/src/llvm-project/clang/lib/Edit/EditedSource.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | //===- EditedSource.cpp - Collection of source edits ----------------------===// |
2 | | // |
3 | | // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions. |
4 | | // See https://llvm.org/LICENSE.txt for license information. |
5 | | // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception |
6 | | // |
7 | | //===----------------------------------------------------------------------===// |
8 | | |
9 | | #include "clang/Edit/EditedSource.h" |
10 | | #include "clang/Basic/CharInfo.h" |
11 | | #include "clang/Basic/LLVM.h" |
12 | | #include "clang/Basic/SourceLocation.h" |
13 | | #include "clang/Basic/SourceManager.h" |
14 | | #include "clang/Edit/Commit.h" |
15 | | #include "clang/Edit/EditsReceiver.h" |
16 | | #include "clang/Edit/FileOffset.h" |
17 | | #include "clang/Lex/Lexer.h" |
18 | | #include "llvm/ADT/STLExtras.h" |
19 | | #include "llvm/ADT/SmallString.h" |
20 | | #include "llvm/ADT/StringRef.h" |
21 | | #include "llvm/ADT/Twine.h" |
22 | | #include <algorithm> |
23 | | #include <cassert> |
24 | | #include <tuple> |
25 | | #include <utility> |
26 | | |
27 | | using namespace clang; |
28 | | using namespace edit; |
29 | | |
30 | 0 | void EditsReceiver::remove(CharSourceRange range) { |
31 | 0 | replace(range, StringRef()); |
32 | 0 | } |
33 | | |
34 | | void EditedSource::deconstructMacroArgLoc(SourceLocation Loc, |
35 | | SourceLocation &ExpansionLoc, |
36 | 0 | MacroArgUse &ArgUse) { |
37 | 0 | assert(SourceMgr.isMacroArgExpansion(Loc)); |
38 | 0 | SourceLocation DefArgLoc = |
39 | 0 | SourceMgr.getImmediateExpansionRange(Loc).getBegin(); |
40 | 0 | SourceLocation ImmediateExpansionLoc = |
41 | 0 | SourceMgr.getImmediateExpansionRange(DefArgLoc).getBegin(); |
42 | 0 | ExpansionLoc = ImmediateExpansionLoc; |
43 | 0 | while (SourceMgr.isMacroBodyExpansion(ExpansionLoc)) |
44 | 0 | ExpansionLoc = |
45 | 0 | SourceMgr.getImmediateExpansionRange(ExpansionLoc).getBegin(); |
46 | 0 | SmallString<20> Buf; |
47 | 0 | StringRef ArgName = Lexer::getSpelling(SourceMgr.getSpellingLoc(DefArgLoc), |
48 | 0 | Buf, SourceMgr, LangOpts); |
49 | 0 | ArgUse = MacroArgUse{nullptr, SourceLocation(), SourceLocation()}; |
50 | 0 | if (!ArgName.empty()) |
51 | 0 | ArgUse = {&IdentTable.get(ArgName), ImmediateExpansionLoc, |
52 | 0 | SourceMgr.getSpellingLoc(DefArgLoc)}; |
53 | 0 | } |
54 | | |
55 | 0 | void EditedSource::startingCommit() {} |
56 | | |
57 | 0 | void EditedSource::finishedCommit() { |
58 | 0 | for (auto &ExpArg : CurrCommitMacroArgExps) { |
59 | 0 | SourceLocation ExpLoc; |
60 | 0 | MacroArgUse ArgUse; |
61 | 0 | std::tie(ExpLoc, ArgUse) = ExpArg; |
62 | 0 | auto &ArgUses = ExpansionToArgMap[ExpLoc]; |
63 | 0 | if (!llvm::is_contained(ArgUses, ArgUse)) |
64 | 0 | ArgUses.push_back(ArgUse); |
65 | 0 | } |
66 | 0 | CurrCommitMacroArgExps.clear(); |
67 | 0 | } |
68 | | |
69 | 0 | StringRef EditedSource::copyString(const Twine &twine) { |
70 | 0 | SmallString<128> Data; |
71 | 0 | return copyString(twine.toStringRef(Data)); |
72 | 0 | } |
73 | | |
74 | 0 | bool EditedSource::canInsertInOffset(SourceLocation OrigLoc, FileOffset Offs) { |
75 | 0 | FileEditsTy::iterator FA = getActionForOffset(Offs); |
76 | 0 | if (FA != FileEdits.end()) { |
77 | 0 | if (FA->first != Offs) |
78 | 0 | return false; // position has been removed. |
79 | 0 | } |
80 | | |
81 | 0 | if (SourceMgr.isMacroArgExpansion(OrigLoc)) { |
82 | 0 | SourceLocation ExpLoc; |
83 | 0 | MacroArgUse ArgUse; |
84 | 0 | deconstructMacroArgLoc(OrigLoc, ExpLoc, ArgUse); |
85 | 0 | auto I = ExpansionToArgMap.find(ExpLoc); |
86 | 0 | if (I != ExpansionToArgMap.end() && |
87 | 0 | llvm::any_of(I->second, [&](const MacroArgUse &U) { |
88 | 0 | return ArgUse.Identifier == U.Identifier && |
89 | 0 | std::tie(ArgUse.ImmediateExpansionLoc, ArgUse.UseLoc) != |
90 | 0 | std::tie(U.ImmediateExpansionLoc, U.UseLoc); |
91 | 0 | })) { |
92 | | // Trying to write in a macro argument input that has already been |
93 | | // written by a previous commit for another expansion of the same macro |
94 | | // argument name. For example: |
95 | | // |
96 | | // \code |
97 | | // #define MAC(x) ((x)+(x)) |
98 | | // MAC(a) |
99 | | // \endcode |
100 | | // |
101 | | // A commit modified the macro argument 'a' due to the first '(x)' |
102 | | // expansion inside the macro definition, and a subsequent commit tried |
103 | | // to modify 'a' again for the second '(x)' expansion. The edits of the |
104 | | // second commit will be rejected. |
105 | 0 | return false; |
106 | 0 | } |
107 | 0 | } |
108 | 0 | return true; |
109 | 0 | } |
110 | | |
111 | | bool EditedSource::commitInsert(SourceLocation OrigLoc, |
112 | | FileOffset Offs, StringRef text, |
113 | 0 | bool beforePreviousInsertions) { |
114 | 0 | if (!canInsertInOffset(OrigLoc, Offs)) |
115 | 0 | return false; |
116 | 0 | if (text.empty()) |
117 | 0 | return true; |
118 | | |
119 | 0 | if (SourceMgr.isMacroArgExpansion(OrigLoc)) { |
120 | 0 | MacroArgUse ArgUse; |
121 | 0 | SourceLocation ExpLoc; |
122 | 0 | deconstructMacroArgLoc(OrigLoc, ExpLoc, ArgUse); |
123 | 0 | if (ArgUse.Identifier) |
124 | 0 | CurrCommitMacroArgExps.emplace_back(ExpLoc, ArgUse); |
125 | 0 | } |
126 | |
|
127 | 0 | FileEdit &FA = FileEdits[Offs]; |
128 | 0 | if (FA.Text.empty()) { |
129 | 0 | FA.Text = copyString(text); |
130 | 0 | return true; |
131 | 0 | } |
132 | | |
133 | 0 | if (beforePreviousInsertions) |
134 | 0 | FA.Text = copyString(Twine(text) + FA.Text); |
135 | 0 | else |
136 | 0 | FA.Text = copyString(Twine(FA.Text) + text); |
137 | |
|
138 | 0 | return true; |
139 | 0 | } |
140 | | |
141 | | bool EditedSource::commitInsertFromRange(SourceLocation OrigLoc, |
142 | | FileOffset Offs, |
143 | | FileOffset InsertFromRangeOffs, unsigned Len, |
144 | 0 | bool beforePreviousInsertions) { |
145 | 0 | if (Len == 0) |
146 | 0 | return true; |
147 | | |
148 | 0 | SmallString<128> StrVec; |
149 | 0 | FileOffset BeginOffs = InsertFromRangeOffs; |
150 | 0 | FileOffset EndOffs = BeginOffs.getWithOffset(Len); |
151 | 0 | FileEditsTy::iterator I = FileEdits.upper_bound(BeginOffs); |
152 | 0 | if (I != FileEdits.begin()) |
153 | 0 | --I; |
154 | |
|
155 | 0 | for (; I != FileEdits.end(); ++I) { |
156 | 0 | FileEdit &FA = I->second; |
157 | 0 | FileOffset B = I->first; |
158 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
159 | |
|
160 | 0 | if (BeginOffs == B) |
161 | 0 | break; |
162 | | |
163 | 0 | if (BeginOffs < E) { |
164 | 0 | if (BeginOffs > B) { |
165 | 0 | BeginOffs = E; |
166 | 0 | ++I; |
167 | 0 | } |
168 | 0 | break; |
169 | 0 | } |
170 | 0 | } |
171 | |
|
172 | 0 | for (; I != FileEdits.end() && EndOffs > I->first; ++I) { |
173 | 0 | FileEdit &FA = I->second; |
174 | 0 | FileOffset B = I->first; |
175 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
176 | |
|
177 | 0 | if (BeginOffs < B) { |
178 | 0 | bool Invalid = false; |
179 | 0 | StringRef text = getSourceText(BeginOffs, B, Invalid); |
180 | 0 | if (Invalid) |
181 | 0 | return false; |
182 | 0 | StrVec += text; |
183 | 0 | } |
184 | 0 | StrVec += FA.Text; |
185 | 0 | BeginOffs = E; |
186 | 0 | } |
187 | | |
188 | 0 | if (BeginOffs < EndOffs) { |
189 | 0 | bool Invalid = false; |
190 | 0 | StringRef text = getSourceText(BeginOffs, EndOffs, Invalid); |
191 | 0 | if (Invalid) |
192 | 0 | return false; |
193 | 0 | StrVec += text; |
194 | 0 | } |
195 | | |
196 | 0 | return commitInsert(OrigLoc, Offs, StrVec, beforePreviousInsertions); |
197 | 0 | } |
198 | | |
199 | | void EditedSource::commitRemove(SourceLocation OrigLoc, |
200 | 0 | FileOffset BeginOffs, unsigned Len) { |
201 | 0 | if (Len == 0) |
202 | 0 | return; |
203 | | |
204 | 0 | FileOffset EndOffs = BeginOffs.getWithOffset(Len); |
205 | 0 | FileEditsTy::iterator I = FileEdits.upper_bound(BeginOffs); |
206 | 0 | if (I != FileEdits.begin()) |
207 | 0 | --I; |
208 | |
|
209 | 0 | for (; I != FileEdits.end(); ++I) { |
210 | 0 | FileEdit &FA = I->second; |
211 | 0 | FileOffset B = I->first; |
212 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
213 | |
|
214 | 0 | if (BeginOffs < E) |
215 | 0 | break; |
216 | 0 | } |
217 | |
|
218 | 0 | FileOffset TopBegin, TopEnd; |
219 | 0 | FileEdit *TopFA = nullptr; |
220 | |
|
221 | 0 | if (I == FileEdits.end()) { |
222 | 0 | FileEditsTy::iterator |
223 | 0 | NewI = FileEdits.insert(I, std::make_pair(BeginOffs, FileEdit())); |
224 | 0 | NewI->second.RemoveLen = Len; |
225 | 0 | return; |
226 | 0 | } |
227 | | |
228 | 0 | FileEdit &FA = I->second; |
229 | 0 | FileOffset B = I->first; |
230 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
231 | 0 | if (BeginOffs < B) { |
232 | 0 | FileEditsTy::iterator |
233 | 0 | NewI = FileEdits.insert(I, std::make_pair(BeginOffs, FileEdit())); |
234 | 0 | TopBegin = BeginOffs; |
235 | 0 | TopEnd = EndOffs; |
236 | 0 | TopFA = &NewI->second; |
237 | 0 | TopFA->RemoveLen = Len; |
238 | 0 | } else { |
239 | 0 | TopBegin = B; |
240 | 0 | TopEnd = E; |
241 | 0 | TopFA = &I->second; |
242 | 0 | if (TopEnd >= EndOffs) |
243 | 0 | return; |
244 | 0 | unsigned diff = EndOffs.getOffset() - TopEnd.getOffset(); |
245 | 0 | TopEnd = EndOffs; |
246 | 0 | TopFA->RemoveLen += diff; |
247 | 0 | if (B == BeginOffs) |
248 | 0 | TopFA->Text = StringRef(); |
249 | 0 | ++I; |
250 | 0 | } |
251 | | |
252 | 0 | while (I != FileEdits.end()) { |
253 | 0 | FileEdit &FA = I->second; |
254 | 0 | FileOffset B = I->first; |
255 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
256 | |
|
257 | 0 | if (B >= TopEnd) |
258 | 0 | break; |
259 | | |
260 | 0 | if (E <= TopEnd) { |
261 | 0 | FileEdits.erase(I++); |
262 | 0 | continue; |
263 | 0 | } |
264 | | |
265 | 0 | if (B < TopEnd) { |
266 | 0 | unsigned diff = E.getOffset() - TopEnd.getOffset(); |
267 | 0 | TopEnd = E; |
268 | 0 | TopFA->RemoveLen += diff; |
269 | 0 | FileEdits.erase(I); |
270 | 0 | } |
271 | |
|
272 | 0 | break; |
273 | 0 | } |
274 | 0 | } |
275 | | |
276 | 0 | bool EditedSource::commit(const Commit &commit) { |
277 | 0 | if (!commit.isCommitable()) |
278 | 0 | return false; |
279 | | |
280 | 0 | struct CommitRAII { |
281 | 0 | EditedSource &Editor; |
282 | |
|
283 | 0 | CommitRAII(EditedSource &Editor) : Editor(Editor) { |
284 | 0 | Editor.startingCommit(); |
285 | 0 | } |
286 | |
|
287 | 0 | ~CommitRAII() { |
288 | 0 | Editor.finishedCommit(); |
289 | 0 | } |
290 | 0 | } CommitRAII(*this); |
291 | |
|
292 | 0 | for (edit::Commit::edit_iterator |
293 | 0 | I = commit.edit_begin(), E = commit.edit_end(); I != E; ++I) { |
294 | 0 | const edit::Commit::Edit &edit = *I; |
295 | 0 | switch (edit.Kind) { |
296 | 0 | case edit::Commit::Act_Insert: |
297 | 0 | commitInsert(edit.OrigLoc, edit.Offset, edit.Text, edit.BeforePrev); |
298 | 0 | break; |
299 | 0 | case edit::Commit::Act_InsertFromRange: |
300 | 0 | commitInsertFromRange(edit.OrigLoc, edit.Offset, |
301 | 0 | edit.InsertFromRangeOffs, edit.Length, |
302 | 0 | edit.BeforePrev); |
303 | 0 | break; |
304 | 0 | case edit::Commit::Act_Remove: |
305 | 0 | commitRemove(edit.OrigLoc, edit.Offset, edit.Length); |
306 | 0 | break; |
307 | 0 | } |
308 | 0 | } |
309 | | |
310 | 0 | return true; |
311 | 0 | } |
312 | | |
313 | | // Returns true if it is ok to make the two given characters adjacent. |
314 | 0 | static bool canBeJoined(char left, char right, const LangOptions &LangOpts) { |
315 | | // FIXME: Should use TokenConcatenation to make sure we don't allow stuff like |
316 | | // making two '<' adjacent. |
317 | 0 | return !(Lexer::isAsciiIdentifierContinueChar(left, LangOpts) && |
318 | 0 | Lexer::isAsciiIdentifierContinueChar(right, LangOpts)); |
319 | 0 | } |
320 | | |
321 | | /// Returns true if it is ok to eliminate the trailing whitespace between |
322 | | /// the given characters. |
323 | | static bool canRemoveWhitespace(char left, char beforeWSpace, char right, |
324 | 0 | const LangOptions &LangOpts) { |
325 | 0 | if (!canBeJoined(left, right, LangOpts)) |
326 | 0 | return false; |
327 | 0 | if (isWhitespace(left) || isWhitespace(right)) |
328 | 0 | return true; |
329 | 0 | if (canBeJoined(beforeWSpace, right, LangOpts)) |
330 | 0 | return false; // the whitespace was intentional, keep it. |
331 | 0 | return true; |
332 | 0 | } |
333 | | |
334 | | /// Check the range that we are going to remove and: |
335 | | /// -Remove any trailing whitespace if possible. |
336 | | /// -Insert a space if removing the range is going to mess up the source tokens. |
337 | | static void adjustRemoval(const SourceManager &SM, const LangOptions &LangOpts, |
338 | | SourceLocation Loc, FileOffset offs, |
339 | 0 | unsigned &len, StringRef &text) { |
340 | 0 | assert(len && text.empty()); |
341 | 0 | SourceLocation BeginTokLoc = Lexer::GetBeginningOfToken(Loc, SM, LangOpts); |
342 | 0 | if (BeginTokLoc != Loc) |
343 | 0 | return; // the range is not at the beginning of a token, keep the range. |
344 | | |
345 | 0 | bool Invalid = false; |
346 | 0 | StringRef buffer = SM.getBufferData(offs.getFID(), &Invalid); |
347 | 0 | if (Invalid) |
348 | 0 | return; |
349 | | |
350 | 0 | unsigned begin = offs.getOffset(); |
351 | 0 | unsigned end = begin + len; |
352 | | |
353 | | // Do not try to extend the removal if we're at the end of the buffer already. |
354 | 0 | if (end == buffer.size()) |
355 | 0 | return; |
356 | | |
357 | 0 | assert(begin < buffer.size() && end < buffer.size() && "Invalid range!"); |
358 | | |
359 | | // FIXME: Remove newline. |
360 | | |
361 | 0 | if (begin == 0) { |
362 | 0 | if (buffer[end] == ' ') |
363 | 0 | ++len; |
364 | 0 | return; |
365 | 0 | } |
366 | | |
367 | 0 | if (buffer[end] == ' ') { |
368 | 0 | assert((end + 1 != buffer.size() || buffer.data()[end + 1] == 0) && |
369 | 0 | "buffer not zero-terminated!"); |
370 | 0 | if (canRemoveWhitespace(/*left=*/buffer[begin-1], |
371 | 0 | /*beforeWSpace=*/buffer[end-1], |
372 | 0 | /*right=*/buffer.data()[end + 1], // zero-terminated |
373 | 0 | LangOpts)) |
374 | 0 | ++len; |
375 | 0 | return; |
376 | 0 | } |
377 | | |
378 | 0 | if (!canBeJoined(buffer[begin-1], buffer[end], LangOpts)) |
379 | 0 | text = " "; |
380 | 0 | } |
381 | | |
382 | | static void applyRewrite(EditsReceiver &receiver, |
383 | | StringRef text, FileOffset offs, unsigned len, |
384 | | const SourceManager &SM, const LangOptions &LangOpts, |
385 | 0 | bool shouldAdjustRemovals) { |
386 | 0 | assert(offs.getFID().isValid()); |
387 | 0 | SourceLocation Loc = SM.getLocForStartOfFile(offs.getFID()); |
388 | 0 | Loc = Loc.getLocWithOffset(offs.getOffset()); |
389 | 0 | assert(Loc.isFileID()); |
390 | | |
391 | 0 | if (text.empty() && shouldAdjustRemovals) |
392 | 0 | adjustRemoval(SM, LangOpts, Loc, offs, len, text); |
393 | |
|
394 | 0 | CharSourceRange range = CharSourceRange::getCharRange(Loc, |
395 | 0 | Loc.getLocWithOffset(len)); |
396 | |
|
397 | 0 | if (text.empty()) { |
398 | 0 | assert(len); |
399 | 0 | receiver.remove(range); |
400 | 0 | return; |
401 | 0 | } |
402 | | |
403 | 0 | if (len) |
404 | 0 | receiver.replace(range, text); |
405 | 0 | else |
406 | 0 | receiver.insert(Loc, text); |
407 | 0 | } |
408 | | |
409 | | void EditedSource::applyRewrites(EditsReceiver &receiver, |
410 | 0 | bool shouldAdjustRemovals) { |
411 | 0 | SmallString<128> StrVec; |
412 | 0 | FileOffset CurOffs, CurEnd; |
413 | 0 | unsigned CurLen; |
414 | |
|
415 | 0 | if (FileEdits.empty()) |
416 | 0 | return; |
417 | | |
418 | 0 | FileEditsTy::iterator I = FileEdits.begin(); |
419 | 0 | CurOffs = I->first; |
420 | 0 | StrVec = I->second.Text; |
421 | 0 | CurLen = I->second.RemoveLen; |
422 | 0 | CurEnd = CurOffs.getWithOffset(CurLen); |
423 | 0 | ++I; |
424 | |
|
425 | 0 | for (FileEditsTy::iterator E = FileEdits.end(); I != E; ++I) { |
426 | 0 | FileOffset offs = I->first; |
427 | 0 | FileEdit act = I->second; |
428 | 0 | assert(offs >= CurEnd); |
429 | | |
430 | 0 | if (offs == CurEnd) { |
431 | 0 | StrVec += act.Text; |
432 | 0 | CurLen += act.RemoveLen; |
433 | 0 | CurEnd.getWithOffset(act.RemoveLen); |
434 | 0 | continue; |
435 | 0 | } |
436 | | |
437 | 0 | applyRewrite(receiver, StrVec, CurOffs, CurLen, SourceMgr, LangOpts, |
438 | 0 | shouldAdjustRemovals); |
439 | 0 | CurOffs = offs; |
440 | 0 | StrVec = act.Text; |
441 | 0 | CurLen = act.RemoveLen; |
442 | 0 | CurEnd = CurOffs.getWithOffset(CurLen); |
443 | 0 | } |
444 | |
|
445 | 0 | applyRewrite(receiver, StrVec, CurOffs, CurLen, SourceMgr, LangOpts, |
446 | 0 | shouldAdjustRemovals); |
447 | 0 | } |
448 | | |
449 | 0 | void EditedSource::clearRewrites() { |
450 | 0 | FileEdits.clear(); |
451 | 0 | StrAlloc.Reset(); |
452 | 0 | } |
453 | | |
454 | | StringRef EditedSource::getSourceText(FileOffset BeginOffs, FileOffset EndOffs, |
455 | 0 | bool &Invalid) { |
456 | 0 | assert(BeginOffs.getFID() == EndOffs.getFID()); |
457 | 0 | assert(BeginOffs <= EndOffs); |
458 | 0 | SourceLocation BLoc = SourceMgr.getLocForStartOfFile(BeginOffs.getFID()); |
459 | 0 | BLoc = BLoc.getLocWithOffset(BeginOffs.getOffset()); |
460 | 0 | assert(BLoc.isFileID()); |
461 | 0 | SourceLocation |
462 | 0 | ELoc = BLoc.getLocWithOffset(EndOffs.getOffset() - BeginOffs.getOffset()); |
463 | 0 | return Lexer::getSourceText(CharSourceRange::getCharRange(BLoc, ELoc), |
464 | 0 | SourceMgr, LangOpts, &Invalid); |
465 | 0 | } |
466 | | |
467 | | EditedSource::FileEditsTy::iterator |
468 | 0 | EditedSource::getActionForOffset(FileOffset Offs) { |
469 | 0 | FileEditsTy::iterator I = FileEdits.upper_bound(Offs); |
470 | 0 | if (I == FileEdits.begin()) |
471 | 0 | return FileEdits.end(); |
472 | 0 | --I; |
473 | 0 | FileEdit &FA = I->second; |
474 | 0 | FileOffset B = I->first; |
475 | 0 | FileOffset E = B.getWithOffset(FA.RemoveLen); |
476 | 0 | if (Offs >= B && Offs < E) |
477 | 0 | return I; |
478 | | |
479 | 0 | return FileEdits.end(); |
480 | 0 | } |