests() const;
314 
315 private:
316   /// Describes a mapping between a continuous subrange of spelled tokens and
317   /// expanded tokens. Represents macro expansions, preprocessor directives,
318   /// conditionally disabled pp regions, etc.
319   ///   #define FOO 1+2
320   ///   #define BAR(a) a + 1
321   ///   FOO    // invocation #1, tokens = {'1','+','2'}, macroTokens = {'FOO'}.
322   ///   BAR(1) // invocation #2, tokens = {'a', '+', '1'},
323   ///                            macroTokens = {'BAR', '(', '1', ')'}.
324   struct Mapping {
325     // Positions in the corresponding spelled token stream. The corresponding
326     // range is never empty.
327     unsigned BeginSpelled = 0;
328     unsigned EndSpelled = 0;
329     // Positions in the expanded token stream. The corresponding range can be
330     // empty.
331     unsigned BeginExpanded = 0;
332     unsigned EndExpanded = 0;
333 
334     /// For debugging purposes.
335     std::string str() const;
336   };
337   /// Spelled tokens of the file with information about the subranges.
338   struct MarkedFile {
339     /// Lexed, but not preprocessed, tokens of the file. These map directly to
340     /// text in the corresponding files and include tokens of all preprocessor
341     /// directives.
342     /// FIXME: spelled tokens don't change across FileID that map to the same
343     ///        FileEntry. We could consider deduplicating them to save memory.
344     std::vector<syntax::Token> SpelledTokens;
345     /// A sorted list to convert between the spelled and expanded token streams.
346     std::vector<Mapping> Mappings;
347     /// The first expanded token produced for this FileID.
348     unsigned BeginExpanded = 0;
349     unsigned EndExpanded = 0;
350   };
351 
352   friend class TokenCollector;
353 
354   /// Maps a single expanded token to its spelled counterpart or a mapping that
355   /// produced it.
356   std::pair<const syntax::Token *, const Mapping *>
357   spelledForExpandedToken(const syntax::Token *Expanded) const;
358 
359   /// Returns a mapping starting before \p Spelled token, or nullptr if no
360   /// such mapping exists.
361   static const Mapping *
362   mappingStartingBeforeSpelled(const MarkedFile &F,
363                                const syntax::Token *Spelled);
364 
365   /// Convert a private Mapping to a public Expansion.
366   Expansion makeExpansion(const MarkedFile &, const Mapping &) const;
367   /// Returns the file that the Spelled tokens are taken from.
368   /// Asserts that they are non-empty, from a tracked file, and in-bounds.
369   const MarkedFile &fileForSpelled(llvm::ArrayRef<syntax::Token> Spelled) const;
370 
371   /// Token stream produced after preprocessing, conceputally this captures the
372   /// same stream as 'clang -E' (excluding the preprocessor directives like
373   /// #file, etc.).
374   std::vector<syntax::Token> ExpandedTokens;
375   // Index of ExpandedTokens for faster lookups by SourceLocation.
376   llvm::DenseMap<SourceLocation, unsigned> ExpandedTokIndex;
377   llvm::DenseMap<FileID, MarkedFile> Files;
378   // The value is never null, pointer instead of reference to avoid disabling
379   // implicit assignment operator.
380   const SourceManager *SourceMgr;
381 };
382 
383 /// The spelled tokens that overlap or touch a spelling location Loc.
384 /// This always returns 0-2 tokens.
385 llvm::ArrayRef<syntax::Token>
386 spelledTokensTouching(SourceLocation Loc, const syntax::TokenBuffer &Tokens);
387 llvm::ArrayRef<syntax::Token>
388 spelledTokensTouching(SourceLocation Loc, llvm::ArrayRef<syntax::Token> Tokens);
389 
390 /// The identifier token that overlaps or touches a spelling location Loc.
391 /// If there is none, returns nullptr.
392 const syntax::Token *
393 spelledIdentifierTouching(SourceLocation Loc,
394                           llvm::ArrayRef<syntax::Token> Tokens);
395 const syntax::Token *
396 spelledIdentifierTouching(SourceLocation Loc,
397                           const syntax::TokenBuffer &Tokens);
398 
399 /// Lex the text buffer, corresponding to \p FID, in raw mode and record the
400 /// resulting spelled tokens. Does minimal post-processing on raw identifiers,
401 /// setting the appropriate token kind (instead of the raw_identifier reported
402 /// by lexer in raw mode). This is a very low-level function, most users should
403 /// prefer to use TokenCollector. Lexing in raw mode produces wildly different
404 /// results from what one might expect when running a C++ frontend, e.g.
405 /// preprocessor does not run at all.
406 /// The result will *not* have a 'eof' token at the end.
407 std::vector<syntax::Token> tokenize(FileID FID, const SourceManager &SM,
408                                     const LangOptions &LO);
409 /// Similar to one above, instead of whole file tokenizes a part of it. Note
410 /// that, the first token might be incomplete if FR.startOffset is not at the
411 /// beginning of a token, and the last token returned will start before the
412 /// FR.endOffset but might end after it.
413 std::vector<syntax::Token>
414 tokenize(const FileRange &FR, const SourceManager &SM, const LangOptions &LO);
415 
416 /// Collects tokens for the main file while running the frontend action. An
417 /// instance of this object should be created on
418 /// FrontendAction::BeginSourceFile() and the results should be consumed after
419 /// FrontendAction::Execute() finishes.
420 class TokenCollector {
421 public:
422   /// Adds the hooks to collect the tokens. Should be called before the
423   /// preprocessing starts, i.e. as a part of BeginSourceFile() or
424   /// CreateASTConsumer().
425   TokenCollector(Preprocessor &P);
426 
427   /// Finalizes token collection. Should be called after preprocessing is
428   /// finished, i.e. after running Execute().
429   LLVM_NODISCARD TokenBuffer consume() &&;
430 
431 private:
432   /// Maps from a start to an end spelling location of transformations
433   /// performed by the preprocessor. These include:
434   ///   1. range from '#' to the last token in the line for PP directives,
435   ///   2. macro name and arguments for macro expansions.
436   /// Note that we record only top-level macro expansions, intermediate
437   /// expansions (e.g. inside macro arguments) are ignored.
438   ///
439   /// Used to find correct boundaries of macro calls and directives when
440   /// building mappings from spelled to expanded tokens.
441   ///
442   /// Logically, at each point of the preprocessor execution there is a stack of
443   /// macro expansions being processed and we could use it to recover the
444   /// location information we need. However, the public preprocessor API only
445   /// exposes the points when macro expansions start (when we push a macro onto
446   /// the stack) and not when they end (when we pop a macro from the stack).
447   /// To workaround this limitation, we rely on source location information
448   /// stored in this map.
449   using PPExpansions = llvm::DenseMap<SourceLocation, SourceLocation>;
450   class Builder;
451   class CollectPPExpansions;
452 
453   std::vector<syntax::Token> Expanded;
454   // FIXME: we only store macro expansions, also add directives(#pragma, etc.)
455   PPExpansions Expansions;
456   Preprocessor &PP;
457   CollectPPExpansions *Collector;
458 };
459 
460 } // namespace syntax
461 } // namespace clang
462 
463 #endif
464