1//===--- BreakableToken.h - Format C++ code ---------------------*- C++ -*-===//
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/// Declares BreakableToken, BreakableStringLiteral, BreakableComment,
11/// BreakableBlockComment and BreakableLineCommentSection classes, that contain
12/// token type-specific logic to break long lines in tokens and reflow content
13/// between tokens.
14///
15//===----------------------------------------------------------------------===//
16
17#ifndef LLVM_CLANG_LIB_FORMAT_BREAKABLETOKEN_H
18#define LLVM_CLANG_LIB_FORMAT_BREAKABLETOKEN_H
19
20#include "Encoding.h"
21#include "WhitespaceManager.h"
22#include "llvm/ADT/StringSet.h"
23
24namespace clang {
25namespace format {
26
27/// Checks if \p Token switches formatting, like /* clang-format off */.
28/// \p Token must be a comment.
29bool switchesFormatting(const FormatToken &Token);
30
31struct FormatStyle;
32
33/// Base class for tokens / ranges of tokens that can allow breaking
34/// within the tokens - for example, to avoid whitespace beyond the column
35/// limit, or to reflow text.
36///
37/// Generally, a breakable token consists of logical lines, addressed by a line
38/// index. For example, in a sequence of line comments, each line comment is its
39/// own logical line; similarly, for a block comment, each line in the block
40/// comment is on its own logical line.
41///
42/// There are two methods to compute the layout of the token:
43/// - getRangeLength measures the number of columns needed for a range of text
44/// within a logical line, and
45/// - getContentStartColumn returns the start column at which we want the
46/// content of a logical line to start (potentially after introducing a line
47/// break).
48///
49/// The mechanism to adapt the layout of the breakable token is organised
50/// around the concept of a \c Split, which is a whitespace range that signifies
51/// a position of the content of a token where a reformatting might be done.
52///
53/// Operating with splits is divided into two operations:
54/// - getSplit, for finding a split starting at a position,
55/// - insertBreak, for executing the split using a whitespace manager.
56///
57/// There is a pair of operations that are used to compress a long whitespace
58/// range with a single space if that will bring the line length under the
59/// column limit:
60/// - getLineLengthAfterCompression, for calculating the size in columns of the
61/// line after a whitespace range has been compressed, and
62/// - compressWhitespace, for executing the whitespace compression using a
63/// whitespace manager; note that the compressed whitespace may be in the
64/// middle of the original line and of the reformatted line.
65///
66/// For tokens where the whitespace before each line needs to be also
67/// reformatted, for example for tokens supporting reflow, there are analogous
68/// operations that might be executed before the main line breaking occurs:
69/// - getReflowSplit, for finding a split such that the content preceding it
70/// needs to be specially reflown,
71/// - reflow, for executing the split using a whitespace manager,
72/// - introducesBreakBefore, for checking if reformatting the beginning
73/// of the content introduces a line break before it,
74/// - adaptStartOfLine, for executing the reflow using a whitespace
75/// manager.
76///
77/// For tokens that require the whitespace after the last line to be
78/// reformatted, for example in multiline jsdoc comments that require the
79/// trailing '*/' to be on a line of itself, there are analogous operations
80/// that might be executed after the last line has been reformatted:
81/// - getSplitAfterLastLine, for finding a split after the last line that needs
82/// to be reflown,
83/// - replaceWhitespaceAfterLastLine, for executing the reflow using a
84/// whitespace manager.
85///
86class BreakableToken {
87public:
88 /// Contains starting character index and length of split.
89 typedef std::pair<StringRef::size_type, unsigned> Split;
90
91 virtual ~BreakableToken() {}
92
93 /// Returns the number of lines in this token in the original code.
94 virtual unsigned getLineCount() const = 0;
95
96 /// Returns the number of columns required to format the text in the
97 /// byte range [\p Offset, \p Offset \c + \p Length).
98 ///
99 /// \p Offset is the byte offset from the start of the content of the line
100 /// at \p LineIndex.
101 ///
102 /// \p StartColumn is the column at which the text starts in the formatted
103 /// file, needed to compute tab stops correctly.
104 virtual unsigned getRangeLength(unsigned LineIndex, unsigned Offset,
105 StringRef::size_type Length,
106 unsigned StartColumn) const = 0;
107
108 /// Returns the number of columns required to format the text following
109 /// the byte \p Offset in the line \p LineIndex, including potentially
110 /// unbreakable sequences of tokens following after the end of the token.
111 ///
112 /// \p Offset is the byte offset from the start of the content of the line
113 /// at \p LineIndex.
114 ///
115 /// \p StartColumn is the column at which the text starts in the formatted
116 /// file, needed to compute tab stops correctly.
117 ///
118 /// For breakable tokens that never use extra space at the end of a line, this
119 /// is equivalent to getRangeLength with a Length of StringRef::npos.
120 virtual unsigned getRemainingLength(unsigned LineIndex, unsigned Offset,
121 unsigned StartColumn) const {
122 return getRangeLength(LineIndex, Offset, Length: StringRef::npos, StartColumn);
123 }
124
125 /// Returns the column at which content in line \p LineIndex starts,
126 /// assuming no reflow.
127 ///
128 /// If \p Break is true, returns the column at which the line should start
129 /// after the line break.
130 /// If \p Break is false, returns the column at which the line itself will
131 /// start.
132 virtual unsigned getContentStartColumn(unsigned LineIndex,
133 bool Break) const = 0;
134
135 /// Returns additional content indent required for the second line after the
136 /// content at line \p LineIndex is broken.
137 ///
138 // (Next lines do not start with `///` since otherwise -Wdocumentation picks
139 // up the example annotations and generates warnings for them)
140 // For example, Javadoc @param annotations require and indent of 4 spaces and
141 // in this example getContentIndex(1) returns 4.
142 // /**
143 // * @param loooooooooooooong line
144 // * continuation
145 // */
146 virtual unsigned getContentIndent(unsigned LineIndex) const { return 0; }
147
148 /// Returns a range (offset, length) at which to break the line at
149 /// \p LineIndex, if previously broken at \p TailOffset. If possible, do not
150 /// violate \p ColumnLimit, assuming the text starting at \p TailOffset in
151 /// the token is formatted starting at ContentStartColumn in the reformatted
152 /// file.
153 virtual Split getSplit(unsigned LineIndex, unsigned TailOffset,
154 unsigned ColumnLimit, unsigned ContentStartColumn,
155 const llvm::Regex &CommentPragmasRegex) const = 0;
156
157 /// Emits the previously retrieved \p Split via \p Whitespaces.
158 virtual void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split,
159 unsigned ContentIndent,
160 WhitespaceManager &Whitespaces) const = 0;
161
162 /// Returns the number of columns needed to format
163 /// \p RemainingTokenColumns, assuming that Split is within the range measured
164 /// by \p RemainingTokenColumns, and that the whitespace in Split is reduced
165 /// to a single space.
166 unsigned getLengthAfterCompression(unsigned RemainingTokenColumns,
167 Split Split) const;
168
169 /// Replaces the whitespace range described by \p Split with a single
170 /// space.
171 virtual void compressWhitespace(unsigned LineIndex, unsigned TailOffset,
172 Split Split,
173 WhitespaceManager &Whitespaces) const = 0;
174
175 /// Returns whether the token supports reflowing text.
176 virtual bool supportsReflow() const { return false; }
177
178 /// Returns a whitespace range (offset, length) of the content at \p
179 /// LineIndex such that the content of that line is reflown to the end of the
180 /// previous one.
181 ///
182 /// Returning (StringRef::npos, 0) indicates reflowing is not possible.
183 ///
184 /// The range will include any whitespace preceding the specified line's
185 /// content.
186 ///
187 /// If the split is not contained within one token, for example when reflowing
188 /// line comments, returns (0, <length>).
189 virtual Split getReflowSplit(unsigned LineIndex,
190 const llvm::Regex &CommentPragmasRegex) const {
191 return Split(StringRef::npos, 0);
192 }
193
194 /// Reflows the current line into the end of the previous one.
195 virtual void reflow(unsigned LineIndex,
196 WhitespaceManager &Whitespaces) const {}
197
198 /// Returns whether there will be a line break at the start of the
199 /// token.
200 virtual bool introducesBreakBeforeToken() const { return false; }
201
202 /// Replaces the whitespace between \p LineIndex-1 and \p LineIndex.
203 virtual void adaptStartOfLine(unsigned LineIndex,
204 WhitespaceManager &Whitespaces) const {}
205
206 /// Returns a whitespace range (offset, length) of the content at
207 /// the last line that needs to be reformatted after the last line has been
208 /// reformatted.
209 ///
210 /// A result having offset == StringRef::npos means that no reformat is
211 /// necessary.
212 virtual Split getSplitAfterLastLine(unsigned TailOffset) const {
213 return Split(StringRef::npos, 0);
214 }
215
216 /// Replaces the whitespace from \p SplitAfterLastLine on the last line
217 /// after the last line has been formatted by performing a reformatting.
218 void replaceWhitespaceAfterLastLine(unsigned TailOffset,
219 Split SplitAfterLastLine,
220 WhitespaceManager &Whitespaces) const {
221 insertBreak(LineIndex: getLineCount() - 1, TailOffset, Split: SplitAfterLastLine,
222 /*ContentIndent=*/ContentIndent: 0, Whitespaces);
223 }
224
225 /// Updates the next token of \p State to the next token after this
226 /// one. This can be used when this token manages a set of underlying tokens
227 /// as a unit and is responsible for the formatting of the them.
228 virtual void updateNextToken(LineState &State) const {}
229
230 /// Adds replacements that are needed when the token is broken. Such as
231 /// wrapping a JavaScript string in parentheses after it gets broken with plus
232 /// signs.
233 virtual void updateAfterBroken(WhitespaceManager &Whitespaces) const {}
234
235protected:
236 BreakableToken(const FormatToken &Tok, bool InPPDirective,
237 encoding::Encoding Encoding, const FormatStyle &Style)
238 : Tok(Tok), InPPDirective(InPPDirective), Encoding(Encoding),
239 Style(Style) {}
240
241 const FormatToken &Tok;
242 const bool InPPDirective;
243 const encoding::Encoding Encoding;
244 const FormatStyle &Style;
245};
246
247class BreakableStringLiteral : public BreakableToken {
248public:
249 /// Creates a breakable token for a single line string literal.
250 ///
251 /// \p StartColumn specifies the column in which the token will start
252 /// after formatting.
253 BreakableStringLiteral(const FormatToken &Tok, unsigned StartColumn,
254 StringRef Prefix, StringRef Postfix,
255 unsigned UnbreakableTailLength, bool InPPDirective,
256 encoding::Encoding Encoding, const FormatStyle &Style);
257
258 Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit,
259 unsigned ContentStartColumn,
260 const llvm::Regex &CommentPragmasRegex) const override;
261 void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split,
262 unsigned ContentIndent,
263 WhitespaceManager &Whitespaces) const override;
264 void compressWhitespace(unsigned LineIndex, unsigned TailOffset, Split Split,
265 WhitespaceManager &Whitespaces) const override {}
266 unsigned getLineCount() const override;
267 unsigned getRangeLength(unsigned LineIndex, unsigned Offset,
268 StringRef::size_type Length,
269 unsigned StartColumn) const override;
270 unsigned getRemainingLength(unsigned LineIndex, unsigned Offset,
271 unsigned StartColumn) const override;
272 unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override;
273
274protected:
275 // The column in which the token starts.
276 unsigned StartColumn;
277 // The prefix a line needs after a break in the token.
278 StringRef Prefix;
279 // The postfix a line needs before introducing a break.
280 StringRef Postfix;
281 // The token text excluding the prefix and postfix.
282 StringRef Line;
283 // Length of the sequence of tokens after this string literal that cannot
284 // contain line breaks.
285 unsigned UnbreakableTailLength;
286};
287
288class BreakableStringLiteralUsingOperators : public BreakableStringLiteral {
289public:
290 enum QuoteStyleType {
291 DoubleQuotes, // The string is quoted with double quotes.
292 SingleQuotes, // The JavaScript string is quoted with single quotes.
293 AtDoubleQuotes, // The C# verbatim string is quoted with the at sign and
294 // double quotes.
295 };
296 /// Creates a breakable token for a single line string literal for C#, Java,
297 /// JavaScript, or Verilog.
298 ///
299 /// \p StartColumn specifies the column in which the token will start
300 /// after formatting.
301 BreakableStringLiteralUsingOperators(
302 const FormatToken &Tok, QuoteStyleType QuoteStyle, bool UnindentPlus,
303 unsigned StartColumn, unsigned UnbreakableTailLength, bool InPPDirective,
304 encoding::Encoding Encoding, const FormatStyle &Style);
305 unsigned getRemainingLength(unsigned LineIndex, unsigned Offset,
306 unsigned StartColumn) const override;
307 unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override;
308 void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split,
309 unsigned ContentIndent,
310 WhitespaceManager &Whitespaces) const override;
311 void updateAfterBroken(WhitespaceManager &Whitespaces) const override;
312
313protected:
314 // Whether braces or parentheses should be inserted around the string to form
315 // a concatenation.
316 bool BracesNeeded;
317 QuoteStyleType QuoteStyle;
318 // The braces or parentheses along with the first character which they
319 // replace, either a quote or at sign.
320 StringRef LeftBraceQuote;
321 StringRef RightBraceQuote;
322 // Width added to the left due to the added brace or parenthesis. Does not
323 // apply to the first line.
324 int ContinuationIndent;
325};
326
327class BreakableComment : public BreakableToken {
328protected:
329 /// Creates a breakable token for a comment.
330 ///
331 /// \p StartColumn specifies the column in which the comment will start after
332 /// formatting.
333 BreakableComment(const FormatToken &Token, unsigned StartColumn,
334 bool InPPDirective, encoding::Encoding Encoding,
335 const FormatStyle &Style);
336
337public:
338 bool supportsReflow() const override { return true; }
339 unsigned getLineCount() const override;
340 Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit,
341 unsigned ContentStartColumn,
342 const llvm::Regex &CommentPragmasRegex) const override;
343 void compressWhitespace(unsigned LineIndex, unsigned TailOffset, Split Split,
344 WhitespaceManager &Whitespaces) const override;
345
346protected:
347 // Returns the token containing the line at LineIndex.
348 const FormatToken &tokenAt(unsigned LineIndex) const;
349
350 // Checks if the content of line LineIndex may be reflown with the previous
351 // line.
352 virtual bool mayReflow(unsigned LineIndex,
353 const llvm::Regex &CommentPragmasRegex) const = 0;
354
355 // Contains the original text of the lines of the block comment.
356 //
357 // In case of a block comments, excludes the leading /* in the first line and
358 // trailing */ in the last line. In case of line comments, excludes the
359 // leading // and spaces.
360 SmallVector<StringRef, 16> Lines;
361
362 // Contains the text of the lines excluding all leading and trailing
363 // whitespace between the lines. Note that the decoration (if present) is also
364 // not considered part of the text.
365 SmallVector<StringRef, 16> Content;
366
367 // Tokens[i] contains a reference to the token containing Lines[i] if the
368 // whitespace range before that token is managed by this block.
369 // Otherwise, Tokens[i] is a null pointer.
370 SmallVector<FormatToken *, 16> Tokens;
371
372 // ContentColumn[i] is the target column at which Content[i] should be.
373 // Note that this excludes a leading "* " or "*" in case of block comments
374 // where all lines have a "*" prefix, or the leading "// " or "//" in case of
375 // line comments.
376 //
377 // In block comments, the first line's target column is always positive. The
378 // remaining lines' target columns are relative to the first line to allow
379 // correct indentation of comments in \c WhitespaceManager. Thus they can be
380 // negative as well (in case the first line needs to be unindented more than
381 // there's actual whitespace in another line).
382 SmallVector<int, 16> ContentColumn;
383
384 // The intended start column of the first line of text from this section.
385 unsigned StartColumn;
386
387 // The prefix to use in front a line that has been reflown up.
388 // For example, when reflowing the second line after the first here:
389 // // comment 1
390 // // comment 2
391 // we expect:
392 // // comment 1 comment 2
393 // and not:
394 // // comment 1comment 2
395 StringRef ReflowPrefix = " ";
396};
397
398class BreakableBlockComment : public BreakableComment {
399public:
400 BreakableBlockComment(const FormatToken &Token, unsigned StartColumn,
401 unsigned OriginalStartColumn, bool FirstInLine,
402 bool InPPDirective, encoding::Encoding Encoding,
403 const FormatStyle &Style, bool UseCRLF);
404
405 Split getSplit(unsigned LineIndex, unsigned TailOffset, unsigned ColumnLimit,
406 unsigned ContentStartColumn,
407 const llvm::Regex &CommentPragmasRegex) const override;
408 unsigned getRangeLength(unsigned LineIndex, unsigned Offset,
409 StringRef::size_type Length,
410 unsigned StartColumn) const override;
411 unsigned getRemainingLength(unsigned LineIndex, unsigned Offset,
412 unsigned StartColumn) const override;
413 unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override;
414 unsigned getContentIndent(unsigned LineIndex) const override;
415 void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split,
416 unsigned ContentIndent,
417 WhitespaceManager &Whitespaces) const override;
418 Split getReflowSplit(unsigned LineIndex,
419 const llvm::Regex &CommentPragmasRegex) const override;
420 void reflow(unsigned LineIndex,
421 WhitespaceManager &Whitespaces) const override;
422 bool introducesBreakBeforeToken() const override;
423 void adaptStartOfLine(unsigned LineIndex,
424 WhitespaceManager &Whitespaces) const override;
425 Split getSplitAfterLastLine(unsigned TailOffset) const override;
426
427 bool mayReflow(unsigned LineIndex,
428 const llvm::Regex &CommentPragmasRegex) const override;
429
430 // Contains Javadoc annotations that require additional indent when continued
431 // on multiple lines.
432 static const llvm::StringSet<> ContentIndentingJavadocAnnotations;
433
434private:
435 // Rearranges the whitespace between Lines[LineIndex-1] and Lines[LineIndex].
436 //
437 // Updates Content[LineIndex-1] and Content[LineIndex] by stripping off
438 // leading and trailing whitespace.
439 //
440 // Sets ContentColumn to the intended column in which the text at
441 // Lines[LineIndex] starts (note that the decoration, if present, is not
442 // considered part of the text).
443 void adjustWhitespace(unsigned LineIndex, int IndentDelta);
444
445 // The column at which the text of a broken line should start.
446 // Note that an optional decoration would go before that column.
447 // IndentAtLineBreak is a uniform position for all lines in a block comment,
448 // regardless of their relative position.
449 // FIXME: Revisit the decision to do this; the main reason was to support
450 // patterns like
451 // /**************//**
452 // * Comment
453 // We could also support such patterns by special casing the first line
454 // instead.
455 unsigned IndentAtLineBreak;
456
457 // This is to distinguish between the case when the last line was empty and
458 // the case when it started with a decoration ("*" or "* ").
459 bool LastLineNeedsDecoration;
460
461 // Either "* " if all lines begin with a "*", or empty.
462 StringRef Decoration;
463
464 // If this block comment has decorations, this is the column of the start of
465 // the decorations.
466 unsigned DecorationColumn;
467
468 // If true, make sure that the opening '/**' and the closing '*/' ends on a
469 // line of itself. Styles like jsdoc require this for multiline comments.
470 bool DelimitersOnNewline;
471
472 // Length of the sequence of tokens after this string literal that cannot
473 // contain line breaks.
474 unsigned UnbreakableTailLength;
475};
476
477class BreakableLineCommentSection : public BreakableComment {
478public:
479 BreakableLineCommentSection(const FormatToken &Token, unsigned StartColumn,
480 bool InPPDirective, encoding::Encoding Encoding,
481 const FormatStyle &Style);
482
483 unsigned getRangeLength(unsigned LineIndex, unsigned Offset,
484 StringRef::size_type Length,
485 unsigned StartColumn) const override;
486 unsigned getContentStartColumn(unsigned LineIndex, bool Break) const override;
487 void insertBreak(unsigned LineIndex, unsigned TailOffset, Split Split,
488 unsigned ContentIndent,
489 WhitespaceManager &Whitespaces) const override;
490 Split getReflowSplit(unsigned LineIndex,
491 const llvm::Regex &CommentPragmasRegex) const override;
492 void reflow(unsigned LineIndex,
493 WhitespaceManager &Whitespaces) const override;
494 void adaptStartOfLine(unsigned LineIndex,
495 WhitespaceManager &Whitespaces) const override;
496 void updateNextToken(LineState &State) const override;
497 bool mayReflow(unsigned LineIndex,
498 const llvm::Regex &CommentPragmasRegex) const override;
499
500private:
501 // OriginalPrefix[i] contains the original prefix of line i, including
502 // trailing whitespace before the start of the content. The indentation
503 // preceding the prefix is not included.
504 // For example, if the line is:
505 // // content
506 // then the original prefix is "// ".
507 SmallVector<StringRef, 16> OriginalPrefix;
508
509 /// Prefix[i] + SpacesToAdd[i] contains the intended leading "//" with
510 /// trailing spaces to account for the indentation of content within the
511 /// comment at line i after formatting. It can be different than the original
512 /// prefix.
513 /// When the original line starts like this:
514 /// //content
515 /// Then the OriginalPrefix[i] is "//", but the Prefix[i] is "// " in the LLVM
516 /// style.
517 /// When the line starts like:
518 /// // content
519 /// And we want to remove the spaces the OriginalPrefix[i] is "// " and
520 /// Prefix[i] is "//".
521 SmallVector<std::string, 16> Prefix;
522
523 /// How many spaces are added or removed from the OriginalPrefix to form
524 /// Prefix.
525 SmallVector<int, 16> PrefixSpaceChange;
526
527 /// The token to which the last line of this breakable token belongs
528 /// to; nullptr if that token is the initial token.
529 ///
530 /// The distinction is because if the token of the last line of this breakable
531 /// token is distinct from the initial token, this breakable token owns the
532 /// whitespace before the token of the last line, and the whitespace manager
533 /// must be able to modify it.
534 FormatToken *LastLineTok = nullptr;
535};
536} // namespace format
537} // namespace clang
538
539#endif
540