Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 1 | #include "clang/AST/CommentLexer.h" |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 2 | #include "clang/Basic/ConvertUTF.h" |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 3 | #include "llvm/ADT/StringSwitch.h" |
| 4 | #include "llvm/Support/ErrorHandling.h" |
| 5 | |
| 6 | namespace clang { |
| 7 | namespace comments { |
| 8 | |
| 9 | void Token::dump(const Lexer &L, const SourceManager &SM) const { |
| 10 | llvm::errs() << "comments::Token Kind=" << Kind << " "; |
| 11 | Loc.dump(SM); |
| 12 | llvm::errs() << " " << Length << " \"" << L.getSpelling(*this, SM) << "\"\n"; |
| 13 | } |
| 14 | |
| 15 | bool Lexer::isVerbatimBlockCommand(StringRef BeginName, |
| 16 | StringRef &EndName) const { |
| 17 | const char *Result = llvm::StringSwitch<const char *>(BeginName) |
| 18 | .Case("code", "endcode") |
| 19 | .Case("verbatim", "endverbatim") |
| 20 | .Case("htmlonly", "endhtmlonly") |
| 21 | .Case("latexonly", "endlatexonly") |
| 22 | .Case("xmlonly", "endxmlonly") |
| 23 | .Case("manonly", "endmanonly") |
| 24 | .Case("rtfonly", "endrtfonly") |
| 25 | |
| 26 | .Case("dot", "enddot") |
| 27 | .Case("msc", "endmsc") |
| 28 | |
| 29 | .Case("f$", "f$") // Inline LaTeX formula |
| 30 | .Case("f[", "f]") // Displayed LaTeX formula |
| 31 | .Case("f{", "f}") // LaTeX environment |
| 32 | |
| 33 | .Default(NULL); |
| 34 | |
| 35 | if (Result) { |
| 36 | EndName = Result; |
| 37 | return true; |
| 38 | } |
| 39 | |
| 40 | for (VerbatimBlockCommandVector::const_iterator |
| 41 | I = VerbatimBlockCommands.begin(), |
| 42 | E = VerbatimBlockCommands.end(); |
| 43 | I != E; ++I) |
| 44 | if (I->BeginName == BeginName) { |
| 45 | EndName = I->EndName; |
| 46 | return true; |
| 47 | } |
| 48 | |
| 49 | return false; |
| 50 | } |
| 51 | |
| 52 | bool Lexer::isVerbatimLineCommand(StringRef Name) const { |
| 53 | bool Result = llvm::StringSwitch<bool>(Name) |
| 54 | .Case("fn", true) |
| 55 | .Case("var", true) |
| 56 | .Case("property", true) |
| 57 | .Case("typedef", true) |
| 58 | |
| 59 | .Case("overload", true) |
| 60 | |
| 61 | .Case("defgroup", true) |
| 62 | .Case("ingroup", true) |
| 63 | .Case("addtogroup", true) |
| 64 | .Case("weakgroup", true) |
| 65 | .Case("name", true) |
| 66 | |
| 67 | .Case("section", true) |
| 68 | .Case("subsection", true) |
| 69 | .Case("subsubsection", true) |
| 70 | .Case("paragraph", true) |
| 71 | |
| 72 | .Case("mainpage", true) |
| 73 | .Case("subpage", true) |
| 74 | .Case("ref", true) |
| 75 | |
| 76 | .Default(false); |
| 77 | |
| 78 | if (Result) |
| 79 | return true; |
| 80 | |
| 81 | for (VerbatimLineCommandVector::const_iterator |
| 82 | I = VerbatimLineCommands.begin(), |
| 83 | E = VerbatimLineCommands.end(); |
| 84 | I != E; ++I) |
| 85 | if (I->Name == Name) |
| 86 | return true; |
| 87 | |
| 88 | return false; |
| 89 | } |
| 90 | |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 91 | namespace { |
| 92 | bool isHTMLNamedCharacterReferenceCharacter(char C) { |
| 93 | return (C >= 'a' && C <= 'z') || |
| 94 | (C >= 'A' && C <= 'Z'); |
| 95 | } |
| 96 | |
| 97 | bool isHTMLDecimalCharacterReferenceCharacter(char C) { |
| 98 | return C >= '0' && C <= '9'; |
| 99 | } |
| 100 | |
| 101 | bool isHTMLHexCharacterReferenceCharacter(char C) { |
| 102 | return (C >= '0' && C <= '9') || |
| 103 | (C >= 'a' && C <= 'f') || |
| 104 | (C >= 'A' && C <= 'F'); |
| 105 | } |
| 106 | } // unnamed namespace |
| 107 | |
| 108 | StringRef Lexer::resolveHTMLNamedCharacterReference(StringRef Name) const { |
| 109 | return llvm::StringSwitch<StringRef>(Name) |
| 110 | .Case("amp", "&") |
| 111 | .Case("lt", "<") |
| 112 | .Case("gt", ">") |
| 113 | .Case("quot", "\"") |
| 114 | .Case("apos", "\'") |
| 115 | .Default(""); |
| 116 | } |
| 117 | |
| 118 | StringRef Lexer::resolveHTMLDecimalCharacterReference(StringRef Name) const { |
| 119 | unsigned CodePoint = 0; |
| 120 | for (unsigned i = 0, e = Name.size(); i != e; ++i) { |
| 121 | assert(isHTMLDecimalCharacterReferenceCharacter(Name[i])); |
| 122 | CodePoint *= 10; |
| 123 | CodePoint += Name[i] - '0'; |
| 124 | } |
| 125 | |
| 126 | char *Resolved = Allocator.Allocate<char>(UNI_MAX_UTF8_BYTES_PER_CODE_POINT); |
| 127 | char *ResolvedPtr = Resolved; |
| 128 | if (ConvertCodePointToUTF8(CodePoint, ResolvedPtr)) |
| 129 | return StringRef(Resolved, ResolvedPtr - Resolved); |
| 130 | else |
| 131 | return StringRef(); |
| 132 | } |
| 133 | |
| 134 | StringRef Lexer::resolveHTMLHexCharacterReference(StringRef Name) const { |
| 135 | unsigned CodePoint = 0; |
| 136 | for (unsigned i = 0, e = Name.size(); i != e; ++i) { |
| 137 | CodePoint *= 16; |
| 138 | const char C = Name[i]; |
| 139 | assert(isHTMLHexCharacterReferenceCharacter(C)); |
| 140 | if (C >= '0' && C <= '9') |
| 141 | CodePoint += Name[i] - '0'; |
| 142 | else if (C >= 'a' && C <= 'f') |
| 143 | CodePoint += Name[i] - 'a' + 10; |
| 144 | else |
| 145 | CodePoint += Name[i] - 'A' + 10; |
| 146 | } |
| 147 | |
| 148 | char *Resolved = Allocator.Allocate<char>(UNI_MAX_UTF8_BYTES_PER_CODE_POINT); |
| 149 | char *ResolvedPtr = Resolved; |
| 150 | if (ConvertCodePointToUTF8(CodePoint, ResolvedPtr)) |
| 151 | return StringRef(Resolved, ResolvedPtr - Resolved); |
| 152 | else |
| 153 | return StringRef(); |
| 154 | } |
| 155 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 156 | void Lexer::skipLineStartingDecorations() { |
| 157 | // This function should be called only for C comments |
| 158 | assert(CommentState == LCS_InsideCComment); |
| 159 | |
| 160 | if (BufferPtr == CommentEnd) |
| 161 | return; |
| 162 | |
| 163 | switch (*BufferPtr) { |
| 164 | case ' ': |
| 165 | case '\t': |
| 166 | case '\f': |
| 167 | case '\v': { |
| 168 | const char *NewBufferPtr = BufferPtr; |
| 169 | NewBufferPtr++; |
| 170 | if (NewBufferPtr == CommentEnd) |
| 171 | return; |
| 172 | |
| 173 | char C = *NewBufferPtr; |
| 174 | while (C == ' ' || C == '\t' || C == '\f' || C == '\v') { |
| 175 | NewBufferPtr++; |
| 176 | if (NewBufferPtr == CommentEnd) |
| 177 | return; |
| 178 | C = *NewBufferPtr; |
| 179 | } |
| 180 | if (C == '*') |
| 181 | BufferPtr = NewBufferPtr + 1; |
| 182 | break; |
| 183 | } |
| 184 | case '*': |
| 185 | BufferPtr++; |
| 186 | break; |
| 187 | } |
| 188 | } |
| 189 | |
| 190 | namespace { |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 191 | /// Returns pointer to the first newline character in the string. |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 192 | const char *findNewline(const char *BufferPtr, const char *BufferEnd) { |
| 193 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 194 | const char C = *BufferPtr; |
| 195 | if (C == '\n' || C == '\r') |
| 196 | return BufferPtr; |
| 197 | } |
| 198 | return BufferEnd; |
| 199 | } |
| 200 | |
| 201 | const char *skipNewline(const char *BufferPtr, const char *BufferEnd) { |
| 202 | if (BufferPtr == BufferEnd) |
| 203 | return BufferPtr; |
| 204 | |
| 205 | if (*BufferPtr == '\n') |
| 206 | BufferPtr++; |
| 207 | else { |
| 208 | assert(*BufferPtr == '\r'); |
| 209 | BufferPtr++; |
| 210 | if (BufferPtr != BufferEnd && *BufferPtr == '\n') |
| 211 | BufferPtr++; |
| 212 | } |
| 213 | return BufferPtr; |
| 214 | } |
| 215 | |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 216 | const char *skipNamedCharacterReference(const char *BufferPtr, |
| 217 | const char *BufferEnd) { |
| 218 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 219 | if (!isHTMLNamedCharacterReferenceCharacter(*BufferPtr)) |
| 220 | return BufferPtr; |
| 221 | } |
| 222 | return BufferEnd; |
| 223 | } |
| 224 | |
| 225 | const char *skipDecimalCharacterReference(const char *BufferPtr, |
| 226 | const char *BufferEnd) { |
| 227 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 228 | if (!isHTMLDecimalCharacterReferenceCharacter(*BufferPtr)) |
| 229 | return BufferPtr; |
| 230 | } |
| 231 | return BufferEnd; |
| 232 | } |
| 233 | |
| 234 | const char *skipHexCharacterReference(const char *BufferPtr, |
| 235 | const char *BufferEnd) { |
| 236 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 237 | if (!isHTMLHexCharacterReferenceCharacter(*BufferPtr)) |
| 238 | return BufferPtr; |
| 239 | } |
| 240 | return BufferEnd; |
| 241 | } |
| 242 | |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 243 | bool isHTMLIdentifierStartingCharacter(char C) { |
| 244 | return (C >= 'a' && C <= 'z') || |
| 245 | (C >= 'A' && C <= 'Z'); |
| 246 | } |
| 247 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 248 | bool isHTMLIdentifierCharacter(char C) { |
| 249 | return (C >= 'a' && C <= 'z') || |
| 250 | (C >= 'A' && C <= 'Z') || |
| 251 | (C >= '0' && C <= '9'); |
| 252 | } |
| 253 | |
| 254 | const char *skipHTMLIdentifier(const char *BufferPtr, const char *BufferEnd) { |
| 255 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 256 | if (!isHTMLIdentifierCharacter(*BufferPtr)) |
| 257 | return BufferPtr; |
| 258 | } |
| 259 | return BufferEnd; |
| 260 | } |
| 261 | |
| 262 | /// Skip HTML string quoted in single or double quotes. Escaping quotes inside |
| 263 | /// string allowed. |
| 264 | /// |
| 265 | /// Returns pointer to closing quote. |
| 266 | const char *skipHTMLQuotedString(const char *BufferPtr, const char *BufferEnd) |
| 267 | { |
| 268 | const char Quote = *BufferPtr; |
| 269 | assert(Quote == '\"' || Quote == '\''); |
| 270 | |
| 271 | BufferPtr++; |
| 272 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 273 | const char C = *BufferPtr; |
| 274 | if (C == Quote && BufferPtr[-1] != '\\') |
| 275 | return BufferPtr; |
| 276 | } |
| 277 | return BufferEnd; |
| 278 | } |
| 279 | |
| 280 | bool isHorizontalWhitespace(char C) { |
| 281 | return C == ' ' || C == '\t' || C == '\f' || C == '\v'; |
| 282 | } |
| 283 | |
| 284 | bool isWhitespace(char C) { |
| 285 | return C == ' ' || C == '\n' || C == '\r' || |
| 286 | C == '\t' || C == '\f' || C == '\v'; |
| 287 | } |
| 288 | |
| 289 | const char *skipWhitespace(const char *BufferPtr, const char *BufferEnd) { |
| 290 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 291 | if (!isWhitespace(*BufferPtr)) |
| 292 | return BufferPtr; |
| 293 | } |
| 294 | return BufferEnd; |
| 295 | } |
| 296 | |
Dmitri Gribenko | 64da4e5 | 2012-07-18 23:01:58 +0000 | [diff] [blame] | 297 | bool isWhitespace(const char *BufferPtr, const char *BufferEnd) { |
| 298 | return skipWhitespace(BufferPtr, BufferEnd) == BufferEnd; |
| 299 | } |
| 300 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 301 | bool isCommandNameCharacter(char C) { |
| 302 | return (C >= 'a' && C <= 'z') || |
| 303 | (C >= 'A' && C <= 'Z') || |
| 304 | (C >= '0' && C <= '9'); |
| 305 | } |
| 306 | |
| 307 | const char *skipCommandName(const char *BufferPtr, const char *BufferEnd) { |
| 308 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 309 | if (!isCommandNameCharacter(*BufferPtr)) |
| 310 | return BufferPtr; |
| 311 | } |
| 312 | return BufferEnd; |
| 313 | } |
| 314 | |
| 315 | /// Return the one past end pointer for BCPL comments. |
| 316 | /// Handles newlines escaped with backslash or trigraph for backslahs. |
| 317 | const char *findBCPLCommentEnd(const char *BufferPtr, const char *BufferEnd) { |
| 318 | const char *CurPtr = BufferPtr; |
| 319 | while (CurPtr != BufferEnd) { |
| 320 | char C = *CurPtr; |
| 321 | while (C != '\n' && C != '\r') { |
| 322 | CurPtr++; |
| 323 | if (CurPtr == BufferEnd) |
| 324 | return BufferEnd; |
| 325 | C = *CurPtr; |
| 326 | } |
| 327 | // We found a newline, check if it is escaped. |
| 328 | const char *EscapePtr = CurPtr - 1; |
| 329 | while(isHorizontalWhitespace(*EscapePtr)) |
| 330 | EscapePtr--; |
| 331 | |
| 332 | if (*EscapePtr == '\\' || |
| 333 | (EscapePtr - 2 >= BufferPtr && EscapePtr[0] == '/' && |
| 334 | EscapePtr[-1] == '?' && EscapePtr[-2] == '?')) { |
| 335 | // We found an escaped newline. |
| 336 | CurPtr = skipNewline(CurPtr, BufferEnd); |
| 337 | } else |
| 338 | return CurPtr; // Not an escaped newline. |
| 339 | } |
| 340 | return BufferEnd; |
| 341 | } |
| 342 | |
| 343 | /// Return the one past end pointer for C comments. |
| 344 | /// Very dumb, does not handle escaped newlines or trigraphs. |
| 345 | const char *findCCommentEnd(const char *BufferPtr, const char *BufferEnd) { |
| 346 | for ( ; BufferPtr != BufferEnd; ++BufferPtr) { |
| 347 | if (*BufferPtr == '*') { |
| 348 | assert(BufferPtr + 1 != BufferEnd); |
| 349 | if (*(BufferPtr + 1) == '/') |
| 350 | return BufferPtr; |
| 351 | } |
| 352 | } |
| 353 | llvm_unreachable("buffer end hit before '*/' was seen"); |
| 354 | } |
| 355 | } // unnamed namespace |
| 356 | |
| 357 | void Lexer::lexCommentText(Token &T) { |
| 358 | assert(CommentState == LCS_InsideBCPLComment || |
| 359 | CommentState == LCS_InsideCComment); |
| 360 | |
| 361 | switch (State) { |
| 362 | case LS_Normal: |
| 363 | break; |
| 364 | case LS_VerbatimBlockFirstLine: |
| 365 | lexVerbatimBlockFirstLine(T); |
| 366 | return; |
| 367 | case LS_VerbatimBlockBody: |
| 368 | lexVerbatimBlockBody(T); |
| 369 | return; |
Dmitri Gribenko | 962668d | 2012-06-27 16:53:58 +0000 | [diff] [blame] | 370 | case LS_VerbatimLineText: |
| 371 | lexVerbatimLineText(T); |
| 372 | return; |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 373 | case LS_HTMLStartTag: |
| 374 | lexHTMLStartTag(T); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 375 | return; |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 376 | case LS_HTMLEndTag: |
| 377 | lexHTMLEndTag(T); |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 378 | return; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 379 | } |
| 380 | |
| 381 | assert(State == LS_Normal); |
| 382 | |
| 383 | const char *TokenPtr = BufferPtr; |
| 384 | assert(TokenPtr < CommentEnd); |
| 385 | while (TokenPtr != CommentEnd) { |
| 386 | switch(*TokenPtr) { |
| 387 | case '\\': |
| 388 | case '@': { |
| 389 | TokenPtr++; |
| 390 | if (TokenPtr == CommentEnd) { |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 391 | formTextToken(T, TokenPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 392 | return; |
| 393 | } |
| 394 | char C = *TokenPtr; |
| 395 | switch (C) { |
| 396 | default: |
| 397 | break; |
| 398 | |
| 399 | case '\\': case '@': case '&': case '$': |
| 400 | case '#': case '<': case '>': case '%': |
| 401 | case '\"': case '.': case ':': |
| 402 | // This is one of \\ \@ \& \$ etc escape sequences. |
| 403 | TokenPtr++; |
| 404 | if (C == ':' && TokenPtr != CommentEnd && *TokenPtr == ':') { |
| 405 | // This is the \:: escape sequence. |
| 406 | TokenPtr++; |
| 407 | } |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 408 | StringRef UnescapedText(BufferPtr + 1, TokenPtr - (BufferPtr + 1)); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 409 | formTokenWithChars(T, TokenPtr, tok::text); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 410 | T.setText(UnescapedText); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 411 | return; |
| 412 | } |
| 413 | |
| 414 | // Don't make zero-length commands. |
| 415 | if (!isCommandNameCharacter(*TokenPtr)) { |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 416 | formTextToken(T, TokenPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 417 | return; |
| 418 | } |
| 419 | |
| 420 | TokenPtr = skipCommandName(TokenPtr, CommentEnd); |
| 421 | unsigned Length = TokenPtr - (BufferPtr + 1); |
| 422 | |
| 423 | // Hardcoded support for lexing LaTeX formula commands |
| 424 | // \f$ \f[ \f] \f{ \f} as a single command. |
| 425 | if (Length == 1 && TokenPtr[-1] == 'f' && TokenPtr != CommentEnd) { |
| 426 | C = *TokenPtr; |
| 427 | if (C == '$' || C == '[' || C == ']' || C == '{' || C == '}') { |
| 428 | TokenPtr++; |
| 429 | Length++; |
| 430 | } |
| 431 | } |
| 432 | |
| 433 | const StringRef CommandName(BufferPtr + 1, Length); |
| 434 | StringRef EndName; |
| 435 | |
| 436 | if (isVerbatimBlockCommand(CommandName, EndName)) { |
| 437 | setupAndLexVerbatimBlock(T, TokenPtr, *BufferPtr, EndName); |
| 438 | return; |
| 439 | } |
| 440 | if (isVerbatimLineCommand(CommandName)) { |
Dmitri Gribenko | 962668d | 2012-06-27 16:53:58 +0000 | [diff] [blame] | 441 | setupAndLexVerbatimLine(T, TokenPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 442 | return; |
| 443 | } |
| 444 | formTokenWithChars(T, TokenPtr, tok::command); |
| 445 | T.setCommandName(CommandName); |
| 446 | return; |
| 447 | } |
| 448 | |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 449 | case '&': |
| 450 | lexHTMLCharacterReference(T); |
| 451 | return; |
| 452 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 453 | case '<': { |
| 454 | TokenPtr++; |
| 455 | if (TokenPtr == CommentEnd) { |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 456 | formTextToken(T, TokenPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 457 | return; |
| 458 | } |
| 459 | const char C = *TokenPtr; |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 460 | if (isHTMLIdentifierStartingCharacter(C)) |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 461 | setupAndLexHTMLStartTag(T); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 462 | else if (C == '/') |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 463 | setupAndLexHTMLEndTag(T); |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 464 | else |
| 465 | formTextToken(T, TokenPtr); |
| 466 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 467 | return; |
| 468 | } |
| 469 | |
| 470 | case '\n': |
| 471 | case '\r': |
| 472 | TokenPtr = skipNewline(TokenPtr, CommentEnd); |
| 473 | formTokenWithChars(T, TokenPtr, tok::newline); |
| 474 | |
| 475 | if (CommentState == LCS_InsideCComment) |
| 476 | skipLineStartingDecorations(); |
| 477 | return; |
| 478 | |
| 479 | default: { |
| 480 | while (true) { |
| 481 | TokenPtr++; |
| 482 | if (TokenPtr == CommentEnd) |
| 483 | break; |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 484 | const char C = *TokenPtr; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 485 | if(C == '\n' || C == '\r' || |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 486 | C == '\\' || C == '@' || C == '&' || C == '<') |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 487 | break; |
| 488 | } |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 489 | formTextToken(T, TokenPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 490 | return; |
| 491 | } |
| 492 | } |
| 493 | } |
| 494 | } |
| 495 | |
| 496 | void Lexer::setupAndLexVerbatimBlock(Token &T, |
| 497 | const char *TextBegin, |
| 498 | char Marker, StringRef EndName) { |
| 499 | VerbatimBlockEndCommandName.clear(); |
| 500 | VerbatimBlockEndCommandName.append(Marker == '\\' ? "\\" : "@"); |
| 501 | VerbatimBlockEndCommandName.append(EndName); |
| 502 | |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 503 | StringRef Name(BufferPtr + 1, TextBegin - (BufferPtr + 1)); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 504 | formTokenWithChars(T, TextBegin, tok::verbatim_block_begin); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 505 | T.setVerbatimBlockName(Name); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 506 | |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 507 | // If there is a newline following the verbatim opening command, skip the |
| 508 | // newline so that we don't create an tok::verbatim_block_line with empty |
| 509 | // text content. |
| 510 | if (BufferPtr != CommentEnd) { |
| 511 | const char C = *BufferPtr; |
| 512 | if (C == '\n' || C == '\r') { |
| 513 | BufferPtr = skipNewline(BufferPtr, CommentEnd); |
| 514 | State = LS_VerbatimBlockBody; |
| 515 | return; |
| 516 | } |
| 517 | } |
| 518 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 519 | State = LS_VerbatimBlockFirstLine; |
| 520 | } |
| 521 | |
| 522 | void Lexer::lexVerbatimBlockFirstLine(Token &T) { |
Dmitri Gribenko | 64da4e5 | 2012-07-18 23:01:58 +0000 | [diff] [blame] | 523 | again: |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 524 | assert(BufferPtr < CommentEnd); |
| 525 | |
| 526 | // FIXME: It would be better to scan the text once, finding either the block |
| 527 | // end command or newline. |
| 528 | // |
| 529 | // Extract current line. |
| 530 | const char *Newline = findNewline(BufferPtr, CommentEnd); |
| 531 | StringRef Line(BufferPtr, Newline - BufferPtr); |
| 532 | |
| 533 | // Look for end command in current line. |
| 534 | size_t Pos = Line.find(VerbatimBlockEndCommandName); |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 535 | const char *TextEnd; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 536 | const char *NextLine; |
| 537 | if (Pos == StringRef::npos) { |
| 538 | // Current line is completely verbatim. |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 539 | TextEnd = Newline; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 540 | NextLine = skipNewline(Newline, CommentEnd); |
| 541 | } else if (Pos == 0) { |
| 542 | // Current line contains just an end command. |
| 543 | const char *End = BufferPtr + VerbatimBlockEndCommandName.size(); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 544 | StringRef Name(BufferPtr + 1, End - (BufferPtr + 1)); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 545 | formTokenWithChars(T, End, tok::verbatim_block_end); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 546 | T.setVerbatimBlockName(Name); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 547 | State = LS_Normal; |
| 548 | return; |
| 549 | } else { |
| 550 | // There is some text, followed by end command. Extract text first. |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 551 | TextEnd = BufferPtr + Pos; |
| 552 | NextLine = TextEnd; |
Dmitri Gribenko | 64da4e5 | 2012-07-18 23:01:58 +0000 | [diff] [blame] | 553 | // If there is only whitespace before end command, skip whitespace. |
| 554 | if (isWhitespace(BufferPtr, TextEnd)) { |
| 555 | BufferPtr = TextEnd; |
| 556 | goto again; |
| 557 | } |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 558 | } |
| 559 | |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 560 | StringRef Text(BufferPtr, TextEnd - BufferPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 561 | formTokenWithChars(T, NextLine, tok::verbatim_block_line); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 562 | T.setVerbatimBlockText(Text); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 563 | |
| 564 | State = LS_VerbatimBlockBody; |
| 565 | } |
| 566 | |
| 567 | void Lexer::lexVerbatimBlockBody(Token &T) { |
| 568 | assert(State == LS_VerbatimBlockBody); |
| 569 | |
| 570 | if (CommentState == LCS_InsideCComment) |
| 571 | skipLineStartingDecorations(); |
| 572 | |
| 573 | lexVerbatimBlockFirstLine(T); |
| 574 | } |
| 575 | |
Dmitri Gribenko | 962668d | 2012-06-27 16:53:58 +0000 | [diff] [blame] | 576 | void Lexer::setupAndLexVerbatimLine(Token &T, const char *TextBegin) { |
| 577 | const StringRef Name(BufferPtr + 1, TextBegin - BufferPtr - 1); |
| 578 | formTokenWithChars(T, TextBegin, tok::verbatim_line_name); |
| 579 | T.setVerbatimLineName(Name); |
| 580 | |
| 581 | State = LS_VerbatimLineText; |
| 582 | } |
| 583 | |
| 584 | void Lexer::lexVerbatimLineText(Token &T) { |
| 585 | assert(State == LS_VerbatimLineText); |
| 586 | |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 587 | // Extract current line. |
| 588 | const char *Newline = findNewline(BufferPtr, CommentEnd); |
Dmitri Gribenko | 962668d | 2012-06-27 16:53:58 +0000 | [diff] [blame] | 589 | const StringRef Text(BufferPtr, Newline - BufferPtr); |
| 590 | formTokenWithChars(T, Newline, tok::verbatim_line_text); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 591 | T.setVerbatimLineText(Text); |
Dmitri Gribenko | 962668d | 2012-06-27 16:53:58 +0000 | [diff] [blame] | 592 | |
| 593 | State = LS_Normal; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 594 | } |
| 595 | |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 596 | void Lexer::lexHTMLCharacterReference(Token &T) { |
| 597 | const char *TokenPtr = BufferPtr; |
| 598 | assert(*TokenPtr == '&'); |
| 599 | TokenPtr++; |
| 600 | if (TokenPtr == CommentEnd) { |
| 601 | formTextToken(T, TokenPtr); |
| 602 | return; |
| 603 | } |
| 604 | const char *NamePtr; |
| 605 | bool isNamed = false; |
| 606 | bool isDecimal = false; |
| 607 | char C = *TokenPtr; |
| 608 | if (isHTMLNamedCharacterReferenceCharacter(C)) { |
| 609 | NamePtr = TokenPtr; |
| 610 | TokenPtr = skipNamedCharacterReference(TokenPtr, CommentEnd); |
| 611 | isNamed = true; |
| 612 | } else if (C == '#') { |
| 613 | TokenPtr++; |
| 614 | if (TokenPtr == CommentEnd) { |
| 615 | formTextToken(T, TokenPtr); |
| 616 | return; |
| 617 | } |
| 618 | C = *TokenPtr; |
| 619 | if (isHTMLDecimalCharacterReferenceCharacter(C)) { |
| 620 | NamePtr = TokenPtr; |
| 621 | TokenPtr = skipDecimalCharacterReference(TokenPtr, CommentEnd); |
| 622 | isDecimal = true; |
| 623 | } else if (C == 'x' || C == 'X') { |
| 624 | TokenPtr++; |
| 625 | NamePtr = TokenPtr; |
| 626 | TokenPtr = skipHexCharacterReference(TokenPtr, CommentEnd); |
| 627 | } else { |
| 628 | formTextToken(T, TokenPtr); |
| 629 | return; |
| 630 | } |
| 631 | } else { |
| 632 | formTextToken(T, TokenPtr); |
| 633 | return; |
| 634 | } |
| 635 | if (NamePtr == TokenPtr || TokenPtr == CommentEnd || |
| 636 | *TokenPtr != ';') { |
| 637 | formTextToken(T, TokenPtr); |
| 638 | return; |
| 639 | } |
| 640 | StringRef Name(NamePtr, TokenPtr - NamePtr); |
| 641 | TokenPtr++; // Skip semicolon. |
| 642 | StringRef Resolved; |
| 643 | if (isNamed) |
| 644 | Resolved = resolveHTMLNamedCharacterReference(Name); |
| 645 | else if (isDecimal) |
| 646 | Resolved = resolveHTMLDecimalCharacterReference(Name); |
| 647 | else |
| 648 | Resolved = resolveHTMLHexCharacterReference(Name); |
| 649 | |
| 650 | if (Resolved.empty()) { |
| 651 | formTextToken(T, TokenPtr); |
| 652 | return; |
| 653 | } |
| 654 | formTokenWithChars(T, TokenPtr, tok::text); |
| 655 | T.setText(Resolved); |
| 656 | return; |
| 657 | } |
| 658 | |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 659 | void Lexer::setupAndLexHTMLStartTag(Token &T) { |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 660 | assert(BufferPtr[0] == '<' && |
| 661 | isHTMLIdentifierStartingCharacter(BufferPtr[1])); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 662 | const char *TagNameEnd = skipHTMLIdentifier(BufferPtr + 2, CommentEnd); |
| 663 | |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 664 | StringRef Name(BufferPtr + 1, TagNameEnd - (BufferPtr + 1)); |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 665 | formTokenWithChars(T, TagNameEnd, tok::html_start_tag); |
| 666 | T.setHTMLTagStartName(Name); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 667 | |
| 668 | BufferPtr = skipWhitespace(BufferPtr, CommentEnd); |
| 669 | |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 670 | const char C = *BufferPtr; |
| 671 | if (BufferPtr != CommentEnd && |
Dmitri Gribenko | a5ef44f | 2012-07-11 21:38:39 +0000 | [diff] [blame] | 672 | (C == '>' || C == '/' || isHTMLIdentifierStartingCharacter(C))) |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 673 | State = LS_HTMLStartTag; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 674 | } |
| 675 | |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 676 | void Lexer::lexHTMLStartTag(Token &T) { |
| 677 | assert(State == LS_HTMLStartTag); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 678 | |
| 679 | const char *TokenPtr = BufferPtr; |
| 680 | char C = *TokenPtr; |
| 681 | if (isHTMLIdentifierCharacter(C)) { |
| 682 | TokenPtr = skipHTMLIdentifier(TokenPtr, CommentEnd); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 683 | StringRef Ident(BufferPtr, TokenPtr - BufferPtr); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 684 | formTokenWithChars(T, TokenPtr, tok::html_ident); |
Dmitri Gribenko | f5e0aea | 2012-06-27 16:30:35 +0000 | [diff] [blame] | 685 | T.setHTMLIdent(Ident); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 686 | } else { |
| 687 | switch (C) { |
| 688 | case '=': |
| 689 | TokenPtr++; |
| 690 | formTokenWithChars(T, TokenPtr, tok::html_equals); |
| 691 | break; |
| 692 | case '\"': |
| 693 | case '\'': { |
| 694 | const char *OpenQuote = TokenPtr; |
| 695 | TokenPtr = skipHTMLQuotedString(TokenPtr, CommentEnd); |
| 696 | const char *ClosingQuote = TokenPtr; |
| 697 | if (TokenPtr != CommentEnd) // Skip closing quote. |
| 698 | TokenPtr++; |
| 699 | formTokenWithChars(T, TokenPtr, tok::html_quoted_string); |
| 700 | T.setHTMLQuotedString(StringRef(OpenQuote + 1, |
| 701 | ClosingQuote - (OpenQuote + 1))); |
| 702 | break; |
| 703 | } |
| 704 | case '>': |
| 705 | TokenPtr++; |
| 706 | formTokenWithChars(T, TokenPtr, tok::html_greater); |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 707 | State = LS_Normal; |
| 708 | return; |
Dmitri Gribenko | a5ef44f | 2012-07-11 21:38:39 +0000 | [diff] [blame] | 709 | case '/': |
| 710 | TokenPtr++; |
| 711 | if (TokenPtr != CommentEnd && *TokenPtr == '>') { |
| 712 | TokenPtr++; |
| 713 | formTokenWithChars(T, TokenPtr, tok::html_slash_greater); |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 714 | } else |
| 715 | formTextToken(T, TokenPtr); |
| 716 | |
Dmitri Gribenko | a5ef44f | 2012-07-11 21:38:39 +0000 | [diff] [blame] | 717 | State = LS_Normal; |
| 718 | return; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 719 | } |
| 720 | } |
| 721 | |
| 722 | // Now look ahead and return to normal state if we don't see any HTML tokens |
| 723 | // ahead. |
| 724 | BufferPtr = skipWhitespace(BufferPtr, CommentEnd); |
| 725 | if (BufferPtr == CommentEnd) { |
| 726 | State = LS_Normal; |
| 727 | return; |
| 728 | } |
| 729 | |
| 730 | C = *BufferPtr; |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 731 | if (!isHTMLIdentifierStartingCharacter(C) && |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 732 | C != '=' && C != '\"' && C != '\'' && C != '>') { |
| 733 | State = LS_Normal; |
| 734 | return; |
| 735 | } |
| 736 | } |
| 737 | |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 738 | void Lexer::setupAndLexHTMLEndTag(Token &T) { |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 739 | assert(BufferPtr[0] == '<' && BufferPtr[1] == '/'); |
| 740 | |
| 741 | const char *TagNameBegin = skipWhitespace(BufferPtr + 2, CommentEnd); |
| 742 | const char *TagNameEnd = skipHTMLIdentifier(TagNameBegin, CommentEnd); |
| 743 | |
| 744 | const char *End = skipWhitespace(TagNameEnd, CommentEnd); |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 745 | |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 746 | formTokenWithChars(T, End, tok::html_end_tag); |
| 747 | T.setHTMLTagEndName(StringRef(TagNameBegin, TagNameEnd - TagNameBegin)); |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 748 | |
| 749 | if (BufferPtr != CommentEnd && *BufferPtr == '>') |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 750 | State = LS_HTMLEndTag; |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 751 | } |
| 752 | |
Dmitri Gribenko | 3f38bf2 | 2012-07-13 00:44:24 +0000 | [diff] [blame] | 753 | void Lexer::lexHTMLEndTag(Token &T) { |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 754 | assert(BufferPtr != CommentEnd && *BufferPtr == '>'); |
| 755 | |
| 756 | formTokenWithChars(T, BufferPtr + 1, tok::html_greater); |
| 757 | State = LS_Normal; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 758 | } |
| 759 | |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 760 | Lexer::Lexer(llvm::BumpPtrAllocator &Allocator, |
| 761 | SourceLocation FileLoc, const CommentOptions &CommOpts, |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 762 | const char *BufferStart, const char *BufferEnd): |
Dmitri Gribenko | 477a9f5 | 2012-07-27 20:37:06 +0000 | [diff] [blame^] | 763 | Allocator(Allocator), |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 764 | BufferStart(BufferStart), BufferEnd(BufferEnd), |
| 765 | FileLoc(FileLoc), CommOpts(CommOpts), BufferPtr(BufferStart), |
| 766 | CommentState(LCS_BeforeComment), State(LS_Normal) { |
| 767 | } |
| 768 | |
| 769 | void Lexer::lex(Token &T) { |
| 770 | again: |
| 771 | switch (CommentState) { |
| 772 | case LCS_BeforeComment: |
| 773 | if (BufferPtr == BufferEnd) { |
| 774 | formTokenWithChars(T, BufferPtr, tok::eof); |
| 775 | return; |
| 776 | } |
| 777 | |
| 778 | assert(*BufferPtr == '/'); |
| 779 | BufferPtr++; // Skip first slash. |
| 780 | switch(*BufferPtr) { |
| 781 | case '/': { // BCPL comment. |
| 782 | BufferPtr++; // Skip second slash. |
| 783 | |
| 784 | if (BufferPtr != BufferEnd) { |
| 785 | // Skip Doxygen magic marker, if it is present. |
| 786 | // It might be missing because of a typo //< or /*<, or because we |
| 787 | // merged this non-Doxygen comment into a bunch of Doxygen comments |
| 788 | // around it: /** ... */ /* ... */ /** ... */ |
| 789 | const char C = *BufferPtr; |
| 790 | if (C == '/' || C == '!') |
| 791 | BufferPtr++; |
| 792 | } |
| 793 | |
| 794 | // Skip less-than symbol that marks trailing comments. |
| 795 | // Skip it even if the comment is not a Doxygen one, because //< and /*< |
| 796 | // are frequent typos. |
| 797 | if (BufferPtr != BufferEnd && *BufferPtr == '<') |
| 798 | BufferPtr++; |
| 799 | |
| 800 | CommentState = LCS_InsideBCPLComment; |
Dmitri Gribenko | 8d3ba23 | 2012-07-06 00:28:32 +0000 | [diff] [blame] | 801 | if (State != LS_VerbatimBlockBody && State != LS_VerbatimBlockFirstLine) |
| 802 | State = LS_Normal; |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 803 | CommentEnd = findBCPLCommentEnd(BufferPtr, BufferEnd); |
| 804 | goto again; |
| 805 | } |
| 806 | case '*': { // C comment. |
| 807 | BufferPtr++; // Skip star. |
| 808 | |
| 809 | // Skip Doxygen magic marker. |
| 810 | const char C = *BufferPtr; |
| 811 | if ((C == '*' && *(BufferPtr + 1) != '/') || C == '!') |
| 812 | BufferPtr++; |
| 813 | |
| 814 | // Skip less-than symbol that marks trailing comments. |
| 815 | if (BufferPtr != BufferEnd && *BufferPtr == '<') |
| 816 | BufferPtr++; |
| 817 | |
| 818 | CommentState = LCS_InsideCComment; |
| 819 | State = LS_Normal; |
| 820 | CommentEnd = findCCommentEnd(BufferPtr, BufferEnd); |
| 821 | goto again; |
| 822 | } |
| 823 | default: |
| 824 | llvm_unreachable("second character of comment should be '/' or '*'"); |
| 825 | } |
| 826 | |
| 827 | case LCS_BetweenComments: { |
| 828 | // Consecutive comments are extracted only if there is only whitespace |
| 829 | // between them. So we can search for the start of the next comment. |
| 830 | const char *EndWhitespace = BufferPtr; |
| 831 | while(EndWhitespace != BufferEnd && *EndWhitespace != '/') |
| 832 | EndWhitespace++; |
| 833 | |
| 834 | // Turn any whitespace between comments (and there is only whitespace |
Dmitri Gribenko | a99ec10 | 2012-07-09 21:32:40 +0000 | [diff] [blame] | 835 | // between them -- guaranteed by comment extraction) into a newline. We |
| 836 | // have two newlines between C comments in total (first one was synthesized |
| 837 | // after a comment). |
Dmitri Gribenko | 2d44d77 | 2012-06-26 20:39:18 +0000 | [diff] [blame] | 838 | formTokenWithChars(T, EndWhitespace, tok::newline); |
| 839 | |
| 840 | CommentState = LCS_BeforeComment; |
| 841 | break; |
| 842 | } |
| 843 | |
| 844 | case LCS_InsideBCPLComment: |
| 845 | case LCS_InsideCComment: |
| 846 | if (BufferPtr != CommentEnd) { |
| 847 | lexCommentText(T); |
| 848 | break; |
| 849 | } else { |
| 850 | // Skip C comment closing sequence. |
| 851 | if (CommentState == LCS_InsideCComment) { |
| 852 | assert(BufferPtr[0] == '*' && BufferPtr[1] == '/'); |
| 853 | BufferPtr += 2; |
| 854 | assert(BufferPtr <= BufferEnd); |
| 855 | |
| 856 | // Synthenize newline just after the C comment, regardless if there is |
| 857 | // actually a newline. |
| 858 | formTokenWithChars(T, BufferPtr, tok::newline); |
| 859 | |
| 860 | CommentState = LCS_BetweenComments; |
| 861 | break; |
| 862 | } else { |
| 863 | // Don't synthesized a newline after BCPL comment. |
| 864 | CommentState = LCS_BetweenComments; |
| 865 | goto again; |
| 866 | } |
| 867 | } |
| 868 | } |
| 869 | } |
| 870 | |
| 871 | StringRef Lexer::getSpelling(const Token &Tok, |
| 872 | const SourceManager &SourceMgr, |
| 873 | bool *Invalid) const { |
| 874 | SourceLocation Loc = Tok.getLocation(); |
| 875 | std::pair<FileID, unsigned> LocInfo = SourceMgr.getDecomposedLoc(Loc); |
| 876 | |
| 877 | bool InvalidTemp = false; |
| 878 | StringRef File = SourceMgr.getBufferData(LocInfo.first, &InvalidTemp); |
| 879 | if (InvalidTemp) { |
| 880 | *Invalid = true; |
| 881 | return StringRef(); |
| 882 | } |
| 883 | |
| 884 | const char *Begin = File.data() + LocInfo.second; |
| 885 | return StringRef(Begin, Tok.getLength()); |
| 886 | } |
| 887 | |
| 888 | void Lexer::addVerbatimBlockCommand(StringRef BeginName, StringRef EndName) { |
| 889 | VerbatimBlockCommand VBC; |
| 890 | VBC.BeginName = BeginName; |
| 891 | VBC.EndName = EndName; |
| 892 | VerbatimBlockCommands.push_back(VBC); |
| 893 | } |
| 894 | |
| 895 | void Lexer::addVerbatimLineCommand(StringRef Name) { |
| 896 | VerbatimLineCommand VLC; |
| 897 | VLC.Name = Name; |
| 898 | VerbatimLineCommands.push_back(VLC); |
| 899 | } |
| 900 | |
| 901 | } // end namespace comments |
| 902 | } // end namespace clang |
| 903 | |