Rewriter.cpp revision 10c8d9e63bcc96d55f788e7c08b72ce626c8aeec 
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/Rewriter.h"
16#include "clang/AST/Stmt.h"
17#include "clang/AST/Decl.h"
18#include "clang/Lex/Lexer.h"
19#include "clang/Basic/SourceManager.h"
20#include "llvm/Support/raw_ostream.h"
21using namespace clang;
22
23llvm::raw_ostream &RewriteBuffer::write(llvm::raw_ostream &os) const {
24  // FIXME: eliminate the copy by writing out each chunk at a time
25  os << std::string(begin(), end());
26  return os;
27}
28
29/// \brief Return true if this character is non-new-line whitespace:
30/// ' ', '\t', '\f', '\v', '\r'.
31static inline bool isWhitespace(unsigned char c) {
32  switch (c) {
33  case ' ':
34  case '\t':
35  case '\f':
36  case '\v':
37  case '\r':
38    return true;
39  default:
40    return false;
41  }
42}
43
44void RewriteBuffer::RemoveText(unsigned OrigOffset, unsigned Size,
45                               bool removeLineIfEmpty) {
46  // Nothing to remove, exit early.
47  if (Size == 0) return;
48
49  unsigned RealOffset = getMappedOffset(OrigOffset, true);
50  assert(RealOffset+Size < Buffer.size() && "Invalid location");
51
52  // Remove the dead characters.
53  Buffer.erase(RealOffset, Size);
54
55  // Add a delta so that future changes are offset correctly.
56  AddReplaceDelta(OrigOffset, -Size);
57
58  if (removeLineIfEmpty) {
59    // Find the line that the remove occurred and if it is completely empty
60    // remove the line as well.
61
62    iterator curLineStart = begin();
63    unsigned curLineStartOffs = 0;
64    iterator posI = begin();
65    for (unsigned i = 0; i != RealOffset; ++i) {
66      if (*posI == '\n') {
67        curLineStart = posI;
68        ++curLineStart;
69        curLineStartOffs = i + 1;
70      }
71      ++posI;
72    }
73
74    unsigned lineSize = 0;
75    posI = curLineStart;
76    while (posI != end() && isWhitespace(*posI)) {
77      ++posI;
78      ++lineSize;
79    }
80    if (posI != end() && *posI == '\n') {
81      Buffer.erase(curLineStartOffs, lineSize + 1/* + '\n'*/);
82      AddReplaceDelta(curLineStartOffs, -(lineSize + 1/* + '\n'*/));
83    }
84  }
85}
86
87void RewriteBuffer::InsertText(unsigned OrigOffset, llvm::StringRef Str,
88                               bool InsertAfter) {
89
90  // Nothing to insert, exit early.
91  if (Str.empty()) return;
92
93  unsigned RealOffset = getMappedOffset(OrigOffset, InsertAfter);
94  Buffer.insert(RealOffset, Str.begin(), Str.end());
95
96  // Add a delta so that future changes are offset correctly.
97  AddInsertDelta(OrigOffset, Str.size());
98}
99
100/// ReplaceText - This method replaces a range of characters in the input
101/// buffer with a new string.  This is effectively a combined "remove+insert"
102/// operation.
103void RewriteBuffer::ReplaceText(unsigned OrigOffset, unsigned OrigLength,
104                                llvm::StringRef NewStr) {
105  unsigned RealOffset = getMappedOffset(OrigOffset, true);
106  Buffer.erase(RealOffset, OrigLength);
107  Buffer.insert(RealOffset, NewStr.begin(), NewStr.end());
108  if (OrigLength != NewStr.size())
109    AddReplaceDelta(OrigOffset, NewStr.size() - OrigLength);
110}
111
112
113//===----------------------------------------------------------------------===//
114// Rewriter class
115//===----------------------------------------------------------------------===//
116
117/// getRangeSize - Return the size in bytes of the specified range if they
118/// are in the same file.  If not, this returns -1.
119int Rewriter::getRangeSize(const CharSourceRange &Range,
120                           RewriteOptions opts) const {
121  if (!isRewritable(Range.getBegin()) ||
122      !isRewritable(Range.getEnd())) return -1;
123
124  FileID StartFileID, EndFileID;
125  unsigned StartOff, EndOff;
126
127  StartOff = getLocationOffsetAndFileID(Range.getBegin(), StartFileID);
128  EndOff   = getLocationOffsetAndFileID(Range.getEnd(), EndFileID);
129
130  if (StartFileID != EndFileID)
131    return -1;
132
133  // If edits have been made to this buffer, the delta between the range may
134  // have changed.
135  std::map<FileID, RewriteBuffer>::const_iterator I =
136    RewriteBuffers.find(StartFileID);
137  if (I != RewriteBuffers.end()) {
138    const RewriteBuffer &RB = I->second;
139    EndOff = RB.getMappedOffset(EndOff, opts.IncludeInsertsAtEndOfRange);
140    StartOff = RB.getMappedOffset(StartOff, !opts.IncludeInsertsAtBeginOfRange);
141  }
142
143
144  // Adjust the end offset to the end of the last token, instead of being the
145  // start of the last token if this is a token range.
146  if (Range.isTokenRange())
147    EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
148
149  return EndOff-StartOff;
150}
151
152int Rewriter::getRangeSize(SourceRange Range, RewriteOptions opts) const {
153  return getRangeSize(CharSourceRange::getTokenRange(Range), opts);
154}
155
156
157/// getRewrittenText - Return the rewritten form of the text in the specified
158/// range.  If the start or end of the range was unrewritable or if they are
159/// in different buffers, this returns an empty string.
160///
161/// Note that this method is not particularly efficient.
162///
163std::string Rewriter::getRewrittenText(SourceRange Range) const {
164  if (!isRewritable(Range.getBegin()) ||
165      !isRewritable(Range.getEnd()))
166    return "";
167
168  FileID StartFileID, EndFileID;
169  unsigned StartOff, EndOff;
170  StartOff = getLocationOffsetAndFileID(Range.getBegin(), StartFileID);
171  EndOff   = getLocationOffsetAndFileID(Range.getEnd(), EndFileID);
172
173  if (StartFileID != EndFileID)
174    return ""; // Start and end in different buffers.
175
176  // If edits have been made to this buffer, the delta between the range may
177  // have changed.
178  std::map<FileID, RewriteBuffer>::const_iterator I =
179    RewriteBuffers.find(StartFileID);
180  if (I == RewriteBuffers.end()) {
181    // If the buffer hasn't been rewritten, just return the text from the input.
182    const char *Ptr = SourceMgr->getCharacterData(Range.getBegin());
183
184    // Adjust the end offset to the end of the last token, instead of being the
185    // start of the last token.
186    EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
187    return std::string(Ptr, Ptr+EndOff-StartOff);
188  }
189
190  const RewriteBuffer &RB = I->second;
191  EndOff = RB.getMappedOffset(EndOff, true);
192  StartOff = RB.getMappedOffset(StartOff);
193
194  // Adjust the end offset to the end of the last token, instead of being the
195  // start of the last token.
196  EndOff += Lexer::MeasureTokenLength(Range.getEnd(), *SourceMgr, *LangOpts);
197
198  // Advance the iterators to the right spot, yay for linear time algorithms.
199  RewriteBuffer::iterator Start = RB.begin();
200  std::advance(Start, StartOff);
201  RewriteBuffer::iterator End = Start;
202  std::advance(End, EndOff-StartOff);
203
204  return std::string(Start, End);
205}
206
207unsigned Rewriter::getLocationOffsetAndFileID(SourceLocation Loc,
208                                              FileID &FID) const {
209  assert(Loc.isValid() && "Invalid location");
210  std::pair<FileID,unsigned> V = SourceMgr->getDecomposedLoc(Loc);
211  FID = V.first;
212  return V.second;
213}
214
215
216/// getEditBuffer - Get or create a RewriteBuffer for the specified FileID.
217///
218RewriteBuffer &Rewriter::getEditBuffer(FileID FID) {
219  std::map<FileID, RewriteBuffer>::iterator I =
220    RewriteBuffers.lower_bound(FID);
221  if (I != RewriteBuffers.end() && I->first == FID)
222    return I->second;
223  I = RewriteBuffers.insert(I, std::make_pair(FID, RewriteBuffer()));
224
225  llvm::StringRef MB = SourceMgr->getBufferData(FID);
226  I->second.Initialize(MB.begin(), MB.end());
227
228  return I->second;
229}
230
231/// InsertText - Insert the specified string at the specified location in the
232/// original buffer.
233bool Rewriter::InsertText(SourceLocation Loc, llvm::StringRef Str,
234                          bool InsertAfter) {
235  if (!isRewritable(Loc)) return true;
236  FileID FID;
237  unsigned StartOffs = getLocationOffsetAndFileID(Loc, FID);
238  getEditBuffer(FID).InsertText(StartOffs, Str, InsertAfter);
239  return false;
240}
241
242bool Rewriter::InsertTextAfterToken(SourceLocation Loc, llvm::StringRef Str) {
243  if (!isRewritable(Loc)) return true;
244  FileID FID;
245  unsigned StartOffs = getLocationOffsetAndFileID(Loc, FID);
246  RewriteOptions rangeOpts;
247  rangeOpts.IncludeInsertsAtBeginOfRange = false;
248  StartOffs += getRangeSize(SourceRange(Loc, Loc), rangeOpts);
249  getEditBuffer(FID).InsertText(StartOffs, Str, /*InsertAfter*/true);
250  return false;
251}
252
253/// RemoveText - Remove the specified text region.
254bool Rewriter::RemoveText(SourceLocation Start, unsigned Length,
255                          RewriteOptions opts) {
256  if (!isRewritable(Start)) return true;
257  FileID FID;
258  unsigned StartOffs = getLocationOffsetAndFileID(Start, FID);
259  getEditBuffer(FID).RemoveText(StartOffs, Length, opts.RemoveLineIfEmpty);
260  return false;
261}
262
263/// ReplaceText - This method replaces a range of characters in the input
264/// buffer with a new string.  This is effectively a combined "remove/insert"
265/// operation.
266bool Rewriter::ReplaceText(SourceLocation Start, unsigned OrigLength,
267                           llvm::StringRef NewStr) {
268  if (!isRewritable(Start)) return true;
269  FileID StartFileID;
270  unsigned StartOffs = getLocationOffsetAndFileID(Start, StartFileID);
271
272  getEditBuffer(StartFileID).ReplaceText(StartOffs, OrigLength, NewStr);
273  return false;
274}
275
276bool Rewriter::ReplaceText(SourceRange range, SourceRange replacementRange) {
277  if (!isRewritable(range.getBegin())) return true;
278  if (!isRewritable(range.getEnd())) return true;
279  if (replacementRange.isInvalid()) return true;
280  SourceLocation start = range.getBegin();
281  unsigned origLength = getRangeSize(range);
282  unsigned newLength = getRangeSize(replacementRange);
283  FileID FID;
284  unsigned newOffs = getLocationOffsetAndFileID(replacementRange.getBegin(),
285                                                FID);
286  llvm::StringRef MB = SourceMgr->getBufferData(FID);
287  return ReplaceText(start, origLength, MB.substr(newOffs, newLength));
288}
289
290/// ReplaceStmt - This replaces a Stmt/Expr with another, using the pretty
291/// printer to generate the replacement code.  This returns true if the input
292/// could not be rewritten, or false if successful.
293bool Rewriter::ReplaceStmt(Stmt *From, Stmt *To) {
294  // Measaure the old text.
295  int Size = getRangeSize(From->getSourceRange());
296  if (Size == -1)
297    return true;
298
299  // Get the new text.
300  std::string SStr;
301  llvm::raw_string_ostream S(SStr);
302  To->printPretty(S, 0, PrintingPolicy(*LangOpts));
303  const std::string &Str = S.str();
304
305  ReplaceText(From->getLocStart(), Size, Str);
306  return false;
307}
308
309std::string Rewriter::ConvertToString(Stmt *From) {
310  std::string SStr;
311  llvm::raw_string_ostream S(SStr);
312  From->printPretty(S, 0, PrintingPolicy(*LangOpts));
313  return SStr;
314}
315
316bool Rewriter::IncreaseIndentation(CharSourceRange range,
317                                   SourceLocation parentIndent) {
318  using llvm::StringRef;
319
320  if (!isRewritable(range.getBegin())) return true;
321  if (!isRewritable(range.getEnd())) return true;
322  if (!isRewritable(parentIndent)) return true;
323
324  FileID StartFileID, EndFileID, parentFileID;
325  unsigned StartOff, EndOff, parentOff;
326
327  StartOff = getLocationOffsetAndFileID(range.getBegin(), StartFileID);
328  EndOff   = getLocationOffsetAndFileID(range.getEnd(), EndFileID);
329  parentOff = getLocationOffsetAndFileID(parentIndent, parentFileID);
330
331  if (StartFileID != EndFileID || StartFileID != parentFileID)
332    return true;
333  if (StartOff >= EndOff || parentOff >= StartOff)
334    return true;
335
336  FileID FID = StartFileID;
337  StringRef MB = SourceMgr->getBufferData(FID);
338
339  unsigned parentLineNo = SourceMgr->getLineNumber(FID, parentOff) - 1;
340  unsigned startLineNo = SourceMgr->getLineNumber(FID, StartOff) - 1;
341  unsigned endLineNo = SourceMgr->getLineNumber(FID, EndOff) - 1;
342
343  const SrcMgr::ContentCache *
344      Content = SourceMgr->getSLocEntry(FID).getFile().getContentCache();
345
346  // Find where the line starts for the three offsets.
347  unsigned parentLineOffs = Content->SourceLineCache[parentLineNo];
348  unsigned startLineOffs = Content->SourceLineCache[startLineNo];
349  unsigned endLineOffs = Content->SourceLineCache[endLineNo];
350
351  if (startLineOffs == endLineOffs || startLineOffs == parentLineOffs)
352    return true;
353
354  // Find the whitespace at the start of each line.
355  StringRef parentSpace, startSpace, endSpace;
356  {
357    unsigned i = parentLineOffs;
358    while (isWhitespace(MB[i]))
359      ++i;
360    parentSpace = MB.substr(parentLineOffs, i-parentLineOffs);
361
362    i = startLineOffs;
363    while (isWhitespace(MB[i]))
364      ++i;
365    startSpace = MB.substr(startLineOffs, i-startLineOffs);
366
367    i = endLineOffs;
368    while (isWhitespace(MB[i]))
369      ++i;
370    endSpace = MB.substr(endLineOffs, i-endLineOffs);
371  }
372  if (parentSpace.size() >= startSpace.size())
373    return true;
374  if (!startSpace.startswith(parentSpace))
375    return true;
376
377  llvm::StringRef indent = startSpace.substr(parentSpace.size());
378
379  // Indent the lines between start/end offsets.
380  RewriteBuffer &RB = getEditBuffer(FID);
381  for (unsigned i = startLineOffs; i != endLineOffs; ++i) {
382    if (MB[i] == '\n') {
383      unsigned startOfLine = i+1;
384      if (startOfLine == endLineOffs)
385        break;
386      StringRef origIndent;
387      unsigned ws = startOfLine;
388      while (isWhitespace(MB[ws]))
389        ++ws;
390      origIndent = MB.substr(startOfLine, ws-startOfLine);
391      if (origIndent.startswith(startSpace))
392        RB.InsertText(startOfLine, indent, /*InsertAfter=*/false);
393    }
394  }
395
396  return false;
397}
398