1//===--- Rewriter.cpp - Code rewriting interface --------------------------===//
2//
3//                     The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9//
10//  This file defines the Rewriter class, which is used for code
11//  transformations.
12//
13//===----------------------------------------------------------------------===//
14
15#include "clang/Rewrite/Core/Rewriter.h"
16#include "clang/Basic/Diagnostic.h"
17#include "clang/Basic/DiagnosticIDs.h"
18#include "clang/Basic/FileManager.h"
19#include "clang/Basic/SourceManager.h"
20#include "clang/Lex/Lexer.h"
21#include "llvm/ADT/SmallString.h"
22#include "llvm/Config/llvm-config.h"
23#include "llvm/Support/FileSystem.h"
24#include "llvm/Support/raw_ostream.h"
25using namespace clang;
26
27raw_ostream &RewriteBuffer::write(raw_ostream &os) const {
28  // Walk RewriteRope chunks efficiently using MoveToNextPiece() instead of the
29  // character iterator.
30  for (RopePieceBTreeIterator I = begin(), E = end(); I != E;
31       I.MoveToNextPiece())
32    os << I.piece();
33  return os;
34}
35
36/// \brief Return true if this character is non-new-line whitespace:
37/// ' ', '\\t', '\\f', '\\v', '\\r'.
38static inline bool isWhitespace(unsigned char c) {
39  switch (c) {
40  case ' ':
41  case '\t':
42  case '\f':
43  case '\v':
44  case '\r':
45    return true;
46  default:
47    return false;
48  }
49}
50
51void RewriteBuffer::RemoveText(unsigned OrigOffset, unsigned Size,
52                               bool removeLineIfEmpty) {
53  // Nothing to remove, exit early.
54  if (Size == 0) return;
55
56  unsigned RealOffset = getMappedOffset(OrigOffset, true);
57  assert(RealOffset+Size <= Buffer.size() && "Invalid location");
58
59  // Remove the dead characters.
60  Buffer.erase(RealOffset, Size);
61
62  // Add a delta so that future changes are offset correctly.
63  AddReplaceDelta(OrigOffset, -Size);
64
65  if (removeLineIfEmpty) {
66    // Find the line that the remove occurred and if it is completely empty
67    // remove the line as well.
68
69    iterator curLineStart = begin();
70    unsigned curLineStartOffs = 0;
71    iterator posI = begin();
72    for (unsigned i = 0; i != RealOffset; ++i) {
73      if (*posI == '\n') {
74        curLineStart = posI;
75        ++curLineStart;
76        curLineStartOffs = i + 1;
77      }
78      ++posI;
79    }
80
81    unsigned lineSize = 0;
82    posI = curLineStart;
83    while (posI != end() && isWhitespace(*posI)) {
84      ++posI;
85      ++lineSize;
86    }
87    if (posI != end() && *posI == '\n') {
88      Buffer.erase(curLineStartOffs, lineSize + 1/* + '\n'*/);
89      AddReplaceDelta(curLineStartOffs, -(lineSize + 1/* + '\n'*/));
90    }
91  }
92}
93
94void RewriteBuffer::InsertText(unsigned OrigOffset, StringRef Str,
95                               bool InsertAfter) {
96
97  // Nothing to insert, exit early.
98  if (Str.empty()) return;
99
100  unsigned RealOffset = getMappedOffset(OrigOffset, InsertAfter);
101  Buffer.insert(RealOffset, Str.begin(), Str.end());
102
103  // Add a delta so that future changes are offset correctly.
104  AddInsertDelta(OrigOffset, Str.size());
105}
106
107/// ReplaceText - This method replaces a range of characters in the input
108/// buffer with a new string.  This is effectively a combined "remove+insert"
109/// operation.
110void RewriteBuffer::ReplaceText(unsigned OrigOffset, unsigned OrigLength,
111                                StringRef NewStr) {
112  unsigned RealOffset = getMappedOffset(OrigOffset, true);
113  Buffer.erase(RealOffset, OrigLength);
114  Buffer.insert(RealOffset, NewStr.begin(), NewStr.end());
115  if (OrigLength != NewStr.size())
116    AddReplaceDelta(OrigOffset, NewStr.size() - OrigLength);
117}
118
119
120//===----------------------------------------------------------------------===//
121// Rewriter class
122//===----------------------------------------------------------------------===//
123
124/// getRangeSize - Return the size in bytes of the specified range if they
125/// are in the same file.  If not, this returns -1.
126int Rewriter::getRangeSize(const CharSourceRange &Range,
127                           RewriteOptions opts) const {
128  if (!isRewritable(Range.getBegin()) ||
129      !isRewritable(Range.getEnd())) return -1;
130
131  FileID StartFileID, EndFileID;
132  unsigned StartOff, EndOff;
133
134  StartOff = getLocationOffsetAndFileID(Range.getBegin(), StartFileID);
135  EndOff   = getLocationOffsetAndFileID(Range.getEnd(), EndFileID);
136
137  if (StartFileID != EndFileID)
138    return -1;
139
140  // If edits have been made to this buffer, the delta between the range may
141  // have changed.
142  std::map<FileID, RewriteBuffer>::const_iterator I =
143    RewriteBuffers.find(StartFileID);
144  if (I != RewriteBuffers.end()) {
145    const RewriteBuffer &RB = I->second;
146    EndOff = RB.getMappedOffset(EndOff, opts.IncludeInsertsAtEndOfRange);
147    StartOff = RB.getMappedOffset(StartOff, !opts.IncludeInsertsAtBeginOfRange);
148  }
149
150
151  // Adjust the end offset to the end of the last token, instead of being the
152  // start of the last token if this is a token range.
153  if (Range.isTokenRange())
154    EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
155
156  return EndOff-StartOff;
157}
158
159int Rewriter::getRangeSize(SourceRange Range, RewriteOptions opts) const {
160  return getRangeSize(CharSourceRange::getTokenRange(Range), opts);
161}
162
163
164/// getRewrittenText - Return the rewritten form of the text in the specified
165/// range.  If the start or end of the range was unrewritable or if they are
166/// in different buffers, this returns an empty string.
167///
168/// Note that this method is not particularly efficient.
169///
170std::string Rewriter::getRewrittenText(SourceRange Range) const {
171  if (!isRewritable(Range.getBegin()) ||
172      !isRewritable(Range.getEnd()))
173    return "";
174
175  FileID StartFileID, EndFileID;
176  unsigned StartOff, EndOff;
177  StartOff = getLocationOffsetAndFileID(Range.getBegin(), StartFileID);
178  EndOff   = getLocationOffsetAndFileID(Range.getEnd(), EndFileID);
179
180  if (StartFileID != EndFileID)
181    return ""; // Start and end in different buffers.
182
183  // If edits have been made to this buffer, the delta between the range may
184  // have changed.
185  std::map<FileID, RewriteBuffer>::const_iterator I =
186    RewriteBuffers.find(StartFileID);
187  if (I == RewriteBuffers.end()) {
188    // If the buffer hasn't been rewritten, just return the text from the input.
189    const char *Ptr = SourceMgr->getCharacterData(Range.getBegin());
190
191    // Adjust the end offset to the end of the last token, instead of being the
192    // start of the last token.
193    EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
194    return std::string(Ptr, Ptr+EndOff-StartOff);
195  }
196
197  const RewriteBuffer &RB = I->second;
198  EndOff = RB.getMappedOffset(EndOff, true);
199  StartOff = RB.getMappedOffset(StartOff);
200
201  // Adjust the end offset to the end of the last token, instead of being the
202  // start of the last token.
203  EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
204
205  // Advance the iterators to the right spot, yay for linear time algorithms.
206  RewriteBuffer::iterator Start = RB.begin();
207  std::advance(Start, StartOff);
208  RewriteBuffer::iterator End = Start;
209  std::advance(End, EndOff-StartOff);
210
211  return std::string(Start, End);
212}
213
214unsigned Rewriter::getLocationOffsetAndFileID(SourceLocation Loc,
215                                              FileID &FID) const {
216  assert(Loc.isValid() && "Invalid location");
217  std::pair<FileID,unsigned> V = SourceMgr->getDecomposedLoc(Loc);
218  FID = V.first;
219  return V.second;
220}
221
222
223/// getEditBuffer - Get or create a RewriteBuffer for the specified FileID.
224///
225RewriteBuffer &Rewriter::getEditBuffer(FileID FID) {
226  std::map<FileID, RewriteBuffer>::iterator I =
227    RewriteBuffers.lower_bound(FID);
228  if (I != RewriteBuffers.end() && I->first == FID)
229    return I->second;
230  I = RewriteBuffers.insert(I, std::make_pair(FID, RewriteBuffer()));
231
232  StringRef MB = SourceMgr->getBufferData(FID);
233  I->second.Initialize(MB.begin(), MB.end());
234
235  return I->second;
236}
237
238/// InsertText - Insert the specified string at the specified location in the
239/// original buffer.
240bool Rewriter::InsertText(SourceLocation Loc, StringRef Str,
241                          bool InsertAfter, bool indentNewLines) {
242  if (!isRewritable(Loc)) return true;
243  FileID FID;
244  unsigned StartOffs = getLocationOffsetAndFileID(Loc, FID);
245
246  SmallString<128> indentedStr;
247  if (indentNewLines && Str.find('\n') != StringRef::npos) {
248    StringRef MB = SourceMgr->getBufferData(FID);
249
250    unsigned lineNo = SourceMgr->getLineNumber(FID, StartOffs) - 1;
251    const SrcMgr::ContentCache *
252        Content = SourceMgr->getSLocEntry(FID).getFile().getContentCache();
253    unsigned lineOffs = Content->SourceLineCache[lineNo];
254
255    // Find the whitespace at the start of the line.
256    StringRef indentSpace;
257    {
258      unsigned i = lineOffs;
259      while (isWhitespace(MB[i]))
260        ++i;
261      indentSpace = MB.substr(lineOffs, i-lineOffs);
262    }
263
264    SmallVector<StringRef, 4> lines;
265    Str.split(lines, "\n");
266
267    for (unsigned i = 0, e = lines.size(); i != e; ++i) {
268      indentedStr += lines[i];
269      if (i < e-1) {
270        indentedStr += '\n';
271        indentedStr += indentSpace;
272      }
273    }
274    Str = indentedStr.str();
275  }
276
277  getEditBuffer(FID).InsertText(StartOffs, Str, InsertAfter);
278  return false;
279}
280
281bool Rewriter::InsertTextAfterToken(SourceLocation Loc, StringRef Str) {
282  if (!isRewritable(Loc)) return true;
283  FileID FID;
284  unsigned StartOffs = getLocationOffsetAndFileID(Loc, FID);
285  RewriteOptions rangeOpts;
286  rangeOpts.IncludeInsertsAtBeginOfRange = false;
287  StartOffs += getRangeSize(SourceRange(Loc, Loc), rangeOpts);
288  getEditBuffer(FID).InsertText(StartOffs, Str, /*InsertAfter*/true);
289  return false;
290}
291
292/// RemoveText - Remove the specified text region.
293bool Rewriter::RemoveText(SourceLocation Start, unsigned Length,
294                          RewriteOptions opts) {
295  if (!isRewritable(Start)) return true;
296  FileID FID;
297  unsigned StartOffs = getLocationOffsetAndFileID(Start, FID);
298  getEditBuffer(FID).RemoveText(StartOffs, Length, opts.RemoveLineIfEmpty);
299  return false;
300}
301
302/// ReplaceText - This method replaces a range of characters in the input
303/// buffer with a new string.  This is effectively a combined "remove/insert"
304/// operation.
305bool Rewriter::ReplaceText(SourceLocation Start, unsigned OrigLength,
306                           StringRef NewStr) {
307  if (!isRewritable(Start)) return true;
308  FileID StartFileID;
309  unsigned StartOffs = getLocationOffsetAndFileID(Start, StartFileID);
310
311  getEditBuffer(StartFileID).ReplaceText(StartOffs, OrigLength, NewStr);
312  return false;
313}
314
315bool Rewriter::ReplaceText(SourceRange range, SourceRange replacementRange) {
316  if (!isRewritable(range.getBegin())) return true;
317  if (!isRewritable(range.getEnd())) return true;
318  if (replacementRange.isInvalid()) return true;
319  SourceLocation start = range.getBegin();
320  unsigned origLength = getRangeSize(range);
321  unsigned newLength = getRangeSize(replacementRange);
322  FileID FID;
323  unsigned newOffs = getLocationOffsetAndFileID(replacementRange.getBegin(),
324                                                FID);
325  StringRef MB = SourceMgr->getBufferData(FID);
326  return ReplaceText(start, origLength, MB.substr(newOffs, newLength));
327}
328
329bool Rewriter::IncreaseIndentation(CharSourceRange range,
330                                   SourceLocation parentIndent) {
331  if (range.isInvalid()) return true;
332  if (!isRewritable(range.getBegin())) return true;
333  if (!isRewritable(range.getEnd())) return true;
334  if (!isRewritable(parentIndent)) return true;
335
336  FileID StartFileID, EndFileID, parentFileID;
337  unsigned StartOff, EndOff, parentOff;
338
339  StartOff = getLocationOffsetAndFileID(range.getBegin(), StartFileID);
340  EndOff   = getLocationOffsetAndFileID(range.getEnd(), EndFileID);
341  parentOff = getLocationOffsetAndFileID(parentIndent, parentFileID);
342
343  if (StartFileID != EndFileID || StartFileID != parentFileID)
344    return true;
345  if (StartOff > EndOff)
346    return true;
347
348  FileID FID = StartFileID;
349  StringRef MB = SourceMgr->getBufferData(FID);
350
351  unsigned parentLineNo = SourceMgr->getLineNumber(FID, parentOff) - 1;
352  unsigned startLineNo = SourceMgr->getLineNumber(FID, StartOff) - 1;
353  unsigned endLineNo = SourceMgr->getLineNumber(FID, EndOff) - 1;
354
355  const SrcMgr::ContentCache *
356      Content = SourceMgr->getSLocEntry(FID).getFile().getContentCache();
357
358  // Find where the lines start.
359  unsigned parentLineOffs = Content->SourceLineCache[parentLineNo];
360  unsigned startLineOffs = Content->SourceLineCache[startLineNo];
361
362  // Find the whitespace at the start of each line.
363  StringRef parentSpace, startSpace;
364  {
365    unsigned i = parentLineOffs;
366    while (isWhitespace(MB[i]))
367      ++i;
368    parentSpace = MB.substr(parentLineOffs, i-parentLineOffs);
369
370    i = startLineOffs;
371    while (isWhitespace(MB[i]))
372      ++i;
373    startSpace = MB.substr(startLineOffs, i-startLineOffs);
374  }
375  if (parentSpace.size() >= startSpace.size())
376    return true;
377  if (!startSpace.startswith(parentSpace))
378    return true;
379
380  StringRef indent = startSpace.substr(parentSpace.size());
381
382  // Indent the lines between start/end offsets.
383  RewriteBuffer &RB = getEditBuffer(FID);
384  for (unsigned lineNo = startLineNo; lineNo <= endLineNo; ++lineNo) {
385    unsigned offs = Content->SourceLineCache[lineNo];
386    unsigned i = offs;
387    while (isWhitespace(MB[i]))
388      ++i;
389    StringRef origIndent = MB.substr(offs, i-offs);
390    if (origIndent.startswith(startSpace))
391      RB.InsertText(offs, indent, /*InsertAfter=*/false);
392  }
393
394  return false;
395}
396
397namespace {
398// A wrapper for a file stream that atomically overwrites the target.
399//
400// Creates a file output stream for a temporary file in the constructor,
401// which is later accessible via getStream() if ok() return true.
402// Flushes the stream and moves the temporary file to the target location
403// in the destructor.
404class AtomicallyMovedFile {
405public:
406  AtomicallyMovedFile(DiagnosticsEngine &Diagnostics, StringRef Filename,
407                      bool &AllWritten)
408    : Diagnostics(Diagnostics), Filename(Filename), AllWritten(AllWritten) {
409    TempFilename = Filename;
410    TempFilename += "-%%%%%%%%";
411    int FD;
412    if (llvm::sys::fs::createUniqueFile(TempFilename.str(), FD, TempFilename)) {
413      AllWritten = false;
414      Diagnostics.Report(clang::diag::err_unable_to_make_temp)
415        << TempFilename;
416    } else {
417      FileStream.reset(new llvm::raw_fd_ostream(FD, /*shouldClose=*/true));
418    }
419  }
420
421  ~AtomicallyMovedFile() {
422    if (!ok()) return;
423
424    FileStream->flush();
425#ifdef LLVM_ON_WIN32
426    // Win32 does not allow rename/removing opened files.
427    FileStream.reset();
428#endif
429    if (std::error_code ec =
430            llvm::sys::fs::rename(TempFilename.str(), Filename)) {
431      AllWritten = false;
432      Diagnostics.Report(clang::diag::err_unable_to_rename_temp)
433        << TempFilename << Filename << ec.message();
434      // If the remove fails, there's not a lot we can do - this is already an
435      // error.
436      llvm::sys::fs::remove(TempFilename.str());
437    }
438  }
439
440  bool ok() { return (bool)FileStream; }
441  raw_ostream &getStream() { return *FileStream; }
442
443private:
444  DiagnosticsEngine &Diagnostics;
445  StringRef Filename;
446  SmallString<128> TempFilename;
447  std::unique_ptr<llvm::raw_fd_ostream> FileStream;
448  bool &AllWritten;
449};
450} // end anonymous namespace
451
452bool Rewriter::overwriteChangedFiles() {
453  bool AllWritten = true;
454  for (buffer_iterator I = buffer_begin(), E = buffer_end(); I != E; ++I) {
455    const FileEntry *Entry =
456        getSourceMgr().getFileEntryForID(I->first);
457    AtomicallyMovedFile File(getSourceMgr().getDiagnostics(), Entry->getName(),
458                             AllWritten);
459    if (File.ok()) {
460      I->second.write(File.getStream());
461    }
462  }
463  return !AllWritten;
464}
465