19 #include "llvm/ADT/STLExtras.h"
20 #include "llvm/Support/Debug.h"
23 #define DEBUG_TYPE "format-token-breaker"
28 static const char *
const Blanks =
" \t\v\f\r";
43 unsigned ContentStartColumn,
47 if (ColumnLimit <= ContentStartColumn + 1)
50 unsigned MaxSplit = ColumnLimit - ContentStartColumn + 1;
51 unsigned MaxSplitBytes = 0;
53 for (
unsigned NumChars = 0;
54 NumChars < MaxSplit && MaxSplitBytes < Text.size();) {
55 unsigned BytesInChar =
59 ContentStartColumn, TabWidth,
Encoding);
60 MaxSplitBytes += BytesInChar;
63 StringRef::size_type SpaceOffset = Text.find_last_of(
Blanks, MaxSplitBytes);
64 if (SpaceOffset == StringRef::npos ||
66 Text.find_last_not_of(
Blanks, SpaceOffset) == StringRef::npos) {
69 StringRef::size_type FirstNonWhitespace = Text.find_first_not_of(
Blanks);
70 if (FirstNonWhitespace == StringRef::npos)
73 SpaceOffset = Text.find_first_of(
74 Blanks, std::max<unsigned>(MaxSplitBytes, FirstNonWhitespace));
76 if (SpaceOffset != StringRef::npos && SpaceOffset != 0) {
77 StringRef BeforeCut = Text.substr(0, SpaceOffset).rtrim(
Blanks);
78 StringRef AfterCut = Text.substr(SpaceOffset).ltrim(
Blanks);
80 AfterCut.begin() - BeforeCut.end());
91 if (ColumnLimit <= UsedColumns)
93 unsigned MaxSplit = ColumnLimit - UsedColumns;
94 StringRef::size_type SpaceOffset = 0;
95 StringRef::size_type SlashOffset = 0;
96 StringRef::size_type WordStartOffset = 0;
97 StringRef::size_type SplitPoint = 0;
98 for (
unsigned Chars = 0;;) {
100 if (Text[0] ==
'\\') {
106 Text.substr(0, Advance), UsedColumns + Chars, TabWidth,
Encoding);
109 if (Chars > MaxSplit || Text.size() <= Advance)
113 SpaceOffset = SplitPoint;
115 SlashOffset = SplitPoint;
117 WordStartOffset = SplitPoint;
119 SplitPoint += Advance;
120 Text = Text.substr(Advance);
123 if (SpaceOffset != 0)
125 if (SlashOffset != 0)
127 if (WordStartOffset != 0)
137 unsigned LineIndex,
unsigned Offset, StringRef::size_type Length)
const {
145 const FormatToken &Tok,
unsigned IndentLevel,
unsigned StartColumn,
146 StringRef Prefix, StringRef Postfix,
bool InPPDirective,
148 :
BreakableToken(Tok, IndentLevel, InPPDirective, Encoding, Style),
149 StartColumn(StartColumn), Prefix(Prefix), Postfix(Postfix) {
152 Prefix.size(), Tok.
TokenText.size() - Prefix.size() - Postfix.size());
156 const FormatToken &Tok,
unsigned IndentLevel,
unsigned StartColumn,
157 StringRef Prefix, StringRef Postfix,
bool InPPDirective,
160 InPPDirective, Encoding, Style) {}
164 unsigned ColumnLimit)
const {
178 if (
Prefix.startswith(
"@"))
186 static const char *
const KnownPrefixes[] = {
"///",
"//",
"//!"};
187 StringRef LongestPrefix;
188 for (StringRef KnownPrefix : KnownPrefixes) {
189 if (Comment.startswith(KnownPrefix)) {
190 size_t PrefixLength = KnownPrefix.size();
191 while (PrefixLength < Comment.size() && Comment[PrefixLength] ==
' ')
193 if (PrefixLength > LongestPrefix.size())
194 LongestPrefix = Comment.substr(0, PrefixLength);
197 return LongestPrefix;
205 InPPDirective, Encoding, Style) {
220 unsigned ColumnLimit)
const {
229 Tok, OriginalPrefix.size() + TailOffset + Split.first, Split.second,
237 Tok, OriginalPrefix.size() + TailOffset + Split.first, Split.second,
"",
244 if (OriginalPrefix !=
Prefix) {
254 unsigned OriginalStartColumn,
bool FirstInLine,
bool InPPDirective,
256 :
BreakableToken(Token, IndentLevel, InPPDirective, Encoding, Style) {
258 assert(TokenText.startswith(
"/*") && TokenText.endswith(
"*/"));
259 TokenText.substr(2, TokenText.size() - 4).split(Lines,
"\n");
261 int IndentDelta = StartColumn - OriginalStartColumn;
262 LeadingWhitespace.resize(Lines.size());
263 StartOfLineColumn.resize(Lines.size());
264 StartOfLineColumn[0] = StartColumn + 2;
265 for (
size_t i = 1; i < Lines.size(); ++i)
266 adjustWhitespace(i, IndentDelta);
269 if (Lines.size() == 1 && !FirstInLine) {
277 for (
size_t i = 1, e = Lines.size(); i < e && !Decoration.empty(); ++i) {
279 if (i + 1 == e && Lines[i].empty())
281 if (!Lines[i].empty() && i + 1 != e && Decoration.startswith(Lines[i]))
283 while (!Lines[i].startswith(Decoration))
284 Decoration = Decoration.substr(0, Decoration.size() - 1);
287 LastLineNeedsDecoration =
true;
288 IndentAtLineBreak = StartOfLineColumn[0] + 1;
289 for (
size_t i = 1; i < Lines.size(); ++i) {
290 if (Lines[i].empty()) {
291 if (i + 1 == Lines.size()) {
295 LastLineNeedsDecoration =
false;
296 }
else if (Decoration.empty()) {
299 StartOfLineColumn[i] = 0;
307 unsigned DecorationSize =
308 Decoration.startswith(Lines[i]) ? Lines[i].size() : Decoration.size();
309 StartOfLineColumn[i] += DecorationSize;
310 Lines[i] = Lines[i].substr(DecorationSize);
311 LeadingWhitespace[i] += DecorationSize;
312 if (!Decoration.startswith(Lines[i]))
314 std::min<int>(IndentAtLineBreak, std::max(0, StartOfLineColumn[i]));
316 IndentAtLineBreak = std::max<unsigned>(IndentAtLineBreak, Decoration.size());
318 llvm::dbgs() <<
"IndentAtLineBreak " << IndentAtLineBreak <<
"\n";
319 for (
size_t i = 0; i < Lines.size(); ++i) {
320 llvm::dbgs() << i <<
" |" << Lines[i] <<
"| " << LeadingWhitespace[i]
326 void BreakableBlockComment::adjustWhitespace(
unsigned LineIndex,
333 size_t EndOfPreviousLine = Lines[LineIndex - 1].size();
339 Lines[LineIndex - 1].find_last_not_of(
Blanks, EndOfPreviousLine);
340 if (EndOfPreviousLine == StringRef::npos)
341 EndOfPreviousLine = 0;
345 size_t StartOfLine = Lines[LineIndex].find_first_not_of(
Blanks);
346 if (StartOfLine == StringRef::npos)
347 StartOfLine = Lines[LineIndex].rtrim(
"\r\n").size();
349 StringRef Whitespace = Lines[LineIndex].substr(0, StartOfLine);
351 Lines[LineIndex - 1] = Lines[LineIndex - 1].substr(0, EndOfPreviousLine);
352 Lines[LineIndex] = Lines[LineIndex].substr(StartOfLine);
355 LeadingWhitespace[LineIndex] =
356 Lines[LineIndex].begin() - Lines[LineIndex - 1].end();
359 StartOfLineColumn[LineIndex] =
367 unsigned LineIndex,
unsigned Offset, StringRef::size_type Length)
const {
368 unsigned ContentStartColumn = getContentStartColumn(LineIndex, Offset);
369 return ContentStartColumn +
374 (LineIndex + 1 == Lines.size() ? 2 : 0);
379 unsigned ColumnLimit)
const {
381 getContentStartColumn(LineIndex, TailOffset),
388 StringRef Text = Lines[LineIndex].substr(TailOffset);
389 StringRef Prefix = Decoration;
390 if (LineIndex + 1 == Lines.size() &&
391 Text.size() == Split.first + Split.second) {
396 unsigned BreakOffsetInToken =
398 unsigned CharsToRemove = Split.second;
399 assert(IndentAtLineBreak >= Decoration.size());
402 IndentLevel, IndentAtLineBreak - Decoration.size());
408 StringRef Text = Lines[LineIndex].substr(TailOffset);
409 unsigned BreakOffsetInToken =
411 unsigned CharsToRemove = Split.second;
413 Tok, BreakOffsetInToken, CharsToRemove,
"",
"",
false,
421 StringRef Prefix = Decoration;
422 if (Lines[LineIndex].empty()) {
423 if (LineIndex + 1 == Lines.size()) {
424 if (!LastLineNeedsDecoration) {
429 }
else if (!Decoration.empty()) {
432 Prefix = Prefix.substr(0, 1);
435 if (StartOfLineColumn[LineIndex] == 1) {
437 Prefix = Prefix.substr(0, 1);
441 unsigned WhitespaceOffsetInToken = Lines[LineIndex].data() -
443 LeadingWhitespace[LineIndex];
445 Tok, WhitespaceOffsetInToken, LeadingWhitespace[LineIndex],
"", Prefix,
447 StartOfLineColumn[LineIndex] - Prefix.size());
451 BreakableBlockComment::getContentStartColumn(
unsigned LineIndex,
452 unsigned TailOffset)
const {
455 return IndentAtLineBreak;
456 return std::max(0, StartOfLineColumn[LineIndex]);
Declares BreakableToken, BreakableStringLiteral, and BreakableBlockComment classes, that contain token type-specific logic to break long lines in tokens.
static LLVM_READONLY bool isAlphanumeric(unsigned char c)
Return true if this character is an ASCII letter or digit: [a-zA-Z0-9].