clang  10.0.0svn
BreakableToken.cpp
Go to the documentation of this file.
1 //===--- BreakableToken.cpp - Format C++ code -----------------------------===//
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 /// \file
10 /// Contains implementation of BreakableToken class and classes derived
11 /// from it.
12 ///
13 //===----------------------------------------------------------------------===//
14 
15 #include "BreakableToken.h"
16 #include "ContinuationIndenter.h"
17 #include "clang/Basic/CharInfo.h"
18 #include "clang/Format/Format.h"
19 #include "llvm/ADT/STLExtras.h"
20 #include "llvm/Support/Debug.h"
21 #include <algorithm>
22 
23 #define DEBUG_TYPE "format-token-breaker"
24 
25 namespace clang {
26 namespace format {
27 
28 static const char *const Blanks = " \t\v\f\r";
29 static bool IsBlank(char C) {
30  switch (C) {
31  case ' ':
32  case '\t':
33  case '\v':
34  case '\f':
35  case '\r':
36  return true;
37  default:
38  return false;
39  }
40 }
41 
42 static StringRef getLineCommentIndentPrefix(StringRef Comment,
43  const FormatStyle &Style) {
44  static const char *const KnownCStylePrefixes[] = {"///<", "//!<", "///", "//",
45  "//!"};
46  static const char *const KnownTextProtoPrefixes[] = {"//", "#", "##", "###",
47  "####"};
48  ArrayRef<const char *> KnownPrefixes(KnownCStylePrefixes);
49  if (Style.Language == FormatStyle::LK_TextProto)
50  KnownPrefixes = KnownTextProtoPrefixes;
51 
52  StringRef LongestPrefix;
53  for (StringRef KnownPrefix : KnownPrefixes) {
54  if (Comment.startswith(KnownPrefix)) {
55  size_t PrefixLength = KnownPrefix.size();
56  while (PrefixLength < Comment.size() && Comment[PrefixLength] == ' ')
57  ++PrefixLength;
58  if (PrefixLength > LongestPrefix.size())
59  LongestPrefix = Comment.substr(0, PrefixLength);
60  }
61  }
62  return LongestPrefix;
63 }
64 
65 static BreakableToken::Split
66 getCommentSplit(StringRef Text, unsigned ContentStartColumn,
67  unsigned ColumnLimit, unsigned TabWidth,
69  bool DecorationEndsWithStar = false) {
70  LLVM_DEBUG(llvm::dbgs() << "Comment split: \"" << Text
71  << "\", Column limit: " << ColumnLimit
72  << ", Content start: " << ContentStartColumn << "\n");
73  if (ColumnLimit <= ContentStartColumn + 1)
74  return BreakableToken::Split(StringRef::npos, 0);
75 
76  unsigned MaxSplit = ColumnLimit - ContentStartColumn + 1;
77  unsigned MaxSplitBytes = 0;
78 
79  for (unsigned NumChars = 0;
80  NumChars < MaxSplit && MaxSplitBytes < Text.size();) {
81  unsigned BytesInChar =
82  encoding::getCodePointNumBytes(Text[MaxSplitBytes], Encoding);
83  NumChars +=
84  encoding::columnWidthWithTabs(Text.substr(MaxSplitBytes, BytesInChar),
85  ContentStartColumn, TabWidth, Encoding);
86  MaxSplitBytes += BytesInChar;
87  }
88 
89  StringRef::size_type SpaceOffset = Text.find_last_of(Blanks, MaxSplitBytes);
90 
91  static auto *const kNumberedListRegexp = new llvm::Regex("^[1-9][0-9]?\\.");
92  while (SpaceOffset != StringRef::npos) {
93  // Do not split before a number followed by a dot: this would be interpreted
94  // as a numbered list, which would prevent re-flowing in subsequent passes.
95  if (kNumberedListRegexp->match(Text.substr(SpaceOffset).ltrim(Blanks)))
96  SpaceOffset = Text.find_last_of(Blanks, SpaceOffset);
97  // In JavaScript, some @tags can be followed by {, and machinery that parses
98  // these comments will fail to understand the comment if followed by a line
99  // break. So avoid ever breaking before a {.
100  else if (Style.Language == FormatStyle::LK_JavaScript &&
101  SpaceOffset + 1 < Text.size() && Text[SpaceOffset + 1] == '{')
102  SpaceOffset = Text.find_last_of(Blanks, SpaceOffset);
103  else
104  break;
105  }
106 
107  if (SpaceOffset == StringRef::npos ||
108  // Don't break at leading whitespace.
109  Text.find_last_not_of(Blanks, SpaceOffset) == StringRef::npos) {
110  // Make sure that we don't break at leading whitespace that
111  // reaches past MaxSplit.
112  StringRef::size_type FirstNonWhitespace = Text.find_first_not_of(Blanks);
113  if (FirstNonWhitespace == StringRef::npos)
114  // If the comment is only whitespace, we cannot split.
115  return BreakableToken::Split(StringRef::npos, 0);
116  SpaceOffset = Text.find_first_of(
117  Blanks, std::max<unsigned>(MaxSplitBytes, FirstNonWhitespace));
118  }
119  if (SpaceOffset != StringRef::npos && SpaceOffset != 0) {
120  // adaptStartOfLine will break after lines starting with /** if the comment
121  // is broken anywhere. Avoid emitting this break twice here.
122  // Example: in /** longtextcomesherethatbreaks */ (with ColumnLimit 20) will
123  // insert a break after /**, so this code must not insert the same break.
124  if (SpaceOffset == 1 && Text[SpaceOffset - 1] == '*')
125  return BreakableToken::Split(StringRef::npos, 0);
126  StringRef BeforeCut = Text.substr(0, SpaceOffset).rtrim(Blanks);
127  StringRef AfterCut = Text.substr(SpaceOffset);
128  // Don't trim the leading blanks if it would create a */ after the break.
129  if (!DecorationEndsWithStar || AfterCut.size() <= 1 || AfterCut[1] != '/')
130  AfterCut = AfterCut.ltrim(Blanks);
131  return BreakableToken::Split(BeforeCut.size(),
132  AfterCut.begin() - BeforeCut.end());
133  }
134  return BreakableToken::Split(StringRef::npos, 0);
135 }
136 
137 static BreakableToken::Split
138 getStringSplit(StringRef Text, unsigned UsedColumns, unsigned ColumnLimit,
139  unsigned TabWidth, encoding::Encoding Encoding) {
140  // FIXME: Reduce unit test case.
141  if (Text.empty())
142  return BreakableToken::Split(StringRef::npos, 0);
143  if (ColumnLimit <= UsedColumns)
144  return BreakableToken::Split(StringRef::npos, 0);
145  unsigned MaxSplit = ColumnLimit - UsedColumns;
146  StringRef::size_type SpaceOffset = 0;
147  StringRef::size_type SlashOffset = 0;
148  StringRef::size_type WordStartOffset = 0;
149  StringRef::size_type SplitPoint = 0;
150  for (unsigned Chars = 0;;) {
151  unsigned Advance;
152  if (Text[0] == '\\') {
153  Advance = encoding::getEscapeSequenceLength(Text);
154  Chars += Advance;
155  } else {
156  Advance = encoding::getCodePointNumBytes(Text[0], Encoding);
158  Text.substr(0, Advance), UsedColumns + Chars, TabWidth, Encoding);
159  }
160 
161  if (Chars > MaxSplit || Text.size() <= Advance)
162  break;
163 
164  if (IsBlank(Text[0]))
165  SpaceOffset = SplitPoint;
166  if (Text[0] == '/')
167  SlashOffset = SplitPoint;
168  if (Advance == 1 && !isAlphanumeric(Text[0]))
169  WordStartOffset = SplitPoint;
170 
171  SplitPoint += Advance;
172  Text = Text.substr(Advance);
173  }
174 
175  if (SpaceOffset != 0)
176  return BreakableToken::Split(SpaceOffset + 1, 0);
177  if (SlashOffset != 0)
178  return BreakableToken::Split(SlashOffset + 1, 0);
179  if (WordStartOffset != 0)
180  return BreakableToken::Split(WordStartOffset + 1, 0);
181  if (SplitPoint != 0)
182  return BreakableToken::Split(SplitPoint, 0);
183  return BreakableToken::Split(StringRef::npos, 0);
184 }
185 
187  assert((Token.is(TT_BlockComment) || Token.is(TT_LineComment)) &&
188  "formatting regions are switched by comment tokens");
189  StringRef Content = Token.TokenText.substr(2).ltrim();
190  return Content.startswith("clang-format on") ||
191  Content.startswith("clang-format off");
192 }
193 
194 unsigned
195 BreakableToken::getLengthAfterCompression(unsigned RemainingTokenColumns,
196  Split Split) const {
197  // Example: consider the content
198  // lala lala
199  // - RemainingTokenColumns is the original number of columns, 10;
200  // - Split is (4, 2), denoting the two spaces between the two words;
201  //
202  // We compute the number of columns when the split is compressed into a single
203  // space, like:
204  // lala lala
205  //
206  // FIXME: Correctly measure the length of whitespace in Split.second so it
207  // works with tabs.
208  return RemainingTokenColumns + 1 - Split.second;
209 }
210 
211 unsigned BreakableStringLiteral::getLineCount() const { return 1; }
212 
213 unsigned BreakableStringLiteral::getRangeLength(unsigned LineIndex,
214  unsigned Offset,
215  StringRef::size_type Length,
216  unsigned StartColumn) const {
217  llvm_unreachable("Getting the length of a part of the string literal "
218  "indicates that the code tries to reflow it.");
219 }
220 
221 unsigned
223  unsigned StartColumn) const {
224  return UnbreakableTailLength + Postfix.size() +
225  encoding::columnWidthWithTabs(Line.substr(Offset, StringRef::npos),
226  StartColumn, Style.TabWidth, Encoding);
227 }
228 
230  bool Break) const {
231  return StartColumn + Prefix.size();
232 }
233 
235  const FormatToken &Tok, unsigned StartColumn, StringRef Prefix,
236  StringRef Postfix, unsigned UnbreakableTailLength, bool InPPDirective,
238  : BreakableToken(Tok, InPPDirective, Encoding, Style),
239  StartColumn(StartColumn), Prefix(Prefix), Postfix(Postfix),
240  UnbreakableTailLength(UnbreakableTailLength) {
241  assert(Tok.TokenText.startswith(Prefix) && Tok.TokenText.endswith(Postfix));
242  Line = Tok.TokenText.substr(
243  Prefix.size(), Tok.TokenText.size() - Prefix.size() - Postfix.size());
244 }
245 
246 BreakableToken::Split BreakableStringLiteral::getSplit(
247  unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit,
248  unsigned ContentStartColumn, llvm::Regex &CommentPragmasRegex) const {
249  return getStringSplit(Line.substr(TailOffset), ContentStartColumn,
250  ColumnLimit - Postfix.size(), Style.TabWidth, Encoding);
251 }
252 
253 void BreakableStringLiteral::insertBreak(unsigned LineIndex,
254  unsigned TailOffset, Split Split,
255  unsigned ContentIndent,
256  WhitespaceManager &Whitespaces) const {
257  Whitespaces.replaceWhitespaceInToken(
258  Tok, Prefix.size() + TailOffset + Split.first, Split.second, Postfix,
260 }
261 
263  unsigned StartColumn, bool InPPDirective,
265  const FormatStyle &Style)
266  : BreakableToken(Token, InPPDirective, Encoding, Style),
267  StartColumn(StartColumn) {}
268 
269 unsigned BreakableComment::getLineCount() const { return Lines.size(); }
270 
271 BreakableToken::Split
272 BreakableComment::getSplit(unsigned LineIndex, unsigned TailOffset,
273  unsigned ColumnLimit, unsigned ContentStartColumn,
274  llvm::Regex &CommentPragmasRegex) const {
275  // Don't break lines matching the comment pragmas regex.
276  if (CommentPragmasRegex.match(Content[LineIndex]))
277  return Split(StringRef::npos, 0);
278  return getCommentSplit(Content[LineIndex].substr(TailOffset),
279  ContentStartColumn, ColumnLimit, Style.TabWidth,
280  Encoding, Style);
281 }
282 
284  unsigned LineIndex, unsigned TailOffset, Split Split,
285  WhitespaceManager &Whitespaces) const {
286  StringRef Text = Content[LineIndex].substr(TailOffset);
287  // Text is relative to the content line, but Whitespaces operates relative to
288  // the start of the corresponding token, so compute the start of the Split
289  // that needs to be compressed into a single space relative to the start of
290  // its token.
291  unsigned BreakOffsetInToken =
292  Text.data() - tokenAt(LineIndex).TokenText.data() + Split.first;
293  unsigned CharsToRemove = Split.second;
294  Whitespaces.replaceWhitespaceInToken(
295  tokenAt(LineIndex), BreakOffsetInToken, CharsToRemove, "", "",
296  /*InPPDirective=*/false, /*Newlines=*/0, /*Spaces=*/1);
297 }
298 
299 const FormatToken &BreakableComment::tokenAt(unsigned LineIndex) const {
300  return Tokens[LineIndex] ? *Tokens[LineIndex] : Tok;
301 }
302 
303 static bool mayReflowContent(StringRef Content) {
304  Content = Content.trim(Blanks);
305  // Lines starting with '@' commonly have special meaning.
306  // Lines starting with '-', '-#', '+' or '*' are bulleted/numbered lists.
307  bool hasSpecialMeaningPrefix = false;
308  for (StringRef Prefix :
309  {"@", "TODO", "FIXME", "XXX", "-# ", "- ", "+ ", "* "}) {
310  if (Content.startswith(Prefix)) {
311  hasSpecialMeaningPrefix = true;
312  break;
313  }
314  }
315 
316  // Numbered lists may also start with a number followed by '.'
317  // To avoid issues if a line starts with a number which is actually the end
318  // of a previous line, we only consider numbers with up to 2 digits.
319  static auto *const kNumberedListRegexp = new llvm::Regex("^[1-9][0-9]?\\. ");
320  hasSpecialMeaningPrefix =
321  hasSpecialMeaningPrefix || kNumberedListRegexp->match(Content);
322 
323  // Simple heuristic for what to reflow: content should contain at least two
324  // characters and either the first or second character must be
325  // non-punctuation.
326  return Content.size() >= 2 && !hasSpecialMeaningPrefix &&
327  !Content.endswith("\\") &&
328  // Note that this is UTF-8 safe, since if isPunctuation(Content[0]) is
329  // true, then the first code point must be 1 byte long.
330  (!isPunctuation(Content[0]) || !isPunctuation(Content[1]));
331 }
332 
334  const FormatToken &Token, unsigned StartColumn,
335  unsigned OriginalStartColumn, bool FirstInLine, bool InPPDirective,
336  encoding::Encoding Encoding, const FormatStyle &Style, bool UseCRLF)
337  : BreakableComment(Token, StartColumn, InPPDirective, Encoding, Style),
338  DelimitersOnNewline(false),
339  UnbreakableTailLength(Token.UnbreakableTailLength) {
340  assert(Tok.is(TT_BlockComment) &&
341  "block comment section must start with a block comment");
342 
343  StringRef TokenText(Tok.TokenText);
344  assert(TokenText.startswith("/*") && TokenText.endswith("*/"));
345  TokenText.substr(2, TokenText.size() - 4)
346  .split(Lines, UseCRLF ? "\r\n" : "\n");
347 
348  int IndentDelta = StartColumn - OriginalStartColumn;
349  Content.resize(Lines.size());
350  Content[0] = Lines[0];
351  ContentColumn.resize(Lines.size());
352  // Account for the initial '/*'.
353  ContentColumn[0] = StartColumn + 2;
354  Tokens.resize(Lines.size());
355  for (size_t i = 1; i < Lines.size(); ++i)
356  adjustWhitespace(i, IndentDelta);
357 
358  // Align decorations with the column of the star on the first line,
359  // that is one column after the start "/*".
360  DecorationColumn = StartColumn + 1;
361 
362  // Account for comment decoration patterns like this:
363  //
364  // /*
365  // ** blah blah blah
366  // */
367  if (Lines.size() >= 2 && Content[1].startswith("**") &&
368  static_cast<unsigned>(ContentColumn[1]) == StartColumn) {
369  DecorationColumn = StartColumn;
370  }
371 
372  Decoration = "* ";
373  if (Lines.size() == 1 && !FirstInLine) {
374  // Comments for which FirstInLine is false can start on arbitrary column,
375  // and available horizontal space can be too small to align consecutive
376  // lines with the first one.
377  // FIXME: We could, probably, align them to current indentation level, but
378  // now we just wrap them without stars.
379  Decoration = "";
380  }
381  for (size_t i = 1, e = Lines.size(); i < e && !Decoration.empty(); ++i) {
382  // If the last line is empty, the closing "*/" will have a star.
383  if (i + 1 == e && Content[i].empty())
384  break;
385  if (!Content[i].empty() && i + 1 != e && Decoration.startswith(Content[i]))
386  continue;
387  while (!Content[i].startswith(Decoration))
388  Decoration = Decoration.substr(0, Decoration.size() - 1);
389  }
390 
391  LastLineNeedsDecoration = true;
392  IndentAtLineBreak = ContentColumn[0] + 1;
393  for (size_t i = 1, e = Lines.size(); i < e; ++i) {
394  if (Content[i].empty()) {
395  if (i + 1 == e) {
396  // Empty last line means that we already have a star as a part of the
397  // trailing */. We also need to preserve whitespace, so that */ is
398  // correctly indented.
399  LastLineNeedsDecoration = false;
400  // Align the star in the last '*/' with the stars on the previous lines.
401  if (e >= 2 && !Decoration.empty()) {
402  ContentColumn[i] = DecorationColumn;
403  }
404  } else if (Decoration.empty()) {
405  // For all other lines, set the start column to 0 if they're empty, so
406  // we do not insert trailing whitespace anywhere.
407  ContentColumn[i] = 0;
408  }
409  continue;
410  }
411 
412  // The first line already excludes the star.
413  // The last line excludes the star if LastLineNeedsDecoration is false.
414  // For all other lines, adjust the line to exclude the star and
415  // (optionally) the first whitespace.
416  unsigned DecorationSize = Decoration.startswith(Content[i])
417  ? Content[i].size()
418  : Decoration.size();
419  if (DecorationSize) {
420  ContentColumn[i] = DecorationColumn + DecorationSize;
421  }
422  Content[i] = Content[i].substr(DecorationSize);
423  if (!Decoration.startswith(Content[i]))
424  IndentAtLineBreak =
425  std::min<int>(IndentAtLineBreak, std::max(0, ContentColumn[i]));
426  }
427  IndentAtLineBreak = std::max<unsigned>(IndentAtLineBreak, Decoration.size());
428 
429  // Detect a multiline jsdoc comment and set DelimitersOnNewline in that case.
430  if (Style.Language == FormatStyle::LK_JavaScript ||
431  Style.Language == FormatStyle::LK_Java) {
432  if ((Lines[0] == "*" || Lines[0].startswith("* ")) && Lines.size() > 1) {
433  // This is a multiline jsdoc comment.
434  DelimitersOnNewline = true;
435  } else if (Lines[0].startswith("* ") && Lines.size() == 1) {
436  // Detect a long single-line comment, like:
437  // /** long long long */
438  // Below, '2' is the width of '*/'.
439  unsigned EndColumn =
440  ContentColumn[0] +
442  Style.TabWidth, Encoding) +
443  2;
444  DelimitersOnNewline = EndColumn > Style.ColumnLimit;
445  }
446  }
447 
448  LLVM_DEBUG({
449  llvm::dbgs() << "IndentAtLineBreak " << IndentAtLineBreak << "\n";
450  llvm::dbgs() << "DelimitersOnNewline " << DelimitersOnNewline << "\n";
451  for (size_t i = 0; i < Lines.size(); ++i) {
452  llvm::dbgs() << i << " |" << Content[i] << "| "
453  << "CC=" << ContentColumn[i] << "| "
454  << "IN=" << (Content[i].data() - Lines[i].data()) << "\n";
455  }
456  });
457 }
458 
459 BreakableToken::Split BreakableBlockComment::getSplit(
460  unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit,
461  unsigned ContentStartColumn, llvm::Regex &CommentPragmasRegex) const {
462  // Don't break lines matching the comment pragmas regex.
463  if (CommentPragmasRegex.match(Content[LineIndex]))
464  return Split(StringRef::npos, 0);
465  return getCommentSplit(Content[LineIndex].substr(TailOffset),
466  ContentStartColumn, ColumnLimit, Style.TabWidth,
467  Encoding, Style, Decoration.endswith("*"));
468 }
469 
470 void BreakableBlockComment::adjustWhitespace(unsigned LineIndex,
471  int IndentDelta) {
472  // When in a preprocessor directive, the trailing backslash in a block comment
473  // is not needed, but can serve a purpose of uniformity with necessary escaped
474  // newlines outside the comment. In this case we remove it here before
475  // trimming the trailing whitespace. The backslash will be re-added later when
476  // inserting a line break.
477  size_t EndOfPreviousLine = Lines[LineIndex - 1].size();
478  if (InPPDirective && Lines[LineIndex - 1].endswith("\\"))
479  --EndOfPreviousLine;
480 
481  // Calculate the end of the non-whitespace text in the previous line.
482  EndOfPreviousLine =
483  Lines[LineIndex - 1].find_last_not_of(Blanks, EndOfPreviousLine);
484  if (EndOfPreviousLine == StringRef::npos)
485  EndOfPreviousLine = 0;
486  else
487  ++EndOfPreviousLine;
488  // Calculate the start of the non-whitespace text in the current line.
489  size_t StartOfLine = Lines[LineIndex].find_first_not_of(Blanks);
490  if (StartOfLine == StringRef::npos)
491  StartOfLine = Lines[LineIndex].size();
492 
493  StringRef Whitespace = Lines[LineIndex].substr(0, StartOfLine);
494  // Adjust Lines to only contain relevant text.
495  size_t PreviousContentOffset =
496  Content[LineIndex - 1].data() - Lines[LineIndex - 1].data();
497  Content[LineIndex - 1] = Lines[LineIndex - 1].substr(
498  PreviousContentOffset, EndOfPreviousLine - PreviousContentOffset);
499  Content[LineIndex] = Lines[LineIndex].substr(StartOfLine);
500 
501  // Adjust the start column uniformly across all lines.
502  ContentColumn[LineIndex] =
503  encoding::columnWidthWithTabs(Whitespace, 0, Style.TabWidth, Encoding) +
504  IndentDelta;
505 }
506 
507 unsigned BreakableBlockComment::getRangeLength(unsigned LineIndex,
508  unsigned Offset,
509  StringRef::size_type Length,
510  unsigned StartColumn) const {
511  unsigned LineLength =
512  encoding::columnWidthWithTabs(Content[LineIndex].substr(Offset, Length),
513  StartColumn, Style.TabWidth, Encoding);
514  // FIXME: This should go into getRemainingLength instead, but we currently
515  // break tests when putting it there. Investigate how to fix those tests.
516  // The last line gets a "*/" postfix.
517  if (LineIndex + 1 == Lines.size()) {
518  LineLength += 2;
519  // We never need a decoration when breaking just the trailing "*/" postfix.
520  // Note that checking that Length == 0 is not enough, since Length could
521  // also be StringRef::npos.
522  if (Content[LineIndex].substr(Offset, StringRef::npos).empty()) {
523  LineLength -= Decoration.size();
524  }
525  }
526  return LineLength;
527 }
528 
529 unsigned BreakableBlockComment::getRemainingLength(unsigned LineIndex,
530  unsigned Offset,
531  unsigned StartColumn) const {
532  return UnbreakableTailLength +
533  getRangeLength(LineIndex, Offset, StringRef::npos, StartColumn);
534 }
535 
537  bool Break) const {
538  if (Break)
539  return IndentAtLineBreak;
540  return std::max(0, ContentColumn[LineIndex]);
541 }
542 
543 const llvm::StringSet<>
545  "@param", "@return", "@returns", "@throws", "@type", "@template",
546  "@see", "@deprecated", "@define", "@exports", "@mods", "@private",
547 };
548 
549 unsigned BreakableBlockComment::getContentIndent(unsigned LineIndex) const {
550  if (Style.Language != FormatStyle::LK_Java &&
551  Style.Language != FormatStyle::LK_JavaScript)
552  return 0;
553  // The content at LineIndex 0 of a comment like:
554  // /** line 0 */
555  // is "* line 0", so we need to skip over the decoration in that case.
556  StringRef ContentWithNoDecoration = Content[LineIndex];
557  if (LineIndex == 0 && ContentWithNoDecoration.startswith("*")) {
558  ContentWithNoDecoration = ContentWithNoDecoration.substr(1).ltrim(Blanks);
559  }
560  StringRef FirstWord = ContentWithNoDecoration.substr(
561  0, ContentWithNoDecoration.find_first_of(Blanks));
562  if (ContentIndentingJavadocAnnotations.find(FirstWord) !=
564  return Style.ContinuationIndentWidth;
565  return 0;
566 }
567 
568 void BreakableBlockComment::insertBreak(unsigned LineIndex, unsigned TailOffset,
569  Split Split, unsigned ContentIndent,
570  WhitespaceManager &Whitespaces) const {
571  StringRef Text = Content[LineIndex].substr(TailOffset);
572  StringRef Prefix = Decoration;
573  // We need this to account for the case when we have a decoration "* " for all
574  // the lines except for the last one, where the star in "*/" acts as a
575  // decoration.
576  unsigned LocalIndentAtLineBreak = IndentAtLineBreak;
577  if (LineIndex + 1 == Lines.size() &&
578  Text.size() == Split.first + Split.second) {
579  // For the last line we need to break before "*/", but not to add "* ".
580  Prefix = "";
581  if (LocalIndentAtLineBreak >= 2)
582  LocalIndentAtLineBreak -= 2;
583  }
584  // The split offset is from the beginning of the line. Convert it to an offset
585  // from the beginning of the token text.
586  unsigned BreakOffsetInToken =
587  Text.data() - tokenAt(LineIndex).TokenText.data() + Split.first;
588  unsigned CharsToRemove = Split.second;
589  assert(LocalIndentAtLineBreak >= Prefix.size());
590  std::string PrefixWithTrailingIndent = Prefix;
591  for (unsigned I = 0; I < ContentIndent; ++I)
592  PrefixWithTrailingIndent += " ";
593  Whitespaces.replaceWhitespaceInToken(
594  tokenAt(LineIndex), BreakOffsetInToken, CharsToRemove, "",
595  PrefixWithTrailingIndent, InPPDirective, /*Newlines=*/1,
596  /*Spaces=*/LocalIndentAtLineBreak + ContentIndent -
597  PrefixWithTrailingIndent.size());
598 }
599 
600 BreakableToken::Split
602  llvm::Regex &CommentPragmasRegex) const {
603  if (!mayReflow(LineIndex, CommentPragmasRegex))
604  return Split(StringRef::npos, 0);
605 
606  // If we're reflowing into a line with content indent, only reflow the next
607  // line if its starting whitespace matches the content indent.
608  size_t Trimmed = Content[LineIndex].find_first_not_of(Blanks);
609  if (LineIndex) {
610  unsigned PreviousContentIndent = getContentIndent(LineIndex - 1);
611  if (PreviousContentIndent && Trimmed != StringRef::npos &&
612  Trimmed != PreviousContentIndent)
613  return Split(StringRef::npos, 0);
614  }
615 
616  return Split(0, Trimmed != StringRef::npos ? Trimmed : 0);
617 }
618 
620  // A break is introduced when we want delimiters on newline.
621  return DelimitersOnNewline &&
622  Lines[0].substr(1).find_first_not_of(Blanks) != StringRef::npos;
623 }
624 
625 void BreakableBlockComment::reflow(unsigned LineIndex,
626  WhitespaceManager &Whitespaces) const {
627  StringRef TrimmedContent = Content[LineIndex].ltrim(Blanks);
628  // Here we need to reflow.
629  assert(Tokens[LineIndex - 1] == Tokens[LineIndex] &&
630  "Reflowing whitespace within a token");
631  // This is the offset of the end of the last line relative to the start of
632  // the token text in the token.
633  unsigned WhitespaceOffsetInToken = Content[LineIndex - 1].data() +
634  Content[LineIndex - 1].size() -
635  tokenAt(LineIndex).TokenText.data();
636  unsigned WhitespaceLength = TrimmedContent.data() -
637  tokenAt(LineIndex).TokenText.data() -
638  WhitespaceOffsetInToken;
639  Whitespaces.replaceWhitespaceInToken(
640  tokenAt(LineIndex), WhitespaceOffsetInToken,
641  /*ReplaceChars=*/WhitespaceLength, /*PreviousPostfix=*/"",
642  /*CurrentPrefix=*/ReflowPrefix, InPPDirective, /*Newlines=*/0,
643  /*Spaces=*/0);
644 }
645 
647  unsigned LineIndex, WhitespaceManager &Whitespaces) const {
648  if (LineIndex == 0) {
649  if (DelimitersOnNewline) {
650  // Since we're breaking at index 1 below, the break position and the
651  // break length are the same.
652  // Note: this works because getCommentSplit is careful never to split at
653  // the beginning of a line.
654  size_t BreakLength = Lines[0].substr(1).find_first_not_of(Blanks);
655  if (BreakLength != StringRef::npos)
656  insertBreak(LineIndex, 0, Split(1, BreakLength), /*ContentIndent=*/0,
657  Whitespaces);
658  }
659  return;
660  }
661  // Here no reflow with the previous line will happen.
662  // Fix the decoration of the line at LineIndex.
663  StringRef Prefix = Decoration;
664  if (Content[LineIndex].empty()) {
665  if (LineIndex + 1 == Lines.size()) {
666  if (!LastLineNeedsDecoration) {
667  // If the last line was empty, we don't need a prefix, as the */ will
668  // line up with the decoration (if it exists).
669  Prefix = "";
670  }
671  } else if (!Decoration.empty()) {
672  // For other empty lines, if we do have a decoration, adapt it to not
673  // contain a trailing whitespace.
674  Prefix = Prefix.substr(0, 1);
675  }
676  } else {
677  if (ContentColumn[LineIndex] == 1) {
678  // This line starts immediately after the decorating *.
679  Prefix = Prefix.substr(0, 1);
680  }
681  }
682  // This is the offset of the end of the last line relative to the start of the
683  // token text in the token.
684  unsigned WhitespaceOffsetInToken = Content[LineIndex - 1].data() +
685  Content[LineIndex - 1].size() -
686  tokenAt(LineIndex).TokenText.data();
687  unsigned WhitespaceLength = Content[LineIndex].data() -
688  tokenAt(LineIndex).TokenText.data() -
689  WhitespaceOffsetInToken;
690  Whitespaces.replaceWhitespaceInToken(
691  tokenAt(LineIndex), WhitespaceOffsetInToken, WhitespaceLength, "", Prefix,
692  InPPDirective, /*Newlines=*/1, ContentColumn[LineIndex] - Prefix.size());
693 }
694 
695 BreakableToken::Split
697  if (DelimitersOnNewline) {
698  // Replace the trailing whitespace of the last line with a newline.
699  // In case the last line is empty, the ending '*/' is already on its own
700  // line.
701  StringRef Line = Content.back().substr(TailOffset);
702  StringRef TrimmedLine = Line.rtrim(Blanks);
703  if (!TrimmedLine.empty())
704  return Split(TrimmedLine.size(), Line.size() - TrimmedLine.size());
705  }
706  return Split(StringRef::npos, 0);
707 }
708 
709 bool BreakableBlockComment::mayReflow(unsigned LineIndex,
710  llvm::Regex &CommentPragmasRegex) const {
711  // Content[LineIndex] may exclude the indent after the '*' decoration. In that
712  // case, we compute the start of the comment pragma manually.
713  StringRef IndentContent = Content[LineIndex];
714  if (Lines[LineIndex].ltrim(Blanks).startswith("*")) {
715  IndentContent = Lines[LineIndex].ltrim(Blanks).substr(1);
716  }
717  return LineIndex > 0 && !CommentPragmasRegex.match(IndentContent) &&
718  mayReflowContent(Content[LineIndex]) && !Tok.Finalized &&
719  !switchesFormatting(tokenAt(LineIndex));
720 }
721 
723  const FormatToken &Token, unsigned StartColumn,
724  unsigned OriginalStartColumn, bool FirstInLine, bool InPPDirective,
726  : BreakableComment(Token, StartColumn, InPPDirective, Encoding, Style) {
727  assert(Tok.is(TT_LineComment) &&
728  "line comment section must start with a line comment");
729  FormatToken *LineTok = nullptr;
730  for (const FormatToken *CurrentTok = &Tok;
731  CurrentTok && CurrentTok->is(TT_LineComment);
732  CurrentTok = CurrentTok->Next) {
733  LastLineTok = LineTok;
734  StringRef TokenText(CurrentTok->TokenText);
735  assert((TokenText.startswith("//") || TokenText.startswith("#")) &&
736  "unsupported line comment prefix, '//' and '#' are supported");
737  size_t FirstLineIndex = Lines.size();
738  TokenText.split(Lines, "\n");
739  Content.resize(Lines.size());
740  ContentColumn.resize(Lines.size());
741  OriginalContentColumn.resize(Lines.size());
742  Tokens.resize(Lines.size());
743  Prefix.resize(Lines.size());
744  OriginalPrefix.resize(Lines.size());
745  for (size_t i = FirstLineIndex, e = Lines.size(); i < e; ++i) {
746  Lines[i] = Lines[i].ltrim(Blanks);
747  // We need to trim the blanks in case this is not the first line in a
748  // multiline comment. Then the indent is included in Lines[i].
749  StringRef IndentPrefix =
750  getLineCommentIndentPrefix(Lines[i].ltrim(Blanks), Style);
751  assert((TokenText.startswith("//") || TokenText.startswith("#")) &&
752  "unsupported line comment prefix, '//' and '#' are supported");
753  OriginalPrefix[i] = Prefix[i] = IndentPrefix;
754  if (Lines[i].size() > Prefix[i].size() &&
755  isAlphanumeric(Lines[i][Prefix[i].size()])) {
756  if (Prefix[i] == "//")
757  Prefix[i] = "// ";
758  else if (Prefix[i] == "///")
759  Prefix[i] = "/// ";
760  else if (Prefix[i] == "//!")
761  Prefix[i] = "//! ";
762  else if (Prefix[i] == "///<")
763  Prefix[i] = "///< ";
764  else if (Prefix[i] == "//!<")
765  Prefix[i] = "//!< ";
766  else if (Prefix[i] == "#" &&
767  Style.Language == FormatStyle::LK_TextProto)
768  Prefix[i] = "# ";
769  }
770 
771  Tokens[i] = LineTok;
772  Content[i] = Lines[i].substr(IndentPrefix.size());
773  OriginalContentColumn[i] =
774  StartColumn + encoding::columnWidthWithTabs(OriginalPrefix[i],
775  StartColumn,
776  Style.TabWidth, Encoding);
777  ContentColumn[i] =
778  StartColumn + encoding::columnWidthWithTabs(Prefix[i], StartColumn,
779  Style.TabWidth, Encoding);
780 
781  // Calculate the end of the non-whitespace text in this line.
782  size_t EndOfLine = Content[i].find_last_not_of(Blanks);
783  if (EndOfLine == StringRef::npos)
784  EndOfLine = Content[i].size();
785  else
786  ++EndOfLine;
787  Content[i] = Content[i].substr(0, EndOfLine);
788  }
789  LineTok = CurrentTok->Next;
790  if (CurrentTok->Next && !CurrentTok->Next->ContinuesLineCommentSection) {
791  // A line comment section needs to broken by a line comment that is
792  // preceded by at least two newlines. Note that we put this break here
793  // instead of breaking at a previous stage during parsing, since that
794  // would split the contents of the enum into two unwrapped lines in this
795  // example, which is undesirable:
796  // enum A {
797  // a, // comment about a
798  //
799  // // comment about b
800  // b
801  // };
802  //
803  // FIXME: Consider putting separate line comment sections as children to
804  // the unwrapped line instead.
805  break;
806  }
807  }
808 }
809 
810 unsigned
812  StringRef::size_type Length,
813  unsigned StartColumn) const {
815  Content[LineIndex].substr(Offset, Length), StartColumn, Style.TabWidth,
816  Encoding);
817 }
818 
820  bool Break) const {
821  if (Break)
822  return OriginalContentColumn[LineIndex];
823  return ContentColumn[LineIndex];
824 }
825 
827  unsigned LineIndex, unsigned TailOffset, Split Split,
828  unsigned ContentIndent, WhitespaceManager &Whitespaces) const {
829  StringRef Text = Content[LineIndex].substr(TailOffset);
830  // Compute the offset of the split relative to the beginning of the token
831  // text.
832  unsigned BreakOffsetInToken =
833  Text.data() - tokenAt(LineIndex).TokenText.data() + Split.first;
834  unsigned CharsToRemove = Split.second;
835  // Compute the size of the new indent, including the size of the new prefix of
836  // the newly broken line.
837  unsigned IndentAtLineBreak = OriginalContentColumn[LineIndex] +
838  Prefix[LineIndex].size() -
839  OriginalPrefix[LineIndex].size();
840  assert(IndentAtLineBreak >= Prefix[LineIndex].size());
841  Whitespaces.replaceWhitespaceInToken(
842  tokenAt(LineIndex), BreakOffsetInToken, CharsToRemove, "",
843  Prefix[LineIndex], InPPDirective, /*Newlines=*/1,
844  /*Spaces=*/IndentAtLineBreak - Prefix[LineIndex].size());
845 }
846 
848  unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const {
849  if (!mayReflow(LineIndex, CommentPragmasRegex))
850  return Split(StringRef::npos, 0);
851 
852  size_t Trimmed = Content[LineIndex].find_first_not_of(Blanks);
853 
854  // In a line comment section each line is a separate token; thus, after a
855  // split we replace all whitespace before the current line comment token
856  // (which does not need to be included in the split), plus the start of the
857  // line up to where the content starts.
858  return Split(0, Trimmed != StringRef::npos ? Trimmed : 0);
859 }
860 
861 void BreakableLineCommentSection::reflow(unsigned LineIndex,
862  WhitespaceManager &Whitespaces) const {
863  if (LineIndex > 0 && Tokens[LineIndex] != Tokens[LineIndex - 1]) {
864  // Reflow happens between tokens. Replace the whitespace between the
865  // tokens by the empty string.
866  Whitespaces.replaceWhitespace(
867  *Tokens[LineIndex], /*Newlines=*/0, /*Spaces=*/0,
868  /*StartOfTokenColumn=*/StartColumn, /*InPPDirective=*/false);
869  } else if (LineIndex > 0) {
870  // In case we're reflowing after the '\' in:
871  //
872  // // line comment \
873  // // line 2
874  //
875  // the reflow happens inside the single comment token (it is a single line
876  // comment with an unescaped newline).
877  // Replace the whitespace between the '\' and '//' with the empty string.
878  //
879  // Offset points to after the '\' relative to start of the token.
880  unsigned Offset = Lines[LineIndex - 1].data() +
881  Lines[LineIndex - 1].size() -
882  tokenAt(LineIndex - 1).TokenText.data();
883  // WhitespaceLength is the number of chars between the '\' and the '//' on
884  // the next line.
885  unsigned WhitespaceLength =
886  Lines[LineIndex].data() - tokenAt(LineIndex).TokenText.data() - Offset;
887  Whitespaces.replaceWhitespaceInToken(*Tokens[LineIndex], Offset,
888  /*ReplaceChars=*/WhitespaceLength,
889  /*PreviousPostfix=*/"",
890  /*CurrentPrefix=*/"",
891  /*InPPDirective=*/false,
892  /*Newlines=*/0,
893  /*Spaces=*/0);
894  }
895  // Replace the indent and prefix of the token with the reflow prefix.
896  unsigned Offset =
897  Lines[LineIndex].data() - tokenAt(LineIndex).TokenText.data();
898  unsigned WhitespaceLength =
899  Content[LineIndex].data() - Lines[LineIndex].data();
900  Whitespaces.replaceWhitespaceInToken(*Tokens[LineIndex], Offset,
901  /*ReplaceChars=*/WhitespaceLength,
902  /*PreviousPostfix=*/"",
903  /*CurrentPrefix=*/ReflowPrefix,
904  /*InPPDirective=*/false,
905  /*Newlines=*/0,
906  /*Spaces=*/0);
907 }
908 
910  unsigned LineIndex, WhitespaceManager &Whitespaces) const {
911  // If this is the first line of a token, we need to inform Whitespace Manager
912  // about it: either adapt the whitespace range preceding it, or mark it as an
913  // untouchable token.
914  // This happens for instance here:
915  // // line 1 \
916  // // line 2
917  if (LineIndex > 0 && Tokens[LineIndex] != Tokens[LineIndex - 1]) {
918  // This is the first line for the current token, but no reflow with the
919  // previous token is necessary. However, we still may need to adjust the
920  // start column. Note that ContentColumn[LineIndex] is the expected
921  // content column after a possible update to the prefix, hence the prefix
922  // length change is included.
923  unsigned LineColumn =
924  ContentColumn[LineIndex] -
925  (Content[LineIndex].data() - Lines[LineIndex].data()) +
926  (OriginalPrefix[LineIndex].size() - Prefix[LineIndex].size());
927 
928  // We always want to create a replacement instead of adding an untouchable
929  // token, even if LineColumn is the same as the original column of the
930  // token. This is because WhitespaceManager doesn't align trailing
931  // comments if they are untouchable.
932  Whitespaces.replaceWhitespace(*Tokens[LineIndex],
933  /*Newlines=*/1,
934  /*Spaces=*/LineColumn,
935  /*StartOfTokenColumn=*/LineColumn,
936  /*InPPDirective=*/false);
937  }
938  if (OriginalPrefix[LineIndex] != Prefix[LineIndex]) {
939  // Adjust the prefix if necessary.
940 
941  // Take care of the space possibly introduced after a decoration.
942  assert(Prefix[LineIndex] == (OriginalPrefix[LineIndex] + " ").str() &&
943  "Expecting a line comment prefix to differ from original by at most "
944  "a space");
945  Whitespaces.replaceWhitespaceInToken(
946  tokenAt(LineIndex), OriginalPrefix[LineIndex].size(), 0, "", "",
947  /*InPPDirective=*/false, /*Newlines=*/0, /*Spaces=*/1);
948  }
949 }
950 
952  if (LastLineTok) {
953  State.NextToken = LastLineTok->Next;
954  }
955 }
956 
958  unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const {
959  // Line comments have the indent as part of the prefix, so we need to
960  // recompute the start of the line.
961  StringRef IndentContent = Content[LineIndex];
962  if (Lines[LineIndex].startswith("//")) {
963  IndentContent = Lines[LineIndex].substr(2);
964  }
965  // FIXME: Decide whether we want to reflow non-regular indents:
966  // Currently, we only reflow when the OriginalPrefix[LineIndex] matches the
967  // OriginalPrefix[LineIndex-1]. That means we don't reflow
968  // // text that protrudes
969  // // into text with different indent
970  // We do reflow in that case in block comments.
971  return LineIndex > 0 && !CommentPragmasRegex.match(IndentContent) &&
972  mayReflowContent(Content[LineIndex]) && !Tok.Finalized &&
973  !switchesFormatting(tokenAt(LineIndex)) &&
974  OriginalPrefix[LineIndex] == OriginalPrefix[LineIndex - 1];
975 }
976 
977 } // namespace format
978 } // namespace clang
unsigned getLineCount() const override
bool switchesFormatting(const FormatToken &Token)
Checks if Token switches formatting, like /* clang-format off.
Declares BreakableToken, BreakableStringLiteral, BreakableComment, BreakableBlockComment and Breakabl...
Split getSplitAfterLastLine(unsigned TailOffset) const override
static const char *const Blanks
static const llvm::StringSet ContentIndentingJavadocAnnotations
void adaptStartOfLine(unsigned LineIndex, WhitespaceManager &Whitespaces) const override
bool introducesBreakBeforeToken() const override
unsigned getRemainingLength(unsigned LineIndex, unsigned Offset, unsigned StartColumn) const override
void reflow(unsigned LineIndex, WhitespaceManager &Whitespaces) const override
Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit, unsigned ContentStartColumn, llvm::Regex &CommentPragmasRegex) const override
const FormatToken & tokenAt(unsigned LineIndex) const
FormatToken * Next
The next token in the unwrapped line.
Definition: FormatToken.h:298
static StringRef getLineCommentIndentPrefix(StringRef Comment, const FormatStyle &Style)
void replaceWhitespaceInToken(const FormatToken &Tok, unsigned Offset, unsigned ReplaceChars, StringRef PreviousPostfix, StringRef CurrentPrefix, bool InPPDirective, unsigned Newlines, int Spaces)
Inserts or replaces whitespace in the middle of a token.
SmallVector< int, 16 > ContentColumn
Split getReflowSplit(unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const override
LineState State
void updateNextToken(LineState &State) const override
unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override
Token - This structure provides full information about a lexed token.
Definition: Token.h:34
unsigned columnWidthWithTabs(StringRef Text, unsigned StartColumn, unsigned TabWidth, Encoding Encoding)
Returns the number of columns required to display the Text, starting from the StartColumn on a termin...
Definition: Encoding.h:61
__DEVICE__ int max(int __a, int __b)
Manages the whitespaces around tokens and their replacements.
void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split, unsigned ContentIndent, WhitespaceManager &Whitespaces) const override
const FormatToken & Tok
bool mayReflow(unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const override
static BreakableToken::Split getCommentSplit(StringRef Text, unsigned ContentStartColumn, unsigned ColumnLimit, unsigned TabWidth, encoding::Encoding Encoding, const FormatStyle &Style, bool DecorationEndsWithStar=false)
Split getReflowSplit(unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const override
The current state when indenting a unwrapped line.
unsigned Offset
Definition: Format.cpp:1809
void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split, unsigned ContentIndent, WhitespaceManager &Whitespaces) const override
SmallVector< StringRef, 16 > Content
const AnnotatedLine * Line
unsigned getEscapeSequenceLength(StringRef Text)
Gets the length of an escape sequence inside a C++ string literal.
Definition: Encoding.h:97
BreakableBlockComment(const FormatToken &Token, unsigned StartColumn, unsigned OriginalStartColumn, bool FirstInLine, bool InPPDirective, encoding::Encoding Encoding, const FormatStyle &Style, bool UseCRLF)
A wrapper around a Token storing information about the whitespace characters preceding it...
Definition: FormatToken.h:129
LLVM_READONLY bool isAlphanumeric(unsigned char c)
Return true if this character is an ASCII letter or digit: [a-zA-Z0-9].
Definition: CharInfo.h:117
unsigned getCodePointNumBytes(char FirstChar, Encoding Encoding)
Gets the number of bytes in a sequence representing a single codepoint and starting with FirstChar in...
Definition: Encoding.h:78
void replaceWhitespace(FormatToken &Tok, unsigned Newlines, unsigned Spaces, unsigned StartOfTokenColumn, bool InPPDirective=false)
Replaces the whitespace in front of Tok.
SmallVector< FormatToken *, 16 > Tokens
#define false
Definition: stdbool.h:17
void adaptStartOfLine(unsigned LineIndex, WhitespaceManager &Whitespaces) const override
bool is(tok::TokenKind Kind) const
Definition: FormatToken.h:312
const bool InPPDirective
Various functions to configurably format source code.
unsigned getRangeLength(unsigned LineIndex, unsigned Offset, StringRef::size_type Length, unsigned StartColumn) const override
void compressWhitespace(unsigned LineIndex, unsigned TailOffset, Split Split, WhitespaceManager &Whitespaces) const override
BreakableComment(const FormatToken &Token, unsigned StartColumn, bool InPPDirective, encoding::Encoding Encoding, const FormatStyle &Style)
Creates a breakable token for a comment.
unsigned getRangeLength(unsigned LineIndex, unsigned Offset, StringRef::size_type Length, unsigned StartColumn) const override
StringRef TokenText
The raw text of the token.
Definition: FormatToken.h:177
Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit, unsigned ContentStartColumn, llvm::Regex &CommentPragmasRegex) const override
BreakableStringLiteral(const FormatToken &Tok, unsigned StartColumn, StringRef Prefix, StringRef Postfix, unsigned UnbreakableTailLength, bool InPPDirective, encoding::Encoding Encoding, const FormatStyle &Style)
Creates a breakable token for a single line string literal.
BreakableLineCommentSection(const FormatToken &Token, unsigned StartColumn, unsigned OriginalStartColumn, bool FirstInLine, bool InPPDirective, encoding::Encoding Encoding, const FormatStyle &Style)
The FormatStyle is used to configure the formatting to follow specific guidelines.
Definition: Format.h:49
static bool mayReflowContent(StringRef Content)
Dataflow Directional Tag Classes.
unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override
bool Finalized
If true, this token has been fully formatted (indented and potentially re-formatted inside)...
Definition: FormatToken.h:310
unsigned getLengthAfterCompression(unsigned RemainingTokenColumns, Split Split) const
Returns the number of columns needed to format RemainingTokenColumns, assuming that Split is within t...
unsigned getRangeLength(unsigned LineIndex, unsigned Offset, StringRef::size_type Length, unsigned StartColumn) const override
void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split, unsigned ContentIndent, WhitespaceManager &Whitespaces) const override
unsigned getRemainingLength(unsigned LineIndex, unsigned Offset, unsigned StartColumn) const override
void reflow(unsigned LineIndex, WhitespaceManager &Whitespaces) const override
bool mayReflow(unsigned LineIndex, llvm::Regex &CommentPragmasRegex) const override
static bool IsBlank(char C)
Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit, unsigned ContentStartColumn, llvm::Regex &CommentPragmasRegex) const override
LLVM_READONLY bool isPunctuation(unsigned char c)
Return true if this character is an ASCII punctuation character.
Definition: CharInfo.h:131
unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override
static BreakableToken::Split getStringSplit(StringRef Text, unsigned UsedColumns, unsigned ColumnLimit, unsigned TabWidth, encoding::Encoding Encoding)
unsigned getContentIndent(unsigned LineIndex) const override
unsigned getLineCount() const override
BreakableToken(const FormatToken &Tok, bool InPPDirective, encoding::Encoding Encoding, const FormatStyle &Style)
StringRef Text
Definition: Format.cpp:1808
FormatToken * NextToken
The token that needs to be next formatted.
This file implements an indenter that manages the indentation of continuations.
SmallVector< StringRef, 16 > Lines
const encoding::Encoding Encoding
const FormatStyle & Style