| OLD | NEW |
|---|
| (Empty) |
| 1 // Copyright 2011 the V8 project authors. All rights reserved. | |
| 2 // Use of this source code is governed by a BSD-style license that can be | |
| 3 // found in the LICENSE file. | |
| 4 | |
| 5 // Features shared by parsing and pre-parsing scanners. | |
| 6 | |
| 7 #ifndef V8_SCANNER_H_ | |
| 8 #define V8_SCANNER_H_ | |
| 9 | |
| 10 #include "src/allocation.h" | |
| 11 #include "src/base/logging.h" | |
| 12 #include "src/char-predicates.h" | |
| 13 #include "src/globals.h" | |
| 14 #include "src/hashmap.h" | |
| 15 #include "src/list.h" | |
| 16 #include "src/token.h" | |
| 17 #include "src/unicode.h" | |
| 18 #include "src/unicode-decoder.h" | |
| 19 #include "src/utils.h" | |
| 20 | |
| 21 namespace v8 { | |
| 22 namespace internal { | |
| 23 | |
| 24 | |
| 25 class AstRawString; | |
| 26 class AstValueFactory; | |
| 27 class ParserRecorder; | |
| 28 class UnicodeCache; | |
| 29 | |
| 30 | |
| 31 // Returns the value (0 .. 15) of a hexadecimal character c. | |
| 32 // If c is not a legal hexadecimal character, returns a value < 0. | |
| 33 inline int HexValue(uc32 c) { | |
| 34 c -= '0'; | |
| 35 if (static_cast<unsigned>(c) <= 9) return c; | |
| 36 c = (c | 0x20) - ('a' - '0'); // detect 0x11..0x16 and 0x31..0x36. | |
| 37 if (static_cast<unsigned>(c) <= 5) return c + 10; | |
| 38 return -1; | |
| 39 } | |
| 40 | |
| 41 | |
| 42 // --------------------------------------------------------------------- | |
| 43 // Buffered stream of UTF-16 code units, using an internal UTF-16 buffer. | |
| 44 // A code unit is a 16 bit value representing either a 16 bit code point | |
| 45 // or one part of a surrogate pair that make a single 21 bit code point. | |
| 46 | |
| 47 class Utf16CharacterStream { | |
| 48 public: | |
| 49 Utf16CharacterStream() : pos_(0) { } | |
| 50 virtual ~Utf16CharacterStream() { } | |
| 51 | |
| 52 // Returns and advances past the next UTF-16 code unit in the input | |
| 53 // stream. If there are no more code units, it returns a negative | |
| 54 // value. | |
| 55 inline uc32 Advance() { | |
| 56 if (buffer_cursor_ < buffer_end_ || ReadBlock()) { | |
| 57 pos_++; | |
| 58 return static_cast<uc32>(*(buffer_cursor_++)); | |
| 59 } | |
| 60 // Note: currently the following increment is necessary to avoid a | |
| 61 // parser problem! The scanner treats the final kEndOfInput as | |
| 62 // a code unit with a position, and does math relative to that | |
| 63 // position. | |
| 64 pos_++; | |
| 65 | |
| 66 return kEndOfInput; | |
| 67 } | |
| 68 | |
| 69 // Return the current position in the code unit stream. | |
| 70 // Starts at zero. | |
| 71 inline size_t pos() const { return pos_; } | |
| 72 | |
| 73 // Skips forward past the next code_unit_count UTF-16 code units | |
| 74 // in the input, or until the end of input if that comes sooner. | |
| 75 // Returns the number of code units actually skipped. If less | |
| 76 // than code_unit_count, | |
| 77 inline size_t SeekForward(size_t code_unit_count) { | |
| 78 size_t buffered_chars = buffer_end_ - buffer_cursor_; | |
| 79 if (code_unit_count <= buffered_chars) { | |
| 80 buffer_cursor_ += code_unit_count; | |
| 81 pos_ += code_unit_count; | |
| 82 return code_unit_count; | |
| 83 } | |
| 84 return SlowSeekForward(code_unit_count); | |
| 85 } | |
| 86 | |
| 87 // Pushes back the most recently read UTF-16 code unit (or negative | |
| 88 // value if at end of input), i.e., the value returned by the most recent | |
| 89 // call to Advance. | |
| 90 // Must not be used right after calling SeekForward. | |
| 91 virtual void PushBack(int32_t code_unit) = 0; | |
| 92 | |
| 93 virtual bool SetBookmark(); | |
| 94 virtual void ResetToBookmark(); | |
| 95 | |
| 96 protected: | |
| 97 static const uc32 kEndOfInput = -1; | |
| 98 | |
| 99 // Ensures that the buffer_cursor_ points to the code_unit at | |
| 100 // position pos_ of the input, if possible. If the position | |
| 101 // is at or after the end of the input, return false. If there | |
| 102 // are more code_units available, return true. | |
| 103 virtual bool ReadBlock() = 0; | |
| 104 virtual size_t SlowSeekForward(size_t code_unit_count) = 0; | |
| 105 | |
| 106 const uint16_t* buffer_cursor_; | |
| 107 const uint16_t* buffer_end_; | |
| 108 size_t pos_; | |
| 109 }; | |
| 110 | |
| 111 | |
| 112 // --------------------------------------------------------------------- | |
| 113 // DuplicateFinder discovers duplicate symbols. | |
| 114 | |
| 115 class DuplicateFinder { | |
| 116 public: | |
| 117 explicit DuplicateFinder(UnicodeCache* constants) | |
| 118 : unicode_constants_(constants), | |
| 119 backing_store_(16), | |
| 120 map_(&Match) { } | |
| 121 | |
| 122 int AddOneByteSymbol(Vector<const uint8_t> key, int value); | |
| 123 int AddTwoByteSymbol(Vector<const uint16_t> key, int value); | |
| 124 // Add a a number literal by converting it (if necessary) | |
| 125 // to the string that ToString(ToNumber(literal)) would generate. | |
| 126 // and then adding that string with AddOneByteSymbol. | |
| 127 // This string is the actual value used as key in an object literal, | |
| 128 // and the one that must be different from the other keys. | |
| 129 int AddNumber(Vector<const uint8_t> key, int value); | |
| 130 | |
| 131 private: | |
| 132 int AddSymbol(Vector<const uint8_t> key, bool is_one_byte, int value); | |
| 133 // Backs up the key and its length in the backing store. | |
| 134 // The backup is stored with a base 127 encoding of the | |
| 135 // length (plus a bit saying whether the string is one byte), | |
| 136 // followed by the bytes of the key. | |
| 137 uint8_t* BackupKey(Vector<const uint8_t> key, bool is_one_byte); | |
| 138 | |
| 139 // Compare two encoded keys (both pointing into the backing store) | |
| 140 // for having the same base-127 encoded lengths and representation. | |
| 141 // and then having the same 'length' bytes following. | |
| 142 static bool Match(void* first, void* second); | |
| 143 // Creates a hash from a sequence of bytes. | |
| 144 static uint32_t Hash(Vector<const uint8_t> key, bool is_one_byte); | |
| 145 // Checks whether a string containing a JS number is its canonical | |
| 146 // form. | |
| 147 static bool IsNumberCanonical(Vector<const uint8_t> key); | |
| 148 | |
| 149 // Size of buffer. Sufficient for using it to call DoubleToCString in | |
| 150 // from conversions.h. | |
| 151 static const int kBufferSize = 100; | |
| 152 | |
| 153 UnicodeCache* unicode_constants_; | |
| 154 // Backing store used to store strings used as hashmap keys. | |
| 155 SequenceCollector<unsigned char> backing_store_; | |
| 156 HashMap map_; | |
| 157 // Buffer used for string->number->canonical string conversions. | |
| 158 char number_buffer_[kBufferSize]; | |
| 159 }; | |
| 160 | |
| 161 | |
| 162 // ---------------------------------------------------------------------------- | |
| 163 // LiteralBuffer - Collector of chars of literals. | |
| 164 | |
| 165 class LiteralBuffer { | |
| 166 public: | |
| 167 LiteralBuffer() : is_one_byte_(true), position_(0), backing_store_() { } | |
| 168 | |
| 169 ~LiteralBuffer() { backing_store_.Dispose(); } | |
| 170 | |
| 171 INLINE(void AddChar(uint32_t code_unit)) { | |
| 172 if (position_ >= backing_store_.length()) ExpandBuffer(); | |
| 173 if (is_one_byte_) { | |
| 174 if (code_unit <= unibrow::Latin1::kMaxChar) { | |
| 175 backing_store_[position_] = static_cast<byte>(code_unit); | |
| 176 position_ += kOneByteSize; | |
| 177 return; | |
| 178 } | |
| 179 ConvertToTwoByte(); | |
| 180 } | |
| 181 if (code_unit <= unibrow::Utf16::kMaxNonSurrogateCharCode) { | |
| 182 *reinterpret_cast<uint16_t*>(&backing_store_[position_]) = code_unit; | |
| 183 position_ += kUC16Size; | |
| 184 } else { | |
| 185 *reinterpret_cast<uint16_t*>(&backing_store_[position_]) = | |
| 186 unibrow::Utf16::LeadSurrogate(code_unit); | |
| 187 position_ += kUC16Size; | |
| 188 if (position_ >= backing_store_.length()) ExpandBuffer(); | |
| 189 *reinterpret_cast<uint16_t*>(&backing_store_[position_]) = | |
| 190 unibrow::Utf16::TrailSurrogate(code_unit); | |
| 191 position_ += kUC16Size; | |
| 192 } | |
| 193 } | |
| 194 | |
| 195 bool is_one_byte() const { return is_one_byte_; } | |
| 196 | |
| 197 bool is_contextual_keyword(Vector<const char> keyword) const { | |
| 198 return is_one_byte() && keyword.length() == position_ && | |
| 199 (memcmp(keyword.start(), backing_store_.start(), position_) == 0); | |
| 200 } | |
| 201 | |
| 202 Vector<const uint16_t> two_byte_literal() const { | |
| 203 DCHECK(!is_one_byte_); | |
| 204 DCHECK((position_ & 0x1) == 0); | |
| 205 return Vector<const uint16_t>( | |
| 206 reinterpret_cast<const uint16_t*>(backing_store_.start()), | |
| 207 position_ >> 1); | |
| 208 } | |
| 209 | |
| 210 Vector<const uint8_t> one_byte_literal() const { | |
| 211 DCHECK(is_one_byte_); | |
| 212 return Vector<const uint8_t>( | |
| 213 reinterpret_cast<const uint8_t*>(backing_store_.start()), | |
| 214 position_); | |
| 215 } | |
| 216 | |
| 217 int length() const { | |
| 218 return is_one_byte_ ? position_ : (position_ >> 1); | |
| 219 } | |
| 220 | |
| 221 void ReduceLength(int delta) { | |
| 222 position_ -= delta * (is_one_byte_ ? kOneByteSize : kUC16Size); | |
| 223 } | |
| 224 | |
| 225 void Reset() { | |
| 226 position_ = 0; | |
| 227 is_one_byte_ = true; | |
| 228 } | |
| 229 | |
| 230 Handle<String> Internalize(Isolate* isolate) const; | |
| 231 | |
| 232 void CopyFrom(const LiteralBuffer* other) { | |
| 233 if (other == nullptr) { | |
| 234 Reset(); | |
| 235 } else { | |
| 236 is_one_byte_ = other->is_one_byte_; | |
| 237 position_ = other->position_; | |
| 238 backing_store_.Dispose(); | |
| 239 backing_store_ = other->backing_store_.Clone(); | |
| 240 } | |
| 241 } | |
| 242 | |
| 243 private: | |
| 244 static const int kInitialCapacity = 16; | |
| 245 static const int kGrowthFactory = 4; | |
| 246 static const int kMinConversionSlack = 256; | |
| 247 static const int kMaxGrowth = 1 * MB; | |
| 248 inline int NewCapacity(int min_capacity) { | |
| 249 int capacity = Max(min_capacity, backing_store_.length()); | |
| 250 int new_capacity = Min(capacity * kGrowthFactory, capacity + kMaxGrowth); | |
| 251 return new_capacity; | |
| 252 } | |
| 253 | |
| 254 void ExpandBuffer() { | |
| 255 Vector<byte> new_store = Vector<byte>::New(NewCapacity(kInitialCapacity)); | |
| 256 MemCopy(new_store.start(), backing_store_.start(), position_); | |
| 257 backing_store_.Dispose(); | |
| 258 backing_store_ = new_store; | |
| 259 } | |
| 260 | |
| 261 void ConvertToTwoByte() { | |
| 262 DCHECK(is_one_byte_); | |
| 263 Vector<byte> new_store; | |
| 264 int new_content_size = position_ * kUC16Size; | |
| 265 if (new_content_size >= backing_store_.length()) { | |
| 266 // Ensure room for all currently read code units as UC16 as well | |
| 267 // as the code unit about to be stored. | |
| 268 new_store = Vector<byte>::New(NewCapacity(new_content_size)); | |
| 269 } else { | |
| 270 new_store = backing_store_; | |
| 271 } | |
| 272 uint8_t* src = backing_store_.start(); | |
| 273 uint16_t* dst = reinterpret_cast<uint16_t*>(new_store.start()); | |
| 274 for (int i = position_ - 1; i >= 0; i--) { | |
| 275 dst[i] = src[i]; | |
| 276 } | |
| 277 if (new_store.start() != backing_store_.start()) { | |
| 278 backing_store_.Dispose(); | |
| 279 backing_store_ = new_store; | |
| 280 } | |
| 281 position_ = new_content_size; | |
| 282 is_one_byte_ = false; | |
| 283 } | |
| 284 | |
| 285 bool is_one_byte_; | |
| 286 int position_; | |
| 287 Vector<byte> backing_store_; | |
| 288 | |
| 289 DISALLOW_COPY_AND_ASSIGN(LiteralBuffer); | |
| 290 }; | |
| 291 | |
| 292 | |
| 293 // ---------------------------------------------------------------------------- | |
| 294 // JavaScript Scanner. | |
| 295 | |
| 296 class Scanner { | |
| 297 public: | |
| 298 // Scoped helper for literal recording. Automatically drops the literal | |
| 299 // if aborting the scanning before it's complete. | |
| 300 class LiteralScope { | |
| 301 public: | |
| 302 explicit LiteralScope(Scanner* self) : scanner_(self), complete_(false) { | |
| 303 scanner_->StartLiteral(); | |
| 304 } | |
| 305 ~LiteralScope() { | |
| 306 if (!complete_) scanner_->DropLiteral(); | |
| 307 } | |
| 308 void Complete() { | |
| 309 complete_ = true; | |
| 310 } | |
| 311 | |
| 312 private: | |
| 313 Scanner* scanner_; | |
| 314 bool complete_; | |
| 315 }; | |
| 316 | |
| 317 // Scoped helper for a re-settable bookmark. | |
| 318 class BookmarkScope { | |
| 319 public: | |
| 320 explicit BookmarkScope(Scanner* scanner) : scanner_(scanner) { | |
| 321 DCHECK_NOT_NULL(scanner_); | |
| 322 } | |
| 323 ~BookmarkScope() { scanner_->DropBookmark(); } | |
| 324 | |
| 325 bool Set() { return scanner_->SetBookmark(); } | |
| 326 void Reset() { scanner_->ResetToBookmark(); } | |
| 327 bool HasBeenSet() { return scanner_->BookmarkHasBeenSet(); } | |
| 328 bool HasBeenReset() { return scanner_->BookmarkHasBeenReset(); } | |
| 329 | |
| 330 private: | |
| 331 Scanner* scanner_; | |
| 332 | |
| 333 DISALLOW_COPY_AND_ASSIGN(BookmarkScope); | |
| 334 }; | |
| 335 | |
| 336 // Representation of an interval of source positions. | |
| 337 struct Location { | |
| 338 Location(int b, int e) : beg_pos(b), end_pos(e) { } | |
| 339 Location() : beg_pos(0), end_pos(0) { } | |
| 340 | |
| 341 bool IsValid() const { | |
| 342 return beg_pos >= 0 && end_pos >= beg_pos; | |
| 343 } | |
| 344 | |
| 345 static Location invalid() { return Location(-1, -1); } | |
| 346 | |
| 347 int beg_pos; | |
| 348 int end_pos; | |
| 349 }; | |
| 350 | |
| 351 // -1 is outside of the range of any real source code. | |
| 352 static const int kNoOctalLocation = -1; | |
| 353 | |
| 354 explicit Scanner(UnicodeCache* scanner_contants); | |
| 355 | |
| 356 void Initialize(Utf16CharacterStream* source); | |
| 357 | |
| 358 // Returns the next token and advances input. | |
| 359 Token::Value Next(); | |
| 360 // Returns the token following peek() | |
| 361 Token::Value PeekAhead(); | |
| 362 // Returns the current token again. | |
| 363 Token::Value current_token() { return current_.token; } | |
| 364 // Returns the location information for the current token | |
| 365 // (the token last returned by Next()). | |
| 366 Location location() const { return current_.location; } | |
| 367 | |
| 368 // Similar functions for the upcoming token. | |
| 369 | |
| 370 // One token look-ahead (past the token returned by Next()). | |
| 371 Token::Value peek() const { return next_.token; } | |
| 372 | |
| 373 Location peek_location() const { return next_.location; } | |
| 374 | |
| 375 bool literal_contains_escapes() const { | |
| 376 return LiteralContainsEscapes(current_); | |
| 377 } | |
| 378 bool next_literal_contains_escapes() const { | |
| 379 return LiteralContainsEscapes(next_); | |
| 380 } | |
| 381 bool is_literal_contextual_keyword(Vector<const char> keyword) { | |
| 382 DCHECK_NOT_NULL(current_.literal_chars); | |
| 383 return current_.literal_chars->is_contextual_keyword(keyword); | |
| 384 } | |
| 385 bool is_next_contextual_keyword(Vector<const char> keyword) { | |
| 386 DCHECK_NOT_NULL(next_.literal_chars); | |
| 387 return next_.literal_chars->is_contextual_keyword(keyword); | |
| 388 } | |
| 389 | |
| 390 const AstRawString* CurrentSymbol(AstValueFactory* ast_value_factory); | |
| 391 const AstRawString* NextSymbol(AstValueFactory* ast_value_factory); | |
| 392 const AstRawString* CurrentRawSymbol(AstValueFactory* ast_value_factory); | |
| 393 | |
| 394 double DoubleValue(); | |
| 395 bool ContainsDot(); | |
| 396 bool LiteralMatches(const char* data, int length, bool allow_escapes = true) { | |
| 397 if (is_literal_one_byte() && | |
| 398 literal_length() == length && | |
| 399 (allow_escapes || !literal_contains_escapes())) { | |
| 400 const char* token = | |
| 401 reinterpret_cast<const char*>(literal_one_byte_string().start()); | |
| 402 return !strncmp(token, data, length); | |
| 403 } | |
| 404 return false; | |
| 405 } | |
| 406 inline bool UnescapedLiteralMatches(const char* data, int length) { | |
| 407 return LiteralMatches(data, length, false); | |
| 408 } | |
| 409 | |
| 410 void IsGetOrSet(bool* is_get, bool* is_set) { | |
| 411 if (is_literal_one_byte() && | |
| 412 literal_length() == 3 && | |
| 413 !literal_contains_escapes()) { | |
| 414 const char* token = | |
| 415 reinterpret_cast<const char*>(literal_one_byte_string().start()); | |
| 416 *is_get = strncmp(token, "get", 3) == 0; | |
| 417 *is_set = !*is_get && strncmp(token, "set", 3) == 0; | |
| 418 } | |
| 419 } | |
| 420 | |
| 421 int FindSymbol(DuplicateFinder* finder, int value); | |
| 422 | |
| 423 UnicodeCache* unicode_cache() { return unicode_cache_; } | |
| 424 | |
| 425 // Returns the location of the last seen octal literal. | |
| 426 Location octal_position() const { return octal_pos_; } | |
| 427 void clear_octal_position() { octal_pos_ = Location::invalid(); } | |
| 428 | |
| 429 // Returns the value of the last smi that was scanned. | |
| 430 int smi_value() const { return current_.smi_value_; } | |
| 431 | |
| 432 // Seek forward to the given position. This operation does not | |
| 433 // work in general, for instance when there are pushed back | |
| 434 // characters, but works for seeking forward until simple delimiter | |
| 435 // tokens, which is what it is used for. | |
| 436 void SeekForward(int pos); | |
| 437 | |
| 438 // Returns true if there was a line terminator before the peek'ed token, | |
| 439 // possibly inside a multi-line comment. | |
| 440 bool HasAnyLineTerminatorBeforeNext() const { | |
| 441 return has_line_terminator_before_next_ || | |
| 442 has_multiline_comment_before_next_; | |
| 443 } | |
| 444 | |
| 445 // Scans the input as a regular expression pattern, previous | |
| 446 // character(s) must be /(=). Returns true if a pattern is scanned. | |
| 447 bool ScanRegExpPattern(bool seen_equal); | |
| 448 // Scans the input as regular expression flags. Returns the flags on success. | |
| 449 Maybe<RegExp::Flags> ScanRegExpFlags(); | |
| 450 | |
| 451 // Scans the input as a template literal | |
| 452 Token::Value ScanTemplateStart(); | |
| 453 Token::Value ScanTemplateContinuation(); | |
| 454 | |
| 455 const LiteralBuffer* source_url() const { return &source_url_; } | |
| 456 const LiteralBuffer* source_mapping_url() const { | |
| 457 return &source_mapping_url_; | |
| 458 } | |
| 459 | |
| 460 bool IdentifierIsFutureStrictReserved(const AstRawString* string) const; | |
| 461 | |
| 462 private: | |
| 463 // The current and look-ahead token. | |
| 464 struct TokenDesc { | |
| 465 Token::Value token; | |
| 466 Location location; | |
| 467 LiteralBuffer* literal_chars; | |
| 468 LiteralBuffer* raw_literal_chars; | |
| 469 int smi_value_; | |
| 470 }; | |
| 471 | |
| 472 static const int kCharacterLookaheadBufferSize = 1; | |
| 473 | |
| 474 // Scans octal escape sequence. Also accepts "\0" decimal escape sequence. | |
| 475 template <bool capture_raw> | |
| 476 uc32 ScanOctalEscape(uc32 c, int length); | |
| 477 | |
| 478 // Call this after setting source_ to the input. | |
| 479 void Init() { | |
| 480 // Set c0_ (one character ahead) | |
| 481 STATIC_ASSERT(kCharacterLookaheadBufferSize == 1); | |
| 482 Advance(); | |
| 483 // Initialize current_ to not refer to a literal. | |
| 484 current_.literal_chars = NULL; | |
| 485 current_.raw_literal_chars = NULL; | |
| 486 next_next_.token = Token::UNINITIALIZED; | |
| 487 } | |
| 488 | |
| 489 // Support BookmarkScope functionality. | |
| 490 bool SetBookmark(); | |
| 491 void ResetToBookmark(); | |
| 492 bool BookmarkHasBeenSet(); | |
| 493 bool BookmarkHasBeenReset(); | |
| 494 void DropBookmark(); | |
| 495 static void CopyTokenDesc(TokenDesc* to, TokenDesc* from); | |
| 496 | |
| 497 // Literal buffer support | |
| 498 inline void StartLiteral() { | |
| 499 LiteralBuffer* free_buffer = | |
| 500 (current_.literal_chars == &literal_buffer0_) | |
| 501 ? &literal_buffer1_ | |
| 502 : (current_.literal_chars == &literal_buffer1_) ? &literal_buffer2_ | |
| 503 : &literal_buffer0_; | |
| 504 free_buffer->Reset(); | |
| 505 next_.literal_chars = free_buffer; | |
| 506 } | |
| 507 | |
| 508 inline void StartRawLiteral() { | |
| 509 LiteralBuffer* free_buffer = | |
| 510 (current_.raw_literal_chars == &raw_literal_buffer0_) | |
| 511 ? &raw_literal_buffer1_ | |
| 512 : (current_.raw_literal_chars == &raw_literal_buffer1_) | |
| 513 ? &raw_literal_buffer2_ | |
| 514 : &raw_literal_buffer0_; | |
| 515 free_buffer->Reset(); | |
| 516 next_.raw_literal_chars = free_buffer; | |
| 517 } | |
| 518 | |
| 519 INLINE(void AddLiteralChar(uc32 c)) { | |
| 520 DCHECK_NOT_NULL(next_.literal_chars); | |
| 521 next_.literal_chars->AddChar(c); | |
| 522 } | |
| 523 | |
| 524 INLINE(void AddRawLiteralChar(uc32 c)) { | |
| 525 DCHECK_NOT_NULL(next_.raw_literal_chars); | |
| 526 next_.raw_literal_chars->AddChar(c); | |
| 527 } | |
| 528 | |
| 529 INLINE(void ReduceRawLiteralLength(int delta)) { | |
| 530 DCHECK_NOT_NULL(next_.raw_literal_chars); | |
| 531 next_.raw_literal_chars->ReduceLength(delta); | |
| 532 } | |
| 533 | |
| 534 // Stops scanning of a literal and drop the collected characters, | |
| 535 // e.g., due to an encountered error. | |
| 536 inline void DropLiteral() { | |
| 537 next_.literal_chars = NULL; | |
| 538 next_.raw_literal_chars = NULL; | |
| 539 } | |
| 540 | |
| 541 inline void AddLiteralCharAdvance() { | |
| 542 AddLiteralChar(c0_); | |
| 543 Advance(); | |
| 544 } | |
| 545 | |
| 546 // Low-level scanning support. | |
| 547 template <bool capture_raw = false, bool check_surrogate = true> | |
| 548 void Advance() { | |
| 549 if (capture_raw) { | |
| 550 AddRawLiteralChar(c0_); | |
| 551 } | |
| 552 c0_ = source_->Advance(); | |
| 553 if (check_surrogate) HandleLeadSurrogate(); | |
| 554 } | |
| 555 | |
| 556 void HandleLeadSurrogate() { | |
| 557 if (unibrow::Utf16::IsLeadSurrogate(c0_)) { | |
| 558 uc32 c1 = source_->Advance(); | |
| 559 if (!unibrow::Utf16::IsTrailSurrogate(c1)) { | |
| 560 source_->PushBack(c1); | |
| 561 } else { | |
| 562 c0_ = unibrow::Utf16::CombineSurrogatePair(c0_, c1); | |
| 563 } | |
| 564 } | |
| 565 } | |
| 566 | |
| 567 void PushBack(uc32 ch) { | |
| 568 if (ch > static_cast<uc32>(unibrow::Utf16::kMaxNonSurrogateCharCode)) { | |
| 569 source_->PushBack(unibrow::Utf16::TrailSurrogate(c0_)); | |
| 570 source_->PushBack(unibrow::Utf16::LeadSurrogate(c0_)); | |
| 571 } else { | |
| 572 source_->PushBack(c0_); | |
| 573 } | |
| 574 c0_ = ch; | |
| 575 } | |
| 576 | |
| 577 inline Token::Value Select(Token::Value tok) { | |
| 578 Advance(); | |
| 579 return tok; | |
| 580 } | |
| 581 | |
| 582 inline Token::Value Select(uc32 next, Token::Value then, Token::Value else_) { | |
| 583 Advance(); | |
| 584 if (c0_ == next) { | |
| 585 Advance(); | |
| 586 return then; | |
| 587 } else { | |
| 588 return else_; | |
| 589 } | |
| 590 } | |
| 591 | |
| 592 // Returns the literal string, if any, for the current token (the | |
| 593 // token last returned by Next()). The string is 0-terminated. | |
| 594 // Literal strings are collected for identifiers, strings, numbers as well | |
| 595 // as for template literals. For template literals we also collect the raw | |
| 596 // form. | |
| 597 // These functions only give the correct result if the literal was scanned | |
| 598 // when a LiteralScope object is alive. | |
| 599 Vector<const uint8_t> literal_one_byte_string() { | |
| 600 DCHECK_NOT_NULL(current_.literal_chars); | |
| 601 return current_.literal_chars->one_byte_literal(); | |
| 602 } | |
| 603 Vector<const uint16_t> literal_two_byte_string() { | |
| 604 DCHECK_NOT_NULL(current_.literal_chars); | |
| 605 return current_.literal_chars->two_byte_literal(); | |
| 606 } | |
| 607 bool is_literal_one_byte() { | |
| 608 DCHECK_NOT_NULL(current_.literal_chars); | |
| 609 return current_.literal_chars->is_one_byte(); | |
| 610 } | |
| 611 int literal_length() const { | |
| 612 DCHECK_NOT_NULL(current_.literal_chars); | |
| 613 return current_.literal_chars->length(); | |
| 614 } | |
| 615 // Returns the literal string for the next token (the token that | |
| 616 // would be returned if Next() were called). | |
| 617 Vector<const uint8_t> next_literal_one_byte_string() { | |
| 618 DCHECK_NOT_NULL(next_.literal_chars); | |
| 619 return next_.literal_chars->one_byte_literal(); | |
| 620 } | |
| 621 Vector<const uint16_t> next_literal_two_byte_string() { | |
| 622 DCHECK_NOT_NULL(next_.literal_chars); | |
| 623 return next_.literal_chars->two_byte_literal(); | |
| 624 } | |
| 625 bool is_next_literal_one_byte() { | |
| 626 DCHECK_NOT_NULL(next_.literal_chars); | |
| 627 return next_.literal_chars->is_one_byte(); | |
| 628 } | |
| 629 Vector<const uint8_t> raw_literal_one_byte_string() { | |
| 630 DCHECK_NOT_NULL(current_.raw_literal_chars); | |
| 631 return current_.raw_literal_chars->one_byte_literal(); | |
| 632 } | |
| 633 Vector<const uint16_t> raw_literal_two_byte_string() { | |
| 634 DCHECK_NOT_NULL(current_.raw_literal_chars); | |
| 635 return current_.raw_literal_chars->two_byte_literal(); | |
| 636 } | |
| 637 bool is_raw_literal_one_byte() { | |
| 638 DCHECK_NOT_NULL(current_.raw_literal_chars); | |
| 639 return current_.raw_literal_chars->is_one_byte(); | |
| 640 } | |
| 641 | |
| 642 template <bool capture_raw> | |
| 643 uc32 ScanHexNumber(int expected_length); | |
| 644 // Scan a number of any length but not bigger than max_value. For example, the | |
| 645 // number can be 000000001, so it's very long in characters but its value is | |
| 646 // small. | |
| 647 template <bool capture_raw> | |
| 648 uc32 ScanUnlimitedLengthHexNumber(int max_value); | |
| 649 | |
| 650 // Scans a single JavaScript token. | |
| 651 void Scan(); | |
| 652 | |
| 653 bool SkipWhiteSpace(); | |
| 654 Token::Value SkipSingleLineComment(); | |
| 655 Token::Value SkipSourceURLComment(); | |
| 656 void TryToParseSourceURLComment(); | |
| 657 Token::Value SkipMultiLineComment(); | |
| 658 // Scans a possible HTML comment -- begins with '<!'. | |
| 659 Token::Value ScanHtmlComment(); | |
| 660 | |
| 661 void ScanDecimalDigits(); | |
| 662 Token::Value ScanNumber(bool seen_period); | |
| 663 Token::Value ScanIdentifierOrKeyword(); | |
| 664 Token::Value ScanIdentifierSuffix(LiteralScope* literal, bool escaped); | |
| 665 | |
| 666 Token::Value ScanString(); | |
| 667 | |
| 668 // Scans an escape-sequence which is part of a string and adds the | |
| 669 // decoded character to the current literal. Returns true if a pattern | |
| 670 // is scanned. | |
| 671 template <bool capture_raw, bool in_template_literal> | |
| 672 bool ScanEscape(); | |
| 673 | |
| 674 // Decodes a Unicode escape-sequence which is part of an identifier. | |
| 675 // If the escape sequence cannot be decoded the result is kBadChar. | |
| 676 uc32 ScanIdentifierUnicodeEscape(); | |
| 677 // Helper for the above functions. | |
| 678 template <bool capture_raw> | |
| 679 uc32 ScanUnicodeEscape(); | |
| 680 | |
| 681 Token::Value ScanTemplateSpan(); | |
| 682 | |
| 683 // Return the current source position. | |
| 684 int source_pos() { | |
| 685 return static_cast<int>(source_->pos()) - kCharacterLookaheadBufferSize; | |
| 686 } | |
| 687 | |
| 688 static bool LiteralContainsEscapes(const TokenDesc& token) { | |
| 689 Location location = token.location; | |
| 690 int source_length = (location.end_pos - location.beg_pos); | |
| 691 if (token.token == Token::STRING) { | |
| 692 // Subtract delimiters. | |
| 693 source_length -= 2; | |
| 694 } | |
| 695 return token.literal_chars->length() != source_length; | |
| 696 } | |
| 697 | |
| 698 UnicodeCache* unicode_cache_; | |
| 699 | |
| 700 // Buffers collecting literal strings, numbers, etc. | |
| 701 LiteralBuffer literal_buffer0_; | |
| 702 LiteralBuffer literal_buffer1_; | |
| 703 LiteralBuffer literal_buffer2_; | |
| 704 | |
| 705 // Values parsed from magic comments. | |
| 706 LiteralBuffer source_url_; | |
| 707 LiteralBuffer source_mapping_url_; | |
| 708 | |
| 709 // Buffer to store raw string values | |
| 710 LiteralBuffer raw_literal_buffer0_; | |
| 711 LiteralBuffer raw_literal_buffer1_; | |
| 712 LiteralBuffer raw_literal_buffer2_; | |
| 713 | |
| 714 TokenDesc current_; // desc for current token (as returned by Next()) | |
| 715 TokenDesc next_; // desc for next token (one token look-ahead) | |
| 716 TokenDesc next_next_; // desc for the token after next (after PeakAhead()) | |
| 717 | |
| 718 // Variables for Scanner::BookmarkScope and the *Bookmark implementation. | |
| 719 // These variables contain the scanner state when a bookmark is set. | |
| 720 // | |
| 721 // We will use bookmark_c0_ as a 'control' variable, where: | |
| 722 // - bookmark_c0_ >= 0: A bookmark has been set and this contains c0_. | |
| 723 // - bookmark_c0_ == -1: No bookmark has been set. | |
| 724 // - bookmark_c0_ == -2: The bookmark has been applied (ResetToBookmark). | |
| 725 // | |
| 726 // Which state is being bookmarked? The parser state is distributed over | |
| 727 // several variables, roughly like this: | |
| 728 // ... 1234 + 5678 ..... [character stream] | |
| 729 // [current_] [next_] c0_ | [scanner state] | |
| 730 // So when the scanner is logically at the beginning of an expression | |
| 731 // like "1234 + 4567", then: | |
| 732 // - current_ contains "1234" | |
| 733 // - next_ contains "+" | |
| 734 // - c0_ contains ' ' (the space between "+" and "5678", | |
| 735 // - the source_ character stream points to the beginning of "5678". | |
| 736 // To be able to restore this state, we will keep copies of current_, next_, | |
| 737 // and c0_; we'll ask the stream to bookmark itself, and we'll copy the | |
| 738 // contents of current_'s and next_'s literal buffers to bookmark_*_literal_. | |
| 739 static const uc32 kNoBookmark = -1; | |
| 740 static const uc32 kBookmarkWasApplied = -2; | |
| 741 uc32 bookmark_c0_; | |
| 742 TokenDesc bookmark_current_; | |
| 743 TokenDesc bookmark_next_; | |
| 744 LiteralBuffer bookmark_current_literal_; | |
| 745 LiteralBuffer bookmark_current_raw_literal_; | |
| 746 LiteralBuffer bookmark_next_literal_; | |
| 747 LiteralBuffer bookmark_next_raw_literal_; | |
| 748 | |
| 749 // Input stream. Must be initialized to an Utf16CharacterStream. | |
| 750 Utf16CharacterStream* source_; | |
| 751 | |
| 752 | |
| 753 // Start position of the octal literal last scanned. | |
| 754 Location octal_pos_; | |
| 755 | |
| 756 // One Unicode character look-ahead; c0_ < 0 at the end of the input. | |
| 757 uc32 c0_; | |
| 758 | |
| 759 // Whether there is a line terminator whitespace character after | |
| 760 // the current token, and before the next. Does not count newlines | |
| 761 // inside multiline comments. | |
| 762 bool has_line_terminator_before_next_; | |
| 763 // Whether there is a multi-line comment that contains a | |
| 764 // line-terminator after the current token, and before the next. | |
| 765 bool has_multiline_comment_before_next_; | |
| 766 }; | |
| 767 | |
| 768 } // namespace internal | |
| 769 } // namespace v8 | |
| 770 | |
| 771 #endif // V8_SCANNER_H_ | |
| OLD | NEW |
|---|
Chromium Code Reviews
Issue 1481613002:
Create ast/ and parsing/ subdirectories and move appropriate files (Closed)
Base URL: https://chromium.googlesource.com/v8/v8.git@master