1 //===--- UnwrappedLineFormatter.cpp - Format C++ code ---------------------===//
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 #include "NamespaceEndCommentsFixer.h"
11 #include "UnwrappedLineFormatter.h"
12 #include "WhitespaceManager.h"
13 #include "llvm/Support/Debug.h"
14 #include <queue>
15
16 #define DEBUG_TYPE "format-formatter"
17
18 namespace clang {
19 namespace format {
20
21 namespace {
22
startsExternCBlock(const AnnotatedLine & Line)23 bool startsExternCBlock(const AnnotatedLine &Line) {
24 const FormatToken *Next = Line.First->getNextNonComment();
25 const FormatToken *NextNext = Next ? Next->getNextNonComment() : nullptr;
26 return Line.startsWith(tok::kw_extern) && Next && Next->isStringLiteral() &&
27 NextNext && NextNext->is(tok::l_brace);
28 }
29
30 /// Tracks the indent level of \c AnnotatedLines across levels.
31 ///
32 /// \c nextLine must be called for each \c AnnotatedLine, after which \c
33 /// getIndent() will return the indent for the last line \c nextLine was called
34 /// with.
35 /// If the line is not formatted (and thus the indent does not change), calling
36 /// \c adjustToUnmodifiedLine after the call to \c nextLine will cause
37 /// subsequent lines on the same level to be indented at the same level as the
38 /// given line.
39 class LevelIndentTracker {
40 public:
LevelIndentTracker(const FormatStyle & Style,const AdditionalKeywords & Keywords,unsigned StartLevel,int AdditionalIndent)41 LevelIndentTracker(const FormatStyle &Style,
42 const AdditionalKeywords &Keywords, unsigned StartLevel,
43 int AdditionalIndent)
44 : Style(Style), Keywords(Keywords), AdditionalIndent(AdditionalIndent) {
45 for (unsigned i = 0; i != StartLevel; ++i)
46 IndentForLevel.push_back(Style.IndentWidth * i + AdditionalIndent);
47 }
48
49 /// Returns the indent for the current line.
getIndent() const50 unsigned getIndent() const { return Indent; }
51
52 /// Update the indent state given that \p Line is going to be formatted
53 /// next.
nextLine(const AnnotatedLine & Line)54 void nextLine(const AnnotatedLine &Line) {
55 Offset = getIndentOffset(*Line.First);
56 // Update the indent level cache size so that we can rely on it
57 // having the right size in adjustToUnmodifiedline.
58 while (IndentForLevel.size() <= Line.Level)
59 IndentForLevel.push_back(-1);
60 if (Line.InPPDirective) {
61 Indent = Line.Level * Style.IndentWidth + AdditionalIndent;
62 } else {
63 IndentForLevel.resize(Line.Level + 1);
64 Indent = getIndent(IndentForLevel, Line.Level);
65 }
66 if (static_cast<int>(Indent) + Offset >= 0)
67 Indent += Offset;
68 }
69
70 /// Update the indent state given that \p Line indent should be
71 /// skipped.
skipLine(const AnnotatedLine & Line)72 void skipLine(const AnnotatedLine &Line) {
73 while (IndentForLevel.size() <= Line.Level)
74 IndentForLevel.push_back(Indent);
75 }
76
77 /// Update the level indent to adapt to the given \p Line.
78 ///
79 /// When a line is not formatted, we move the subsequent lines on the same
80 /// level to the same indent.
81 /// Note that \c nextLine must have been called before this method.
adjustToUnmodifiedLine(const AnnotatedLine & Line)82 void adjustToUnmodifiedLine(const AnnotatedLine &Line) {
83 unsigned LevelIndent = Line.First->OriginalColumn;
84 if (static_cast<int>(LevelIndent) - Offset >= 0)
85 LevelIndent -= Offset;
86 if ((!Line.First->is(tok::comment) || IndentForLevel[Line.Level] == -1) &&
87 !Line.InPPDirective)
88 IndentForLevel[Line.Level] = LevelIndent;
89 }
90
91 private:
92 /// Get the offset of the line relatively to the level.
93 ///
94 /// For example, 'public:' labels in classes are offset by 1 or 2
95 /// characters to the left from their level.
getIndentOffset(const FormatToken & RootToken)96 int getIndentOffset(const FormatToken &RootToken) {
97 if (Style.Language == FormatStyle::LK_Java ||
98 Style.Language == FormatStyle::LK_JavaScript)
99 return 0;
100 if (RootToken.isAccessSpecifier(false) ||
101 RootToken.isObjCAccessSpecifier() ||
102 (RootToken.isOneOf(Keywords.kw_signals, Keywords.kw_qsignals) &&
103 RootToken.Next && RootToken.Next->is(tok::colon)))
104 return Style.AccessModifierOffset;
105 return 0;
106 }
107
108 /// Get the indent of \p Level from \p IndentForLevel.
109 ///
110 /// \p IndentForLevel must contain the indent for the level \c l
111 /// at \p IndentForLevel[l], or a value < 0 if the indent for
112 /// that level is unknown.
getIndent(ArrayRef<int> IndentForLevel,unsigned Level)113 unsigned getIndent(ArrayRef<int> IndentForLevel, unsigned Level) {
114 if (IndentForLevel[Level] != -1)
115 return IndentForLevel[Level];
116 if (Level == 0)
117 return 0;
118 return getIndent(IndentForLevel, Level - 1) + Style.IndentWidth;
119 }
120
121 const FormatStyle &Style;
122 const AdditionalKeywords &Keywords;
123 const unsigned AdditionalIndent;
124
125 /// The indent in characters for each level.
126 std::vector<int> IndentForLevel;
127
128 /// Offset of the current line relative to the indent level.
129 ///
130 /// For example, the 'public' keywords is often indented with a negative
131 /// offset.
132 int Offset = 0;
133
134 /// The current line's indent.
135 unsigned Indent = 0;
136 };
137
isNamespaceDeclaration(const AnnotatedLine * Line)138 bool isNamespaceDeclaration(const AnnotatedLine *Line) {
139 const FormatToken *NamespaceTok = Line->First;
140 return NamespaceTok && NamespaceTok->getNamespaceToken();
141 }
142
isEndOfNamespace(const AnnotatedLine * Line,const SmallVectorImpl<AnnotatedLine * > & AnnotatedLines)143 bool isEndOfNamespace(const AnnotatedLine *Line,
144 const SmallVectorImpl<AnnotatedLine *> &AnnotatedLines) {
145 if (!Line->startsWith(tok::r_brace))
146 return false;
147 size_t StartLineIndex = Line->MatchingOpeningBlockLineIndex;
148 if (StartLineIndex == UnwrappedLine::kInvalidIndex)
149 return false;
150 assert(StartLineIndex < AnnotatedLines.size());
151 return isNamespaceDeclaration(AnnotatedLines[StartLineIndex]);
152 }
153
154 class LineJoiner {
155 public:
LineJoiner(const FormatStyle & Style,const AdditionalKeywords & Keywords,const SmallVectorImpl<AnnotatedLine * > & Lines)156 LineJoiner(const FormatStyle &Style, const AdditionalKeywords &Keywords,
157 const SmallVectorImpl<AnnotatedLine *> &Lines)
158 : Style(Style), Keywords(Keywords), End(Lines.end()), Next(Lines.begin()),
159 AnnotatedLines(Lines) {}
160
161 /// Returns the next line, merging multiple lines into one if possible.
getNextMergedLine(bool DryRun,LevelIndentTracker & IndentTracker)162 const AnnotatedLine *getNextMergedLine(bool DryRun,
163 LevelIndentTracker &IndentTracker) {
164 if (Next == End)
165 return nullptr;
166 const AnnotatedLine *Current = *Next;
167 IndentTracker.nextLine(*Current);
168 unsigned MergedLines = tryFitMultipleLinesInOne(IndentTracker, Next, End);
169 if (MergedLines > 0 && Style.ColumnLimit == 0)
170 // Disallow line merging if there is a break at the start of one of the
171 // input lines.
172 for (unsigned i = 0; i < MergedLines; ++i)
173 if (Next[i + 1]->First->NewlinesBefore > 0)
174 MergedLines = 0;
175 if (!DryRun)
176 for (unsigned i = 0; i < MergedLines; ++i)
177 join(*Next[0], *Next[i + 1]);
178 Next = Next + MergedLines + 1;
179 return Current;
180 }
181
182 private:
183 /// Calculates how many lines can be merged into 1 starting at \p I.
184 unsigned
tryFitMultipleLinesInOne(LevelIndentTracker & IndentTracker,SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E)185 tryFitMultipleLinesInOne(LevelIndentTracker &IndentTracker,
186 SmallVectorImpl<AnnotatedLine *>::const_iterator I,
187 SmallVectorImpl<AnnotatedLine *>::const_iterator E) {
188 const unsigned Indent = IndentTracker.getIndent();
189
190 // Can't join the last line with anything.
191 if (I + 1 == E)
192 return 0;
193 // We can never merge stuff if there are trailing line comments.
194 const AnnotatedLine *TheLine = *I;
195 if (TheLine->Last->is(TT_LineComment))
196 return 0;
197 if (I[1]->Type == LT_Invalid || I[1]->First->MustBreakBefore)
198 return 0;
199 if (TheLine->InPPDirective &&
200 (!I[1]->InPPDirective || I[1]->First->HasUnescapedNewline))
201 return 0;
202
203 if (Style.ColumnLimit > 0 && Indent > Style.ColumnLimit)
204 return 0;
205
206 unsigned Limit =
207 Style.ColumnLimit == 0 ? UINT_MAX : Style.ColumnLimit - Indent;
208 // If we already exceed the column limit, we set 'Limit' to 0. The different
209 // tryMerge..() functions can then decide whether to still do merging.
210 Limit = TheLine->Last->TotalLength > Limit
211 ? 0
212 : Limit - TheLine->Last->TotalLength;
213
214 if (TheLine->Last->is(TT_FunctionLBrace) &&
215 TheLine->First == TheLine->Last &&
216 !Style.BraceWrapping.SplitEmptyFunction &&
217 I[1]->First->is(tok::r_brace))
218 return tryMergeSimpleBlock(I, E, Limit);
219
220 // Handle empty record blocks where the brace has already been wrapped
221 if (TheLine->Last->is(tok::l_brace) && TheLine->First == TheLine->Last &&
222 I != AnnotatedLines.begin()) {
223 bool EmptyBlock = I[1]->First->is(tok::r_brace);
224
225 const FormatToken *Tok = I[-1]->First;
226 if (Tok && Tok->is(tok::comment))
227 Tok = Tok->getNextNonComment();
228
229 if (Tok && Tok->getNamespaceToken())
230 return !Style.BraceWrapping.SplitEmptyNamespace && EmptyBlock
231 ? tryMergeSimpleBlock(I, E, Limit)
232 : 0;
233
234 if (Tok && Tok->is(tok::kw_typedef))
235 Tok = Tok->getNextNonComment();
236 if (Tok && Tok->isOneOf(tok::kw_class, tok::kw_struct, tok::kw_union,
237 tok::kw_extern, Keywords.kw_interface))
238 return !Style.BraceWrapping.SplitEmptyRecord && EmptyBlock
239 ? tryMergeSimpleBlock(I, E, Limit)
240 : 0;
241 }
242
243 // FIXME: TheLine->Level != 0 might or might not be the right check to do.
244 // If necessary, change to something smarter.
245 bool MergeShortFunctions =
246 Style.AllowShortFunctionsOnASingleLine == FormatStyle::SFS_All ||
247 (Style.AllowShortFunctionsOnASingleLine >= FormatStyle::SFS_Empty &&
248 I[1]->First->is(tok::r_brace)) ||
249 (Style.AllowShortFunctionsOnASingleLine & FormatStyle::SFS_InlineOnly &&
250 TheLine->Level != 0);
251
252 if (Style.CompactNamespaces) {
253 if (isNamespaceDeclaration(TheLine)) {
254 int i = 0;
255 unsigned closingLine = TheLine->MatchingClosingBlockLineIndex - 1;
256 for (; I + 1 + i != E && isNamespaceDeclaration(I[i + 1]) &&
257 closingLine == I[i + 1]->MatchingClosingBlockLineIndex &&
258 I[i + 1]->Last->TotalLength < Limit;
259 i++, closingLine--) {
260 // No extra indent for compacted namespaces
261 IndentTracker.skipLine(*I[i + 1]);
262
263 Limit -= I[i + 1]->Last->TotalLength;
264 }
265 return i;
266 }
267
268 if (isEndOfNamespace(TheLine, AnnotatedLines)) {
269 int i = 0;
270 unsigned openingLine = TheLine->MatchingOpeningBlockLineIndex - 1;
271 for (; I + 1 + i != E && isEndOfNamespace(I[i + 1], AnnotatedLines) &&
272 openingLine == I[i + 1]->MatchingOpeningBlockLineIndex;
273 i++, openingLine--) {
274 // No space between consecutive braces
275 I[i + 1]->First->SpacesRequiredBefore = !I[i]->Last->is(tok::r_brace);
276
277 // Indent like the outer-most namespace
278 IndentTracker.nextLine(*I[i + 1]);
279 }
280 return i;
281 }
282 }
283
284 // Try to merge a function block with left brace unwrapped
285 if (TheLine->Last->is(TT_FunctionLBrace) &&
286 TheLine->First != TheLine->Last) {
287 return MergeShortFunctions ? tryMergeSimpleBlock(I, E, Limit) : 0;
288 }
289 // Try to merge a control statement block with left brace unwrapped
290 if (TheLine->Last->is(tok::l_brace) && TheLine->First != TheLine->Last &&
291 TheLine->First->isOneOf(tok::kw_if, tok::kw_while, tok::kw_for)) {
292 return Style.AllowShortBlocksOnASingleLine
293 ? tryMergeSimpleBlock(I, E, Limit)
294 : 0;
295 }
296 // Try to merge a control statement block with left brace wrapped
297 if (I[1]->First->is(tok::l_brace) &&
298 TheLine->First->isOneOf(tok::kw_if, tok::kw_while, tok::kw_for)) {
299 return Style.BraceWrapping.AfterControlStatement
300 ? tryMergeSimpleBlock(I, E, Limit)
301 : 0;
302 }
303 // Try to merge either empty or one-line block if is precedeed by control
304 // statement token
305 if (TheLine->First->is(tok::l_brace) && TheLine->First == TheLine->Last &&
306 I != AnnotatedLines.begin() &&
307 I[-1]->First->isOneOf(tok::kw_if, tok::kw_while, tok::kw_for)) {
308 unsigned MergedLines = 0;
309 if (Style.AllowShortBlocksOnASingleLine) {
310 MergedLines = tryMergeSimpleBlock(I - 1, E, Limit);
311 // If we managed to merge the block, discard the first merged line
312 // since we are merging starting from I.
313 if (MergedLines > 0)
314 --MergedLines;
315 }
316 return MergedLines;
317 }
318 // Don't merge block with left brace wrapped after ObjC special blocks
319 if (TheLine->First->is(tok::l_brace) && I != AnnotatedLines.begin() &&
320 I[-1]->First->is(tok::at) && I[-1]->First->Next) {
321 tok::ObjCKeywordKind kwId = I[-1]->First->Next->Tok.getObjCKeywordID();
322 if (kwId == clang::tok::objc_autoreleasepool ||
323 kwId == clang::tok::objc_synchronized)
324 return 0;
325 }
326 // Try to merge a block with left brace wrapped that wasn't yet covered
327 if (TheLine->Last->is(tok::l_brace)) {
328 return !Style.BraceWrapping.AfterFunction ||
329 (I[1]->First->is(tok::r_brace) &&
330 !Style.BraceWrapping.SplitEmptyRecord)
331 ? tryMergeSimpleBlock(I, E, Limit)
332 : 0;
333 }
334 // Try to merge a function block with left brace wrapped
335 if (I[1]->First->is(TT_FunctionLBrace) &&
336 Style.BraceWrapping.AfterFunction) {
337 if (I[1]->Last->is(TT_LineComment))
338 return 0;
339
340 // Check for Limit <= 2 to account for the " {".
341 if (Limit <= 2 || (Style.ColumnLimit == 0 && containsMustBreak(TheLine)))
342 return 0;
343 Limit -= 2;
344
345 unsigned MergedLines = 0;
346 if (MergeShortFunctions ||
347 (Style.AllowShortFunctionsOnASingleLine >= FormatStyle::SFS_Empty &&
348 I[1]->First == I[1]->Last && I + 2 != E &&
349 I[2]->First->is(tok::r_brace))) {
350 MergedLines = tryMergeSimpleBlock(I + 1, E, Limit);
351 // If we managed to merge the block, count the function header, which is
352 // on a separate line.
353 if (MergedLines > 0)
354 ++MergedLines;
355 }
356 return MergedLines;
357 }
358 if (TheLine->First->is(tok::kw_if)) {
359 return Style.AllowShortIfStatementsOnASingleLine
360 ? tryMergeSimpleControlStatement(I, E, Limit)
361 : 0;
362 }
363 if (TheLine->First->isOneOf(tok::kw_for, tok::kw_while)) {
364 return Style.AllowShortLoopsOnASingleLine
365 ? tryMergeSimpleControlStatement(I, E, Limit)
366 : 0;
367 }
368 if (TheLine->First->isOneOf(tok::kw_case, tok::kw_default)) {
369 return Style.AllowShortCaseLabelsOnASingleLine
370 ? tryMergeShortCaseLabels(I, E, Limit)
371 : 0;
372 }
373 if (TheLine->InPPDirective &&
374 (TheLine->First->HasUnescapedNewline || TheLine->First->IsFirst)) {
375 return tryMergeSimplePPDirective(I, E, Limit);
376 }
377 return 0;
378 }
379
380 unsigned
tryMergeSimplePPDirective(SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E,unsigned Limit)381 tryMergeSimplePPDirective(SmallVectorImpl<AnnotatedLine *>::const_iterator I,
382 SmallVectorImpl<AnnotatedLine *>::const_iterator E,
383 unsigned Limit) {
384 if (Limit == 0)
385 return 0;
386 if (I + 2 != E && I[2]->InPPDirective && !I[2]->First->HasUnescapedNewline)
387 return 0;
388 if (1 + I[1]->Last->TotalLength > Limit)
389 return 0;
390 return 1;
391 }
392
tryMergeSimpleControlStatement(SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E,unsigned Limit)393 unsigned tryMergeSimpleControlStatement(
394 SmallVectorImpl<AnnotatedLine *>::const_iterator I,
395 SmallVectorImpl<AnnotatedLine *>::const_iterator E, unsigned Limit) {
396 if (Limit == 0)
397 return 0;
398 if (Style.BraceWrapping.AfterControlStatement &&
399 (I[1]->First->is(tok::l_brace) && !Style.AllowShortBlocksOnASingleLine))
400 return 0;
401 if (I[1]->InPPDirective != (*I)->InPPDirective ||
402 (I[1]->InPPDirective && I[1]->First->HasUnescapedNewline))
403 return 0;
404 Limit = limitConsideringMacros(I + 1, E, Limit);
405 AnnotatedLine &Line = **I;
406 if (Line.Last->isNot(tok::r_paren))
407 return 0;
408 if (1 + I[1]->Last->TotalLength > Limit)
409 return 0;
410 if (I[1]->First->isOneOf(tok::semi, tok::kw_if, tok::kw_for, tok::kw_while,
411 TT_LineComment))
412 return 0;
413 // Only inline simple if's (no nested if or else).
414 if (I + 2 != E && Line.startsWith(tok::kw_if) &&
415 I[2]->First->is(tok::kw_else))
416 return 0;
417 return 1;
418 }
419
420 unsigned
tryMergeShortCaseLabels(SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E,unsigned Limit)421 tryMergeShortCaseLabels(SmallVectorImpl<AnnotatedLine *>::const_iterator I,
422 SmallVectorImpl<AnnotatedLine *>::const_iterator E,
423 unsigned Limit) {
424 if (Limit == 0 || I + 1 == E ||
425 I[1]->First->isOneOf(tok::kw_case, tok::kw_default))
426 return 0;
427 unsigned NumStmts = 0;
428 unsigned Length = 0;
429 bool EndsWithComment = false;
430 bool InPPDirective = I[0]->InPPDirective;
431 const unsigned Level = I[0]->Level;
432 for (; NumStmts < 3; ++NumStmts) {
433 if (I + 1 + NumStmts == E)
434 break;
435 const AnnotatedLine *Line = I[1 + NumStmts];
436 if (Line->InPPDirective != InPPDirective)
437 break;
438 if (Line->First->isOneOf(tok::kw_case, tok::kw_default, tok::r_brace))
439 break;
440 if (Line->First->isOneOf(tok::kw_if, tok::kw_for, tok::kw_switch,
441 tok::kw_while) ||
442 EndsWithComment)
443 return 0;
444 if (Line->First->is(tok::comment)) {
445 if (Level != Line->Level)
446 return 0;
447 SmallVectorImpl<AnnotatedLine *>::const_iterator J = I + 2 + NumStmts;
448 for (; J != E; ++J) {
449 Line = *J;
450 if (Line->InPPDirective != InPPDirective)
451 break;
452 if (Line->First->isOneOf(tok::kw_case, tok::kw_default, tok::r_brace))
453 break;
454 if (Line->First->isNot(tok::comment) || Level != Line->Level)
455 return 0;
456 }
457 break;
458 }
459 if (Line->Last->is(tok::comment))
460 EndsWithComment = true;
461 Length += I[1 + NumStmts]->Last->TotalLength + 1; // 1 for the space.
462 }
463 if (NumStmts == 0 || NumStmts == 3 || Length > Limit)
464 return 0;
465 return NumStmts;
466 }
467
468 unsigned
tryMergeSimpleBlock(SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E,unsigned Limit)469 tryMergeSimpleBlock(SmallVectorImpl<AnnotatedLine *>::const_iterator I,
470 SmallVectorImpl<AnnotatedLine *>::const_iterator E,
471 unsigned Limit) {
472 AnnotatedLine &Line = **I;
473
474 // Don't merge ObjC @ keywords and methods.
475 // FIXME: If an option to allow short exception handling clauses on a single
476 // line is added, change this to not return for @try and friends.
477 if (Style.Language != FormatStyle::LK_Java &&
478 Line.First->isOneOf(tok::at, tok::minus, tok::plus))
479 return 0;
480
481 // Check that the current line allows merging. This depends on whether we
482 // are in a control flow statements as well as several style flags.
483 if (Line.First->isOneOf(tok::kw_else, tok::kw_case) ||
484 (Line.First->Next && Line.First->Next->is(tok::kw_else)))
485 return 0;
486 if (Line.First->isOneOf(tok::kw_if, tok::kw_while, tok::kw_do, tok::kw_try,
487 tok::kw___try, tok::kw_catch, tok::kw___finally,
488 tok::kw_for, tok::r_brace, Keywords.kw___except)) {
489 if (!Style.AllowShortBlocksOnASingleLine)
490 return 0;
491 // Don't merge when we can't except the case when
492 // the control statement block is empty
493 if (!Style.AllowShortIfStatementsOnASingleLine &&
494 Line.startsWith(tok::kw_if) &&
495 !Style.BraceWrapping.AfterControlStatement &&
496 !I[1]->First->is(tok::r_brace))
497 return 0;
498 if (!Style.AllowShortIfStatementsOnASingleLine &&
499 Line.startsWith(tok::kw_if) &&
500 Style.BraceWrapping.AfterControlStatement && I + 2 != E &&
501 !I[2]->First->is(tok::r_brace))
502 return 0;
503 if (!Style.AllowShortLoopsOnASingleLine &&
504 Line.First->isOneOf(tok::kw_while, tok::kw_do, tok::kw_for) &&
505 !Style.BraceWrapping.AfterControlStatement &&
506 !I[1]->First->is(tok::r_brace))
507 return 0;
508 if (!Style.AllowShortLoopsOnASingleLine &&
509 Line.First->isOneOf(tok::kw_while, tok::kw_do, tok::kw_for) &&
510 Style.BraceWrapping.AfterControlStatement && I + 2 != E &&
511 !I[2]->First->is(tok::r_brace))
512 return 0;
513 // FIXME: Consider an option to allow short exception handling clauses on
514 // a single line.
515 // FIXME: This isn't covered by tests.
516 // FIXME: For catch, __except, __finally the first token on the line
517 // is '}', so this isn't correct here.
518 if (Line.First->isOneOf(tok::kw_try, tok::kw___try, tok::kw_catch,
519 Keywords.kw___except, tok::kw___finally))
520 return 0;
521 }
522
523 if (Line.Last->is(tok::l_brace)) {
524 FormatToken *Tok = I[1]->First;
525 if (Tok->is(tok::r_brace) && !Tok->MustBreakBefore &&
526 (Tok->getNextNonComment() == nullptr ||
527 Tok->getNextNonComment()->is(tok::semi))) {
528 // We merge empty blocks even if the line exceeds the column limit.
529 Tok->SpacesRequiredBefore = 0;
530 Tok->CanBreakBefore = true;
531 return 1;
532 } else if (Limit != 0 && !Line.startsWith(tok::kw_namespace) &&
533 !startsExternCBlock(Line)) {
534 // We don't merge short records.
535 FormatToken *RecordTok = Line.First;
536 // Skip record modifiers.
537 while (RecordTok->Next &&
538 RecordTok->isOneOf(tok::kw_typedef, tok::kw_export,
539 Keywords.kw_declare, Keywords.kw_abstract,
540 tok::kw_default))
541 RecordTok = RecordTok->Next;
542 if (RecordTok &&
543 RecordTok->isOneOf(tok::kw_class, tok::kw_union, tok::kw_struct,
544 Keywords.kw_interface))
545 return 0;
546
547 // Check that we still have three lines and they fit into the limit.
548 if (I + 2 == E || I[2]->Type == LT_Invalid)
549 return 0;
550 Limit = limitConsideringMacros(I + 2, E, Limit);
551
552 if (!nextTwoLinesFitInto(I, Limit))
553 return 0;
554
555 // Second, check that the next line does not contain any braces - if it
556 // does, readability declines when putting it into a single line.
557 if (I[1]->Last->is(TT_LineComment))
558 return 0;
559 do {
560 if (Tok->is(tok::l_brace) && Tok->BlockKind != BK_BracedInit)
561 return 0;
562 Tok = Tok->Next;
563 } while (Tok);
564
565 // Last, check that the third line starts with a closing brace.
566 Tok = I[2]->First;
567 if (Tok->isNot(tok::r_brace))
568 return 0;
569
570 // Don't merge "if (a) { .. } else {".
571 if (Tok->Next && Tok->Next->is(tok::kw_else))
572 return 0;
573
574 return 2;
575 }
576 } else if (I[1]->First->is(tok::l_brace)) {
577 if (I[1]->Last->is(TT_LineComment))
578 return 0;
579
580 // Check for Limit <= 2 to account for the " {".
581 if (Limit <= 2 || (Style.ColumnLimit == 0 && containsMustBreak(*I)))
582 return 0;
583 Limit -= 2;
584 unsigned MergedLines = 0;
585 if (Style.AllowShortBlocksOnASingleLine ||
586 (I[1]->First == I[1]->Last && I + 2 != E &&
587 I[2]->First->is(tok::r_brace))) {
588 MergedLines = tryMergeSimpleBlock(I + 1, E, Limit);
589 // If we managed to merge the block, count the statement header, which
590 // is on a separate line.
591 if (MergedLines > 0)
592 ++MergedLines;
593 }
594 return MergedLines;
595 }
596 return 0;
597 }
598
599 /// Returns the modified column limit for \p I if it is inside a macro and
600 /// needs a trailing '\'.
601 unsigned
limitConsideringMacros(SmallVectorImpl<AnnotatedLine * >::const_iterator I,SmallVectorImpl<AnnotatedLine * >::const_iterator E,unsigned Limit)602 limitConsideringMacros(SmallVectorImpl<AnnotatedLine *>::const_iterator I,
603 SmallVectorImpl<AnnotatedLine *>::const_iterator E,
604 unsigned Limit) {
605 if (I[0]->InPPDirective && I + 1 != E &&
606 !I[1]->First->HasUnescapedNewline && !I[1]->First->is(tok::eof)) {
607 return Limit < 2 ? 0 : Limit - 2;
608 }
609 return Limit;
610 }
611
nextTwoLinesFitInto(SmallVectorImpl<AnnotatedLine * >::const_iterator I,unsigned Limit)612 bool nextTwoLinesFitInto(SmallVectorImpl<AnnotatedLine *>::const_iterator I,
613 unsigned Limit) {
614 if (I[1]->First->MustBreakBefore || I[2]->First->MustBreakBefore)
615 return false;
616 return 1 + I[1]->Last->TotalLength + 1 + I[2]->Last->TotalLength <= Limit;
617 }
618
containsMustBreak(const AnnotatedLine * Line)619 bool containsMustBreak(const AnnotatedLine *Line) {
620 for (const FormatToken *Tok = Line->First; Tok; Tok = Tok->Next) {
621 if (Tok->MustBreakBefore)
622 return true;
623 }
624 return false;
625 }
626
join(AnnotatedLine & A,const AnnotatedLine & B)627 void join(AnnotatedLine &A, const AnnotatedLine &B) {
628 assert(!A.Last->Next);
629 assert(!B.First->Previous);
630 if (B.Affected)
631 A.Affected = true;
632 A.Last->Next = B.First;
633 B.First->Previous = A.Last;
634 B.First->CanBreakBefore = true;
635 unsigned LengthA = A.Last->TotalLength + B.First->SpacesRequiredBefore;
636 for (FormatToken *Tok = B.First; Tok; Tok = Tok->Next) {
637 Tok->TotalLength += LengthA;
638 A.Last = Tok;
639 }
640 }
641
642 const FormatStyle &Style;
643 const AdditionalKeywords &Keywords;
644 const SmallVectorImpl<AnnotatedLine *>::const_iterator End;
645
646 SmallVectorImpl<AnnotatedLine *>::const_iterator Next;
647 const SmallVectorImpl<AnnotatedLine *> &AnnotatedLines;
648 };
649
markFinalized(FormatToken * Tok)650 static void markFinalized(FormatToken *Tok) {
651 for (; Tok; Tok = Tok->Next) {
652 Tok->Finalized = true;
653 for (AnnotatedLine *Child : Tok->Children)
654 markFinalized(Child->First);
655 }
656 }
657
658 #ifndef NDEBUG
printLineState(const LineState & State)659 static void printLineState(const LineState &State) {
660 llvm::dbgs() << "State: ";
661 for (const ParenState &P : State.Stack) {
662 llvm::dbgs() << (P.Tok ? P.Tok->TokenText : "F") << "|" << P.Indent << "|"
663 << P.LastSpace << "|" << P.NestedBlockIndent << " ";
664 }
665 llvm::dbgs() << State.NextToken->TokenText << "\n";
666 }
667 #endif
668
669 /// Base class for classes that format one \c AnnotatedLine.
670 class LineFormatter {
671 public:
LineFormatter(ContinuationIndenter * Indenter,WhitespaceManager * Whitespaces,const FormatStyle & Style,UnwrappedLineFormatter * BlockFormatter)672 LineFormatter(ContinuationIndenter *Indenter, WhitespaceManager *Whitespaces,
673 const FormatStyle &Style,
674 UnwrappedLineFormatter *BlockFormatter)
675 : Indenter(Indenter), Whitespaces(Whitespaces), Style(Style),
676 BlockFormatter(BlockFormatter) {}
~LineFormatter()677 virtual ~LineFormatter() {}
678
679 /// Formats an \c AnnotatedLine and returns the penalty.
680 ///
681 /// If \p DryRun is \c false, directly applies the changes.
682 virtual unsigned formatLine(const AnnotatedLine &Line,
683 unsigned FirstIndent,
684 unsigned FirstStartColumn,
685 bool DryRun) = 0;
686
687 protected:
688 /// If the \p State's next token is an r_brace closing a nested block,
689 /// format the nested block before it.
690 ///
691 /// Returns \c true if all children could be placed successfully and adapts
692 /// \p Penalty as well as \p State. If \p DryRun is false, also directly
693 /// creates changes using \c Whitespaces.
694 ///
695 /// The crucial idea here is that children always get formatted upon
696 /// encountering the closing brace right after the nested block. Now, if we
697 /// are currently trying to keep the "}" on the same line (i.e. \p NewLine is
698 /// \c false), the entire block has to be kept on the same line (which is only
699 /// possible if it fits on the line, only contains a single statement, etc.
700 ///
701 /// If \p NewLine is true, we format the nested block on separate lines, i.e.
702 /// break after the "{", format all lines with correct indentation and the put
703 /// the closing "}" on yet another new line.
704 ///
705 /// This enables us to keep the simple structure of the
706 /// \c UnwrappedLineFormatter, where we only have two options for each token:
707 /// break or don't break.
formatChildren(LineState & State,bool NewLine,bool DryRun,unsigned & Penalty)708 bool formatChildren(LineState &State, bool NewLine, bool DryRun,
709 unsigned &Penalty) {
710 const FormatToken *LBrace = State.NextToken->getPreviousNonComment();
711 FormatToken &Previous = *State.NextToken->Previous;
712 if (!LBrace || LBrace->isNot(tok::l_brace) ||
713 LBrace->BlockKind != BK_Block || Previous.Children.size() == 0)
714 // The previous token does not open a block. Nothing to do. We don't
715 // assert so that we can simply call this function for all tokens.
716 return true;
717
718 if (NewLine) {
719 int AdditionalIndent = State.Stack.back().Indent -
720 Previous.Children[0]->Level * Style.IndentWidth;
721
722 Penalty +=
723 BlockFormatter->format(Previous.Children, DryRun, AdditionalIndent,
724 /*FixBadIndentation=*/true);
725 return true;
726 }
727
728 if (Previous.Children[0]->First->MustBreakBefore)
729 return false;
730
731 // Cannot merge into one line if this line ends on a comment.
732 if (Previous.is(tok::comment))
733 return false;
734
735 // Cannot merge multiple statements into a single line.
736 if (Previous.Children.size() > 1)
737 return false;
738
739 const AnnotatedLine *Child = Previous.Children[0];
740 // We can't put the closing "}" on a line with a trailing comment.
741 if (Child->Last->isTrailingComment())
742 return false;
743
744 // If the child line exceeds the column limit, we wouldn't want to merge it.
745 // We add +2 for the trailing " }".
746 if (Style.ColumnLimit > 0 &&
747 Child->Last->TotalLength + State.Column + 2 > Style.ColumnLimit)
748 return false;
749
750 if (!DryRun) {
751 Whitespaces->replaceWhitespace(
752 *Child->First, /*Newlines=*/0, /*Spaces=*/1,
753 /*StartOfTokenColumn=*/State.Column, State.Line->InPPDirective);
754 }
755 Penalty +=
756 formatLine(*Child, State.Column + 1, /*FirstStartColumn=*/0, DryRun);
757
758 State.Column += 1 + Child->Last->TotalLength;
759 return true;
760 }
761
762 ContinuationIndenter *Indenter;
763
764 private:
765 WhitespaceManager *Whitespaces;
766 const FormatStyle &Style;
767 UnwrappedLineFormatter *BlockFormatter;
768 };
769
770 /// Formatter that keeps the existing line breaks.
771 class NoColumnLimitLineFormatter : public LineFormatter {
772 public:
NoColumnLimitLineFormatter(ContinuationIndenter * Indenter,WhitespaceManager * Whitespaces,const FormatStyle & Style,UnwrappedLineFormatter * BlockFormatter)773 NoColumnLimitLineFormatter(ContinuationIndenter *Indenter,
774 WhitespaceManager *Whitespaces,
775 const FormatStyle &Style,
776 UnwrappedLineFormatter *BlockFormatter)
777 : LineFormatter(Indenter, Whitespaces, Style, BlockFormatter) {}
778
779 /// Formats the line, simply keeping all of the input's line breaking
780 /// decisions.
formatLine(const AnnotatedLine & Line,unsigned FirstIndent,unsigned FirstStartColumn,bool DryRun)781 unsigned formatLine(const AnnotatedLine &Line, unsigned FirstIndent,
782 unsigned FirstStartColumn, bool DryRun) override {
783 assert(!DryRun);
784 LineState State = Indenter->getInitialState(FirstIndent, FirstStartColumn,
785 &Line, /*DryRun=*/false);
786 while (State.NextToken) {
787 bool Newline =
788 Indenter->mustBreak(State) ||
789 (Indenter->canBreak(State) && State.NextToken->NewlinesBefore > 0);
790 unsigned Penalty = 0;
791 formatChildren(State, Newline, /*DryRun=*/false, Penalty);
792 Indenter->addTokenToState(State, Newline, /*DryRun=*/false);
793 }
794 return 0;
795 }
796 };
797
798 /// Formatter that puts all tokens into a single line without breaks.
799 class NoLineBreakFormatter : public LineFormatter {
800 public:
NoLineBreakFormatter(ContinuationIndenter * Indenter,WhitespaceManager * Whitespaces,const FormatStyle & Style,UnwrappedLineFormatter * BlockFormatter)801 NoLineBreakFormatter(ContinuationIndenter *Indenter,
802 WhitespaceManager *Whitespaces, const FormatStyle &Style,
803 UnwrappedLineFormatter *BlockFormatter)
804 : LineFormatter(Indenter, Whitespaces, Style, BlockFormatter) {}
805
806 /// Puts all tokens into a single line.
formatLine(const AnnotatedLine & Line,unsigned FirstIndent,unsigned FirstStartColumn,bool DryRun)807 unsigned formatLine(const AnnotatedLine &Line, unsigned FirstIndent,
808 unsigned FirstStartColumn, bool DryRun) override {
809 unsigned Penalty = 0;
810 LineState State =
811 Indenter->getInitialState(FirstIndent, FirstStartColumn, &Line, DryRun);
812 while (State.NextToken) {
813 formatChildren(State, /*Newline=*/false, DryRun, Penalty);
814 Indenter->addTokenToState(
815 State, /*Newline=*/State.NextToken->MustBreakBefore, DryRun);
816 }
817 return Penalty;
818 }
819 };
820
821 /// Finds the best way to break lines.
822 class OptimizingLineFormatter : public LineFormatter {
823 public:
OptimizingLineFormatter(ContinuationIndenter * Indenter,WhitespaceManager * Whitespaces,const FormatStyle & Style,UnwrappedLineFormatter * BlockFormatter)824 OptimizingLineFormatter(ContinuationIndenter *Indenter,
825 WhitespaceManager *Whitespaces,
826 const FormatStyle &Style,
827 UnwrappedLineFormatter *BlockFormatter)
828 : LineFormatter(Indenter, Whitespaces, Style, BlockFormatter) {}
829
830 /// Formats the line by finding the best line breaks with line lengths
831 /// below the column limit.
formatLine(const AnnotatedLine & Line,unsigned FirstIndent,unsigned FirstStartColumn,bool DryRun)832 unsigned formatLine(const AnnotatedLine &Line, unsigned FirstIndent,
833 unsigned FirstStartColumn, bool DryRun) override {
834 LineState State =
835 Indenter->getInitialState(FirstIndent, FirstStartColumn, &Line, DryRun);
836
837 // If the ObjC method declaration does not fit on a line, we should format
838 // it with one arg per line.
839 if (State.Line->Type == LT_ObjCMethodDecl)
840 State.Stack.back().BreakBeforeParameter = true;
841
842 // Find best solution in solution space.
843 return analyzeSolutionSpace(State, DryRun);
844 }
845
846 private:
847 struct CompareLineStatePointers {
operator ()clang::format::__anon738017f00111::OptimizingLineFormatter::CompareLineStatePointers848 bool operator()(LineState *obj1, LineState *obj2) const {
849 return *obj1 < *obj2;
850 }
851 };
852
853 /// A pair of <penalty, count> that is used to prioritize the BFS on.
854 ///
855 /// In case of equal penalties, we want to prefer states that were inserted
856 /// first. During state generation we make sure that we insert states first
857 /// that break the line as late as possible.
858 typedef std::pair<unsigned, unsigned> OrderedPenalty;
859
860 /// An edge in the solution space from \c Previous->State to \c State,
861 /// inserting a newline dependent on the \c NewLine.
862 struct StateNode {
StateNodeclang::format::__anon738017f00111::OptimizingLineFormatter::StateNode863 StateNode(const LineState &State, bool NewLine, StateNode *Previous)
864 : State(State), NewLine(NewLine), Previous(Previous) {}
865 LineState State;
866 bool NewLine;
867 StateNode *Previous;
868 };
869
870 /// An item in the prioritized BFS search queue. The \c StateNode's
871 /// \c State has the given \c OrderedPenalty.
872 typedef std::pair<OrderedPenalty, StateNode *> QueueItem;
873
874 /// The BFS queue type.
875 typedef std::priority_queue<QueueItem, std::vector<QueueItem>,
876 std::greater<QueueItem>>
877 QueueType;
878
879 /// Analyze the entire solution space starting from \p InitialState.
880 ///
881 /// This implements a variant of Dijkstra's algorithm on the graph that spans
882 /// the solution space (\c LineStates are the nodes). The algorithm tries to
883 /// find the shortest path (the one with lowest penalty) from \p InitialState
884 /// to a state where all tokens are placed. Returns the penalty.
885 ///
886 /// If \p DryRun is \c false, directly applies the changes.
analyzeSolutionSpace(LineState & InitialState,bool DryRun)887 unsigned analyzeSolutionSpace(LineState &InitialState, bool DryRun) {
888 std::set<LineState *, CompareLineStatePointers> Seen;
889
890 // Increasing count of \c StateNode items we have created. This is used to
891 // create a deterministic order independent of the container.
892 unsigned Count = 0;
893 QueueType Queue;
894
895 // Insert start element into queue.
896 StateNode *Node =
897 new (Allocator.Allocate()) StateNode(InitialState, false, nullptr);
898 Queue.push(QueueItem(OrderedPenalty(0, Count), Node));
899 ++Count;
900
901 unsigned Penalty = 0;
902
903 // While not empty, take first element and follow edges.
904 while (!Queue.empty()) {
905 Penalty = Queue.top().first.first;
906 StateNode *Node = Queue.top().second;
907 if (!Node->State.NextToken) {
908 LLVM_DEBUG(llvm::dbgs()
909 << "\n---\nPenalty for line: " << Penalty << "\n");
910 break;
911 }
912 Queue.pop();
913
914 // Cut off the analysis of certain solutions if the analysis gets too
915 // complex. See description of IgnoreStackForComparison.
916 if (Count > 50000)
917 Node->State.IgnoreStackForComparison = true;
918
919 if (!Seen.insert(&Node->State).second)
920 // State already examined with lower penalty.
921 continue;
922
923 FormatDecision LastFormat = Node->State.NextToken->Decision;
924 if (LastFormat == FD_Unformatted || LastFormat == FD_Continue)
925 addNextStateToQueue(Penalty, Node, /*NewLine=*/false, &Count, &Queue);
926 if (LastFormat == FD_Unformatted || LastFormat == FD_Break)
927 addNextStateToQueue(Penalty, Node, /*NewLine=*/true, &Count, &Queue);
928 }
929
930 if (Queue.empty()) {
931 // We were unable to find a solution, do nothing.
932 // FIXME: Add diagnostic?
933 LLVM_DEBUG(llvm::dbgs() << "Could not find a solution.\n");
934 return 0;
935 }
936
937 // Reconstruct the solution.
938 if (!DryRun)
939 reconstructPath(InitialState, Queue.top().second);
940
941 LLVM_DEBUG(llvm::dbgs()
942 << "Total number of analyzed states: " << Count << "\n");
943 LLVM_DEBUG(llvm::dbgs() << "---\n");
944
945 return Penalty;
946 }
947
948 /// Add the following state to the analysis queue \c Queue.
949 ///
950 /// Assume the current state is \p PreviousNode and has been reached with a
951 /// penalty of \p Penalty. Insert a line break if \p NewLine is \c true.
addNextStateToQueue(unsigned Penalty,StateNode * PreviousNode,bool NewLine,unsigned * Count,QueueType * Queue)952 void addNextStateToQueue(unsigned Penalty, StateNode *PreviousNode,
953 bool NewLine, unsigned *Count, QueueType *Queue) {
954 if (NewLine && !Indenter->canBreak(PreviousNode->State))
955 return;
956 if (!NewLine && Indenter->mustBreak(PreviousNode->State))
957 return;
958
959 StateNode *Node = new (Allocator.Allocate())
960 StateNode(PreviousNode->State, NewLine, PreviousNode);
961 if (!formatChildren(Node->State, NewLine, /*DryRun=*/true, Penalty))
962 return;
963
964 Penalty += Indenter->addTokenToState(Node->State, NewLine, true);
965
966 Queue->push(QueueItem(OrderedPenalty(Penalty, *Count), Node));
967 ++(*Count);
968 }
969
970 /// Applies the best formatting by reconstructing the path in the
971 /// solution space that leads to \c Best.
reconstructPath(LineState & State,StateNode * Best)972 void reconstructPath(LineState &State, StateNode *Best) {
973 std::deque<StateNode *> Path;
974 // We do not need a break before the initial token.
975 while (Best->Previous) {
976 Path.push_front(Best);
977 Best = Best->Previous;
978 }
979 for (std::deque<StateNode *>::iterator I = Path.begin(), E = Path.end();
980 I != E; ++I) {
981 unsigned Penalty = 0;
982 formatChildren(State, (*I)->NewLine, /*DryRun=*/false, Penalty);
983 Penalty += Indenter->addTokenToState(State, (*I)->NewLine, false);
984
985 LLVM_DEBUG({
986 printLineState((*I)->Previous->State);
987 if ((*I)->NewLine) {
988 llvm::dbgs() << "Penalty for placing "
989 << (*I)->Previous->State.NextToken->Tok.getName() << ": "
990 << Penalty << "\n";
991 }
992 });
993 }
994 }
995
996 llvm::SpecificBumpPtrAllocator<StateNode> Allocator;
997 };
998
999 } // anonymous namespace
1000
1001 unsigned
format(const SmallVectorImpl<AnnotatedLine * > & Lines,bool DryRun,int AdditionalIndent,bool FixBadIndentation,unsigned FirstStartColumn,unsigned NextStartColumn,unsigned LastStartColumn)1002 UnwrappedLineFormatter::format(const SmallVectorImpl<AnnotatedLine *> &Lines,
1003 bool DryRun, int AdditionalIndent,
1004 bool FixBadIndentation,
1005 unsigned FirstStartColumn,
1006 unsigned NextStartColumn,
1007 unsigned LastStartColumn) {
1008 LineJoiner Joiner(Style, Keywords, Lines);
1009
1010 // Try to look up already computed penalty in DryRun-mode.
1011 std::pair<const SmallVectorImpl<AnnotatedLine *> *, unsigned> CacheKey(
1012 &Lines, AdditionalIndent);
1013 auto CacheIt = PenaltyCache.find(CacheKey);
1014 if (DryRun && CacheIt != PenaltyCache.end())
1015 return CacheIt->second;
1016
1017 assert(!Lines.empty());
1018 unsigned Penalty = 0;
1019 LevelIndentTracker IndentTracker(Style, Keywords, Lines[0]->Level,
1020 AdditionalIndent);
1021 const AnnotatedLine *PreviousLine = nullptr;
1022 const AnnotatedLine *NextLine = nullptr;
1023
1024 // The minimum level of consecutive lines that have been formatted.
1025 unsigned RangeMinLevel = UINT_MAX;
1026
1027 bool FirstLine = true;
1028 for (const AnnotatedLine *Line =
1029 Joiner.getNextMergedLine(DryRun, IndentTracker);
1030 Line; Line = NextLine, FirstLine = false) {
1031 const AnnotatedLine &TheLine = *Line;
1032 unsigned Indent = IndentTracker.getIndent();
1033
1034 // We continue formatting unchanged lines to adjust their indent, e.g. if a
1035 // scope was added. However, we need to carefully stop doing this when we
1036 // exit the scope of affected lines to prevent indenting a the entire
1037 // remaining file if it currently missing a closing brace.
1038 bool PreviousRBrace =
1039 PreviousLine && PreviousLine->startsWith(tok::r_brace);
1040 bool ContinueFormatting =
1041 TheLine.Level > RangeMinLevel ||
1042 (TheLine.Level == RangeMinLevel && !PreviousRBrace &&
1043 !TheLine.startsWith(tok::r_brace));
1044
1045 bool FixIndentation = (FixBadIndentation || ContinueFormatting) &&
1046 Indent != TheLine.First->OriginalColumn;
1047 bool ShouldFormat = TheLine.Affected || FixIndentation;
1048 // We cannot format this line; if the reason is that the line had a
1049 // parsing error, remember that.
1050 if (ShouldFormat && TheLine.Type == LT_Invalid && Status) {
1051 Status->FormatComplete = false;
1052 Status->Line =
1053 SourceMgr.getSpellingLineNumber(TheLine.First->Tok.getLocation());
1054 }
1055
1056 if (ShouldFormat && TheLine.Type != LT_Invalid) {
1057 if (!DryRun) {
1058 bool LastLine = Line->First->is(tok::eof);
1059 formatFirstToken(TheLine, PreviousLine, Lines, Indent,
1060 LastLine ? LastStartColumn : NextStartColumn + Indent);
1061 }
1062
1063 NextLine = Joiner.getNextMergedLine(DryRun, IndentTracker);
1064 unsigned ColumnLimit = getColumnLimit(TheLine.InPPDirective, NextLine);
1065 bool FitsIntoOneLine =
1066 TheLine.Last->TotalLength + Indent <= ColumnLimit ||
1067 (TheLine.Type == LT_ImportStatement &&
1068 (Style.Language != FormatStyle::LK_JavaScript ||
1069 !Style.JavaScriptWrapImports));
1070 if (Style.ColumnLimit == 0)
1071 NoColumnLimitLineFormatter(Indenter, Whitespaces, Style, this)
1072 .formatLine(TheLine, NextStartColumn + Indent,
1073 FirstLine ? FirstStartColumn : 0, DryRun);
1074 else if (FitsIntoOneLine)
1075 Penalty += NoLineBreakFormatter(Indenter, Whitespaces, Style, this)
1076 .formatLine(TheLine, NextStartColumn + Indent,
1077 FirstLine ? FirstStartColumn : 0, DryRun);
1078 else
1079 Penalty += OptimizingLineFormatter(Indenter, Whitespaces, Style, this)
1080 .formatLine(TheLine, NextStartColumn + Indent,
1081 FirstLine ? FirstStartColumn : 0, DryRun);
1082 RangeMinLevel = std::min(RangeMinLevel, TheLine.Level);
1083 } else {
1084 // If no token in the current line is affected, we still need to format
1085 // affected children.
1086 if (TheLine.ChildrenAffected)
1087 for (const FormatToken *Tok = TheLine.First; Tok; Tok = Tok->Next)
1088 if (!Tok->Children.empty())
1089 format(Tok->Children, DryRun);
1090
1091 // Adapt following lines on the current indent level to the same level
1092 // unless the current \c AnnotatedLine is not at the beginning of a line.
1093 bool StartsNewLine =
1094 TheLine.First->NewlinesBefore > 0 || TheLine.First->IsFirst;
1095 if (StartsNewLine)
1096 IndentTracker.adjustToUnmodifiedLine(TheLine);
1097 if (!DryRun) {
1098 bool ReformatLeadingWhitespace =
1099 StartsNewLine && ((PreviousLine && PreviousLine->Affected) ||
1100 TheLine.LeadingEmptyLinesAffected);
1101 // Format the first token.
1102 if (ReformatLeadingWhitespace)
1103 formatFirstToken(TheLine, PreviousLine, Lines,
1104 TheLine.First->OriginalColumn,
1105 TheLine.First->OriginalColumn);
1106 else
1107 Whitespaces->addUntouchableToken(*TheLine.First,
1108 TheLine.InPPDirective);
1109
1110 // Notify the WhitespaceManager about the unchanged whitespace.
1111 for (FormatToken *Tok = TheLine.First->Next; Tok; Tok = Tok->Next)
1112 Whitespaces->addUntouchableToken(*Tok, TheLine.InPPDirective);
1113 }
1114 NextLine = Joiner.getNextMergedLine(DryRun, IndentTracker);
1115 RangeMinLevel = UINT_MAX;
1116 }
1117 if (!DryRun)
1118 markFinalized(TheLine.First);
1119 PreviousLine = &TheLine;
1120 }
1121 PenaltyCache[CacheKey] = Penalty;
1122 return Penalty;
1123 }
1124
formatFirstToken(const AnnotatedLine & Line,const AnnotatedLine * PreviousLine,const SmallVectorImpl<AnnotatedLine * > & Lines,unsigned Indent,unsigned NewlineIndent)1125 void UnwrappedLineFormatter::formatFirstToken(
1126 const AnnotatedLine &Line, const AnnotatedLine *PreviousLine,
1127 const SmallVectorImpl<AnnotatedLine *> &Lines, unsigned Indent,
1128 unsigned NewlineIndent) {
1129 FormatToken &RootToken = *Line.First;
1130 if (RootToken.is(tok::eof)) {
1131 unsigned Newlines = std::min(RootToken.NewlinesBefore, 1u);
1132 unsigned TokenIndent = Newlines ? NewlineIndent : 0;
1133 Whitespaces->replaceWhitespace(RootToken, Newlines, TokenIndent,
1134 TokenIndent);
1135 return;
1136 }
1137 unsigned Newlines =
1138 std::min(RootToken.NewlinesBefore, Style.MaxEmptyLinesToKeep + 1);
1139 // Remove empty lines before "}" where applicable.
1140 if (RootToken.is(tok::r_brace) &&
1141 (!RootToken.Next ||
1142 (RootToken.Next->is(tok::semi) && !RootToken.Next->Next)) &&
1143 // Do not remove empty lines before namespace closing "}".
1144 !getNamespaceToken(&Line, Lines))
1145 Newlines = std::min(Newlines, 1u);
1146 // Remove empty lines at the start of nested blocks (lambdas/arrow functions)
1147 if (PreviousLine == nullptr && Line.Level > 0)
1148 Newlines = std::min(Newlines, 1u);
1149 if (Newlines == 0 && !RootToken.IsFirst)
1150 Newlines = 1;
1151 if (RootToken.IsFirst && !RootToken.HasUnescapedNewline)
1152 Newlines = 0;
1153
1154 // Remove empty lines after "{".
1155 if (!Style.KeepEmptyLinesAtTheStartOfBlocks && PreviousLine &&
1156 PreviousLine->Last->is(tok::l_brace) &&
1157 PreviousLine->First->isNot(tok::kw_namespace) &&
1158 !startsExternCBlock(*PreviousLine))
1159 Newlines = 1;
1160
1161 // Insert extra new line before access specifiers.
1162 if (PreviousLine && PreviousLine->Last->isOneOf(tok::semi, tok::r_brace) &&
1163 RootToken.isAccessSpecifier() && RootToken.NewlinesBefore == 1)
1164 ++Newlines;
1165
1166 // Remove empty lines after access specifiers.
1167 if (PreviousLine && PreviousLine->First->isAccessSpecifier() &&
1168 (!PreviousLine->InPPDirective || !RootToken.HasUnescapedNewline))
1169 Newlines = std::min(1u, Newlines);
1170
1171 if (Newlines)
1172 Indent = NewlineIndent;
1173
1174 // Preprocessor directives get indented after the hash, if indented.
1175 if (Line.Type == LT_PreprocessorDirective || Line.Type == LT_ImportStatement)
1176 Indent = 0;
1177
1178 Whitespaces->replaceWhitespace(RootToken, Newlines, Indent, Indent,
1179 Line.InPPDirective &&
1180 !RootToken.HasUnescapedNewline);
1181 }
1182
1183 unsigned
getColumnLimit(bool InPPDirective,const AnnotatedLine * NextLine) const1184 UnwrappedLineFormatter::getColumnLimit(bool InPPDirective,
1185 const AnnotatedLine *NextLine) const {
1186 // In preprocessor directives reserve two chars for trailing " \" if the
1187 // next line continues the preprocessor directive.
1188 bool ContinuesPPDirective =
1189 InPPDirective &&
1190 // If there is no next line, this is likely a child line and the parent
1191 // continues the preprocessor directive.
1192 (!NextLine ||
1193 (NextLine->InPPDirective &&
1194 // If there is an unescaped newline between this line and the next, the
1195 // next line starts a new preprocessor directive.
1196 !NextLine->First->HasUnescapedNewline));
1197 return Style.ColumnLimit - (ContinuesPPDirective ? 2 : 0);
1198 }
1199
1200 } // namespace format
1201 } // namespace clang
1202