Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 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_PARSING_SCANNER_H_ |
| 8 | #define V8_PARSING_SCANNER_H_ |
| 9 | |
| 10 | #include "src/allocation.h" |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 11 | #include "src/base/hashmap.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 12 | #include "src/base/logging.h" |
| 13 | #include "src/char-predicates.h" |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 14 | #include "src/collector.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 15 | #include "src/globals.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 16 | #include "src/list.h" |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 17 | #include "src/messages.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 18 | #include "src/parsing/token.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 19 | #include "src/unicode-decoder.h" |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 20 | #include "src/unicode.h" |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 21 | |
| 22 | namespace v8 { |
| 23 | namespace internal { |
| 24 | |
| 25 | |
| 26 | class AstRawString; |
| 27 | class AstValueFactory; |
| 28 | class ParserRecorder; |
| 29 | class UnicodeCache; |
| 30 | |
| 31 | |
| 32 | // --------------------------------------------------------------------- |
| 33 | // Buffered stream of UTF-16 code units, using an internal UTF-16 buffer. |
| 34 | // A code unit is a 16 bit value representing either a 16 bit code point |
| 35 | // or one part of a surrogate pair that make a single 21 bit code point. |
| 36 | |
| 37 | class Utf16CharacterStream { |
| 38 | public: |
| 39 | Utf16CharacterStream() : pos_(0) { } |
| 40 | virtual ~Utf16CharacterStream() { } |
| 41 | |
| 42 | // Returns and advances past the next UTF-16 code unit in the input |
| 43 | // stream. If there are no more code units, it returns a negative |
| 44 | // value. |
| 45 | inline uc32 Advance() { |
| 46 | if (buffer_cursor_ < buffer_end_ || ReadBlock()) { |
| 47 | pos_++; |
| 48 | return static_cast<uc32>(*(buffer_cursor_++)); |
| 49 | } |
| 50 | // Note: currently the following increment is necessary to avoid a |
| 51 | // parser problem! The scanner treats the final kEndOfInput as |
| 52 | // a code unit with a position, and does math relative to that |
| 53 | // position. |
| 54 | pos_++; |
| 55 | |
| 56 | return kEndOfInput; |
| 57 | } |
| 58 | |
| 59 | // Return the current position in the code unit stream. |
| 60 | // Starts at zero. |
| 61 | inline size_t pos() const { return pos_; } |
| 62 | |
| 63 | // Skips forward past the next code_unit_count UTF-16 code units |
| 64 | // in the input, or until the end of input if that comes sooner. |
| 65 | // Returns the number of code units actually skipped. If less |
| 66 | // than code_unit_count, |
| 67 | inline size_t SeekForward(size_t code_unit_count) { |
| 68 | size_t buffered_chars = buffer_end_ - buffer_cursor_; |
| 69 | if (code_unit_count <= buffered_chars) { |
| 70 | buffer_cursor_ += code_unit_count; |
| 71 | pos_ += code_unit_count; |
| 72 | return code_unit_count; |
| 73 | } |
| 74 | return SlowSeekForward(code_unit_count); |
| 75 | } |
| 76 | |
| 77 | // Pushes back the most recently read UTF-16 code unit (or negative |
| 78 | // value if at end of input), i.e., the value returned by the most recent |
| 79 | // call to Advance. |
| 80 | // Must not be used right after calling SeekForward. |
| 81 | virtual void PushBack(int32_t code_unit) = 0; |
| 82 | |
| 83 | virtual bool SetBookmark(); |
| 84 | virtual void ResetToBookmark(); |
| 85 | |
| 86 | protected: |
| 87 | static const uc32 kEndOfInput = -1; |
| 88 | |
| 89 | // Ensures that the buffer_cursor_ points to the code_unit at |
| 90 | // position pos_ of the input, if possible. If the position |
| 91 | // is at or after the end of the input, return false. If there |
| 92 | // are more code_units available, return true. |
| 93 | virtual bool ReadBlock() = 0; |
| 94 | virtual size_t SlowSeekForward(size_t code_unit_count) = 0; |
| 95 | |
| 96 | const uint16_t* buffer_cursor_; |
| 97 | const uint16_t* buffer_end_; |
| 98 | size_t pos_; |
| 99 | }; |
| 100 | |
| 101 | |
| 102 | // --------------------------------------------------------------------- |
| 103 | // DuplicateFinder discovers duplicate symbols. |
| 104 | |
| 105 | class DuplicateFinder { |
| 106 | public: |
| 107 | explicit DuplicateFinder(UnicodeCache* constants) |
| 108 | : unicode_constants_(constants), |
| 109 | backing_store_(16), |
| 110 | map_(&Match) { } |
| 111 | |
| 112 | int AddOneByteSymbol(Vector<const uint8_t> key, int value); |
| 113 | int AddTwoByteSymbol(Vector<const uint16_t> key, int value); |
| 114 | // Add a a number literal by converting it (if necessary) |
| 115 | // to the string that ToString(ToNumber(literal)) would generate. |
| 116 | // and then adding that string with AddOneByteSymbol. |
| 117 | // This string is the actual value used as key in an object literal, |
| 118 | // and the one that must be different from the other keys. |
| 119 | int AddNumber(Vector<const uint8_t> key, int value); |
| 120 | |
| 121 | private: |
| 122 | int AddSymbol(Vector<const uint8_t> key, bool is_one_byte, int value); |
| 123 | // Backs up the key and its length in the backing store. |
| 124 | // The backup is stored with a base 127 encoding of the |
| 125 | // length (plus a bit saying whether the string is one byte), |
| 126 | // followed by the bytes of the key. |
| 127 | uint8_t* BackupKey(Vector<const uint8_t> key, bool is_one_byte); |
| 128 | |
| 129 | // Compare two encoded keys (both pointing into the backing store) |
| 130 | // for having the same base-127 encoded lengths and representation. |
| 131 | // and then having the same 'length' bytes following. |
| 132 | static bool Match(void* first, void* second); |
| 133 | // Creates a hash from a sequence of bytes. |
| 134 | static uint32_t Hash(Vector<const uint8_t> key, bool is_one_byte); |
| 135 | // Checks whether a string containing a JS number is its canonical |
| 136 | // form. |
| 137 | static bool IsNumberCanonical(Vector<const uint8_t> key); |
| 138 | |
| 139 | // Size of buffer. Sufficient for using it to call DoubleToCString in |
| 140 | // from conversions.h. |
| 141 | static const int kBufferSize = 100; |
| 142 | |
| 143 | UnicodeCache* unicode_constants_; |
| 144 | // Backing store used to store strings used as hashmap keys. |
| 145 | SequenceCollector<unsigned char> backing_store_; |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 146 | base::HashMap map_; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 147 | // Buffer used for string->number->canonical string conversions. |
| 148 | char number_buffer_[kBufferSize]; |
| 149 | }; |
| 150 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 151 | // ---------------------------------------------------------------------------- |
| 152 | // LiteralBuffer - Collector of chars of literals. |
| 153 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 154 | const int kMaxAscii = 127; |
| 155 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 156 | class LiteralBuffer { |
| 157 | public: |
| 158 | LiteralBuffer() : is_one_byte_(true), position_(0), backing_store_() { } |
| 159 | |
| 160 | ~LiteralBuffer() { backing_store_.Dispose(); } |
| 161 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 162 | INLINE(void AddChar(char code_unit)) { |
| 163 | if (position_ >= backing_store_.length()) ExpandBuffer(); |
| 164 | DCHECK(is_one_byte_); |
| 165 | DCHECK(0 <= code_unit && code_unit <= kMaxAscii); |
| 166 | backing_store_[position_] = static_cast<byte>(code_unit); |
| 167 | position_ += kOneByteSize; |
| 168 | return; |
| 169 | } |
| 170 | |
| 171 | INLINE(void AddChar(uc32 code_unit)) { |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 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_; |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 238 | if (position_ < backing_store_.length()) { |
| 239 | std::copy(other->backing_store_.begin(), |
| 240 | other->backing_store_.begin() + position_, |
| 241 | backing_store_.begin()); |
| 242 | } else { |
| 243 | backing_store_.Dispose(); |
| 244 | backing_store_ = other->backing_store_.Clone(); |
| 245 | } |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 246 | } |
| 247 | } |
| 248 | |
| 249 | private: |
| 250 | static const int kInitialCapacity = 16; |
| 251 | static const int kGrowthFactory = 4; |
| 252 | static const int kMinConversionSlack = 256; |
| 253 | static const int kMaxGrowth = 1 * MB; |
| 254 | inline int NewCapacity(int min_capacity) { |
| 255 | int capacity = Max(min_capacity, backing_store_.length()); |
| 256 | int new_capacity = Min(capacity * kGrowthFactory, capacity + kMaxGrowth); |
| 257 | return new_capacity; |
| 258 | } |
| 259 | |
| 260 | void ExpandBuffer() { |
| 261 | Vector<byte> new_store = Vector<byte>::New(NewCapacity(kInitialCapacity)); |
| 262 | MemCopy(new_store.start(), backing_store_.start(), position_); |
| 263 | backing_store_.Dispose(); |
| 264 | backing_store_ = new_store; |
| 265 | } |
| 266 | |
| 267 | void ConvertToTwoByte() { |
| 268 | DCHECK(is_one_byte_); |
| 269 | Vector<byte> new_store; |
| 270 | int new_content_size = position_ * kUC16Size; |
| 271 | if (new_content_size >= backing_store_.length()) { |
| 272 | // Ensure room for all currently read code units as UC16 as well |
| 273 | // as the code unit about to be stored. |
| 274 | new_store = Vector<byte>::New(NewCapacity(new_content_size)); |
| 275 | } else { |
| 276 | new_store = backing_store_; |
| 277 | } |
| 278 | uint8_t* src = backing_store_.start(); |
| 279 | uint16_t* dst = reinterpret_cast<uint16_t*>(new_store.start()); |
| 280 | for (int i = position_ - 1; i >= 0; i--) { |
| 281 | dst[i] = src[i]; |
| 282 | } |
| 283 | if (new_store.start() != backing_store_.start()) { |
| 284 | backing_store_.Dispose(); |
| 285 | backing_store_ = new_store; |
| 286 | } |
| 287 | position_ = new_content_size; |
| 288 | is_one_byte_ = false; |
| 289 | } |
| 290 | |
| 291 | bool is_one_byte_; |
| 292 | int position_; |
| 293 | Vector<byte> backing_store_; |
| 294 | |
| 295 | DISALLOW_COPY_AND_ASSIGN(LiteralBuffer); |
| 296 | }; |
| 297 | |
| 298 | |
| 299 | // ---------------------------------------------------------------------------- |
| 300 | // JavaScript Scanner. |
| 301 | |
| 302 | class Scanner { |
| 303 | public: |
| 304 | // Scoped helper for literal recording. Automatically drops the literal |
| 305 | // if aborting the scanning before it's complete. |
| 306 | class LiteralScope { |
| 307 | public: |
| 308 | explicit LiteralScope(Scanner* self) : scanner_(self), complete_(false) { |
| 309 | scanner_->StartLiteral(); |
| 310 | } |
| 311 | ~LiteralScope() { |
| 312 | if (!complete_) scanner_->DropLiteral(); |
| 313 | } |
| 314 | void Complete() { |
| 315 | complete_ = true; |
| 316 | } |
| 317 | |
| 318 | private: |
| 319 | Scanner* scanner_; |
| 320 | bool complete_; |
| 321 | }; |
| 322 | |
| 323 | // Scoped helper for a re-settable bookmark. |
| 324 | class BookmarkScope { |
| 325 | public: |
| 326 | explicit BookmarkScope(Scanner* scanner) : scanner_(scanner) { |
| 327 | DCHECK_NOT_NULL(scanner_); |
| 328 | } |
| 329 | ~BookmarkScope() { scanner_->DropBookmark(); } |
| 330 | |
| 331 | bool Set() { return scanner_->SetBookmark(); } |
| 332 | void Reset() { scanner_->ResetToBookmark(); } |
| 333 | bool HasBeenSet() { return scanner_->BookmarkHasBeenSet(); } |
| 334 | bool HasBeenReset() { return scanner_->BookmarkHasBeenReset(); } |
| 335 | |
| 336 | private: |
| 337 | Scanner* scanner_; |
| 338 | |
| 339 | DISALLOW_COPY_AND_ASSIGN(BookmarkScope); |
| 340 | }; |
| 341 | |
| 342 | // Representation of an interval of source positions. |
| 343 | struct Location { |
| 344 | Location(int b, int e) : beg_pos(b), end_pos(e) { } |
| 345 | Location() : beg_pos(0), end_pos(0) { } |
| 346 | |
| 347 | bool IsValid() const { |
| 348 | return beg_pos >= 0 && end_pos >= beg_pos; |
| 349 | } |
| 350 | |
| 351 | static Location invalid() { return Location(-1, -1); } |
| 352 | |
| 353 | int beg_pos; |
| 354 | int end_pos; |
| 355 | }; |
| 356 | |
| 357 | // -1 is outside of the range of any real source code. |
| 358 | static const int kNoOctalLocation = -1; |
| 359 | |
| 360 | explicit Scanner(UnicodeCache* scanner_contants); |
| 361 | |
| 362 | void Initialize(Utf16CharacterStream* source); |
| 363 | |
| 364 | // Returns the next token and advances input. |
| 365 | Token::Value Next(); |
| 366 | // Returns the token following peek() |
| 367 | Token::Value PeekAhead(); |
| 368 | // Returns the current token again. |
| 369 | Token::Value current_token() { return current_.token; } |
| 370 | // Returns the location information for the current token |
| 371 | // (the token last returned by Next()). |
| 372 | Location location() const { return current_.location; } |
| 373 | |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 374 | bool has_error() const { return scanner_error_ != MessageTemplate::kNone; } |
| 375 | MessageTemplate::Template error() const { return scanner_error_; } |
| 376 | Location error_location() const { return scanner_error_location_; } |
| 377 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 378 | // Similar functions for the upcoming token. |
| 379 | |
| 380 | // One token look-ahead (past the token returned by Next()). |
| 381 | Token::Value peek() const { return next_.token; } |
| 382 | |
| 383 | Location peek_location() const { return next_.location; } |
| 384 | |
| 385 | bool literal_contains_escapes() const { |
| 386 | return LiteralContainsEscapes(current_); |
| 387 | } |
| 388 | bool next_literal_contains_escapes() const { |
| 389 | return LiteralContainsEscapes(next_); |
| 390 | } |
| 391 | bool is_literal_contextual_keyword(Vector<const char> keyword) { |
| 392 | DCHECK_NOT_NULL(current_.literal_chars); |
| 393 | return current_.literal_chars->is_contextual_keyword(keyword); |
| 394 | } |
| 395 | bool is_next_contextual_keyword(Vector<const char> keyword) { |
| 396 | DCHECK_NOT_NULL(next_.literal_chars); |
| 397 | return next_.literal_chars->is_contextual_keyword(keyword); |
| 398 | } |
| 399 | |
| 400 | const AstRawString* CurrentSymbol(AstValueFactory* ast_value_factory); |
| 401 | const AstRawString* NextSymbol(AstValueFactory* ast_value_factory); |
| 402 | const AstRawString* CurrentRawSymbol(AstValueFactory* ast_value_factory); |
| 403 | |
| 404 | double DoubleValue(); |
| 405 | bool ContainsDot(); |
| 406 | bool LiteralMatches(const char* data, int length, bool allow_escapes = true) { |
| 407 | if (is_literal_one_byte() && |
| 408 | literal_length() == length && |
| 409 | (allow_escapes || !literal_contains_escapes())) { |
| 410 | const char* token = |
| 411 | reinterpret_cast<const char*>(literal_one_byte_string().start()); |
| 412 | return !strncmp(token, data, length); |
| 413 | } |
| 414 | return false; |
| 415 | } |
| 416 | inline bool UnescapedLiteralMatches(const char* data, int length) { |
| 417 | return LiteralMatches(data, length, false); |
| 418 | } |
| 419 | |
| 420 | void IsGetOrSet(bool* is_get, bool* is_set) { |
| 421 | if (is_literal_one_byte() && |
| 422 | literal_length() == 3 && |
| 423 | !literal_contains_escapes()) { |
| 424 | const char* token = |
| 425 | reinterpret_cast<const char*>(literal_one_byte_string().start()); |
| 426 | *is_get = strncmp(token, "get", 3) == 0; |
| 427 | *is_set = !*is_get && strncmp(token, "set", 3) == 0; |
| 428 | } |
| 429 | } |
| 430 | |
| 431 | int FindSymbol(DuplicateFinder* finder, int value); |
| 432 | |
| 433 | UnicodeCache* unicode_cache() { return unicode_cache_; } |
| 434 | |
| 435 | // Returns the location of the last seen octal literal. |
| 436 | Location octal_position() const { return octal_pos_; } |
| 437 | void clear_octal_position() { octal_pos_ = Location::invalid(); } |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 438 | // Returns the location of the last seen decimal literal with a leading zero. |
| 439 | Location decimal_with_leading_zero_position() const { |
| 440 | return decimal_with_leading_zero_pos_; |
| 441 | } |
| 442 | void clear_decimal_with_leading_zero_position() { |
| 443 | decimal_with_leading_zero_pos_ = Location::invalid(); |
| 444 | } |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 445 | |
| 446 | // Returns the value of the last smi that was scanned. |
| 447 | int smi_value() const { return current_.smi_value_; } |
| 448 | |
| 449 | // Seek forward to the given position. This operation does not |
| 450 | // work in general, for instance when there are pushed back |
| 451 | // characters, but works for seeking forward until simple delimiter |
| 452 | // tokens, which is what it is used for. |
| 453 | void SeekForward(int pos); |
| 454 | |
| 455 | // Returns true if there was a line terminator before the peek'ed token, |
| 456 | // possibly inside a multi-line comment. |
| 457 | bool HasAnyLineTerminatorBeforeNext() const { |
| 458 | return has_line_terminator_before_next_ || |
| 459 | has_multiline_comment_before_next_; |
| 460 | } |
| 461 | |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 462 | bool HasAnyLineTerminatorAfterNext() { |
| 463 | Token::Value ensure_next_next = PeekAhead(); |
| 464 | USE(ensure_next_next); |
| 465 | return has_line_terminator_after_next_; |
| 466 | } |
| 467 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 468 | // Scans the input as a regular expression pattern, previous |
| 469 | // character(s) must be /(=). Returns true if a pattern is scanned. |
| 470 | bool ScanRegExpPattern(bool seen_equal); |
| 471 | // Scans the input as regular expression flags. Returns the flags on success. |
| 472 | Maybe<RegExp::Flags> ScanRegExpFlags(); |
| 473 | |
| 474 | // Scans the input as a template literal |
| 475 | Token::Value ScanTemplateStart(); |
| 476 | Token::Value ScanTemplateContinuation(); |
| 477 | |
| 478 | const LiteralBuffer* source_url() const { return &source_url_; } |
| 479 | const LiteralBuffer* source_mapping_url() const { |
| 480 | return &source_mapping_url_; |
| 481 | } |
| 482 | |
| 483 | bool IdentifierIsFutureStrictReserved(const AstRawString* string) const; |
| 484 | |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 485 | bool FoundHtmlComment() const { return found_html_comment_; } |
| 486 | |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 487 | #define DECLARE_ACCESSORS(name) \ |
| 488 | inline bool allow_##name() const { return allow_##name##_; } \ |
| 489 | inline void set_allow_##name(bool allow) { allow_##name##_ = allow; } |
| 490 | DECLARE_ACCESSORS(harmony_exponentiation_operator) |
| 491 | #undef ACCESSOR |
| 492 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 493 | private: |
| 494 | // The current and look-ahead token. |
| 495 | struct TokenDesc { |
| 496 | Token::Value token; |
| 497 | Location location; |
| 498 | LiteralBuffer* literal_chars; |
| 499 | LiteralBuffer* raw_literal_chars; |
| 500 | int smi_value_; |
| 501 | }; |
| 502 | |
| 503 | static const int kCharacterLookaheadBufferSize = 1; |
| 504 | |
| 505 | // Scans octal escape sequence. Also accepts "\0" decimal escape sequence. |
| 506 | template <bool capture_raw> |
| 507 | uc32 ScanOctalEscape(uc32 c, int length); |
| 508 | |
| 509 | // Call this after setting source_ to the input. |
| 510 | void Init() { |
| 511 | // Set c0_ (one character ahead) |
| 512 | STATIC_ASSERT(kCharacterLookaheadBufferSize == 1); |
| 513 | Advance(); |
| 514 | // Initialize current_ to not refer to a literal. |
| 515 | current_.literal_chars = NULL; |
| 516 | current_.raw_literal_chars = NULL; |
| 517 | next_next_.token = Token::UNINITIALIZED; |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 518 | found_html_comment_ = false; |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 519 | scanner_error_ = MessageTemplate::kNone; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 520 | } |
| 521 | |
| 522 | // Support BookmarkScope functionality. |
| 523 | bool SetBookmark(); |
| 524 | void ResetToBookmark(); |
| 525 | bool BookmarkHasBeenSet(); |
| 526 | bool BookmarkHasBeenReset(); |
| 527 | void DropBookmark(); |
| 528 | static void CopyTokenDesc(TokenDesc* to, TokenDesc* from); |
| 529 | |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 530 | void ReportScannerError(const Location& location, |
| 531 | MessageTemplate::Template error) { |
| 532 | if (has_error()) return; |
| 533 | scanner_error_ = error; |
| 534 | scanner_error_location_ = location; |
| 535 | } |
| 536 | |
| 537 | void ReportScannerError(int pos, MessageTemplate::Template error) { |
| 538 | if (has_error()) return; |
| 539 | scanner_error_ = error; |
| 540 | scanner_error_location_ = Location(pos, pos + 1); |
| 541 | } |
| 542 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 543 | // Literal buffer support |
| 544 | inline void StartLiteral() { |
| 545 | LiteralBuffer* free_buffer = |
| 546 | (current_.literal_chars == &literal_buffer0_) |
| 547 | ? &literal_buffer1_ |
| 548 | : (current_.literal_chars == &literal_buffer1_) ? &literal_buffer2_ |
| 549 | : &literal_buffer0_; |
| 550 | free_buffer->Reset(); |
| 551 | next_.literal_chars = free_buffer; |
| 552 | } |
| 553 | |
| 554 | inline void StartRawLiteral() { |
| 555 | LiteralBuffer* free_buffer = |
| 556 | (current_.raw_literal_chars == &raw_literal_buffer0_) |
| 557 | ? &raw_literal_buffer1_ |
| 558 | : (current_.raw_literal_chars == &raw_literal_buffer1_) |
| 559 | ? &raw_literal_buffer2_ |
| 560 | : &raw_literal_buffer0_; |
| 561 | free_buffer->Reset(); |
| 562 | next_.raw_literal_chars = free_buffer; |
| 563 | } |
| 564 | |
| 565 | INLINE(void AddLiteralChar(uc32 c)) { |
| 566 | DCHECK_NOT_NULL(next_.literal_chars); |
| 567 | next_.literal_chars->AddChar(c); |
| 568 | } |
| 569 | |
Ben Murdoch | 61f157c | 2016-09-16 13:49:30 +0100 | [diff] [blame] | 570 | INLINE(void AddLiteralChar(char c)) { |
| 571 | DCHECK_NOT_NULL(next_.literal_chars); |
| 572 | next_.literal_chars->AddChar(c); |
| 573 | } |
| 574 | |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 575 | INLINE(void AddRawLiteralChar(uc32 c)) { |
| 576 | DCHECK_NOT_NULL(next_.raw_literal_chars); |
| 577 | next_.raw_literal_chars->AddChar(c); |
| 578 | } |
| 579 | |
| 580 | INLINE(void ReduceRawLiteralLength(int delta)) { |
| 581 | DCHECK_NOT_NULL(next_.raw_literal_chars); |
| 582 | next_.raw_literal_chars->ReduceLength(delta); |
| 583 | } |
| 584 | |
| 585 | // Stops scanning of a literal and drop the collected characters, |
| 586 | // e.g., due to an encountered error. |
| 587 | inline void DropLiteral() { |
| 588 | next_.literal_chars = NULL; |
| 589 | next_.raw_literal_chars = NULL; |
| 590 | } |
| 591 | |
| 592 | inline void AddLiteralCharAdvance() { |
| 593 | AddLiteralChar(c0_); |
| 594 | Advance(); |
| 595 | } |
| 596 | |
| 597 | // Low-level scanning support. |
| 598 | template <bool capture_raw = false, bool check_surrogate = true> |
| 599 | void Advance() { |
| 600 | if (capture_raw) { |
| 601 | AddRawLiteralChar(c0_); |
| 602 | } |
| 603 | c0_ = source_->Advance(); |
| 604 | if (check_surrogate) HandleLeadSurrogate(); |
| 605 | } |
| 606 | |
| 607 | void HandleLeadSurrogate() { |
| 608 | if (unibrow::Utf16::IsLeadSurrogate(c0_)) { |
| 609 | uc32 c1 = source_->Advance(); |
| 610 | if (!unibrow::Utf16::IsTrailSurrogate(c1)) { |
| 611 | source_->PushBack(c1); |
| 612 | } else { |
| 613 | c0_ = unibrow::Utf16::CombineSurrogatePair(c0_, c1); |
| 614 | } |
| 615 | } |
| 616 | } |
| 617 | |
| 618 | void PushBack(uc32 ch) { |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 619 | if (c0_ > static_cast<uc32>(unibrow::Utf16::kMaxNonSurrogateCharCode)) { |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 620 | source_->PushBack(unibrow::Utf16::TrailSurrogate(c0_)); |
| 621 | source_->PushBack(unibrow::Utf16::LeadSurrogate(c0_)); |
| 622 | } else { |
| 623 | source_->PushBack(c0_); |
| 624 | } |
| 625 | c0_ = ch; |
| 626 | } |
| 627 | |
| 628 | inline Token::Value Select(Token::Value tok) { |
| 629 | Advance(); |
| 630 | return tok; |
| 631 | } |
| 632 | |
| 633 | inline Token::Value Select(uc32 next, Token::Value then, Token::Value else_) { |
| 634 | Advance(); |
| 635 | if (c0_ == next) { |
| 636 | Advance(); |
| 637 | return then; |
| 638 | } else { |
| 639 | return else_; |
| 640 | } |
| 641 | } |
| 642 | |
| 643 | // Returns the literal string, if any, for the current token (the |
| 644 | // token last returned by Next()). The string is 0-terminated. |
| 645 | // Literal strings are collected for identifiers, strings, numbers as well |
| 646 | // as for template literals. For template literals we also collect the raw |
| 647 | // form. |
| 648 | // These functions only give the correct result if the literal was scanned |
| 649 | // when a LiteralScope object is alive. |
| 650 | Vector<const uint8_t> literal_one_byte_string() { |
| 651 | DCHECK_NOT_NULL(current_.literal_chars); |
| 652 | return current_.literal_chars->one_byte_literal(); |
| 653 | } |
| 654 | Vector<const uint16_t> literal_two_byte_string() { |
| 655 | DCHECK_NOT_NULL(current_.literal_chars); |
| 656 | return current_.literal_chars->two_byte_literal(); |
| 657 | } |
| 658 | bool is_literal_one_byte() { |
| 659 | DCHECK_NOT_NULL(current_.literal_chars); |
| 660 | return current_.literal_chars->is_one_byte(); |
| 661 | } |
| 662 | int literal_length() const { |
| 663 | DCHECK_NOT_NULL(current_.literal_chars); |
| 664 | return current_.literal_chars->length(); |
| 665 | } |
| 666 | // Returns the literal string for the next token (the token that |
| 667 | // would be returned if Next() were called). |
| 668 | Vector<const uint8_t> next_literal_one_byte_string() { |
| 669 | DCHECK_NOT_NULL(next_.literal_chars); |
| 670 | return next_.literal_chars->one_byte_literal(); |
| 671 | } |
| 672 | Vector<const uint16_t> next_literal_two_byte_string() { |
| 673 | DCHECK_NOT_NULL(next_.literal_chars); |
| 674 | return next_.literal_chars->two_byte_literal(); |
| 675 | } |
| 676 | bool is_next_literal_one_byte() { |
| 677 | DCHECK_NOT_NULL(next_.literal_chars); |
| 678 | return next_.literal_chars->is_one_byte(); |
| 679 | } |
| 680 | Vector<const uint8_t> raw_literal_one_byte_string() { |
| 681 | DCHECK_NOT_NULL(current_.raw_literal_chars); |
| 682 | return current_.raw_literal_chars->one_byte_literal(); |
| 683 | } |
| 684 | Vector<const uint16_t> raw_literal_two_byte_string() { |
| 685 | DCHECK_NOT_NULL(current_.raw_literal_chars); |
| 686 | return current_.raw_literal_chars->two_byte_literal(); |
| 687 | } |
| 688 | bool is_raw_literal_one_byte() { |
| 689 | DCHECK_NOT_NULL(current_.raw_literal_chars); |
| 690 | return current_.raw_literal_chars->is_one_byte(); |
| 691 | } |
| 692 | |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 693 | template <bool capture_raw, bool unicode = false> |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 694 | uc32 ScanHexNumber(int expected_length); |
| 695 | // Scan a number of any length but not bigger than max_value. For example, the |
| 696 | // number can be 000000001, so it's very long in characters but its value is |
| 697 | // small. |
| 698 | template <bool capture_raw> |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 699 | uc32 ScanUnlimitedLengthHexNumber(int max_value, int beg_pos); |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 700 | |
| 701 | // Scans a single JavaScript token. |
| 702 | void Scan(); |
| 703 | |
| 704 | bool SkipWhiteSpace(); |
| 705 | Token::Value SkipSingleLineComment(); |
| 706 | Token::Value SkipSourceURLComment(); |
| 707 | void TryToParseSourceURLComment(); |
| 708 | Token::Value SkipMultiLineComment(); |
| 709 | // Scans a possible HTML comment -- begins with '<!'. |
| 710 | Token::Value ScanHtmlComment(); |
| 711 | |
| 712 | void ScanDecimalDigits(); |
| 713 | Token::Value ScanNumber(bool seen_period); |
| 714 | Token::Value ScanIdentifierOrKeyword(); |
| 715 | Token::Value ScanIdentifierSuffix(LiteralScope* literal, bool escaped); |
| 716 | |
| 717 | Token::Value ScanString(); |
| 718 | |
| 719 | // Scans an escape-sequence which is part of a string and adds the |
| 720 | // decoded character to the current literal. Returns true if a pattern |
| 721 | // is scanned. |
| 722 | template <bool capture_raw, bool in_template_literal> |
| 723 | bool ScanEscape(); |
| 724 | |
| 725 | // Decodes a Unicode escape-sequence which is part of an identifier. |
| 726 | // If the escape sequence cannot be decoded the result is kBadChar. |
| 727 | uc32 ScanIdentifierUnicodeEscape(); |
| 728 | // Helper for the above functions. |
| 729 | template <bool capture_raw> |
| 730 | uc32 ScanUnicodeEscape(); |
| 731 | |
| 732 | Token::Value ScanTemplateSpan(); |
| 733 | |
| 734 | // Return the current source position. |
| 735 | int source_pos() { |
| 736 | return static_cast<int>(source_->pos()) - kCharacterLookaheadBufferSize; |
| 737 | } |
| 738 | |
| 739 | static bool LiteralContainsEscapes(const TokenDesc& token) { |
| 740 | Location location = token.location; |
| 741 | int source_length = (location.end_pos - location.beg_pos); |
| 742 | if (token.token == Token::STRING) { |
| 743 | // Subtract delimiters. |
| 744 | source_length -= 2; |
| 745 | } |
| 746 | return token.literal_chars->length() != source_length; |
| 747 | } |
| 748 | |
| 749 | UnicodeCache* unicode_cache_; |
| 750 | |
| 751 | // Buffers collecting literal strings, numbers, etc. |
| 752 | LiteralBuffer literal_buffer0_; |
| 753 | LiteralBuffer literal_buffer1_; |
| 754 | LiteralBuffer literal_buffer2_; |
| 755 | |
| 756 | // Values parsed from magic comments. |
| 757 | LiteralBuffer source_url_; |
| 758 | LiteralBuffer source_mapping_url_; |
| 759 | |
| 760 | // Buffer to store raw string values |
| 761 | LiteralBuffer raw_literal_buffer0_; |
| 762 | LiteralBuffer raw_literal_buffer1_; |
| 763 | LiteralBuffer raw_literal_buffer2_; |
| 764 | |
| 765 | TokenDesc current_; // desc for current token (as returned by Next()) |
| 766 | TokenDesc next_; // desc for next token (one token look-ahead) |
| 767 | TokenDesc next_next_; // desc for the token after next (after PeakAhead()) |
| 768 | |
| 769 | // Variables for Scanner::BookmarkScope and the *Bookmark implementation. |
| 770 | // These variables contain the scanner state when a bookmark is set. |
| 771 | // |
| 772 | // We will use bookmark_c0_ as a 'control' variable, where: |
| 773 | // - bookmark_c0_ >= 0: A bookmark has been set and this contains c0_. |
| 774 | // - bookmark_c0_ == -1: No bookmark has been set. |
| 775 | // - bookmark_c0_ == -2: The bookmark has been applied (ResetToBookmark). |
| 776 | // |
| 777 | // Which state is being bookmarked? The parser state is distributed over |
| 778 | // several variables, roughly like this: |
| 779 | // ... 1234 + 5678 ..... [character stream] |
| 780 | // [current_] [next_] c0_ | [scanner state] |
| 781 | // So when the scanner is logically at the beginning of an expression |
| 782 | // like "1234 + 4567", then: |
| 783 | // - current_ contains "1234" |
| 784 | // - next_ contains "+" |
| 785 | // - c0_ contains ' ' (the space between "+" and "5678", |
| 786 | // - the source_ character stream points to the beginning of "5678". |
| 787 | // To be able to restore this state, we will keep copies of current_, next_, |
| 788 | // and c0_; we'll ask the stream to bookmark itself, and we'll copy the |
| 789 | // contents of current_'s and next_'s literal buffers to bookmark_*_literal_. |
| 790 | static const uc32 kNoBookmark = -1; |
| 791 | static const uc32 kBookmarkWasApplied = -2; |
| 792 | uc32 bookmark_c0_; |
| 793 | TokenDesc bookmark_current_; |
| 794 | TokenDesc bookmark_next_; |
| 795 | LiteralBuffer bookmark_current_literal_; |
| 796 | LiteralBuffer bookmark_current_raw_literal_; |
| 797 | LiteralBuffer bookmark_next_literal_; |
| 798 | LiteralBuffer bookmark_next_raw_literal_; |
| 799 | |
| 800 | // Input stream. Must be initialized to an Utf16CharacterStream. |
| 801 | Utf16CharacterStream* source_; |
| 802 | |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 803 | // Last-seen positions of potentially problematic tokens. |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 804 | Location octal_pos_; |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 805 | Location decimal_with_leading_zero_pos_; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 806 | |
| 807 | // One Unicode character look-ahead; c0_ < 0 at the end of the input. |
| 808 | uc32 c0_; |
| 809 | |
| 810 | // Whether there is a line terminator whitespace character after |
| 811 | // the current token, and before the next. Does not count newlines |
| 812 | // inside multiline comments. |
| 813 | bool has_line_terminator_before_next_; |
| 814 | // Whether there is a multi-line comment that contains a |
| 815 | // line-terminator after the current token, and before the next. |
| 816 | bool has_multiline_comment_before_next_; |
Ben Murdoch | c561043 | 2016-08-08 18:44:38 +0100 | [diff] [blame] | 817 | bool has_line_terminator_after_next_; |
Ben Murdoch | 097c5b2 | 2016-05-18 11:27:45 +0100 | [diff] [blame] | 818 | |
| 819 | // Whether this scanner encountered an HTML comment. |
| 820 | bool found_html_comment_; |
Ben Murdoch | da12d29 | 2016-06-02 14:46:10 +0100 | [diff] [blame] | 821 | |
| 822 | bool allow_harmony_exponentiation_operator_; |
| 823 | |
| 824 | MessageTemplate::Template scanner_error_; |
| 825 | Location scanner_error_location_; |
Ben Murdoch | 4a90d5f | 2016-03-22 12:00:34 +0000 | [diff] [blame] | 826 | }; |
| 827 | |
| 828 | } // namespace internal |
| 829 | } // namespace v8 |
| 830 | |
| 831 | #endif // V8_PARSING_SCANNER_H_ |