clang-tools  10.0.0git
SourceCode.cpp
Go to the documentation of this file.
1 //===--- SourceCode.h - Manipulating source code as strings -----*- 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 #include "SourceCode.h"
9 
10 #include "Context.h"
11 #include "FuzzyMatch.h"
12 #include "Logger.h"
13 #include "Protocol.h"
14 #include "refactor/Tweak.h"
15 #include "clang/AST/ASTContext.h"
16 #include "clang/Basic/LangOptions.h"
17 #include "clang/Basic/SourceLocation.h"
18 #include "clang/Basic/SourceManager.h"
19 #include "clang/Basic/TokenKinds.h"
20 #include "clang/Driver/Types.h"
21 #include "clang/Format/Format.h"
22 #include "clang/Lex/Lexer.h"
23 #include "clang/Lex/Preprocessor.h"
24 #include "clang/Lex/Token.h"
25 #include "clang/Tooling/Core/Replacement.h"
26 #include "llvm/ADT/ArrayRef.h"
27 #include "llvm/ADT/None.h"
28 #include "llvm/ADT/STLExtras.h"
29 #include "llvm/ADT/StringExtras.h"
30 #include "llvm/ADT/StringMap.h"
31 #include "llvm/ADT/StringRef.h"
32 #include "llvm/Support/Compiler.h"
33 #include "llvm/Support/Errc.h"
34 #include "llvm/Support/Error.h"
35 #include "llvm/Support/ErrorHandling.h"
36 #include "llvm/Support/LineIterator.h"
37 #include "llvm/Support/MemoryBuffer.h"
38 #include "llvm/Support/Path.h"
39 #include "llvm/Support/SHA1.h"
40 #include "llvm/Support/VirtualFileSystem.h"
41 #include "llvm/Support/xxhash.h"
42 #include <algorithm>
43 #include <cstddef>
44 #include <string>
45 #include <vector>
46 
47 namespace clang {
48 namespace clangd {
49 
50 // Here be dragons. LSP positions use columns measured in *UTF-16 code units*!
51 // Clangd uses UTF-8 and byte-offsets internally, so conversion is nontrivial.
52 
53 // Iterates over unicode codepoints in the (UTF-8) string. For each,
54 // invokes CB(UTF-8 length, UTF-16 length), and breaks if it returns true.
55 // Returns true if CB returned true, false if we hit the end of string.
56 template <typename Callback>
57 static bool iterateCodepoints(llvm::StringRef U8, const Callback &CB) {
58  // A codepoint takes two UTF-16 code unit if it's astral (outside BMP).
59  // Astral codepoints are encoded as 4 bytes in UTF-8, starting with 11110xxx.
60  for (size_t I = 0; I < U8.size();) {
61  unsigned char C = static_cast<unsigned char>(U8[I]);
62  if (LLVM_LIKELY(!(C & 0x80))) { // ASCII character.
63  if (CB(1, 1))
64  return true;
65  ++I;
66  continue;
67  }
68  // This convenient property of UTF-8 holds for all non-ASCII characters.
69  size_t UTF8Length = llvm::countLeadingOnes(C);
70  // 0xxx is ASCII, handled above. 10xxx is a trailing byte, invalid here.
71  // 11111xxx is not valid UTF-8 at all. Assert because it's probably our bug.
72  assert((UTF8Length >= 2 && UTF8Length <= 4) &&
73  "Invalid UTF-8, or transcoding bug?");
74  I += UTF8Length; // Skip over all trailing bytes.
75  // A codepoint takes two UTF-16 code unit if it's astral (outside BMP).
76  // Astral codepoints are encoded as 4 bytes in UTF-8 (11110xxx ...)
77  if (CB(UTF8Length, UTF8Length == 4 ? 2 : 1))
78  return true;
79  }
80  return false;
81 }
82 
83 // Returns the byte offset into the string that is an offset of \p Units in
84 // the specified encoding.
85 // Conceptually, this converts to the encoding, truncates to CodeUnits,
86 // converts back to UTF-8, and returns the length in bytes.
87 static size_t measureUnits(llvm::StringRef U8, int Units, OffsetEncoding Enc,
88  bool &Valid) {
89  Valid = Units >= 0;
90  if (Units <= 0)
91  return 0;
92  size_t Result = 0;
93  switch (Enc) {
95  Result = Units;
96  break;
98  Valid = iterateCodepoints(U8, [&](int U8Len, int U16Len) {
99  Result += U8Len;
100  Units -= U16Len;
101  return Units <= 0;
102  });
103  if (Units < 0) // Offset in the middle of a surrogate pair.
104  Valid = false;
105  break;
107  Valid = iterateCodepoints(U8, [&](int U8Len, int U16Len) {
108  Result += U8Len;
109  Units--;
110  return Units <= 0;
111  });
112  break;
114  llvm_unreachable("unsupported encoding");
115  }
116  // Don't return an out-of-range index if we overran.
117  if (Result > U8.size()) {
118  Valid = false;
119  return U8.size();
120  }
121  return Result;
122 }
123 
126  auto *Enc = Context::current().get(kCurrentOffsetEncoding);
127  return Enc ? *Enc : OffsetEncoding::UTF16;
128 }
129 
130 // Like most strings in clangd, the input is UTF-8 encoded.
131 size_t lspLength(llvm::StringRef Code) {
132  size_t Count = 0;
133  switch (lspEncoding()) {
135  Count = Code.size();
136  break;
138  iterateCodepoints(Code, [&](int U8Len, int U16Len) {
139  Count += U16Len;
140  return false;
141  });
142  break;
144  iterateCodepoints(Code, [&](int U8Len, int U16Len) {
145  ++Count;
146  return false;
147  });
148  break;
150  llvm_unreachable("unsupported encoding");
151  }
152  return Count;
153 }
154 
155 llvm::Expected<size_t> positionToOffset(llvm::StringRef Code, Position P,
156  bool AllowColumnsBeyondLineLength) {
157  if (P.line < 0)
158  return llvm::make_error<llvm::StringError>(
159  llvm::formatv("Line value can't be negative ({0})", P.line),
160  llvm::errc::invalid_argument);
161  if (P.character < 0)
162  return llvm::make_error<llvm::StringError>(
163  llvm::formatv("Character value can't be negative ({0})", P.character),
164  llvm::errc::invalid_argument);
165  size_t StartOfLine = 0;
166  for (int I = 0; I != P.line; ++I) {
167  size_t NextNL = Code.find('\n', StartOfLine);
168  if (NextNL == llvm::StringRef::npos)
169  return llvm::make_error<llvm::StringError>(
170  llvm::formatv("Line value is out of range ({0})", P.line),
171  llvm::errc::invalid_argument);
172  StartOfLine = NextNL + 1;
173  }
174  StringRef Line =
175  Code.substr(StartOfLine).take_until([](char C) { return C == '\n'; });
176 
177  // P.character may be in UTF-16, transcode if necessary.
178  bool Valid;
179  size_t ByteInLine = measureUnits(Line, P.character, lspEncoding(), Valid);
180  if (!Valid && !AllowColumnsBeyondLineLength)
181  return llvm::make_error<llvm::StringError>(
182  llvm::formatv("{0} offset {1} is invalid for line {2}", lspEncoding(),
183  P.character, P.line),
184  llvm::errc::invalid_argument);
185  return StartOfLine + ByteInLine;
186 }
187 
188 Position offsetToPosition(llvm::StringRef Code, size_t Offset) {
189  Offset = std::min(Code.size(), Offset);
190  llvm::StringRef Before = Code.substr(0, Offset);
191  int Lines = Before.count('\n');
192  size_t PrevNL = Before.rfind('\n');
193  size_t StartOfLine = (PrevNL == llvm::StringRef::npos) ? 0 : (PrevNL + 1);
194  Position Pos;
195  Pos.line = Lines;
196  Pos.character = lspLength(Before.substr(StartOfLine));
197  return Pos;
198 }
199 
200 Position sourceLocToPosition(const SourceManager &SM, SourceLocation Loc) {
201  // We use the SourceManager's line tables, but its column number is in bytes.
202  FileID FID;
203  unsigned Offset;
204  std::tie(FID, Offset) = SM.getDecomposedSpellingLoc(Loc);
205  Position P;
206  P.line = static_cast<int>(SM.getLineNumber(FID, Offset)) - 1;
207  bool Invalid = false;
208  llvm::StringRef Code = SM.getBufferData(FID, &Invalid);
209  if (!Invalid) {
210  auto ColumnInBytes = SM.getColumnNumber(FID, Offset) - 1;
211  auto LineSoFar = Code.substr(Offset - ColumnInBytes, ColumnInBytes);
212  P.character = lspLength(LineSoFar);
213  }
214  return P;
215 }
216 
217 bool isSpelledInSource(SourceLocation Loc, const SourceManager &SM) {
218  if (Loc.isMacroID()) {
219  std::string PrintLoc = SM.getSpellingLoc(Loc).printToString(SM);
220  if (llvm::StringRef(PrintLoc).startswith("<scratch") ||
221  llvm::StringRef(PrintLoc).startswith("<command line>"))
222  return false;
223  }
224  return true;
225 }
226 
227 llvm::Optional<Range> getTokenRange(const SourceManager &SM,
228  const LangOptions &LangOpts,
229  SourceLocation TokLoc) {
230  if (!TokLoc.isValid())
231  return llvm::None;
232  SourceLocation End = Lexer::getLocForEndOfToken(TokLoc, 0, SM, LangOpts);
233  if (!End.isValid())
234  return llvm::None;
235  return halfOpenToRange(SM, CharSourceRange::getCharRange(TokLoc, End));
236 }
237 
238 namespace {
239 
240 enum TokenFlavor { Identifier, Operator, Whitespace, Other };
241 
242 bool isOverloadedOperator(const Token &Tok) {
243  switch (Tok.getKind()) {
244 #define OVERLOADED_OPERATOR(Name, Spelling, Token, Unary, Binary, MemOnly) \
245  case tok::Token:
246 #define OVERLOADED_OPERATOR_MULTI(Name, Spelling, Unary, Binary, MemOnly)
247 #include "clang/Basic/OperatorKinds.def"
248  return true;
249 
250  default:
251  break;
252  }
253  return false;
254 }
255 
256 TokenFlavor getTokenFlavor(SourceLocation Loc, const SourceManager &SM,
257  const LangOptions &LangOpts) {
258  Token Tok;
259  Tok.setKind(tok::NUM_TOKENS);
260  if (Lexer::getRawToken(Loc, Tok, SM, LangOpts,
261  /*IgnoreWhiteSpace*/ false))
262  return Other;
263 
264  // getRawToken will return false without setting Tok when the token is
265  // whitespace, so if the flag is not set, we are sure this is a whitespace.
266  if (Tok.is(tok::TokenKind::NUM_TOKENS))
267  return Whitespace;
268  if (Tok.is(tok::TokenKind::raw_identifier))
269  return Identifier;
270  if (isOverloadedOperator(Tok))
271  return Operator;
272  return Other;
273 }
274 
275 } // namespace
276 
277 SourceLocation getBeginningOfIdentifier(const Position &Pos,
278  const SourceManager &SM,
279  const LangOptions &LangOpts) {
280  FileID FID = SM.getMainFileID();
281  auto Offset = positionToOffset(SM.getBufferData(FID), Pos);
282  if (!Offset) {
283  log("getBeginningOfIdentifier: {0}", Offset.takeError());
284  return SourceLocation();
285  }
286 
287  // GetBeginningOfToken(InputLoc) is almost what we want, but does the wrong
288  // thing if the cursor is at the end of the token (identifier or operator).
289  // The cases are:
290  // 1) at the beginning of the token
291  // 2) at the middle of the token
292  // 3) at the end of the token
293  // 4) anywhere outside the identifier or operator
294  // To distinguish all cases, we lex both at the
295  // GetBeginningOfToken(InputLoc-1) and GetBeginningOfToken(InputLoc), for
296  // cases 1 and 4, we just return the original location.
297  SourceLocation InputLoc = SM.getComposedLoc(FID, *Offset);
298  if (*Offset == 0) // Case 1 or 4.
299  return InputLoc;
300  SourceLocation Before = SM.getComposedLoc(FID, *Offset - 1);
301  SourceLocation BeforeTokBeginning =
302  Lexer::GetBeginningOfToken(Before, SM, LangOpts);
303  TokenFlavor BeforeKind = getTokenFlavor(BeforeTokBeginning, SM, LangOpts);
304 
305  SourceLocation CurrentTokBeginning =
306  Lexer::GetBeginningOfToken(InputLoc, SM, LangOpts);
307  TokenFlavor CurrentKind = getTokenFlavor(CurrentTokBeginning, SM, LangOpts);
308 
309  // At the middle of the token.
310  if (BeforeTokBeginning == CurrentTokBeginning) {
311  // For interesting token, we return the beginning of the token.
312  if (CurrentKind == Identifier || CurrentKind == Operator)
313  return CurrentTokBeginning;
314  // otherwise, we return the original loc.
315  return InputLoc;
316  }
317 
318  // Whitespace is not interesting.
319  if (BeforeKind == Whitespace)
320  return CurrentTokBeginning;
321  if (CurrentKind == Whitespace)
322  return BeforeTokBeginning;
323 
324  // The cursor is at the token boundary, e.g. "Before^Current", we prefer
325  // identifiers to other tokens.
326  if (CurrentKind == Identifier)
327  return CurrentTokBeginning;
328  if (BeforeKind == Identifier)
329  return BeforeTokBeginning;
330  // Then prefer overloaded operators to other tokens.
331  if (CurrentKind == Operator)
332  return CurrentTokBeginning;
333  if (BeforeKind == Operator)
334  return BeforeTokBeginning;
335 
336  // Non-interesting case, we just return the original location.
337  return InputLoc;
338 }
339 
340 bool isValidFileRange(const SourceManager &Mgr, SourceRange R) {
341  if (!R.getBegin().isValid() || !R.getEnd().isValid())
342  return false;
343 
344  FileID BeginFID;
345  size_t BeginOffset = 0;
346  std::tie(BeginFID, BeginOffset) = Mgr.getDecomposedLoc(R.getBegin());
347 
348  FileID EndFID;
349  size_t EndOffset = 0;
350  std::tie(EndFID, EndOffset) = Mgr.getDecomposedLoc(R.getEnd());
351 
352  return BeginFID.isValid() && BeginFID == EndFID && BeginOffset <= EndOffset;
353 }
354 
355 bool halfOpenRangeContains(const SourceManager &Mgr, SourceRange R,
356  SourceLocation L) {
357  assert(isValidFileRange(Mgr, R));
358 
359  FileID BeginFID;
360  size_t BeginOffset = 0;
361  std::tie(BeginFID, BeginOffset) = Mgr.getDecomposedLoc(R.getBegin());
362  size_t EndOffset = Mgr.getFileOffset(R.getEnd());
363 
364  FileID LFid;
365  size_t LOffset;
366  std::tie(LFid, LOffset) = Mgr.getDecomposedLoc(L);
367  return BeginFID == LFid && BeginOffset <= LOffset && LOffset < EndOffset;
368 }
369 
370 bool halfOpenRangeTouches(const SourceManager &Mgr, SourceRange R,
371  SourceLocation L) {
372  return L == R.getEnd() || halfOpenRangeContains(Mgr, R, L);
373 }
374 
375 SourceLocation includeHashLoc(FileID IncludedFile, const SourceManager &SM) {
376  assert(SM.getLocForEndOfFile(IncludedFile).isFileID());
377  FileID IncludingFile;
378  unsigned Offset;
379  std::tie(IncludingFile, Offset) =
380  SM.getDecomposedExpansionLoc(SM.getIncludeLoc(IncludedFile));
381  bool Invalid = false;
382  llvm::StringRef Buf = SM.getBufferData(IncludingFile, &Invalid);
383  if (Invalid)
384  return SourceLocation();
385  // Now buf is "...\n#include <foo>\n..."
386  // and Offset points here: ^
387  // Rewind to the preceding # on the line.
388  assert(Offset < Buf.size());
389  for (;; --Offset) {
390  if (Buf[Offset] == '#')
391  return SM.getComposedLoc(IncludingFile, Offset);
392  if (Buf[Offset] == '\n' || Offset == 0) // no hash, what's going on?
393  return SourceLocation();
394  }
395 }
396 
397 static unsigned getTokenLengthAtLoc(SourceLocation Loc, const SourceManager &SM,
398  const LangOptions &LangOpts) {
399  Token TheTok;
400  if (Lexer::getRawToken(Loc, TheTok, SM, LangOpts))
401  return 0;
402  // FIXME: Here we check whether the token at the location is a greatergreater
403  // (>>) token and consider it as a single greater (>). This is to get it
404  // working for templates but it isn't correct for the right shift operator. We
405  // can avoid this by using half open char ranges in getFileRange() but getting
406  // token ending is not well supported in macroIDs.
407  if (TheTok.is(tok::greatergreater))
408  return 1;
409  return TheTok.getLength();
410 }
411 
412 // Returns location of the last character of the token at a given loc
413 static SourceLocation getLocForTokenEnd(SourceLocation BeginLoc,
414  const SourceManager &SM,
415  const LangOptions &LangOpts) {
416  unsigned Len = getTokenLengthAtLoc(BeginLoc, SM, LangOpts);
417  return BeginLoc.getLocWithOffset(Len ? Len - 1 : 0);
418 }
419 
420 // Returns location of the starting of the token at a given EndLoc
421 static SourceLocation getLocForTokenBegin(SourceLocation EndLoc,
422  const SourceManager &SM,
423  const LangOptions &LangOpts) {
424  return EndLoc.getLocWithOffset(
425  -(signed)getTokenLengthAtLoc(EndLoc, SM, LangOpts));
426 }
427 
428 // Converts a char source range to a token range.
429 static SourceRange toTokenRange(CharSourceRange Range, const SourceManager &SM,
430  const LangOptions &LangOpts) {
431  if (!Range.isTokenRange())
432  Range.setEnd(getLocForTokenBegin(Range.getEnd(), SM, LangOpts));
433  return Range.getAsRange();
434 }
435 // Returns the union of two token ranges.
436 // To find the maximum of the Ends of the ranges, we compare the location of the
437 // last character of the token.
438 static SourceRange unionTokenRange(SourceRange R1, SourceRange R2,
439  const SourceManager &SM,
440  const LangOptions &LangOpts) {
441  SourceLocation Begin =
442  SM.isBeforeInTranslationUnit(R1.getBegin(), R2.getBegin())
443  ? R1.getBegin()
444  : R2.getBegin();
445  SourceLocation End =
446  SM.isBeforeInTranslationUnit(getLocForTokenEnd(R1.getEnd(), SM, LangOpts),
447  getLocForTokenEnd(R2.getEnd(), SM, LangOpts))
448  ? R2.getEnd()
449  : R1.getEnd();
450  return SourceRange(Begin, End);
451 }
452 
453 // Given a range whose endpoints may be in different expansions or files,
454 // tries to find a range within a common file by following up the expansion and
455 // include location in each.
456 static SourceRange rangeInCommonFile(SourceRange R, const SourceManager &SM,
457  const LangOptions &LangOpts) {
458  // Fast path for most common cases.
459  if (SM.isWrittenInSameFile(R.getBegin(), R.getEnd()))
460  return R;
461  // Record the stack of expansion locations for the beginning, keyed by FileID.
462  llvm::DenseMap<FileID, SourceLocation> BeginExpansions;
463  for (SourceLocation Begin = R.getBegin(); Begin.isValid();
464  Begin = Begin.isFileID()
465  ? includeHashLoc(SM.getFileID(Begin), SM)
466  : SM.getImmediateExpansionRange(Begin).getBegin()) {
467  BeginExpansions[SM.getFileID(Begin)] = Begin;
468  }
469  // Move up the stack of expansion locations for the end until we find the
470  // location in BeginExpansions with that has the same file id.
471  for (SourceLocation End = R.getEnd(); End.isValid();
472  End = End.isFileID() ? includeHashLoc(SM.getFileID(End), SM)
473  : toTokenRange(SM.getImmediateExpansionRange(End),
474  SM, LangOpts)
475  .getEnd()) {
476  auto It = BeginExpansions.find(SM.getFileID(End));
477  if (It != BeginExpansions.end()) {
478  if (SM.getFileOffset(It->second) > SM.getFileOffset(End))
479  return SourceLocation();
480  return {It->second, End};
481  }
482  }
483  return SourceRange();
484 }
485 
486 // Find an expansion range (not necessarily immediate) the ends of which are in
487 // the same file id.
488 static SourceRange
489 getExpansionTokenRangeInSameFile(SourceLocation Loc, const SourceManager &SM,
490  const LangOptions &LangOpts) {
491  return rangeInCommonFile(
492  toTokenRange(SM.getImmediateExpansionRange(Loc), SM, LangOpts), SM,
493  LangOpts);
494 }
495 
496 // Returns the file range for a given Location as a Token Range
497 // This is quite similar to getFileLoc in SourceManager as both use
498 // getImmediateExpansionRange and getImmediateSpellingLoc (for macro IDs).
499 // However:
500 // - We want to maintain the full range information as we move from one file to
501 // the next. getFileLoc only uses the BeginLoc of getImmediateExpansionRange.
502 // - We want to split '>>' tokens as the lexer parses the '>>' in nested
503 // template instantiations as a '>>' instead of two '>'s.
504 // There is also getExpansionRange but it simply calls
505 // getImmediateExpansionRange on the begin and ends separately which is wrong.
506 static SourceRange getTokenFileRange(SourceLocation Loc,
507  const SourceManager &SM,
508  const LangOptions &LangOpts) {
509  SourceRange FileRange = Loc;
510  while (!FileRange.getBegin().isFileID()) {
511  if (SM.isMacroArgExpansion(FileRange.getBegin())) {
512  FileRange = unionTokenRange(
513  SM.getImmediateSpellingLoc(FileRange.getBegin()),
514  SM.getImmediateSpellingLoc(FileRange.getEnd()), SM, LangOpts);
515  assert(SM.isWrittenInSameFile(FileRange.getBegin(), FileRange.getEnd()));
516  } else {
517  SourceRange ExpansionRangeForBegin =
518  getExpansionTokenRangeInSameFile(FileRange.getBegin(), SM, LangOpts);
519  SourceRange ExpansionRangeForEnd =
520  getExpansionTokenRangeInSameFile(FileRange.getEnd(), SM, LangOpts);
521  if (ExpansionRangeForBegin.isInvalid() ||
522  ExpansionRangeForEnd.isInvalid())
523  return SourceRange();
524  assert(SM.isWrittenInSameFile(ExpansionRangeForBegin.getBegin(),
525  ExpansionRangeForEnd.getBegin()) &&
526  "Both Expansion ranges should be in same file.");
527  FileRange = unionTokenRange(ExpansionRangeForBegin, ExpansionRangeForEnd,
528  SM, LangOpts);
529  }
530  }
531  return FileRange;
532 }
533 
534 bool isInsideMainFile(SourceLocation Loc, const SourceManager &SM) {
535  return Loc.isValid() && SM.isWrittenInMainFile(SM.getExpansionLoc(Loc));
536 }
537 
538 llvm::Optional<SourceRange> toHalfOpenFileRange(const SourceManager &SM,
539  const LangOptions &LangOpts,
540  SourceRange R) {
541  SourceRange R1 = getTokenFileRange(R.getBegin(), SM, LangOpts);
542  if (!isValidFileRange(SM, R1))
543  return llvm::None;
544 
545  SourceRange R2 = getTokenFileRange(R.getEnd(), SM, LangOpts);
546  if (!isValidFileRange(SM, R2))
547  return llvm::None;
548 
549  SourceRange Result =
550  rangeInCommonFile(unionTokenRange(R1, R2, SM, LangOpts), SM, LangOpts);
551  unsigned TokLen = getTokenLengthAtLoc(Result.getEnd(), SM, LangOpts);
552  // Convert from closed token range to half-open (char) range
553  Result.setEnd(Result.getEnd().getLocWithOffset(TokLen));
554  if (!isValidFileRange(SM, Result))
555  return llvm::None;
556 
557  return Result;
558 }
559 
560 llvm::StringRef toSourceCode(const SourceManager &SM, SourceRange R) {
561  assert(isValidFileRange(SM, R));
562  bool Invalid = false;
563  auto *Buf = SM.getBuffer(SM.getFileID(R.getBegin()), &Invalid);
564  assert(!Invalid);
565 
566  size_t BeginOffset = SM.getFileOffset(R.getBegin());
567  size_t EndOffset = SM.getFileOffset(R.getEnd());
568  return Buf->getBuffer().substr(BeginOffset, EndOffset - BeginOffset);
569 }
570 
571 llvm::Expected<SourceLocation> sourceLocationInMainFile(const SourceManager &SM,
572  Position P) {
573  llvm::StringRef Code = SM.getBuffer(SM.getMainFileID())->getBuffer();
574  auto Offset =
575  positionToOffset(Code, P, /*AllowColumnBeyondLineLength=*/false);
576  if (!Offset)
577  return Offset.takeError();
578  return SM.getLocForStartOfFile(SM.getMainFileID()).getLocWithOffset(*Offset);
579 }
580 
581 Range halfOpenToRange(const SourceManager &SM, CharSourceRange R) {
582  // Clang is 1-based, LSP uses 0-based indexes.
583  Position Begin = sourceLocToPosition(SM, R.getBegin());
584  Position End = sourceLocToPosition(SM, R.getEnd());
585 
586  return {Begin, End};
587 }
588 
589 std::pair<size_t, size_t> offsetToClangLineColumn(llvm::StringRef Code,
590  size_t Offset) {
591  Offset = std::min(Code.size(), Offset);
592  llvm::StringRef Before = Code.substr(0, Offset);
593  int Lines = Before.count('\n');
594  size_t PrevNL = Before.rfind('\n');
595  size_t StartOfLine = (PrevNL == llvm::StringRef::npos) ? 0 : (PrevNL + 1);
596  return {Lines + 1, Offset - StartOfLine + 1};
597 }
598 
599 std::pair<StringRef, StringRef> splitQualifiedName(StringRef QName) {
600  size_t Pos = QName.rfind("::");
601  if (Pos == llvm::StringRef::npos)
602  return {llvm::StringRef(), QName};
603  return {QName.substr(0, Pos + 2), QName.substr(Pos + 2)};
604 }
605 
607  const tooling::Replacement &R) {
608  Range ReplacementRange = {
609  offsetToPosition(Code, R.getOffset()),
610  offsetToPosition(Code, R.getOffset() + R.getLength())};
611  return {ReplacementRange, R.getReplacementText()};
612 }
613 
614 std::vector<TextEdit> replacementsToEdits(llvm::StringRef Code,
615  const tooling::Replacements &Repls) {
616  std::vector<TextEdit> Edits;
617  for (const auto &R : Repls)
618  Edits.push_back(replacementToEdit(Code, R));
619  return Edits;
620 }
621 
622 llvm::Optional<std::string> getCanonicalPath(const FileEntry *F,
623  const SourceManager &SourceMgr) {
624  if (!F)
625  return None;
626 
627  llvm::SmallString<128> FilePath = F->getName();
628  if (!llvm::sys::path::is_absolute(FilePath)) {
629  if (auto EC =
630  SourceMgr.getFileManager().getVirtualFileSystem().makeAbsolute(
631  FilePath)) {
632  elog("Could not turn relative path '{0}' to absolute: {1}", FilePath,
633  EC.message());
634  return None;
635  }
636  }
637 
638  // Handle the symbolic link path case where the current working directory
639  // (getCurrentWorkingDirectory) is a symlink. We always want to the real
640  // file path (instead of the symlink path) for the C++ symbols.
641  //
642  // Consider the following example:
643  //
644  // src dir: /project/src/foo.h
645  // current working directory (symlink): /tmp/build -> /project/src/
646  //
647  // The file path of Symbol is "/project/src/foo.h" instead of
648  // "/tmp/build/foo.h"
649  if (auto Dir = SourceMgr.getFileManager().getDirectory(
650  llvm::sys::path::parent_path(FilePath))) {
651  llvm::SmallString<128> RealPath;
652  llvm::StringRef DirName = SourceMgr.getFileManager().getCanonicalName(*Dir);
653  llvm::sys::path::append(RealPath, DirName,
654  llvm::sys::path::filename(FilePath));
655  return RealPath.str().str();
656  }
657 
658  return FilePath.str().str();
659 }
660 
661 TextEdit toTextEdit(const FixItHint &FixIt, const SourceManager &M,
662  const LangOptions &L) {
663  TextEdit Result;
664  Result.range =
665  halfOpenToRange(M, Lexer::makeFileCharRange(FixIt.RemoveRange, M, L));
666  Result.newText = FixIt.CodeToInsert;
667  return Result;
668 }
669 
670 bool isRangeConsecutive(const Range &Left, const Range &Right) {
671  return Left.end.line == Right.start.line &&
672  Left.end.character == Right.start.character;
673 }
674 
675 FileDigest digest(llvm::StringRef Content) {
676  uint64_t Hash{llvm::xxHash64(Content)};
677  FileDigest Result;
678  for (unsigned I = 0; I < Result.size(); ++I) {
679  Result[I] = uint8_t(Hash);
680  Hash >>= 8;
681  }
682  return Result;
683 }
684 
685 llvm::Optional<FileDigest> digestFile(const SourceManager &SM, FileID FID) {
686  bool Invalid = false;
687  llvm::StringRef Content = SM.getBufferData(FID, &Invalid);
688  if (Invalid)
689  return None;
690  return digest(Content);
691 }
692 
694  llvm::StringRef Content,
695  llvm::vfs::FileSystem *FS) {
696  auto Style = format::getStyle(format::DefaultFormatStyle, File,
697  format::DefaultFallbackStyle, Content, FS);
698  if (!Style) {
699  log("getStyle() failed for file {0}: {1}. Fallback is LLVM style.", File,
700  Style.takeError());
701  Style = format::getLLVMStyle();
702  }
703  return *Style;
704 }
705 
706 llvm::Expected<tooling::Replacements>
707 cleanupAndFormat(StringRef Code, const tooling::Replacements &Replaces,
708  const format::FormatStyle &Style) {
709  auto CleanReplaces = cleanupAroundReplacements(Code, Replaces, Style);
710  if (!CleanReplaces)
711  return CleanReplaces;
712  return formatReplacements(Code, std::move(*CleanReplaces), Style);
713 }
714 
715 static void
716 lex(llvm::StringRef Code, const LangOptions &LangOpts,
717  llvm::function_ref<void(const clang::Token &, const SourceManager &SM)>
718  Action) {
719  // FIXME: InMemoryFileAdapter crashes unless the buffer is null terminated!
720  std::string NullTerminatedCode = Code.str();
721  SourceManagerForFile FileSM("dummy.cpp", NullTerminatedCode);
722  auto &SM = FileSM.get();
723  auto FID = SM.getMainFileID();
724  // Create a raw lexer (with no associated preprocessor object).
725  Lexer Lex(FID, SM.getBuffer(FID), SM, LangOpts);
726  Token Tok;
727 
728  while (!Lex.LexFromRawLexer(Tok))
729  Action(Tok, SM);
730  // LexFromRawLexer returns true after it lexes last token, so we still have
731  // one more token to report.
732  Action(Tok, SM);
733 }
734 
735 llvm::StringMap<unsigned> collectIdentifiers(llvm::StringRef Content,
736  const format::FormatStyle &Style) {
737  llvm::StringMap<unsigned> Identifiers;
738  auto LangOpt = format::getFormattingLangOpts(Style);
739  lex(Content, LangOpt, [&](const clang::Token &Tok, const SourceManager &) {
740  if (Tok.getKind() == tok::raw_identifier)
741  ++Identifiers[Tok.getRawIdentifier()];
742  });
743  return Identifiers;
744 }
745 
746 std::vector<Range> collectIdentifierRanges(llvm::StringRef Identifier,
747  llvm::StringRef Content,
748  const LangOptions &LangOpts) {
749  std::vector<Range> Ranges;
750  lex(Content, LangOpts, [&](const clang::Token &Tok, const SourceManager &SM) {
751  if (Tok.getKind() != tok::raw_identifier)
752  return;
753  if (Tok.getRawIdentifier() != Identifier)
754  return;
755  auto Range = getTokenRange(SM, LangOpts, Tok.getLocation());
756  if (!Range)
757  return;
758  Ranges.push_back(*Range);
759  });
760  return Ranges;
761 }
762 
763 namespace {
764 struct NamespaceEvent {
765  enum {
766  BeginNamespace, // namespace <ns> {. Payload is resolved <ns>.
767  EndNamespace, // } // namespace <ns>. Payload is resolved *outer*
768  // namespace.
769  UsingDirective // using namespace <ns>. Payload is unresolved <ns>.
770  } Trigger;
771  std::string Payload;
772  Position Pos;
773 };
774 // Scans C++ source code for constructs that change the visible namespaces.
775 void parseNamespaceEvents(llvm::StringRef Code,
776  const format::FormatStyle &Style,
777  llvm::function_ref<void(NamespaceEvent)> Callback) {
778 
779  // Stack of enclosing namespaces, e.g. {"clang", "clangd"}
780  std::vector<std::string> Enclosing; // Contains e.g. "clang", "clangd"
781  // Stack counts open braces. true if the brace opened a namespace.
782  std::vector<bool> BraceStack;
783 
784  enum {
785  Default,
786  Namespace, // just saw 'namespace'
787  NamespaceName, // just saw 'namespace' NSName
788  Using, // just saw 'using'
789  UsingNamespace, // just saw 'using namespace'
790  UsingNamespaceName, // just saw 'using namespace' NSName
791  } State = Default;
792  std::string NSName;
793 
794  NamespaceEvent Event;
795  lex(Code, format::getFormattingLangOpts(Style),
796  [&](const clang::Token &Tok,const SourceManager &SM) {
797  Event.Pos = sourceLocToPosition(SM, Tok.getLocation());
798  switch (Tok.getKind()) {
799  case tok::raw_identifier:
800  // In raw mode, this could be a keyword or a name.
801  switch (State) {
802  case UsingNamespace:
803  case UsingNamespaceName:
804  NSName.append(Tok.getRawIdentifier());
805  State = UsingNamespaceName;
806  break;
807  case Namespace:
808  case NamespaceName:
809  NSName.append(Tok.getRawIdentifier());
810  State = NamespaceName;
811  break;
812  case Using:
813  State =
814  (Tok.getRawIdentifier() == "namespace") ? UsingNamespace : Default;
815  break;
816  case Default:
817  NSName.clear();
818  if (Tok.getRawIdentifier() == "namespace")
819  State = Namespace;
820  else if (Tok.getRawIdentifier() == "using")
821  State = Using;
822  break;
823  }
824  break;
825  case tok::coloncolon:
826  // This can come at the beginning or in the middle of a namespace name.
827  switch (State) {
828  case UsingNamespace:
829  case UsingNamespaceName:
830  NSName.append("::");
831  State = UsingNamespaceName;
832  break;
833  case NamespaceName:
834  NSName.append("::");
835  State = NamespaceName;
836  break;
837  case Namespace: // Not legal here.
838  case Using:
839  case Default:
840  State = Default;
841  break;
842  }
843  break;
844  case tok::l_brace:
845  // Record which { started a namespace, so we know when } ends one.
846  if (State == NamespaceName) {
847  // Parsed: namespace <name> {
848  BraceStack.push_back(true);
849  Enclosing.push_back(NSName);
850  Event.Trigger = NamespaceEvent::BeginNamespace;
851  Event.Payload = llvm::join(Enclosing, "::");
852  Callback(Event);
853  } else {
854  // This case includes anonymous namespaces (State = Namespace).
855  // For our purposes, they're not namespaces and we ignore them.
856  BraceStack.push_back(false);
857  }
858  State = Default;
859  break;
860  case tok::r_brace:
861  // If braces are unmatched, we're going to be confused, but don't crash.
862  if (!BraceStack.empty()) {
863  if (BraceStack.back()) {
864  // Parsed: } // namespace
865  Enclosing.pop_back();
866  Event.Trigger = NamespaceEvent::EndNamespace;
867  Event.Payload = llvm::join(Enclosing, "::");
868  Callback(Event);
869  }
870  BraceStack.pop_back();
871  }
872  break;
873  case tok::semi:
874  if (State == UsingNamespaceName) {
875  // Parsed: using namespace <name> ;
876  Event.Trigger = NamespaceEvent::UsingDirective;
877  Event.Payload = std::move(NSName);
878  Callback(Event);
879  }
880  State = Default;
881  break;
882  default:
883  State = Default;
884  break;
885  }
886  });
887 }
888 
889 // Returns the prefix namespaces of NS: {"" ... NS}.
890 llvm::SmallVector<llvm::StringRef, 8> ancestorNamespaces(llvm::StringRef NS) {
891  llvm::SmallVector<llvm::StringRef, 8> Results;
892  Results.push_back(NS.take_front(0));
893  NS.split(Results, "::", /*MaxSplit=*/-1, /*KeepEmpty=*/false);
894  for (llvm::StringRef &R : Results)
895  R = NS.take_front(R.end() - NS.begin());
896  return Results;
897 }
898 
899 } // namespace
900 
901 std::vector<std::string> visibleNamespaces(llvm::StringRef Code,
902  const format::FormatStyle &Style) {
903  std::string Current;
904  // Map from namespace to (resolved) namespaces introduced via using directive.
905  llvm::StringMap<llvm::StringSet<>> UsingDirectives;
906 
907  parseNamespaceEvents(Code, Style, [&](NamespaceEvent Event) {
908  llvm::StringRef NS = Event.Payload;
909  switch (Event.Trigger) {
910  case NamespaceEvent::BeginNamespace:
911  case NamespaceEvent::EndNamespace:
912  Current = std::move(Event.Payload);
913  break;
914  case NamespaceEvent::UsingDirective:
915  if (NS.consume_front("::"))
916  UsingDirectives[Current].insert(NS);
917  else {
918  for (llvm::StringRef Enclosing : ancestorNamespaces(Current)) {
919  if (Enclosing.empty())
920  UsingDirectives[Current].insert(NS);
921  else
922  UsingDirectives[Current].insert((Enclosing + "::" + NS).str());
923  }
924  }
925  break;
926  }
927  });
928 
929  std::vector<std::string> Found;
930  for (llvm::StringRef Enclosing : ancestorNamespaces(Current)) {
931  Found.push_back(Enclosing);
932  auto It = UsingDirectives.find(Enclosing);
933  if (It != UsingDirectives.end())
934  for (const auto &Used : It->second)
935  Found.push_back(Used.getKey());
936  }
937 
938  llvm::sort(Found, [&](const std::string &LHS, const std::string &RHS) {
939  if (Current == RHS)
940  return false;
941  if (Current == LHS)
942  return true;
943  return LHS < RHS;
944  });
945  Found.erase(std::unique(Found.begin(), Found.end()), Found.end());
946  return Found;
947 }
948 
949 llvm::StringSet<> collectWords(llvm::StringRef Content) {
950  // We assume short words are not significant.
951  // We may want to consider other stopwords, e.g. language keywords.
952  // (A very naive implementation showed no benefit, but lexing might do better)
953  static constexpr int MinWordLength = 4;
954 
955  std::vector<CharRole> Roles(Content.size());
956  calculateRoles(Content, Roles);
957 
958  llvm::StringSet<> Result;
959  llvm::SmallString<256> Word;
960  auto Flush = [&] {
961  if (Word.size() >= MinWordLength) {
962  for (char &C : Word)
963  C = llvm::toLower(C);
964  Result.insert(Word);
965  }
966  Word.clear();
967  };
968  for (unsigned I = 0; I < Content.size(); ++I) {
969  switch (Roles[I]) {
970  case Head:
971  Flush();
972  LLVM_FALLTHROUGH;
973  case Tail:
974  Word.push_back(Content[I]);
975  break;
976  case Unknown:
977  case Separator:
978  Flush();
979  break;
980  }
981  }
982  Flush();
983 
984  return Result;
985 }
986 
987 llvm::Optional<DefinedMacro> locateMacroAt(SourceLocation Loc,
988  Preprocessor &PP) {
989  const auto &SM = PP.getSourceManager();
990  const auto &LangOpts = PP.getLangOpts();
991  Token Result;
992  if (Lexer::getRawToken(SM.getSpellingLoc(Loc), Result, SM, LangOpts, false))
993  return None;
994  if (Result.is(tok::raw_identifier))
995  PP.LookUpIdentifierInfo(Result);
996  IdentifierInfo *IdentifierInfo = Result.getIdentifierInfo();
997  if (!IdentifierInfo || !IdentifierInfo->hadMacroDefinition())
998  return None;
999 
1000  std::pair<FileID, unsigned int> DecLoc = SM.getDecomposedExpansionLoc(Loc);
1001  // Get the definition just before the searched location so that a macro
1002  // referenced in a '#undef MACRO' can still be found.
1003  SourceLocation BeforeSearchedLocation =
1004  SM.getMacroArgExpandedLocation(SM.getLocForStartOfFile(DecLoc.first)
1005  .getLocWithOffset(DecLoc.second - 1));
1006  MacroDefinition MacroDef =
1007  PP.getMacroDefinitionAtLoc(IdentifierInfo, BeforeSearchedLocation);
1008  if (auto *MI = MacroDef.getMacroInfo())
1009  return DefinedMacro{IdentifierInfo->getName(), MI};
1010  return None;
1011 }
1012 
1013 llvm::Expected<std::string> Edit::apply() const {
1014  return tooling::applyAllReplacements(InitialCode, Replacements);
1015 }
1016 
1017 std::vector<TextEdit> Edit::asTextEdits() const {
1018  return replacementsToEdits(InitialCode, Replacements);
1019 }
1020 
1021 bool Edit::canApplyTo(llvm::StringRef Code) const {
1022  // Create line iterators, since line numbers are important while applying our
1023  // edit we cannot skip blank lines.
1024  auto LHS = llvm::MemoryBuffer::getMemBuffer(Code);
1025  llvm::line_iterator LHSIt(*LHS, /*SkipBlanks=*/false);
1026 
1027  auto RHS = llvm::MemoryBuffer::getMemBuffer(InitialCode);
1028  llvm::line_iterator RHSIt(*RHS, /*SkipBlanks=*/false);
1029 
1030  // Compare the InitialCode we prepared the edit for with the Code we received
1031  // line by line to make sure there are no differences.
1032  // FIXME: This check is too conservative now, it should be enough to only
1033  // check lines around the replacements contained inside the Edit.
1034  while (!LHSIt.is_at_eof() && !RHSIt.is_at_eof()) {
1035  if (*LHSIt != *RHSIt)
1036  return false;
1037  ++LHSIt;
1038  ++RHSIt;
1039  }
1040 
1041  // After we reach EOF for any of the files we make sure the other one doesn't
1042  // contain any additional content except empty lines, they should not
1043  // interfere with the edit we produced.
1044  while (!LHSIt.is_at_eof()) {
1045  if (!LHSIt->empty())
1046  return false;
1047  ++LHSIt;
1048  }
1049  while (!RHSIt.is_at_eof()) {
1050  if (!RHSIt->empty())
1051  return false;
1052  ++RHSIt;
1053  }
1054  return true;
1055 }
1056 
1058  if (auto NewEdits = cleanupAndFormat(E.InitialCode, E.Replacements, Style))
1059  E.Replacements = std::move(*NewEdits);
1060  else
1061  return NewEdits.takeError();
1062  return llvm::Error::success();
1063 }
1064 
1065 EligibleRegion getEligiblePoints(llvm::StringRef Code,
1066  llvm::StringRef FullyQualifiedName,
1067  const format::FormatStyle &Style) {
1068  EligibleRegion ER;
1069  // Start with global namespace.
1070  std::vector<std::string> Enclosing = {""};
1071  // FIXME: In addition to namespaces try to generate events for function
1072  // definitions as well. One might use a closing parantheses(")" followed by an
1073  // opening brace "{" to trigger the start.
1074  parseNamespaceEvents(Code, Style, [&](NamespaceEvent Event) {
1075  // Using Directives only introduces declarations to current scope, they do
1076  // not change the current namespace, so skip them.
1077  if (Event.Trigger == NamespaceEvent::UsingDirective)
1078  return;
1079  // Do not qualify the global namespace.
1080  if (!Event.Payload.empty())
1081  Event.Payload.append("::");
1082 
1083  std::string CurrentNamespace;
1084  if (Event.Trigger == NamespaceEvent::BeginNamespace) {
1085  Enclosing.emplace_back(std::move(Event.Payload));
1086  CurrentNamespace = Enclosing.back();
1087  // parseNameSpaceEvents reports the beginning position of a token; we want
1088  // to insert after '{', so increment by one.
1089  ++Event.Pos.character;
1090  } else {
1091  // Event.Payload points to outer namespace when exiting a scope, so use
1092  // the namespace we've last entered instead.
1093  CurrentNamespace = std::move(Enclosing.back());
1094  Enclosing.pop_back();
1095  assert(Enclosing.back() == Event.Payload);
1096  }
1097 
1098  // Ignore namespaces that are not a prefix of the target.
1099  if (!FullyQualifiedName.startswith(CurrentNamespace))
1100  return;
1101 
1102  // Prefer the namespace that shares the longest prefix with target.
1103  if (CurrentNamespace.size() > ER.EnclosingNamespace.size()) {
1104  ER.EligiblePoints.clear();
1105  ER.EnclosingNamespace = CurrentNamespace;
1106  }
1107  if (CurrentNamespace.size() == ER.EnclosingNamespace.size())
1108  ER.EligiblePoints.emplace_back(std::move(Event.Pos));
1109  });
1110  // If there were no shared namespaces just return EOF.
1111  if (ER.EligiblePoints.empty()) {
1112  assert(ER.EnclosingNamespace.empty());
1113  ER.EligiblePoints.emplace_back(offsetToPosition(Code, Code.size()));
1114  }
1115  return ER;
1116 }
1117 
1118 bool isHeaderFile(llvm::StringRef FileName,
1119  llvm::Optional<LangOptions> LangOpts) {
1120  // Respect the langOpts, for non-file-extension cases, e.g. standard library
1121  // files.
1122  if (LangOpts && LangOpts->IsHeaderFile)
1123  return true;
1124  namespace types = clang::driver::types;
1125  auto Lang = types::lookupTypeForExtension(
1126  llvm::sys::path::extension(FileName).substr(1));
1127  return Lang != types::TY_INVALID && types::onlyPrecompileType(Lang);
1128 }
1129 
1130 } // namespace clangd
1131 } // namespace clang
SourceLocation Loc
&#39;#&#39; location in the include directive
static SourceRange getExpansionTokenRangeInSameFile(SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:489
llvm::StringSet collectWords(llvm::StringRef Content)
Collects words from the source code.
Definition: SourceCode.cpp:949
std::string Code
llvm::Expected< tooling::Replacements > cleanupAndFormat(StringRef Code, const tooling::Replacements &Replaces, const format::FormatStyle &Style)
Cleanup and format the given replacements.
Definition: SourceCode.cpp:707
Position start
The range&#39;s start position.
Definition: Protocol.h:158
static SourceRange rangeInCommonFile(SourceRange R, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:456
size_t lspLength(llvm::StringRef Code)
Definition: SourceCode.cpp:131
std::array< uint8_t, 8 > FileDigest
Definition: SourceCode.h:38
std::pair< StringRef, StringRef > splitQualifiedName(StringRef QName)
Definition: SourceCode.cpp:599
bool isInsideMainFile(SourceLocation Loc, const SourceManager &SM)
Returns true iff Loc is inside the main file.
Definition: SourceCode.cpp:534
static SourceRange unionTokenRange(SourceRange R1, SourceRange R2, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:438
llvm::Expected< std::string > apply() const
Returns the file contents after changes are applied.
An Event<T> allows events of type T to be broadcast to listeners.
Definition: Function.h:31
SourceLocation getBeginningOfIdentifier(const Position &Pos, const SourceManager &SM, const LangOptions &LangOpts)
Get the beginning SourceLocation at a specified Pos in the main file.
Definition: SourceCode.cpp:277
bool halfOpenRangeContains(const SourceManager &Mgr, SourceRange R, SourceLocation L)
Returns true iff L is contained in R.
Definition: SourceCode.cpp:355
std::vector< CodeCompletionResult > Results
Values in a Context are indexed by typed keys.
Definition: Context.h:40
llvm::unique_function< void(llvm::Expected< T >)> Callback
A Callback<T> is a void function that accepts Expected<T>.
Definition: Function.h:28
MockFSProvider FS
Documents should not be synced at all.
bool halfOpenRangeTouches(const SourceManager &Mgr, SourceRange R, SourceLocation L)
Returns true iff L is contained in R or L is equal to the end point of R.
Definition: SourceCode.cpp:370
std::string InitialCode
Definition: SourceCode.h:206
void elog(const char *Fmt, Ts &&... Vals)
Definition: Logger.h:56
bool isSpelledInSource(SourceLocation Loc, const SourceManager &SM)
Returns true if the token at Loc is spelled in the source code.
Definition: SourceCode.cpp:217
llvm::Expected< SourceLocation > sourceLocationInMainFile(const SourceManager &SM, Position P)
Return the file location, corresponding to P.
Definition: SourceCode.cpp:571
std::vector< std::string > visibleNamespaces(llvm::StringRef Code, const format::FormatStyle &Style)
Heuristically determine namespaces visible at a point, without parsing Code.
Definition: SourceCode.cpp:901
bool isRangeConsecutive(const Range &Left, const Range &Right)
Definition: SourceCode.cpp:670
bool canApplyTo(llvm::StringRef Code) const
Checks whether the Replacements are applicable to given Code.
bool isValidFileRange(const SourceManager &Mgr, SourceRange R)
Returns true iff all of the following conditions hold:
Definition: SourceCode.cpp:340
const Type * get(const Key< Type > &Key) const
Get data stored for a typed Key.
Definition: Context.h:100
std::pair< size_t, size_t > offsetToClangLineColumn(llvm::StringRef Code, size_t Offset)
Definition: SourceCode.cpp:589
TextEdit toTextEdit(const FixItHint &FixIt, const SourceManager &M, const LangOptions &L)
Definition: SourceCode.cpp:661
std::string newText
The string to be inserted.
Definition: Protocol.h:209
std::string QName
Position offsetToPosition(llvm::StringRef Code, size_t Offset)
Turn an offset in Code into a [line, column] pair.
Definition: SourceCode.cpp:188
llvm::Expected< size_t > positionToOffset(llvm::StringRef Code, Position P, bool AllowColumnsBeyondLineLength)
Turn a [line, column] pair into an offset in Code.
Definition: SourceCode.cpp:155
Range range
The range of the text document to be manipulated.
Definition: Protocol.h:205
std::string EnclosingNamespace
Namespace that owns all of the EligiblePoints, e.g.
Definition: SourceCode.h:272
llvm::unique_function< void()> Action
void log(const char *Fmt, Ts &&... Vals)
Definition: Logger.h:62
static size_t measureUnits(llvm::StringRef U8, int Units, OffsetEncoding Enc, bool &Valid)
Definition: SourceCode.cpp:87
EligibleRegion getEligiblePoints(llvm::StringRef Code, llvm::StringRef FullyQualifiedName, const format::FormatStyle &Style)
Returns most eligible region to insert a definition for FullyQualifiedName in the Code...
static SourceRange toTokenRange(CharSourceRange Range, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:429
static const Context & current()
Returns the context for the current thread, creating it if needed.
Definition: Context.cpp:27
enum clang::clangd::@771::NamespaceEvent::@3 Trigger
Key< OffsetEncoding > kCurrentOffsetEncoding
Definition: SourceCode.cpp:124
CharTypeSet calculateRoles(llvm::StringRef Text, llvm::MutableArrayRef< CharRole > Roles)
Definition: FuzzyMatch.cpp:154
static void lex(llvm::StringRef Code, const LangOptions &LangOpts, llvm::function_ref< void(const clang::Token &, const SourceManager &SM)> Action)
Definition: SourceCode.cpp:716
PathRef FileName
Represents locations that can accept a definition.
Definition: SourceCode.h:268
llvm::Optional< Range > getTokenRange(const SourceManager &SM, const LangOptions &LangOpts, SourceLocation TokLoc)
Returns the taken range at TokLoc.
Definition: SourceCode.cpp:227
Position Pos
Definition: SourceCode.cpp:772
llvm::Optional< FileDigest > digestFile(const SourceManager &SM, FileID FID)
Definition: SourceCode.cpp:685
Position sourceLocToPosition(const SourceManager &SM, SourceLocation Loc)
Turn a SourceLocation into a [line, column] pair.
Definition: SourceCode.cpp:200
format::FormatStyle getFormatStyleForFile(llvm::StringRef File, llvm::StringRef Content, llvm::vfs::FileSystem *FS)
Choose the clang-format style we should apply to a certain file.
Definition: SourceCode.cpp:693
static SourceLocation getLocForTokenBegin(SourceLocation EndLoc, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:421
tooling::Replacements Replacements
Definition: SourceCode.h:205
FileDigest digest(llvm::StringRef Content)
Definition: SourceCode.cpp:675
llvm::Optional< SourceRange > toHalfOpenFileRange(const SourceManager &SM, const LangOptions &LangOpts, SourceRange R)
Turns a token range into a half-open range and checks its correctness.
Definition: SourceCode.cpp:538
SourceLocation includeHashLoc(FileID IncludedFile, const SourceManager &SM)
Returns the #include location through which IncludedFIle was loaded.
Definition: SourceCode.cpp:375
int line
Line position in a document (zero-based).
Definition: Protocol.h:129
llvm::StringRef toSourceCode(const SourceManager &SM, SourceRange R)
Returns the source code covered by the source range.
Definition: SourceCode.cpp:560
size_t Offset
int character
Character offset on a line in a document (zero-based).
Definition: Protocol.h:134
static SourceRange getTokenFileRange(SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:506
===– Representation.cpp - ClangDoc Representation --------—*- C++ -*-===//
TextEdit replacementToEdit(llvm::StringRef Code, const tooling::Replacement &R)
Definition: SourceCode.cpp:606
std::string Payload
Definition: SourceCode.cpp:771
llvm::Optional< std::string > getCanonicalPath(const FileEntry *F, const SourceManager &SourceMgr)
Get the canonical path of F.
Definition: SourceCode.cpp:622
static SourceLocation getLocForTokenEnd(SourceLocation BeginLoc, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:413
std::vector< TextEdit > replacementsToEdits(llvm::StringRef Code, const tooling::Replacements &Repls)
Definition: SourceCode.cpp:614
std::vector< TextEdit > asTextEdits() const
Represents Replacements as TextEdits that are available for use in LSP.
bool isHeaderFile(llvm::StringRef FileName, llvm::Optional< LangOptions > LangOpts)
Infers whether this is a header from the FileName and LangOpts (if presents).
const Expr * E
llvm::Optional< FixItHint > FixIt
static std::string join(ArrayRef< SpecialMemberFunctionsCheck::SpecialMemberFunctionKind > SMFS, llvm::StringRef AndOr)
Position end
The range&#39;s end position.
Definition: Protocol.h:161
static OffsetEncoding lspEncoding()
Definition: SourceCode.cpp:125
std::vector< Range > collectIdentifierRanges(llvm::StringRef Identifier, llvm::StringRef Content, const LangOptions &LangOpts)
Collects all ranges of the given identifier in the source code.
Definition: SourceCode.cpp:746
llvm::StringMap< unsigned > collectIdentifiers(llvm::StringRef Content, const format::FormatStyle &Style)
Collects identifiers with counts in the source code.
Definition: SourceCode.cpp:735
llvm::Optional< DefinedMacro > locateMacroAt(SourceLocation Loc, Preprocessor &PP)
Gets the macro at a specified Loc.
Definition: SourceCode.cpp:987
static bool iterateCodepoints(llvm::StringRef U8, const Callback &CB)
Definition: SourceCode.cpp:57
std::string Word
A set of edits generated for a single file.
Definition: SourceCode.h:204
static unsigned getTokenLengthAtLoc(SourceLocation Loc, const SourceManager &SM, const LangOptions &LangOpts)
Definition: SourceCode.cpp:397
llvm::Error reformatEdit(Edit &E, const format::FormatStyle &Style)
Formats the edits and code around it according to Style.
unsigned Lines
std::vector< Position > EligiblePoints
Offsets into the code marking eligible points to insert a function definition.
Definition: SourceCode.h:275
Range halfOpenToRange(const SourceManager &SM, CharSourceRange R)
Definition: SourceCode.cpp:581
static cl::opt< std::string > FormatStyle("format-style", cl::desc(R"( Style for formatting code around applied fixes: - 'none' (default) turns off formatting - 'file' (literally 'file', not a placeholder) uses .clang-format file in the closest parent directory - '{ <json> }' specifies options inline, e.g. -format-style='{BasedOnStyle: llvm, IndentWidth: 8}' - 'llvm', 'google', 'webkit', 'mozilla' See clang-format documentation for the up-to-date information about formatting styles and options. This option overrides the 'FormatStyle` option in .clang-tidy file, if any. )"), cl::init("none"), cl::cat(ClangTidyCategory))