sqlglot.tokens
1from __future__ import annotations 2 3import typing as t 4from enum import auto 5 6from sqlglot.helper import AutoName 7from sqlglot.trie import TrieResult, in_trie, new_trie 8 9 10class TokenType(AutoName): 11 L_PAREN = auto() 12 R_PAREN = auto() 13 L_BRACKET = auto() 14 R_BRACKET = auto() 15 L_BRACE = auto() 16 R_BRACE = auto() 17 COMMA = auto() 18 DOT = auto() 19 DASH = auto() 20 PLUS = auto() 21 COLON = auto() 22 DCOLON = auto() 23 SEMICOLON = auto() 24 STAR = auto() 25 BACKSLASH = auto() 26 SLASH = auto() 27 LT = auto() 28 LTE = auto() 29 GT = auto() 30 GTE = auto() 31 NOT = auto() 32 EQ = auto() 33 NEQ = auto() 34 NULLSAFE_EQ = auto() 35 AND = auto() 36 OR = auto() 37 AMP = auto() 38 DPIPE = auto() 39 PIPE = auto() 40 CARET = auto() 41 TILDA = auto() 42 ARROW = auto() 43 DARROW = auto() 44 FARROW = auto() 45 HASH = auto() 46 HASH_ARROW = auto() 47 DHASH_ARROW = auto() 48 LR_ARROW = auto() 49 LT_AT = auto() 50 AT_GT = auto() 51 DOLLAR = auto() 52 PARAMETER = auto() 53 SESSION_PARAMETER = auto() 54 DAMP = auto() 55 56 BLOCK_START = auto() 57 BLOCK_END = auto() 58 59 SPACE = auto() 60 BREAK = auto() 61 62 STRING = auto() 63 NUMBER = auto() 64 IDENTIFIER = auto() 65 DATABASE = auto() 66 COLUMN = auto() 67 COLUMN_DEF = auto() 68 SCHEMA = auto() 69 TABLE = auto() 70 VAR = auto() 71 BIT_STRING = auto() 72 HEX_STRING = auto() 73 BYTE_STRING = auto() 74 NATIONAL_STRING = auto() 75 RAW_STRING = auto() 76 77 # types 78 BIT = auto() 79 BOOLEAN = auto() 80 TINYINT = auto() 81 UTINYINT = auto() 82 SMALLINT = auto() 83 USMALLINT = auto() 84 INT = auto() 85 UINT = auto() 86 BIGINT = auto() 87 UBIGINT = auto() 88 INT128 = auto() 89 UINT128 = auto() 90 INT256 = auto() 91 UINT256 = auto() 92 FLOAT = auto() 93 DOUBLE = auto() 94 DECIMAL = auto() 95 BIGDECIMAL = auto() 96 CHAR = auto() 97 NCHAR = auto() 98 VARCHAR = auto() 99 NVARCHAR = auto() 100 TEXT = auto() 101 MEDIUMTEXT = auto() 102 LONGTEXT = auto() 103 MEDIUMBLOB = auto() 104 LONGBLOB = auto() 105 BINARY = auto() 106 VARBINARY = auto() 107 JSON = auto() 108 JSONB = auto() 109 TIME = auto() 110 TIMESTAMP = auto() 111 TIMESTAMPTZ = auto() 112 TIMESTAMPLTZ = auto() 113 DATETIME = auto() 114 DATETIME64 = auto() 115 DATE = auto() 116 INT4RANGE = auto() 117 INT4MULTIRANGE = auto() 118 INT8RANGE = auto() 119 INT8MULTIRANGE = auto() 120 NUMRANGE = auto() 121 NUMMULTIRANGE = auto() 122 TSRANGE = auto() 123 TSMULTIRANGE = auto() 124 TSTZRANGE = auto() 125 TSTZMULTIRANGE = auto() 126 DATERANGE = auto() 127 DATEMULTIRANGE = auto() 128 UUID = auto() 129 GEOGRAPHY = auto() 130 NULLABLE = auto() 131 GEOMETRY = auto() 132 HLLSKETCH = auto() 133 HSTORE = auto() 134 SUPER = auto() 135 SERIAL = auto() 136 SMALLSERIAL = auto() 137 BIGSERIAL = auto() 138 XML = auto() 139 UNIQUEIDENTIFIER = auto() 140 USERDEFINED = auto() 141 MONEY = auto() 142 SMALLMONEY = auto() 143 ROWVERSION = auto() 144 IMAGE = auto() 145 VARIANT = auto() 146 OBJECT = auto() 147 INET = auto() 148 ENUM = auto() 149 150 # keywords 151 ALIAS = auto() 152 ALTER = auto() 153 ALWAYS = auto() 154 ALL = auto() 155 ANTI = auto() 156 ANY = auto() 157 APPLY = auto() 158 ARRAY = auto() 159 ASC = auto() 160 ASOF = auto() 161 AUTO_INCREMENT = auto() 162 BEGIN = auto() 163 BETWEEN = auto() 164 CACHE = auto() 165 CASE = auto() 166 CHARACTER_SET = auto() 167 CLUSTER_BY = auto() 168 COLLATE = auto() 169 COMMAND = auto() 170 COMMENT = auto() 171 COMMIT = auto() 172 CONSTRAINT = auto() 173 CREATE = auto() 174 CROSS = auto() 175 CUBE = auto() 176 CURRENT_DATE = auto() 177 CURRENT_DATETIME = auto() 178 CURRENT_TIME = auto() 179 CURRENT_TIMESTAMP = auto() 180 CURRENT_USER = auto() 181 DEFAULT = auto() 182 DELETE = auto() 183 DESC = auto() 184 DESCRIBE = auto() 185 DICTIONARY = auto() 186 DISTINCT = auto() 187 DISTRIBUTE_BY = auto() 188 DIV = auto() 189 DROP = auto() 190 ELSE = auto() 191 END = auto() 192 ESCAPE = auto() 193 EXCEPT = auto() 194 EXECUTE = auto() 195 EXISTS = auto() 196 FALSE = auto() 197 FETCH = auto() 198 FILTER = auto() 199 FINAL = auto() 200 FIRST = auto() 201 FOR = auto() 202 FORCE = auto() 203 FOREIGN_KEY = auto() 204 FORMAT = auto() 205 FROM = auto() 206 FULL = auto() 207 FUNCTION = auto() 208 GLOB = auto() 209 GLOBAL = auto() 210 GROUP_BY = auto() 211 GROUPING_SETS = auto() 212 HAVING = auto() 213 HINT = auto() 214 IF = auto() 215 IGNORE = auto() 216 ILIKE = auto() 217 ILIKE_ANY = auto() 218 IN = auto() 219 INDEX = auto() 220 INNER = auto() 221 INSERT = auto() 222 INTERSECT = auto() 223 INTERVAL = auto() 224 INTO = auto() 225 INTRODUCER = auto() 226 IRLIKE = auto() 227 IS = auto() 228 ISNULL = auto() 229 JOIN = auto() 230 JOIN_MARKER = auto() 231 KEEP = auto() 232 LANGUAGE = auto() 233 LATERAL = auto() 234 LEFT = auto() 235 LIKE = auto() 236 LIKE_ANY = auto() 237 LIMIT = auto() 238 LOAD = auto() 239 LOCK = auto() 240 MAP = auto() 241 MATCH_RECOGNIZE = auto() 242 MERGE = auto() 243 MOD = auto() 244 NATURAL = auto() 245 NEXT = auto() 246 NEXT_VALUE_FOR = auto() 247 NOTNULL = auto() 248 NULL = auto() 249 OFFSET = auto() 250 ON = auto() 251 ORDER_BY = auto() 252 ORDERED = auto() 253 ORDINALITY = auto() 254 OUTER = auto() 255 OVER = auto() 256 OVERLAPS = auto() 257 OVERWRITE = auto() 258 PARTITION = auto() 259 PARTITION_BY = auto() 260 PERCENT = auto() 261 PIVOT = auto() 262 PLACEHOLDER = auto() 263 PRAGMA = auto() 264 PRIMARY_KEY = auto() 265 PROCEDURE = auto() 266 PROPERTIES = auto() 267 PSEUDO_TYPE = auto() 268 QUALIFY = auto() 269 QUOTE = auto() 270 RANGE = auto() 271 RECURSIVE = auto() 272 REPLACE = auto() 273 RETURNING = auto() 274 REFERENCES = auto() 275 RIGHT = auto() 276 RLIKE = auto() 277 ROLLBACK = auto() 278 ROLLUP = auto() 279 ROW = auto() 280 ROWS = auto() 281 SELECT = auto() 282 SEMI = auto() 283 SEPARATOR = auto() 284 SERDE_PROPERTIES = auto() 285 SET = auto() 286 SETTINGS = auto() 287 SHOW = auto() 288 SIMILAR_TO = auto() 289 SOME = auto() 290 SORT_BY = auto() 291 STRUCT = auto() 292 TABLE_SAMPLE = auto() 293 TEMPORARY = auto() 294 TOP = auto() 295 THEN = auto() 296 TRUE = auto() 297 UNCACHE = auto() 298 UNION = auto() 299 UNNEST = auto() 300 UNPIVOT = auto() 301 UPDATE = auto() 302 USE = auto() 303 USING = auto() 304 VALUES = auto() 305 VIEW = auto() 306 VOLATILE = auto() 307 WHEN = auto() 308 WHERE = auto() 309 WINDOW = auto() 310 WITH = auto() 311 UNIQUE = auto() 312 313 314class Token: 315 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 316 317 @classmethod 318 def number(cls, number: int) -> Token: 319 """Returns a NUMBER token with `number` as its text.""" 320 return cls(TokenType.NUMBER, str(number)) 321 322 @classmethod 323 def string(cls, string: str) -> Token: 324 """Returns a STRING token with `string` as its text.""" 325 return cls(TokenType.STRING, string) 326 327 @classmethod 328 def identifier(cls, identifier: str) -> Token: 329 """Returns an IDENTIFIER token with `identifier` as its text.""" 330 return cls(TokenType.IDENTIFIER, identifier) 331 332 @classmethod 333 def var(cls, var: str) -> Token: 334 """Returns an VAR token with `var` as its text.""" 335 return cls(TokenType.VAR, var) 336 337 def __init__( 338 self, 339 token_type: TokenType, 340 text: str, 341 line: int = 1, 342 col: int = 1, 343 start: int = 0, 344 end: int = 0, 345 comments: t.List[str] = [], 346 ) -> None: 347 """Token initializer. 348 349 Args: 350 token_type: The TokenType Enum. 351 text: The text of the token. 352 line: The line that the token ends on. 353 col: The column that the token ends on. 354 start: The start index of the token. 355 end: The ending index of the token. 356 comments: The comments to attach to the token. 357 """ 358 self.token_type = token_type 359 self.text = text 360 self.line = line 361 self.col = col 362 self.start = start 363 self.end = end 364 self.comments = comments 365 366 def __repr__(self) -> str: 367 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 368 return f"<Token {attributes}>" 369 370 371class _Tokenizer(type): 372 def __new__(cls, clsname, bases, attrs): 373 klass = super().__new__(cls, clsname, bases, attrs) 374 375 def _convert_quotes(arr: t.List[str | t.Tuple[str, str]]) -> t.Dict[str, str]: 376 return dict( 377 (item, item) if isinstance(item, str) else (item[0], item[1]) for item in arr 378 ) 379 380 def _quotes_to_format( 381 token_type: TokenType, arr: t.List[str | t.Tuple[str, str]] 382 ) -> t.Dict[str, t.Tuple[str, TokenType]]: 383 return {k: (v, token_type) for k, v in _convert_quotes(arr).items()} 384 385 klass._QUOTES = _convert_quotes(klass.QUOTES) 386 klass._IDENTIFIERS = _convert_quotes(klass.IDENTIFIERS) 387 388 klass._FORMAT_STRINGS = { 389 **{ 390 p + s: (e, TokenType.NATIONAL_STRING) 391 for s, e in klass._QUOTES.items() 392 for p in ("n", "N") 393 }, 394 **_quotes_to_format(TokenType.BIT_STRING, klass.BIT_STRINGS), 395 **_quotes_to_format(TokenType.BYTE_STRING, klass.BYTE_STRINGS), 396 **_quotes_to_format(TokenType.HEX_STRING, klass.HEX_STRINGS), 397 **_quotes_to_format(TokenType.RAW_STRING, klass.RAW_STRINGS), 398 } 399 400 klass._STRING_ESCAPES = set(klass.STRING_ESCAPES) 401 klass._IDENTIFIER_ESCAPES = set(klass.IDENTIFIER_ESCAPES) 402 klass._COMMENTS = { 403 **dict( 404 (comment, None) if isinstance(comment, str) else (comment[0], comment[1]) 405 for comment in klass.COMMENTS 406 ), 407 "{#": "#}", # Ensure Jinja comments are tokenized correctly in all dialects 408 } 409 410 klass._KEYWORD_TRIE = new_trie( 411 key.upper() 412 for key in ( 413 *klass.KEYWORDS, 414 *klass._COMMENTS, 415 *klass._QUOTES, 416 *klass._FORMAT_STRINGS, 417 ) 418 if " " in key or any(single in key for single in klass.SINGLE_TOKENS) 419 ) 420 421 return klass 422 423 424class Tokenizer(metaclass=_Tokenizer): 425 SINGLE_TOKENS = { 426 "(": TokenType.L_PAREN, 427 ")": TokenType.R_PAREN, 428 "[": TokenType.L_BRACKET, 429 "]": TokenType.R_BRACKET, 430 "{": TokenType.L_BRACE, 431 "}": TokenType.R_BRACE, 432 "&": TokenType.AMP, 433 "^": TokenType.CARET, 434 ":": TokenType.COLON, 435 ",": TokenType.COMMA, 436 ".": TokenType.DOT, 437 "-": TokenType.DASH, 438 "=": TokenType.EQ, 439 ">": TokenType.GT, 440 "<": TokenType.LT, 441 "%": TokenType.MOD, 442 "!": TokenType.NOT, 443 "|": TokenType.PIPE, 444 "+": TokenType.PLUS, 445 ";": TokenType.SEMICOLON, 446 "/": TokenType.SLASH, 447 "\\": TokenType.BACKSLASH, 448 "*": TokenType.STAR, 449 "~": TokenType.TILDA, 450 "?": TokenType.PLACEHOLDER, 451 "@": TokenType.PARAMETER, 452 # used for breaking a var like x'y' but nothing else 453 # the token type doesn't matter 454 "'": TokenType.QUOTE, 455 "`": TokenType.IDENTIFIER, 456 '"': TokenType.IDENTIFIER, 457 "#": TokenType.HASH, 458 } 459 460 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 461 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 462 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 463 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 464 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 465 IDENTIFIER_ESCAPES = ['"'] 466 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 467 STRING_ESCAPES = ["'"] 468 VAR_SINGLE_TOKENS: t.Set[str] = set() 469 470 # Autofilled 471 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 472 473 _COMMENTS: t.Dict[str, str] = {} 474 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 475 _IDENTIFIERS: t.Dict[str, str] = {} 476 _IDENTIFIER_ESCAPES: t.Set[str] = set() 477 _QUOTES: t.Dict[str, str] = {} 478 _STRING_ESCAPES: t.Set[str] = set() 479 _KEYWORD_TRIE: t.Dict = {} 480 481 KEYWORDS: t.Dict[str, TokenType] = { 482 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 483 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 484 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 485 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 486 "/*+": TokenType.HINT, 487 "==": TokenType.EQ, 488 "::": TokenType.DCOLON, 489 "||": TokenType.DPIPE, 490 ">=": TokenType.GTE, 491 "<=": TokenType.LTE, 492 "<>": TokenType.NEQ, 493 "!=": TokenType.NEQ, 494 "<=>": TokenType.NULLSAFE_EQ, 495 "->": TokenType.ARROW, 496 "->>": TokenType.DARROW, 497 "=>": TokenType.FARROW, 498 "#>": TokenType.HASH_ARROW, 499 "#>>": TokenType.DHASH_ARROW, 500 "<->": TokenType.LR_ARROW, 501 "&&": TokenType.DAMP, 502 "ALL": TokenType.ALL, 503 "ALWAYS": TokenType.ALWAYS, 504 "AND": TokenType.AND, 505 "ANTI": TokenType.ANTI, 506 "ANY": TokenType.ANY, 507 "ASC": TokenType.ASC, 508 "AS": TokenType.ALIAS, 509 "ASOF": TokenType.ASOF, 510 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 511 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 512 "BEGIN": TokenType.BEGIN, 513 "BETWEEN": TokenType.BETWEEN, 514 "CACHE": TokenType.CACHE, 515 "UNCACHE": TokenType.UNCACHE, 516 "CASE": TokenType.CASE, 517 "CHARACTER SET": TokenType.CHARACTER_SET, 518 "CLUSTER BY": TokenType.CLUSTER_BY, 519 "COLLATE": TokenType.COLLATE, 520 "COLUMN": TokenType.COLUMN, 521 "COMMIT": TokenType.COMMIT, 522 "CONSTRAINT": TokenType.CONSTRAINT, 523 "CREATE": TokenType.CREATE, 524 "CROSS": TokenType.CROSS, 525 "CUBE": TokenType.CUBE, 526 "CURRENT_DATE": TokenType.CURRENT_DATE, 527 "CURRENT_TIME": TokenType.CURRENT_TIME, 528 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 529 "CURRENT_USER": TokenType.CURRENT_USER, 530 "DATABASE": TokenType.DATABASE, 531 "DEFAULT": TokenType.DEFAULT, 532 "DELETE": TokenType.DELETE, 533 "DESC": TokenType.DESC, 534 "DESCRIBE": TokenType.DESCRIBE, 535 "DISTINCT": TokenType.DISTINCT, 536 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 537 "DIV": TokenType.DIV, 538 "DROP": TokenType.DROP, 539 "ELSE": TokenType.ELSE, 540 "END": TokenType.END, 541 "ESCAPE": TokenType.ESCAPE, 542 "EXCEPT": TokenType.EXCEPT, 543 "EXECUTE": TokenType.EXECUTE, 544 "EXISTS": TokenType.EXISTS, 545 "FALSE": TokenType.FALSE, 546 "FETCH": TokenType.FETCH, 547 "FILTER": TokenType.FILTER, 548 "FIRST": TokenType.FIRST, 549 "FULL": TokenType.FULL, 550 "FUNCTION": TokenType.FUNCTION, 551 "FOR": TokenType.FOR, 552 "FOREIGN KEY": TokenType.FOREIGN_KEY, 553 "FORMAT": TokenType.FORMAT, 554 "FROM": TokenType.FROM, 555 "GEOGRAPHY": TokenType.GEOGRAPHY, 556 "GEOMETRY": TokenType.GEOMETRY, 557 "GLOB": TokenType.GLOB, 558 "GROUP BY": TokenType.GROUP_BY, 559 "GROUPING SETS": TokenType.GROUPING_SETS, 560 "HAVING": TokenType.HAVING, 561 "IF": TokenType.IF, 562 "ILIKE": TokenType.ILIKE, 563 "IN": TokenType.IN, 564 "INDEX": TokenType.INDEX, 565 "INET": TokenType.INET, 566 "INNER": TokenType.INNER, 567 "INSERT": TokenType.INSERT, 568 "INTERVAL": TokenType.INTERVAL, 569 "INTERSECT": TokenType.INTERSECT, 570 "INTO": TokenType.INTO, 571 "IS": TokenType.IS, 572 "ISNULL": TokenType.ISNULL, 573 "JOIN": TokenType.JOIN, 574 "KEEP": TokenType.KEEP, 575 "LATERAL": TokenType.LATERAL, 576 "LEFT": TokenType.LEFT, 577 "LIKE": TokenType.LIKE, 578 "LIMIT": TokenType.LIMIT, 579 "LOAD": TokenType.LOAD, 580 "LOCK": TokenType.LOCK, 581 "MERGE": TokenType.MERGE, 582 "NATURAL": TokenType.NATURAL, 583 "NEXT": TokenType.NEXT, 584 "NEXT VALUE FOR": TokenType.NEXT_VALUE_FOR, 585 "NOT": TokenType.NOT, 586 "NOTNULL": TokenType.NOTNULL, 587 "NULL": TokenType.NULL, 588 "OBJECT": TokenType.OBJECT, 589 "OFFSET": TokenType.OFFSET, 590 "ON": TokenType.ON, 591 "OR": TokenType.OR, 592 "ORDER BY": TokenType.ORDER_BY, 593 "ORDINALITY": TokenType.ORDINALITY, 594 "OUTER": TokenType.OUTER, 595 "OVER": TokenType.OVER, 596 "OVERLAPS": TokenType.OVERLAPS, 597 "OVERWRITE": TokenType.OVERWRITE, 598 "PARTITION": TokenType.PARTITION, 599 "PARTITION BY": TokenType.PARTITION_BY, 600 "PARTITIONED BY": TokenType.PARTITION_BY, 601 "PARTITIONED_BY": TokenType.PARTITION_BY, 602 "PERCENT": TokenType.PERCENT, 603 "PIVOT": TokenType.PIVOT, 604 "PRAGMA": TokenType.PRAGMA, 605 "PRIMARY KEY": TokenType.PRIMARY_KEY, 606 "PROCEDURE": TokenType.PROCEDURE, 607 "QUALIFY": TokenType.QUALIFY, 608 "RANGE": TokenType.RANGE, 609 "RECURSIVE": TokenType.RECURSIVE, 610 "REGEXP": TokenType.RLIKE, 611 "REPLACE": TokenType.REPLACE, 612 "RETURNING": TokenType.RETURNING, 613 "REFERENCES": TokenType.REFERENCES, 614 "RIGHT": TokenType.RIGHT, 615 "RLIKE": TokenType.RLIKE, 616 "ROLLBACK": TokenType.ROLLBACK, 617 "ROLLUP": TokenType.ROLLUP, 618 "ROW": TokenType.ROW, 619 "ROWS": TokenType.ROWS, 620 "SCHEMA": TokenType.SCHEMA, 621 "SELECT": TokenType.SELECT, 622 "SEMI": TokenType.SEMI, 623 "SET": TokenType.SET, 624 "SETTINGS": TokenType.SETTINGS, 625 "SHOW": TokenType.SHOW, 626 "SIMILAR TO": TokenType.SIMILAR_TO, 627 "SOME": TokenType.SOME, 628 "SORT BY": TokenType.SORT_BY, 629 "TABLE": TokenType.TABLE, 630 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 631 "TEMP": TokenType.TEMPORARY, 632 "TEMPORARY": TokenType.TEMPORARY, 633 "THEN": TokenType.THEN, 634 "TRUE": TokenType.TRUE, 635 "UNION": TokenType.UNION, 636 "UNNEST": TokenType.UNNEST, 637 "UNPIVOT": TokenType.UNPIVOT, 638 "UPDATE": TokenType.UPDATE, 639 "USE": TokenType.USE, 640 "USING": TokenType.USING, 641 "UUID": TokenType.UUID, 642 "VALUES": TokenType.VALUES, 643 "VIEW": TokenType.VIEW, 644 "VOLATILE": TokenType.VOLATILE, 645 "WHEN": TokenType.WHEN, 646 "WHERE": TokenType.WHERE, 647 "WINDOW": TokenType.WINDOW, 648 "WITH": TokenType.WITH, 649 "APPLY": TokenType.APPLY, 650 "ARRAY": TokenType.ARRAY, 651 "BIT": TokenType.BIT, 652 "BOOL": TokenType.BOOLEAN, 653 "BOOLEAN": TokenType.BOOLEAN, 654 "BYTE": TokenType.TINYINT, 655 "TINYINT": TokenType.TINYINT, 656 "SHORT": TokenType.SMALLINT, 657 "SMALLINT": TokenType.SMALLINT, 658 "INT2": TokenType.SMALLINT, 659 "INTEGER": TokenType.INT, 660 "INT": TokenType.INT, 661 "INT4": TokenType.INT, 662 "LONG": TokenType.BIGINT, 663 "BIGINT": TokenType.BIGINT, 664 "INT8": TokenType.BIGINT, 665 "DEC": TokenType.DECIMAL, 666 "DECIMAL": TokenType.DECIMAL, 667 "BIGDECIMAL": TokenType.BIGDECIMAL, 668 "BIGNUMERIC": TokenType.BIGDECIMAL, 669 "MAP": TokenType.MAP, 670 "NULLABLE": TokenType.NULLABLE, 671 "NUMBER": TokenType.DECIMAL, 672 "NUMERIC": TokenType.DECIMAL, 673 "FIXED": TokenType.DECIMAL, 674 "REAL": TokenType.FLOAT, 675 "FLOAT": TokenType.FLOAT, 676 "FLOAT4": TokenType.FLOAT, 677 "FLOAT8": TokenType.DOUBLE, 678 "DOUBLE": TokenType.DOUBLE, 679 "DOUBLE PRECISION": TokenType.DOUBLE, 680 "JSON": TokenType.JSON, 681 "CHAR": TokenType.CHAR, 682 "CHARACTER": TokenType.CHAR, 683 "NCHAR": TokenType.NCHAR, 684 "VARCHAR": TokenType.VARCHAR, 685 "VARCHAR2": TokenType.VARCHAR, 686 "NVARCHAR": TokenType.NVARCHAR, 687 "NVARCHAR2": TokenType.NVARCHAR, 688 "STR": TokenType.TEXT, 689 "STRING": TokenType.TEXT, 690 "TEXT": TokenType.TEXT, 691 "CLOB": TokenType.TEXT, 692 "LONGVARCHAR": TokenType.TEXT, 693 "BINARY": TokenType.BINARY, 694 "BLOB": TokenType.VARBINARY, 695 "BYTEA": TokenType.VARBINARY, 696 "VARBINARY": TokenType.VARBINARY, 697 "TIME": TokenType.TIME, 698 "TIMESTAMP": TokenType.TIMESTAMP, 699 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 700 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 701 "DATE": TokenType.DATE, 702 "DATETIME": TokenType.DATETIME, 703 "INT4RANGE": TokenType.INT4RANGE, 704 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 705 "INT8RANGE": TokenType.INT8RANGE, 706 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 707 "NUMRANGE": TokenType.NUMRANGE, 708 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 709 "TSRANGE": TokenType.TSRANGE, 710 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 711 "TSTZRANGE": TokenType.TSTZRANGE, 712 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 713 "DATERANGE": TokenType.DATERANGE, 714 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 715 "UNIQUE": TokenType.UNIQUE, 716 "STRUCT": TokenType.STRUCT, 717 "VARIANT": TokenType.VARIANT, 718 "ALTER": TokenType.ALTER, 719 "ANALYZE": TokenType.COMMAND, 720 "CALL": TokenType.COMMAND, 721 "COMMENT": TokenType.COMMENT, 722 "COPY": TokenType.COMMAND, 723 "EXPLAIN": TokenType.COMMAND, 724 "GRANT": TokenType.COMMAND, 725 "OPTIMIZE": TokenType.COMMAND, 726 "PREPARE": TokenType.COMMAND, 727 "TRUNCATE": TokenType.COMMAND, 728 "VACUUM": TokenType.COMMAND, 729 "USER-DEFINED": TokenType.USERDEFINED, 730 } 731 732 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 733 " ": TokenType.SPACE, 734 "\t": TokenType.SPACE, 735 "\n": TokenType.BREAK, 736 "\r": TokenType.BREAK, 737 "\r\n": TokenType.BREAK, 738 } 739 740 COMMANDS = { 741 TokenType.COMMAND, 742 TokenType.EXECUTE, 743 TokenType.FETCH, 744 TokenType.SHOW, 745 } 746 747 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 748 749 # handle numeric literals like in hive (3L = BIGINT) 750 NUMERIC_LITERALS: t.Dict[str, str] = {} 751 ENCODE: t.Optional[str] = None 752 753 COMMENTS = ["--", ("/*", "*/")] 754 755 __slots__ = ( 756 "sql", 757 "size", 758 "tokens", 759 "_start", 760 "_current", 761 "_line", 762 "_col", 763 "_comments", 764 "_char", 765 "_end", 766 "_peek", 767 "_prev_token_line", 768 ) 769 770 def __init__(self) -> None: 771 self.reset() 772 773 def reset(self) -> None: 774 self.sql = "" 775 self.size = 0 776 self.tokens: t.List[Token] = [] 777 self._start = 0 778 self._current = 0 779 self._line = 1 780 self._col = 0 781 self._comments: t.List[str] = [] 782 783 self._char = "" 784 self._end = False 785 self._peek = "" 786 self._prev_token_line = -1 787 788 def tokenize(self, sql: str) -> t.List[Token]: 789 """Returns a list of tokens corresponding to the SQL string `sql`.""" 790 self.reset() 791 self.sql = sql 792 self.size = len(sql) 793 794 try: 795 self._scan() 796 except Exception as e: 797 start = max(self._current - 50, 0) 798 end = min(self._current + 50, self.size - 1) 799 context = self.sql[start:end] 800 raise ValueError(f"Error tokenizing '{context}'") from e 801 802 return self.tokens 803 804 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 805 while self.size and not self._end: 806 self._start = self._current 807 self._advance() 808 809 if self._char is None: 810 break 811 812 if self._char not in self.WHITE_SPACE: 813 if self._char.isdigit(): 814 self._scan_number() 815 elif self._char in self._IDENTIFIERS: 816 self._scan_identifier(self._IDENTIFIERS[self._char]) 817 else: 818 self._scan_keywords() 819 820 if until and until(): 821 break 822 823 if self.tokens and self._comments: 824 self.tokens[-1].comments.extend(self._comments) 825 826 def _chars(self, size: int) -> str: 827 if size == 1: 828 return self._char 829 830 start = self._current - 1 831 end = start + size 832 833 return self.sql[start:end] if end <= self.size else "" 834 835 def _advance(self, i: int = 1, alnum: bool = False) -> None: 836 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 837 self._col = 1 838 self._line += 1 839 else: 840 self._col += i 841 842 self._current += i 843 self._end = self._current >= self.size 844 self._char = self.sql[self._current - 1] 845 self._peek = "" if self._end else self.sql[self._current] 846 847 if alnum and self._char.isalnum(): 848 # Here we use local variables instead of attributes for better performance 849 _col = self._col 850 _current = self._current 851 _end = self._end 852 _peek = self._peek 853 854 while _peek.isalnum(): 855 _col += 1 856 _current += 1 857 _end = _current >= self.size 858 _peek = "" if _end else self.sql[_current] 859 860 self._col = _col 861 self._current = _current 862 self._end = _end 863 self._peek = _peek 864 self._char = self.sql[_current - 1] 865 866 @property 867 def _text(self) -> str: 868 return self.sql[self._start : self._current] 869 870 def peek(self, i: int = 0) -> str: 871 i = self._current + i 872 if i < self.size: 873 return self.sql[i] 874 return "" 875 876 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 877 self._prev_token_line = self._line 878 self.tokens.append( 879 Token( 880 token_type, 881 text=self._text if text is None else text, 882 line=self._line, 883 col=self._col, 884 start=self._start, 885 end=self._current - 1, 886 comments=self._comments, 887 ) 888 ) 889 self._comments = [] 890 891 # If we have either a semicolon or a begin token before the command's token, we'll parse 892 # whatever follows the command's token as a string 893 if ( 894 token_type in self.COMMANDS 895 and self._peek != ";" 896 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 897 ): 898 start = self._current 899 tokens = len(self.tokens) 900 self._scan(lambda: self._peek == ";") 901 self.tokens = self.tokens[:tokens] 902 text = self.sql[start : self._current].strip() 903 if text: 904 self._add(TokenType.STRING, text) 905 906 def _scan_keywords(self) -> None: 907 size = 0 908 word = None 909 chars = self._text 910 char = chars 911 prev_space = False 912 skip = False 913 trie = self._KEYWORD_TRIE 914 single_token = char in self.SINGLE_TOKENS 915 916 while chars: 917 if skip: 918 result = TrieResult.PREFIX 919 else: 920 result, trie = in_trie(trie, char.upper()) 921 922 if result == TrieResult.FAILED: 923 break 924 if result == TrieResult.EXISTS: 925 word = chars 926 927 size += 1 928 end = self._current - 1 + size 929 930 if end < self.size: 931 char = self.sql[end] 932 single_token = single_token or char in self.SINGLE_TOKENS 933 is_space = char in self.WHITE_SPACE 934 935 if not is_space or not prev_space: 936 if is_space: 937 char = " " 938 chars += char 939 prev_space = is_space 940 skip = False 941 else: 942 skip = True 943 else: 944 char = "" 945 chars = " " 946 947 word = None if not single_token and chars[-1] not in self.WHITE_SPACE else word 948 949 if not word: 950 if self._char in self.SINGLE_TOKENS: 951 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 952 return 953 self._scan_var() 954 return 955 956 if self._scan_string(word): 957 return 958 if self._scan_comment(word): 959 return 960 961 self._advance(size - 1) 962 word = word.upper() 963 self._add(self.KEYWORDS[word], text=word) 964 965 def _scan_comment(self, comment_start: str) -> bool: 966 if comment_start not in self._COMMENTS: 967 return False 968 969 comment_start_line = self._line 970 comment_start_size = len(comment_start) 971 comment_end = self._COMMENTS[comment_start] 972 973 if comment_end: 974 # Skip the comment's start delimiter 975 self._advance(comment_start_size) 976 977 comment_end_size = len(comment_end) 978 while not self._end and self._chars(comment_end_size) != comment_end: 979 self._advance(alnum=True) 980 981 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 982 self._advance(comment_end_size - 1) 983 else: 984 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 985 self._advance(alnum=True) 986 self._comments.append(self._text[comment_start_size:]) 987 988 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 989 # Multiple consecutive comments are preserved by appending them to the current comments list. 990 if comment_start_line == self._prev_token_line: 991 self.tokens[-1].comments.extend(self._comments) 992 self._comments = [] 993 self._prev_token_line = self._line 994 995 return True 996 997 def _scan_number(self) -> None: 998 if self._char == "0": 999 peek = self._peek.upper() 1000 if peek == "B": 1001 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1002 elif peek == "X": 1003 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1004 1005 decimal = False 1006 scientific = 0 1007 1008 while True: 1009 if self._peek.isdigit(): 1010 self._advance() 1011 elif self._peek == "." and not decimal: 1012 after = self.peek(1) 1013 if after.isdigit() or not after.isalpha(): 1014 decimal = True 1015 self._advance() 1016 else: 1017 return self._add(TokenType.VAR) 1018 elif self._peek in ("-", "+") and scientific == 1: 1019 scientific += 1 1020 self._advance() 1021 elif self._peek.upper() == "E" and not scientific: 1022 scientific += 1 1023 self._advance() 1024 elif self._peek.isidentifier(): 1025 number_text = self._text 1026 literal = "" 1027 1028 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1029 literal += self._peek.upper() 1030 self._advance() 1031 1032 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1033 1034 if token_type: 1035 self._add(TokenType.NUMBER, number_text) 1036 self._add(TokenType.DCOLON, "::") 1037 return self._add(token_type, literal) 1038 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1039 return self._add(TokenType.VAR) 1040 1041 self._add(TokenType.NUMBER, number_text) 1042 return self._advance(-len(literal)) 1043 else: 1044 return self._add(TokenType.NUMBER) 1045 1046 def _scan_bits(self) -> None: 1047 self._advance() 1048 value = self._extract_value() 1049 try: 1050 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1051 int(value, 2) 1052 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1053 except ValueError: 1054 self._add(TokenType.IDENTIFIER) 1055 1056 def _scan_hex(self) -> None: 1057 self._advance() 1058 value = self._extract_value() 1059 try: 1060 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1061 int(value, 16) 1062 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1063 except ValueError: 1064 self._add(TokenType.IDENTIFIER) 1065 1066 def _extract_value(self) -> str: 1067 while True: 1068 char = self._peek.strip() 1069 if char and char not in self.SINGLE_TOKENS: 1070 self._advance(alnum=True) 1071 else: 1072 break 1073 1074 return self._text 1075 1076 def _scan_string(self, start: str) -> bool: 1077 base = None 1078 token_type = TokenType.STRING 1079 1080 if start in self._QUOTES: 1081 end = self._QUOTES[start] 1082 elif start in self._FORMAT_STRINGS: 1083 end, token_type = self._FORMAT_STRINGS[start] 1084 1085 if token_type == TokenType.HEX_STRING: 1086 base = 16 1087 elif token_type == TokenType.BIT_STRING: 1088 base = 2 1089 else: 1090 return False 1091 1092 self._advance(len(start)) 1093 text = self._extract_string(end) 1094 1095 if base: 1096 try: 1097 int(text, base) 1098 except: 1099 raise RuntimeError( 1100 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1101 ) 1102 else: 1103 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1104 1105 self._add(token_type, text) 1106 return True 1107 1108 def _scan_identifier(self, identifier_end: str) -> None: 1109 self._advance() 1110 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1111 self._add(TokenType.IDENTIFIER, text) 1112 1113 def _scan_var(self) -> None: 1114 while True: 1115 char = self._peek.strip() 1116 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1117 self._advance(alnum=True) 1118 else: 1119 break 1120 1121 self._add( 1122 TokenType.VAR 1123 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1124 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1125 ) 1126 1127 def _extract_string(self, delimiter: str, escapes=None) -> str: 1128 text = "" 1129 delim_size = len(delimiter) 1130 escapes = self._STRING_ESCAPES if escapes is None else escapes 1131 1132 while True: 1133 if self._char in escapes and (self._peek == delimiter or self._peek in escapes): 1134 if self._peek == delimiter: 1135 text += self._peek 1136 else: 1137 text += self._char + self._peek 1138 1139 if self._current + 1 < self.size: 1140 self._advance(2) 1141 else: 1142 raise RuntimeError(f"Missing {delimiter} from {self._line}:{self._current}") 1143 else: 1144 if self._chars(delim_size) == delimiter: 1145 if delim_size > 1: 1146 self._advance(delim_size - 1) 1147 break 1148 1149 if self._end: 1150 raise RuntimeError(f"Missing {delimiter} from {self._line}:{self._start}") 1151 1152 current = self._current - 1 1153 self._advance(alnum=True) 1154 text += self.sql[current : self._current - 1] 1155 1156 return text
11class TokenType(AutoName): 12 L_PAREN = auto() 13 R_PAREN = auto() 14 L_BRACKET = auto() 15 R_BRACKET = auto() 16 L_BRACE = auto() 17 R_BRACE = auto() 18 COMMA = auto() 19 DOT = auto() 20 DASH = auto() 21 PLUS = auto() 22 COLON = auto() 23 DCOLON = auto() 24 SEMICOLON = auto() 25 STAR = auto() 26 BACKSLASH = auto() 27 SLASH = auto() 28 LT = auto() 29 LTE = auto() 30 GT = auto() 31 GTE = auto() 32 NOT = auto() 33 EQ = auto() 34 NEQ = auto() 35 NULLSAFE_EQ = auto() 36 AND = auto() 37 OR = auto() 38 AMP = auto() 39 DPIPE = auto() 40 PIPE = auto() 41 CARET = auto() 42 TILDA = auto() 43 ARROW = auto() 44 DARROW = auto() 45 FARROW = auto() 46 HASH = auto() 47 HASH_ARROW = auto() 48 DHASH_ARROW = auto() 49 LR_ARROW = auto() 50 LT_AT = auto() 51 AT_GT = auto() 52 DOLLAR = auto() 53 PARAMETER = auto() 54 SESSION_PARAMETER = auto() 55 DAMP = auto() 56 57 BLOCK_START = auto() 58 BLOCK_END = auto() 59 60 SPACE = auto() 61 BREAK = auto() 62 63 STRING = auto() 64 NUMBER = auto() 65 IDENTIFIER = auto() 66 DATABASE = auto() 67 COLUMN = auto() 68 COLUMN_DEF = auto() 69 SCHEMA = auto() 70 TABLE = auto() 71 VAR = auto() 72 BIT_STRING = auto() 73 HEX_STRING = auto() 74 BYTE_STRING = auto() 75 NATIONAL_STRING = auto() 76 RAW_STRING = auto() 77 78 # types 79 BIT = auto() 80 BOOLEAN = auto() 81 TINYINT = auto() 82 UTINYINT = auto() 83 SMALLINT = auto() 84 USMALLINT = auto() 85 INT = auto() 86 UINT = auto() 87 BIGINT = auto() 88 UBIGINT = auto() 89 INT128 = auto() 90 UINT128 = auto() 91 INT256 = auto() 92 UINT256 = auto() 93 FLOAT = auto() 94 DOUBLE = auto() 95 DECIMAL = auto() 96 BIGDECIMAL = auto() 97 CHAR = auto() 98 NCHAR = auto() 99 VARCHAR = auto() 100 NVARCHAR = auto() 101 TEXT = auto() 102 MEDIUMTEXT = auto() 103 LONGTEXT = auto() 104 MEDIUMBLOB = auto() 105 LONGBLOB = auto() 106 BINARY = auto() 107 VARBINARY = auto() 108 JSON = auto() 109 JSONB = auto() 110 TIME = auto() 111 TIMESTAMP = auto() 112 TIMESTAMPTZ = auto() 113 TIMESTAMPLTZ = auto() 114 DATETIME = auto() 115 DATETIME64 = auto() 116 DATE = auto() 117 INT4RANGE = auto() 118 INT4MULTIRANGE = auto() 119 INT8RANGE = auto() 120 INT8MULTIRANGE = auto() 121 NUMRANGE = auto() 122 NUMMULTIRANGE = auto() 123 TSRANGE = auto() 124 TSMULTIRANGE = auto() 125 TSTZRANGE = auto() 126 TSTZMULTIRANGE = auto() 127 DATERANGE = auto() 128 DATEMULTIRANGE = auto() 129 UUID = auto() 130 GEOGRAPHY = auto() 131 NULLABLE = auto() 132 GEOMETRY = auto() 133 HLLSKETCH = auto() 134 HSTORE = auto() 135 SUPER = auto() 136 SERIAL = auto() 137 SMALLSERIAL = auto() 138 BIGSERIAL = auto() 139 XML = auto() 140 UNIQUEIDENTIFIER = auto() 141 USERDEFINED = auto() 142 MONEY = auto() 143 SMALLMONEY = auto() 144 ROWVERSION = auto() 145 IMAGE = auto() 146 VARIANT = auto() 147 OBJECT = auto() 148 INET = auto() 149 ENUM = auto() 150 151 # keywords 152 ALIAS = auto() 153 ALTER = auto() 154 ALWAYS = auto() 155 ALL = auto() 156 ANTI = auto() 157 ANY = auto() 158 APPLY = auto() 159 ARRAY = auto() 160 ASC = auto() 161 ASOF = auto() 162 AUTO_INCREMENT = auto() 163 BEGIN = auto() 164 BETWEEN = auto() 165 CACHE = auto() 166 CASE = auto() 167 CHARACTER_SET = auto() 168 CLUSTER_BY = auto() 169 COLLATE = auto() 170 COMMAND = auto() 171 COMMENT = auto() 172 COMMIT = auto() 173 CONSTRAINT = auto() 174 CREATE = auto() 175 CROSS = auto() 176 CUBE = auto() 177 CURRENT_DATE = auto() 178 CURRENT_DATETIME = auto() 179 CURRENT_TIME = auto() 180 CURRENT_TIMESTAMP = auto() 181 CURRENT_USER = auto() 182 DEFAULT = auto() 183 DELETE = auto() 184 DESC = auto() 185 DESCRIBE = auto() 186 DICTIONARY = auto() 187 DISTINCT = auto() 188 DISTRIBUTE_BY = auto() 189 DIV = auto() 190 DROP = auto() 191 ELSE = auto() 192 END = auto() 193 ESCAPE = auto() 194 EXCEPT = auto() 195 EXECUTE = auto() 196 EXISTS = auto() 197 FALSE = auto() 198 FETCH = auto() 199 FILTER = auto() 200 FINAL = auto() 201 FIRST = auto() 202 FOR = auto() 203 FORCE = auto() 204 FOREIGN_KEY = auto() 205 FORMAT = auto() 206 FROM = auto() 207 FULL = auto() 208 FUNCTION = auto() 209 GLOB = auto() 210 GLOBAL = auto() 211 GROUP_BY = auto() 212 GROUPING_SETS = auto() 213 HAVING = auto() 214 HINT = auto() 215 IF = auto() 216 IGNORE = auto() 217 ILIKE = auto() 218 ILIKE_ANY = auto() 219 IN = auto() 220 INDEX = auto() 221 INNER = auto() 222 INSERT = auto() 223 INTERSECT = auto() 224 INTERVAL = auto() 225 INTO = auto() 226 INTRODUCER = auto() 227 IRLIKE = auto() 228 IS = auto() 229 ISNULL = auto() 230 JOIN = auto() 231 JOIN_MARKER = auto() 232 KEEP = auto() 233 LANGUAGE = auto() 234 LATERAL = auto() 235 LEFT = auto() 236 LIKE = auto() 237 LIKE_ANY = auto() 238 LIMIT = auto() 239 LOAD = auto() 240 LOCK = auto() 241 MAP = auto() 242 MATCH_RECOGNIZE = auto() 243 MERGE = auto() 244 MOD = auto() 245 NATURAL = auto() 246 NEXT = auto() 247 NEXT_VALUE_FOR = auto() 248 NOTNULL = auto() 249 NULL = auto() 250 OFFSET = auto() 251 ON = auto() 252 ORDER_BY = auto() 253 ORDERED = auto() 254 ORDINALITY = auto() 255 OUTER = auto() 256 OVER = auto() 257 OVERLAPS = auto() 258 OVERWRITE = auto() 259 PARTITION = auto() 260 PARTITION_BY = auto() 261 PERCENT = auto() 262 PIVOT = auto() 263 PLACEHOLDER = auto() 264 PRAGMA = auto() 265 PRIMARY_KEY = auto() 266 PROCEDURE = auto() 267 PROPERTIES = auto() 268 PSEUDO_TYPE = auto() 269 QUALIFY = auto() 270 QUOTE = auto() 271 RANGE = auto() 272 RECURSIVE = auto() 273 REPLACE = auto() 274 RETURNING = auto() 275 REFERENCES = auto() 276 RIGHT = auto() 277 RLIKE = auto() 278 ROLLBACK = auto() 279 ROLLUP = auto() 280 ROW = auto() 281 ROWS = auto() 282 SELECT = auto() 283 SEMI = auto() 284 SEPARATOR = auto() 285 SERDE_PROPERTIES = auto() 286 SET = auto() 287 SETTINGS = auto() 288 SHOW = auto() 289 SIMILAR_TO = auto() 290 SOME = auto() 291 SORT_BY = auto() 292 STRUCT = auto() 293 TABLE_SAMPLE = auto() 294 TEMPORARY = auto() 295 TOP = auto() 296 THEN = auto() 297 TRUE = auto() 298 UNCACHE = auto() 299 UNION = auto() 300 UNNEST = auto() 301 UNPIVOT = auto() 302 UPDATE = auto() 303 USE = auto() 304 USING = auto() 305 VALUES = auto() 306 VIEW = auto() 307 VOLATILE = auto() 308 WHEN = auto() 309 WHERE = auto() 310 WINDOW = auto() 311 WITH = auto() 312 UNIQUE = auto()
An enumeration.
L_PAREN =
<TokenType.L_PAREN: 'L_PAREN'>
R_PAREN =
<TokenType.R_PAREN: 'R_PAREN'>
L_BRACKET =
<TokenType.L_BRACKET: 'L_BRACKET'>
R_BRACKET =
<TokenType.R_BRACKET: 'R_BRACKET'>
L_BRACE =
<TokenType.L_BRACE: 'L_BRACE'>
R_BRACE =
<TokenType.R_BRACE: 'R_BRACE'>
COMMA =
<TokenType.COMMA: 'COMMA'>
DOT =
<TokenType.DOT: 'DOT'>
DASH =
<TokenType.DASH: 'DASH'>
PLUS =
<TokenType.PLUS: 'PLUS'>
COLON =
<TokenType.COLON: 'COLON'>
DCOLON =
<TokenType.DCOLON: 'DCOLON'>
SEMICOLON =
<TokenType.SEMICOLON: 'SEMICOLON'>
STAR =
<TokenType.STAR: 'STAR'>
BACKSLASH =
<TokenType.BACKSLASH: 'BACKSLASH'>
SLASH =
<TokenType.SLASH: 'SLASH'>
LT =
<TokenType.LT: 'LT'>
LTE =
<TokenType.LTE: 'LTE'>
GT =
<TokenType.GT: 'GT'>
GTE =
<TokenType.GTE: 'GTE'>
NOT =
<TokenType.NOT: 'NOT'>
EQ =
<TokenType.EQ: 'EQ'>
NEQ =
<TokenType.NEQ: 'NEQ'>
NULLSAFE_EQ =
<TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>
AND =
<TokenType.AND: 'AND'>
OR =
<TokenType.OR: 'OR'>
AMP =
<TokenType.AMP: 'AMP'>
DPIPE =
<TokenType.DPIPE: 'DPIPE'>
PIPE =
<TokenType.PIPE: 'PIPE'>
CARET =
<TokenType.CARET: 'CARET'>
TILDA =
<TokenType.TILDA: 'TILDA'>
ARROW =
<TokenType.ARROW: 'ARROW'>
DARROW =
<TokenType.DARROW: 'DARROW'>
FARROW =
<TokenType.FARROW: 'FARROW'>
HASH =
<TokenType.HASH: 'HASH'>
HASH_ARROW =
<TokenType.HASH_ARROW: 'HASH_ARROW'>
DHASH_ARROW =
<TokenType.DHASH_ARROW: 'DHASH_ARROW'>
LR_ARROW =
<TokenType.LR_ARROW: 'LR_ARROW'>
LT_AT =
<TokenType.LT_AT: 'LT_AT'>
AT_GT =
<TokenType.AT_GT: 'AT_GT'>
DOLLAR =
<TokenType.DOLLAR: 'DOLLAR'>
PARAMETER =
<TokenType.PARAMETER: 'PARAMETER'>
SESSION_PARAMETER =
<TokenType.SESSION_PARAMETER: 'SESSION_PARAMETER'>
DAMP =
<TokenType.DAMP: 'DAMP'>
BLOCK_START =
<TokenType.BLOCK_START: 'BLOCK_START'>
BLOCK_END =
<TokenType.BLOCK_END: 'BLOCK_END'>
SPACE =
<TokenType.SPACE: 'SPACE'>
BREAK =
<TokenType.BREAK: 'BREAK'>
STRING =
<TokenType.STRING: 'STRING'>
NUMBER =
<TokenType.NUMBER: 'NUMBER'>
IDENTIFIER =
<TokenType.IDENTIFIER: 'IDENTIFIER'>
DATABASE =
<TokenType.DATABASE: 'DATABASE'>
COLUMN =
<TokenType.COLUMN: 'COLUMN'>
COLUMN_DEF =
<TokenType.COLUMN_DEF: 'COLUMN_DEF'>
SCHEMA =
<TokenType.SCHEMA: 'SCHEMA'>
TABLE =
<TokenType.TABLE: 'TABLE'>
VAR =
<TokenType.VAR: 'VAR'>
BIT_STRING =
<TokenType.BIT_STRING: 'BIT_STRING'>
HEX_STRING =
<TokenType.HEX_STRING: 'HEX_STRING'>
BYTE_STRING =
<TokenType.BYTE_STRING: 'BYTE_STRING'>
NATIONAL_STRING =
<TokenType.NATIONAL_STRING: 'NATIONAL_STRING'>
RAW_STRING =
<TokenType.RAW_STRING: 'RAW_STRING'>
BIT =
<TokenType.BIT: 'BIT'>
BOOLEAN =
<TokenType.BOOLEAN: 'BOOLEAN'>
TINYINT =
<TokenType.TINYINT: 'TINYINT'>
UTINYINT =
<TokenType.UTINYINT: 'UTINYINT'>
SMALLINT =
<TokenType.SMALLINT: 'SMALLINT'>
USMALLINT =
<TokenType.USMALLINT: 'USMALLINT'>
INT =
<TokenType.INT: 'INT'>
UINT =
<TokenType.UINT: 'UINT'>
BIGINT =
<TokenType.BIGINT: 'BIGINT'>
UBIGINT =
<TokenType.UBIGINT: 'UBIGINT'>
INT128 =
<TokenType.INT128: 'INT128'>
UINT128 =
<TokenType.UINT128: 'UINT128'>
INT256 =
<TokenType.INT256: 'INT256'>
UINT256 =
<TokenType.UINT256: 'UINT256'>
FLOAT =
<TokenType.FLOAT: 'FLOAT'>
DOUBLE =
<TokenType.DOUBLE: 'DOUBLE'>
DECIMAL =
<TokenType.DECIMAL: 'DECIMAL'>
BIGDECIMAL =
<TokenType.BIGDECIMAL: 'BIGDECIMAL'>
CHAR =
<TokenType.CHAR: 'CHAR'>
NCHAR =
<TokenType.NCHAR: 'NCHAR'>
VARCHAR =
<TokenType.VARCHAR: 'VARCHAR'>
NVARCHAR =
<TokenType.NVARCHAR: 'NVARCHAR'>
TEXT =
<TokenType.TEXT: 'TEXT'>
MEDIUMTEXT =
<TokenType.MEDIUMTEXT: 'MEDIUMTEXT'>
LONGTEXT =
<TokenType.LONGTEXT: 'LONGTEXT'>
MEDIUMBLOB =
<TokenType.MEDIUMBLOB: 'MEDIUMBLOB'>
LONGBLOB =
<TokenType.LONGBLOB: 'LONGBLOB'>
BINARY =
<TokenType.BINARY: 'BINARY'>
VARBINARY =
<TokenType.VARBINARY: 'VARBINARY'>
JSON =
<TokenType.JSON: 'JSON'>
JSONB =
<TokenType.JSONB: 'JSONB'>
TIME =
<TokenType.TIME: 'TIME'>
TIMESTAMP =
<TokenType.TIMESTAMP: 'TIMESTAMP'>
TIMESTAMPTZ =
<TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>
TIMESTAMPLTZ =
<TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>
DATETIME =
<TokenType.DATETIME: 'DATETIME'>
DATETIME64 =
<TokenType.DATETIME64: 'DATETIME64'>
DATE =
<TokenType.DATE: 'DATE'>
INT4RANGE =
<TokenType.INT4RANGE: 'INT4RANGE'>
INT4MULTIRANGE =
<TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>
INT8RANGE =
<TokenType.INT8RANGE: 'INT8RANGE'>
INT8MULTIRANGE =
<TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>
NUMRANGE =
<TokenType.NUMRANGE: 'NUMRANGE'>
NUMMULTIRANGE =
<TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>
TSRANGE =
<TokenType.TSRANGE: 'TSRANGE'>
TSMULTIRANGE =
<TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>
TSTZRANGE =
<TokenType.TSTZRANGE: 'TSTZRANGE'>
TSTZMULTIRANGE =
<TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>
DATERANGE =
<TokenType.DATERANGE: 'DATERANGE'>
DATEMULTIRANGE =
<TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>
UUID =
<TokenType.UUID: 'UUID'>
GEOGRAPHY =
<TokenType.GEOGRAPHY: 'GEOGRAPHY'>
NULLABLE =
<TokenType.NULLABLE: 'NULLABLE'>
GEOMETRY =
<TokenType.GEOMETRY: 'GEOMETRY'>
HLLSKETCH =
<TokenType.HLLSKETCH: 'HLLSKETCH'>
HSTORE =
<TokenType.HSTORE: 'HSTORE'>
SUPER =
<TokenType.SUPER: 'SUPER'>
SERIAL =
<TokenType.SERIAL: 'SERIAL'>
SMALLSERIAL =
<TokenType.SMALLSERIAL: 'SMALLSERIAL'>
BIGSERIAL =
<TokenType.BIGSERIAL: 'BIGSERIAL'>
XML =
<TokenType.XML: 'XML'>
UNIQUEIDENTIFIER =
<TokenType.UNIQUEIDENTIFIER: 'UNIQUEIDENTIFIER'>
USERDEFINED =
<TokenType.USERDEFINED: 'USERDEFINED'>
MONEY =
<TokenType.MONEY: 'MONEY'>
SMALLMONEY =
<TokenType.SMALLMONEY: 'SMALLMONEY'>
ROWVERSION =
<TokenType.ROWVERSION: 'ROWVERSION'>
IMAGE =
<TokenType.IMAGE: 'IMAGE'>
VARIANT =
<TokenType.VARIANT: 'VARIANT'>
OBJECT =
<TokenType.OBJECT: 'OBJECT'>
INET =
<TokenType.INET: 'INET'>
ENUM =
<TokenType.ENUM: 'ENUM'>
ALIAS =
<TokenType.ALIAS: 'ALIAS'>
ALTER =
<TokenType.ALTER: 'ALTER'>
ALWAYS =
<TokenType.ALWAYS: 'ALWAYS'>
ALL =
<TokenType.ALL: 'ALL'>
ANTI =
<TokenType.ANTI: 'ANTI'>
ANY =
<TokenType.ANY: 'ANY'>
APPLY =
<TokenType.APPLY: 'APPLY'>
ARRAY =
<TokenType.ARRAY: 'ARRAY'>
ASC =
<TokenType.ASC: 'ASC'>
ASOF =
<TokenType.ASOF: 'ASOF'>
AUTO_INCREMENT =
<TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>
BEGIN =
<TokenType.BEGIN: 'BEGIN'>
BETWEEN =
<TokenType.BETWEEN: 'BETWEEN'>
CACHE =
<TokenType.CACHE: 'CACHE'>
CASE =
<TokenType.CASE: 'CASE'>
CHARACTER_SET =
<TokenType.CHARACTER_SET: 'CHARACTER_SET'>
CLUSTER_BY =
<TokenType.CLUSTER_BY: 'CLUSTER_BY'>
COLLATE =
<TokenType.COLLATE: 'COLLATE'>
COMMAND =
<TokenType.COMMAND: 'COMMAND'>
COMMENT =
<TokenType.COMMENT: 'COMMENT'>
COMMIT =
<TokenType.COMMIT: 'COMMIT'>
CONSTRAINT =
<TokenType.CONSTRAINT: 'CONSTRAINT'>
CREATE =
<TokenType.CREATE: 'CREATE'>
CROSS =
<TokenType.CROSS: 'CROSS'>
CUBE =
<TokenType.CUBE: 'CUBE'>
CURRENT_DATE =
<TokenType.CURRENT_DATE: 'CURRENT_DATE'>
CURRENT_DATETIME =
<TokenType.CURRENT_DATETIME: 'CURRENT_DATETIME'>
CURRENT_TIME =
<TokenType.CURRENT_TIME: 'CURRENT_TIME'>
CURRENT_TIMESTAMP =
<TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>
CURRENT_USER =
<TokenType.CURRENT_USER: 'CURRENT_USER'>
DEFAULT =
<TokenType.DEFAULT: 'DEFAULT'>
DELETE =
<TokenType.DELETE: 'DELETE'>
DESC =
<TokenType.DESC: 'DESC'>
DESCRIBE =
<TokenType.DESCRIBE: 'DESCRIBE'>
DICTIONARY =
<TokenType.DICTIONARY: 'DICTIONARY'>
DISTINCT =
<TokenType.DISTINCT: 'DISTINCT'>
DISTRIBUTE_BY =
<TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>
DIV =
<TokenType.DIV: 'DIV'>
DROP =
<TokenType.DROP: 'DROP'>
ELSE =
<TokenType.ELSE: 'ELSE'>
END =
<TokenType.END: 'END'>
ESCAPE =
<TokenType.ESCAPE: 'ESCAPE'>
EXCEPT =
<TokenType.EXCEPT: 'EXCEPT'>
EXECUTE =
<TokenType.EXECUTE: 'EXECUTE'>
EXISTS =
<TokenType.EXISTS: 'EXISTS'>
FALSE =
<TokenType.FALSE: 'FALSE'>
FETCH =
<TokenType.FETCH: 'FETCH'>
FILTER =
<TokenType.FILTER: 'FILTER'>
FINAL =
<TokenType.FINAL: 'FINAL'>
FIRST =
<TokenType.FIRST: 'FIRST'>
FOR =
<TokenType.FOR: 'FOR'>
FORCE =
<TokenType.FORCE: 'FORCE'>
FOREIGN_KEY =
<TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>
FORMAT =
<TokenType.FORMAT: 'FORMAT'>
FROM =
<TokenType.FROM: 'FROM'>
FULL =
<TokenType.FULL: 'FULL'>
FUNCTION =
<TokenType.FUNCTION: 'FUNCTION'>
GLOB =
<TokenType.GLOB: 'GLOB'>
GLOBAL =
<TokenType.GLOBAL: 'GLOBAL'>
GROUP_BY =
<TokenType.GROUP_BY: 'GROUP_BY'>
GROUPING_SETS =
<TokenType.GROUPING_SETS: 'GROUPING_SETS'>
HAVING =
<TokenType.HAVING: 'HAVING'>
HINT =
<TokenType.HINT: 'HINT'>
IF =
<TokenType.IF: 'IF'>
IGNORE =
<TokenType.IGNORE: 'IGNORE'>
ILIKE =
<TokenType.ILIKE: 'ILIKE'>
ILIKE_ANY =
<TokenType.ILIKE_ANY: 'ILIKE_ANY'>
IN =
<TokenType.IN: 'IN'>
INDEX =
<TokenType.INDEX: 'INDEX'>
INNER =
<TokenType.INNER: 'INNER'>
INSERT =
<TokenType.INSERT: 'INSERT'>
INTERSECT =
<TokenType.INTERSECT: 'INTERSECT'>
INTERVAL =
<TokenType.INTERVAL: 'INTERVAL'>
INTO =
<TokenType.INTO: 'INTO'>
INTRODUCER =
<TokenType.INTRODUCER: 'INTRODUCER'>
IRLIKE =
<TokenType.IRLIKE: 'IRLIKE'>
IS =
<TokenType.IS: 'IS'>
ISNULL =
<TokenType.ISNULL: 'ISNULL'>
JOIN =
<TokenType.JOIN: 'JOIN'>
JOIN_MARKER =
<TokenType.JOIN_MARKER: 'JOIN_MARKER'>
KEEP =
<TokenType.KEEP: 'KEEP'>
LANGUAGE =
<TokenType.LANGUAGE: 'LANGUAGE'>
LATERAL =
<TokenType.LATERAL: 'LATERAL'>
LEFT =
<TokenType.LEFT: 'LEFT'>
LIKE =
<TokenType.LIKE: 'LIKE'>
LIKE_ANY =
<TokenType.LIKE_ANY: 'LIKE_ANY'>
LIMIT =
<TokenType.LIMIT: 'LIMIT'>
LOAD =
<TokenType.LOAD: 'LOAD'>
LOCK =
<TokenType.LOCK: 'LOCK'>
MAP =
<TokenType.MAP: 'MAP'>
MATCH_RECOGNIZE =
<TokenType.MATCH_RECOGNIZE: 'MATCH_RECOGNIZE'>
MERGE =
<TokenType.MERGE: 'MERGE'>
MOD =
<TokenType.MOD: 'MOD'>
NATURAL =
<TokenType.NATURAL: 'NATURAL'>
NEXT =
<TokenType.NEXT: 'NEXT'>
NEXT_VALUE_FOR =
<TokenType.NEXT_VALUE_FOR: 'NEXT_VALUE_FOR'>
NOTNULL =
<TokenType.NOTNULL: 'NOTNULL'>
NULL =
<TokenType.NULL: 'NULL'>
OFFSET =
<TokenType.OFFSET: 'OFFSET'>
ON =
<TokenType.ON: 'ON'>
ORDER_BY =
<TokenType.ORDER_BY: 'ORDER_BY'>
ORDERED =
<TokenType.ORDERED: 'ORDERED'>
ORDINALITY =
<TokenType.ORDINALITY: 'ORDINALITY'>
OUTER =
<TokenType.OUTER: 'OUTER'>
OVER =
<TokenType.OVER: 'OVER'>
OVERLAPS =
<TokenType.OVERLAPS: 'OVERLAPS'>
OVERWRITE =
<TokenType.OVERWRITE: 'OVERWRITE'>
PARTITION =
<TokenType.PARTITION: 'PARTITION'>
PARTITION_BY =
<TokenType.PARTITION_BY: 'PARTITION_BY'>
PERCENT =
<TokenType.PERCENT: 'PERCENT'>
PIVOT =
<TokenType.PIVOT: 'PIVOT'>
PLACEHOLDER =
<TokenType.PLACEHOLDER: 'PLACEHOLDER'>
PRAGMA =
<TokenType.PRAGMA: 'PRAGMA'>
PRIMARY_KEY =
<TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>
PROCEDURE =
<TokenType.PROCEDURE: 'PROCEDURE'>
PROPERTIES =
<TokenType.PROPERTIES: 'PROPERTIES'>
PSEUDO_TYPE =
<TokenType.PSEUDO_TYPE: 'PSEUDO_TYPE'>
QUALIFY =
<TokenType.QUALIFY: 'QUALIFY'>
QUOTE =
<TokenType.QUOTE: 'QUOTE'>
RANGE =
<TokenType.RANGE: 'RANGE'>
RECURSIVE =
<TokenType.RECURSIVE: 'RECURSIVE'>
REPLACE =
<TokenType.REPLACE: 'REPLACE'>
RETURNING =
<TokenType.RETURNING: 'RETURNING'>
REFERENCES =
<TokenType.REFERENCES: 'REFERENCES'>
RIGHT =
<TokenType.RIGHT: 'RIGHT'>
RLIKE =
<TokenType.RLIKE: 'RLIKE'>
ROLLBACK =
<TokenType.ROLLBACK: 'ROLLBACK'>
ROLLUP =
<TokenType.ROLLUP: 'ROLLUP'>
ROW =
<TokenType.ROW: 'ROW'>
ROWS =
<TokenType.ROWS: 'ROWS'>
SELECT =
<TokenType.SELECT: 'SELECT'>
SEMI =
<TokenType.SEMI: 'SEMI'>
SEPARATOR =
<TokenType.SEPARATOR: 'SEPARATOR'>
SERDE_PROPERTIES =
<TokenType.SERDE_PROPERTIES: 'SERDE_PROPERTIES'>
SET =
<TokenType.SET: 'SET'>
SETTINGS =
<TokenType.SETTINGS: 'SETTINGS'>
SHOW =
<TokenType.SHOW: 'SHOW'>
SIMILAR_TO =
<TokenType.SIMILAR_TO: 'SIMILAR_TO'>
SOME =
<TokenType.SOME: 'SOME'>
SORT_BY =
<TokenType.SORT_BY: 'SORT_BY'>
STRUCT =
<TokenType.STRUCT: 'STRUCT'>
TABLE_SAMPLE =
<TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>
TEMPORARY =
<TokenType.TEMPORARY: 'TEMPORARY'>
TOP =
<TokenType.TOP: 'TOP'>
THEN =
<TokenType.THEN: 'THEN'>
TRUE =
<TokenType.TRUE: 'TRUE'>
UNCACHE =
<TokenType.UNCACHE: 'UNCACHE'>
UNION =
<TokenType.UNION: 'UNION'>
UNNEST =
<TokenType.UNNEST: 'UNNEST'>
UNPIVOT =
<TokenType.UNPIVOT: 'UNPIVOT'>
UPDATE =
<TokenType.UPDATE: 'UPDATE'>
USE =
<TokenType.USE: 'USE'>
USING =
<TokenType.USING: 'USING'>
VALUES =
<TokenType.VALUES: 'VALUES'>
VIEW =
<TokenType.VIEW: 'VIEW'>
VOLATILE =
<TokenType.VOLATILE: 'VOLATILE'>
WHEN =
<TokenType.WHEN: 'WHEN'>
WHERE =
<TokenType.WHERE: 'WHERE'>
WINDOW =
<TokenType.WINDOW: 'WINDOW'>
WITH =
<TokenType.WITH: 'WITH'>
UNIQUE =
<TokenType.UNIQUE: 'UNIQUE'>
Inherited Members
- enum.Enum
- name
- value
class
Token:
315class Token: 316 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 317 318 @classmethod 319 def number(cls, number: int) -> Token: 320 """Returns a NUMBER token with `number` as its text.""" 321 return cls(TokenType.NUMBER, str(number)) 322 323 @classmethod 324 def string(cls, string: str) -> Token: 325 """Returns a STRING token with `string` as its text.""" 326 return cls(TokenType.STRING, string) 327 328 @classmethod 329 def identifier(cls, identifier: str) -> Token: 330 """Returns an IDENTIFIER token with `identifier` as its text.""" 331 return cls(TokenType.IDENTIFIER, identifier) 332 333 @classmethod 334 def var(cls, var: str) -> Token: 335 """Returns an VAR token with `var` as its text.""" 336 return cls(TokenType.VAR, var) 337 338 def __init__( 339 self, 340 token_type: TokenType, 341 text: str, 342 line: int = 1, 343 col: int = 1, 344 start: int = 0, 345 end: int = 0, 346 comments: t.List[str] = [], 347 ) -> None: 348 """Token initializer. 349 350 Args: 351 token_type: The TokenType Enum. 352 text: The text of the token. 353 line: The line that the token ends on. 354 col: The column that the token ends on. 355 start: The start index of the token. 356 end: The ending index of the token. 357 comments: The comments to attach to the token. 358 """ 359 self.token_type = token_type 360 self.text = text 361 self.line = line 362 self.col = col 363 self.start = start 364 self.end = end 365 self.comments = comments 366 367 def __repr__(self) -> str: 368 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 369 return f"<Token {attributes}>"
Token( token_type: sqlglot.tokens.TokenType, text: str, line: int = 1, col: int = 1, start: int = 0, end: int = 0, comments: List[str] = [])
338 def __init__( 339 self, 340 token_type: TokenType, 341 text: str, 342 line: int = 1, 343 col: int = 1, 344 start: int = 0, 345 end: int = 0, 346 comments: t.List[str] = [], 347 ) -> None: 348 """Token initializer. 349 350 Args: 351 token_type: The TokenType Enum. 352 text: The text of the token. 353 line: The line that the token ends on. 354 col: The column that the token ends on. 355 start: The start index of the token. 356 end: The ending index of the token. 357 comments: The comments to attach to the token. 358 """ 359 self.token_type = token_type 360 self.text = text 361 self.line = line 362 self.col = col 363 self.start = start 364 self.end = end 365 self.comments = comments
Token initializer.
Arguments:
- token_type: The TokenType Enum.
- text: The text of the token.
- line: The line that the token ends on.
- col: The column that the token ends on.
- start: The start index of the token.
- end: The ending index of the token.
- comments: The comments to attach to the token.
318 @classmethod 319 def number(cls, number: int) -> Token: 320 """Returns a NUMBER token with `number` as its text.""" 321 return cls(TokenType.NUMBER, str(number))
Returns a NUMBER token with number
as its text.
323 @classmethod 324 def string(cls, string: str) -> Token: 325 """Returns a STRING token with `string` as its text.""" 326 return cls(TokenType.STRING, string)
Returns a STRING token with string
as its text.
328 @classmethod 329 def identifier(cls, identifier: str) -> Token: 330 """Returns an IDENTIFIER token with `identifier` as its text.""" 331 return cls(TokenType.IDENTIFIER, identifier)
Returns an IDENTIFIER token with identifier
as its text.
class
Tokenizer:
425class Tokenizer(metaclass=_Tokenizer): 426 SINGLE_TOKENS = { 427 "(": TokenType.L_PAREN, 428 ")": TokenType.R_PAREN, 429 "[": TokenType.L_BRACKET, 430 "]": TokenType.R_BRACKET, 431 "{": TokenType.L_BRACE, 432 "}": TokenType.R_BRACE, 433 "&": TokenType.AMP, 434 "^": TokenType.CARET, 435 ":": TokenType.COLON, 436 ",": TokenType.COMMA, 437 ".": TokenType.DOT, 438 "-": TokenType.DASH, 439 "=": TokenType.EQ, 440 ">": TokenType.GT, 441 "<": TokenType.LT, 442 "%": TokenType.MOD, 443 "!": TokenType.NOT, 444 "|": TokenType.PIPE, 445 "+": TokenType.PLUS, 446 ";": TokenType.SEMICOLON, 447 "/": TokenType.SLASH, 448 "\\": TokenType.BACKSLASH, 449 "*": TokenType.STAR, 450 "~": TokenType.TILDA, 451 "?": TokenType.PLACEHOLDER, 452 "@": TokenType.PARAMETER, 453 # used for breaking a var like x'y' but nothing else 454 # the token type doesn't matter 455 "'": TokenType.QUOTE, 456 "`": TokenType.IDENTIFIER, 457 '"': TokenType.IDENTIFIER, 458 "#": TokenType.HASH, 459 } 460 461 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 462 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 463 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 464 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 465 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 466 IDENTIFIER_ESCAPES = ['"'] 467 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 468 STRING_ESCAPES = ["'"] 469 VAR_SINGLE_TOKENS: t.Set[str] = set() 470 471 # Autofilled 472 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 473 474 _COMMENTS: t.Dict[str, str] = {} 475 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 476 _IDENTIFIERS: t.Dict[str, str] = {} 477 _IDENTIFIER_ESCAPES: t.Set[str] = set() 478 _QUOTES: t.Dict[str, str] = {} 479 _STRING_ESCAPES: t.Set[str] = set() 480 _KEYWORD_TRIE: t.Dict = {} 481 482 KEYWORDS: t.Dict[str, TokenType] = { 483 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 484 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 485 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 486 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 487 "/*+": TokenType.HINT, 488 "==": TokenType.EQ, 489 "::": TokenType.DCOLON, 490 "||": TokenType.DPIPE, 491 ">=": TokenType.GTE, 492 "<=": TokenType.LTE, 493 "<>": TokenType.NEQ, 494 "!=": TokenType.NEQ, 495 "<=>": TokenType.NULLSAFE_EQ, 496 "->": TokenType.ARROW, 497 "->>": TokenType.DARROW, 498 "=>": TokenType.FARROW, 499 "#>": TokenType.HASH_ARROW, 500 "#>>": TokenType.DHASH_ARROW, 501 "<->": TokenType.LR_ARROW, 502 "&&": TokenType.DAMP, 503 "ALL": TokenType.ALL, 504 "ALWAYS": TokenType.ALWAYS, 505 "AND": TokenType.AND, 506 "ANTI": TokenType.ANTI, 507 "ANY": TokenType.ANY, 508 "ASC": TokenType.ASC, 509 "AS": TokenType.ALIAS, 510 "ASOF": TokenType.ASOF, 511 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 512 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 513 "BEGIN": TokenType.BEGIN, 514 "BETWEEN": TokenType.BETWEEN, 515 "CACHE": TokenType.CACHE, 516 "UNCACHE": TokenType.UNCACHE, 517 "CASE": TokenType.CASE, 518 "CHARACTER SET": TokenType.CHARACTER_SET, 519 "CLUSTER BY": TokenType.CLUSTER_BY, 520 "COLLATE": TokenType.COLLATE, 521 "COLUMN": TokenType.COLUMN, 522 "COMMIT": TokenType.COMMIT, 523 "CONSTRAINT": TokenType.CONSTRAINT, 524 "CREATE": TokenType.CREATE, 525 "CROSS": TokenType.CROSS, 526 "CUBE": TokenType.CUBE, 527 "CURRENT_DATE": TokenType.CURRENT_DATE, 528 "CURRENT_TIME": TokenType.CURRENT_TIME, 529 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 530 "CURRENT_USER": TokenType.CURRENT_USER, 531 "DATABASE": TokenType.DATABASE, 532 "DEFAULT": TokenType.DEFAULT, 533 "DELETE": TokenType.DELETE, 534 "DESC": TokenType.DESC, 535 "DESCRIBE": TokenType.DESCRIBE, 536 "DISTINCT": TokenType.DISTINCT, 537 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 538 "DIV": TokenType.DIV, 539 "DROP": TokenType.DROP, 540 "ELSE": TokenType.ELSE, 541 "END": TokenType.END, 542 "ESCAPE": TokenType.ESCAPE, 543 "EXCEPT": TokenType.EXCEPT, 544 "EXECUTE": TokenType.EXECUTE, 545 "EXISTS": TokenType.EXISTS, 546 "FALSE": TokenType.FALSE, 547 "FETCH": TokenType.FETCH, 548 "FILTER": TokenType.FILTER, 549 "FIRST": TokenType.FIRST, 550 "FULL": TokenType.FULL, 551 "FUNCTION": TokenType.FUNCTION, 552 "FOR": TokenType.FOR, 553 "FOREIGN KEY": TokenType.FOREIGN_KEY, 554 "FORMAT": TokenType.FORMAT, 555 "FROM": TokenType.FROM, 556 "GEOGRAPHY": TokenType.GEOGRAPHY, 557 "GEOMETRY": TokenType.GEOMETRY, 558 "GLOB": TokenType.GLOB, 559 "GROUP BY": TokenType.GROUP_BY, 560 "GROUPING SETS": TokenType.GROUPING_SETS, 561 "HAVING": TokenType.HAVING, 562 "IF": TokenType.IF, 563 "ILIKE": TokenType.ILIKE, 564 "IN": TokenType.IN, 565 "INDEX": TokenType.INDEX, 566 "INET": TokenType.INET, 567 "INNER": TokenType.INNER, 568 "INSERT": TokenType.INSERT, 569 "INTERVAL": TokenType.INTERVAL, 570 "INTERSECT": TokenType.INTERSECT, 571 "INTO": TokenType.INTO, 572 "IS": TokenType.IS, 573 "ISNULL": TokenType.ISNULL, 574 "JOIN": TokenType.JOIN, 575 "KEEP": TokenType.KEEP, 576 "LATERAL": TokenType.LATERAL, 577 "LEFT": TokenType.LEFT, 578 "LIKE": TokenType.LIKE, 579 "LIMIT": TokenType.LIMIT, 580 "LOAD": TokenType.LOAD, 581 "LOCK": TokenType.LOCK, 582 "MERGE": TokenType.MERGE, 583 "NATURAL": TokenType.NATURAL, 584 "NEXT": TokenType.NEXT, 585 "NEXT VALUE FOR": TokenType.NEXT_VALUE_FOR, 586 "NOT": TokenType.NOT, 587 "NOTNULL": TokenType.NOTNULL, 588 "NULL": TokenType.NULL, 589 "OBJECT": TokenType.OBJECT, 590 "OFFSET": TokenType.OFFSET, 591 "ON": TokenType.ON, 592 "OR": TokenType.OR, 593 "ORDER BY": TokenType.ORDER_BY, 594 "ORDINALITY": TokenType.ORDINALITY, 595 "OUTER": TokenType.OUTER, 596 "OVER": TokenType.OVER, 597 "OVERLAPS": TokenType.OVERLAPS, 598 "OVERWRITE": TokenType.OVERWRITE, 599 "PARTITION": TokenType.PARTITION, 600 "PARTITION BY": TokenType.PARTITION_BY, 601 "PARTITIONED BY": TokenType.PARTITION_BY, 602 "PARTITIONED_BY": TokenType.PARTITION_BY, 603 "PERCENT": TokenType.PERCENT, 604 "PIVOT": TokenType.PIVOT, 605 "PRAGMA": TokenType.PRAGMA, 606 "PRIMARY KEY": TokenType.PRIMARY_KEY, 607 "PROCEDURE": TokenType.PROCEDURE, 608 "QUALIFY": TokenType.QUALIFY, 609 "RANGE": TokenType.RANGE, 610 "RECURSIVE": TokenType.RECURSIVE, 611 "REGEXP": TokenType.RLIKE, 612 "REPLACE": TokenType.REPLACE, 613 "RETURNING": TokenType.RETURNING, 614 "REFERENCES": TokenType.REFERENCES, 615 "RIGHT": TokenType.RIGHT, 616 "RLIKE": TokenType.RLIKE, 617 "ROLLBACK": TokenType.ROLLBACK, 618 "ROLLUP": TokenType.ROLLUP, 619 "ROW": TokenType.ROW, 620 "ROWS": TokenType.ROWS, 621 "SCHEMA": TokenType.SCHEMA, 622 "SELECT": TokenType.SELECT, 623 "SEMI": TokenType.SEMI, 624 "SET": TokenType.SET, 625 "SETTINGS": TokenType.SETTINGS, 626 "SHOW": TokenType.SHOW, 627 "SIMILAR TO": TokenType.SIMILAR_TO, 628 "SOME": TokenType.SOME, 629 "SORT BY": TokenType.SORT_BY, 630 "TABLE": TokenType.TABLE, 631 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 632 "TEMP": TokenType.TEMPORARY, 633 "TEMPORARY": TokenType.TEMPORARY, 634 "THEN": TokenType.THEN, 635 "TRUE": TokenType.TRUE, 636 "UNION": TokenType.UNION, 637 "UNNEST": TokenType.UNNEST, 638 "UNPIVOT": TokenType.UNPIVOT, 639 "UPDATE": TokenType.UPDATE, 640 "USE": TokenType.USE, 641 "USING": TokenType.USING, 642 "UUID": TokenType.UUID, 643 "VALUES": TokenType.VALUES, 644 "VIEW": TokenType.VIEW, 645 "VOLATILE": TokenType.VOLATILE, 646 "WHEN": TokenType.WHEN, 647 "WHERE": TokenType.WHERE, 648 "WINDOW": TokenType.WINDOW, 649 "WITH": TokenType.WITH, 650 "APPLY": TokenType.APPLY, 651 "ARRAY": TokenType.ARRAY, 652 "BIT": TokenType.BIT, 653 "BOOL": TokenType.BOOLEAN, 654 "BOOLEAN": TokenType.BOOLEAN, 655 "BYTE": TokenType.TINYINT, 656 "TINYINT": TokenType.TINYINT, 657 "SHORT": TokenType.SMALLINT, 658 "SMALLINT": TokenType.SMALLINT, 659 "INT2": TokenType.SMALLINT, 660 "INTEGER": TokenType.INT, 661 "INT": TokenType.INT, 662 "INT4": TokenType.INT, 663 "LONG": TokenType.BIGINT, 664 "BIGINT": TokenType.BIGINT, 665 "INT8": TokenType.BIGINT, 666 "DEC": TokenType.DECIMAL, 667 "DECIMAL": TokenType.DECIMAL, 668 "BIGDECIMAL": TokenType.BIGDECIMAL, 669 "BIGNUMERIC": TokenType.BIGDECIMAL, 670 "MAP": TokenType.MAP, 671 "NULLABLE": TokenType.NULLABLE, 672 "NUMBER": TokenType.DECIMAL, 673 "NUMERIC": TokenType.DECIMAL, 674 "FIXED": TokenType.DECIMAL, 675 "REAL": TokenType.FLOAT, 676 "FLOAT": TokenType.FLOAT, 677 "FLOAT4": TokenType.FLOAT, 678 "FLOAT8": TokenType.DOUBLE, 679 "DOUBLE": TokenType.DOUBLE, 680 "DOUBLE PRECISION": TokenType.DOUBLE, 681 "JSON": TokenType.JSON, 682 "CHAR": TokenType.CHAR, 683 "CHARACTER": TokenType.CHAR, 684 "NCHAR": TokenType.NCHAR, 685 "VARCHAR": TokenType.VARCHAR, 686 "VARCHAR2": TokenType.VARCHAR, 687 "NVARCHAR": TokenType.NVARCHAR, 688 "NVARCHAR2": TokenType.NVARCHAR, 689 "STR": TokenType.TEXT, 690 "STRING": TokenType.TEXT, 691 "TEXT": TokenType.TEXT, 692 "CLOB": TokenType.TEXT, 693 "LONGVARCHAR": TokenType.TEXT, 694 "BINARY": TokenType.BINARY, 695 "BLOB": TokenType.VARBINARY, 696 "BYTEA": TokenType.VARBINARY, 697 "VARBINARY": TokenType.VARBINARY, 698 "TIME": TokenType.TIME, 699 "TIMESTAMP": TokenType.TIMESTAMP, 700 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 701 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 702 "DATE": TokenType.DATE, 703 "DATETIME": TokenType.DATETIME, 704 "INT4RANGE": TokenType.INT4RANGE, 705 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 706 "INT8RANGE": TokenType.INT8RANGE, 707 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 708 "NUMRANGE": TokenType.NUMRANGE, 709 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 710 "TSRANGE": TokenType.TSRANGE, 711 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 712 "TSTZRANGE": TokenType.TSTZRANGE, 713 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 714 "DATERANGE": TokenType.DATERANGE, 715 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 716 "UNIQUE": TokenType.UNIQUE, 717 "STRUCT": TokenType.STRUCT, 718 "VARIANT": TokenType.VARIANT, 719 "ALTER": TokenType.ALTER, 720 "ANALYZE": TokenType.COMMAND, 721 "CALL": TokenType.COMMAND, 722 "COMMENT": TokenType.COMMENT, 723 "COPY": TokenType.COMMAND, 724 "EXPLAIN": TokenType.COMMAND, 725 "GRANT": TokenType.COMMAND, 726 "OPTIMIZE": TokenType.COMMAND, 727 "PREPARE": TokenType.COMMAND, 728 "TRUNCATE": TokenType.COMMAND, 729 "VACUUM": TokenType.COMMAND, 730 "USER-DEFINED": TokenType.USERDEFINED, 731 } 732 733 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 734 " ": TokenType.SPACE, 735 "\t": TokenType.SPACE, 736 "\n": TokenType.BREAK, 737 "\r": TokenType.BREAK, 738 "\r\n": TokenType.BREAK, 739 } 740 741 COMMANDS = { 742 TokenType.COMMAND, 743 TokenType.EXECUTE, 744 TokenType.FETCH, 745 TokenType.SHOW, 746 } 747 748 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 749 750 # handle numeric literals like in hive (3L = BIGINT) 751 NUMERIC_LITERALS: t.Dict[str, str] = {} 752 ENCODE: t.Optional[str] = None 753 754 COMMENTS = ["--", ("/*", "*/")] 755 756 __slots__ = ( 757 "sql", 758 "size", 759 "tokens", 760 "_start", 761 "_current", 762 "_line", 763 "_col", 764 "_comments", 765 "_char", 766 "_end", 767 "_peek", 768 "_prev_token_line", 769 ) 770 771 def __init__(self) -> None: 772 self.reset() 773 774 def reset(self) -> None: 775 self.sql = "" 776 self.size = 0 777 self.tokens: t.List[Token] = [] 778 self._start = 0 779 self._current = 0 780 self._line = 1 781 self._col = 0 782 self._comments: t.List[str] = [] 783 784 self._char = "" 785 self._end = False 786 self._peek = "" 787 self._prev_token_line = -1 788 789 def tokenize(self, sql: str) -> t.List[Token]: 790 """Returns a list of tokens corresponding to the SQL string `sql`.""" 791 self.reset() 792 self.sql = sql 793 self.size = len(sql) 794 795 try: 796 self._scan() 797 except Exception as e: 798 start = max(self._current - 50, 0) 799 end = min(self._current + 50, self.size - 1) 800 context = self.sql[start:end] 801 raise ValueError(f"Error tokenizing '{context}'") from e 802 803 return self.tokens 804 805 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 806 while self.size and not self._end: 807 self._start = self._current 808 self._advance() 809 810 if self._char is None: 811 break 812 813 if self._char not in self.WHITE_SPACE: 814 if self._char.isdigit(): 815 self._scan_number() 816 elif self._char in self._IDENTIFIERS: 817 self._scan_identifier(self._IDENTIFIERS[self._char]) 818 else: 819 self._scan_keywords() 820 821 if until and until(): 822 break 823 824 if self.tokens and self._comments: 825 self.tokens[-1].comments.extend(self._comments) 826 827 def _chars(self, size: int) -> str: 828 if size == 1: 829 return self._char 830 831 start = self._current - 1 832 end = start + size 833 834 return self.sql[start:end] if end <= self.size else "" 835 836 def _advance(self, i: int = 1, alnum: bool = False) -> None: 837 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 838 self._col = 1 839 self._line += 1 840 else: 841 self._col += i 842 843 self._current += i 844 self._end = self._current >= self.size 845 self._char = self.sql[self._current - 1] 846 self._peek = "" if self._end else self.sql[self._current] 847 848 if alnum and self._char.isalnum(): 849 # Here we use local variables instead of attributes for better performance 850 _col = self._col 851 _current = self._current 852 _end = self._end 853 _peek = self._peek 854 855 while _peek.isalnum(): 856 _col += 1 857 _current += 1 858 _end = _current >= self.size 859 _peek = "" if _end else self.sql[_current] 860 861 self._col = _col 862 self._current = _current 863 self._end = _end 864 self._peek = _peek 865 self._char = self.sql[_current - 1] 866 867 @property 868 def _text(self) -> str: 869 return self.sql[self._start : self._current] 870 871 def peek(self, i: int = 0) -> str: 872 i = self._current + i 873 if i < self.size: 874 return self.sql[i] 875 return "" 876 877 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 878 self._prev_token_line = self._line 879 self.tokens.append( 880 Token( 881 token_type, 882 text=self._text if text is None else text, 883 line=self._line, 884 col=self._col, 885 start=self._start, 886 end=self._current - 1, 887 comments=self._comments, 888 ) 889 ) 890 self._comments = [] 891 892 # If we have either a semicolon or a begin token before the command's token, we'll parse 893 # whatever follows the command's token as a string 894 if ( 895 token_type in self.COMMANDS 896 and self._peek != ";" 897 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 898 ): 899 start = self._current 900 tokens = len(self.tokens) 901 self._scan(lambda: self._peek == ";") 902 self.tokens = self.tokens[:tokens] 903 text = self.sql[start : self._current].strip() 904 if text: 905 self._add(TokenType.STRING, text) 906 907 def _scan_keywords(self) -> None: 908 size = 0 909 word = None 910 chars = self._text 911 char = chars 912 prev_space = False 913 skip = False 914 trie = self._KEYWORD_TRIE 915 single_token = char in self.SINGLE_TOKENS 916 917 while chars: 918 if skip: 919 result = TrieResult.PREFIX 920 else: 921 result, trie = in_trie(trie, char.upper()) 922 923 if result == TrieResult.FAILED: 924 break 925 if result == TrieResult.EXISTS: 926 word = chars 927 928 size += 1 929 end = self._current - 1 + size 930 931 if end < self.size: 932 char = self.sql[end] 933 single_token = single_token or char in self.SINGLE_TOKENS 934 is_space = char in self.WHITE_SPACE 935 936 if not is_space or not prev_space: 937 if is_space: 938 char = " " 939 chars += char 940 prev_space = is_space 941 skip = False 942 else: 943 skip = True 944 else: 945 char = "" 946 chars = " " 947 948 word = None if not single_token and chars[-1] not in self.WHITE_SPACE else word 949 950 if not word: 951 if self._char in self.SINGLE_TOKENS: 952 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 953 return 954 self._scan_var() 955 return 956 957 if self._scan_string(word): 958 return 959 if self._scan_comment(word): 960 return 961 962 self._advance(size - 1) 963 word = word.upper() 964 self._add(self.KEYWORDS[word], text=word) 965 966 def _scan_comment(self, comment_start: str) -> bool: 967 if comment_start not in self._COMMENTS: 968 return False 969 970 comment_start_line = self._line 971 comment_start_size = len(comment_start) 972 comment_end = self._COMMENTS[comment_start] 973 974 if comment_end: 975 # Skip the comment's start delimiter 976 self._advance(comment_start_size) 977 978 comment_end_size = len(comment_end) 979 while not self._end and self._chars(comment_end_size) != comment_end: 980 self._advance(alnum=True) 981 982 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 983 self._advance(comment_end_size - 1) 984 else: 985 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 986 self._advance(alnum=True) 987 self._comments.append(self._text[comment_start_size:]) 988 989 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 990 # Multiple consecutive comments are preserved by appending them to the current comments list. 991 if comment_start_line == self._prev_token_line: 992 self.tokens[-1].comments.extend(self._comments) 993 self._comments = [] 994 self._prev_token_line = self._line 995 996 return True 997 998 def _scan_number(self) -> None: 999 if self._char == "0": 1000 peek = self._peek.upper() 1001 if peek == "B": 1002 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1003 elif peek == "X": 1004 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1005 1006 decimal = False 1007 scientific = 0 1008 1009 while True: 1010 if self._peek.isdigit(): 1011 self._advance() 1012 elif self._peek == "." and not decimal: 1013 after = self.peek(1) 1014 if after.isdigit() or not after.isalpha(): 1015 decimal = True 1016 self._advance() 1017 else: 1018 return self._add(TokenType.VAR) 1019 elif self._peek in ("-", "+") and scientific == 1: 1020 scientific += 1 1021 self._advance() 1022 elif self._peek.upper() == "E" and not scientific: 1023 scientific += 1 1024 self._advance() 1025 elif self._peek.isidentifier(): 1026 number_text = self._text 1027 literal = "" 1028 1029 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1030 literal += self._peek.upper() 1031 self._advance() 1032 1033 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1034 1035 if token_type: 1036 self._add(TokenType.NUMBER, number_text) 1037 self._add(TokenType.DCOLON, "::") 1038 return self._add(token_type, literal) 1039 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1040 return self._add(TokenType.VAR) 1041 1042 self._add(TokenType.NUMBER, number_text) 1043 return self._advance(-len(literal)) 1044 else: 1045 return self._add(TokenType.NUMBER) 1046 1047 def _scan_bits(self) -> None: 1048 self._advance() 1049 value = self._extract_value() 1050 try: 1051 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1052 int(value, 2) 1053 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1054 except ValueError: 1055 self._add(TokenType.IDENTIFIER) 1056 1057 def _scan_hex(self) -> None: 1058 self._advance() 1059 value = self._extract_value() 1060 try: 1061 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1062 int(value, 16) 1063 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1064 except ValueError: 1065 self._add(TokenType.IDENTIFIER) 1066 1067 def _extract_value(self) -> str: 1068 while True: 1069 char = self._peek.strip() 1070 if char and char not in self.SINGLE_TOKENS: 1071 self._advance(alnum=True) 1072 else: 1073 break 1074 1075 return self._text 1076 1077 def _scan_string(self, start: str) -> bool: 1078 base = None 1079 token_type = TokenType.STRING 1080 1081 if start in self._QUOTES: 1082 end = self._QUOTES[start] 1083 elif start in self._FORMAT_STRINGS: 1084 end, token_type = self._FORMAT_STRINGS[start] 1085 1086 if token_type == TokenType.HEX_STRING: 1087 base = 16 1088 elif token_type == TokenType.BIT_STRING: 1089 base = 2 1090 else: 1091 return False 1092 1093 self._advance(len(start)) 1094 text = self._extract_string(end) 1095 1096 if base: 1097 try: 1098 int(text, base) 1099 except: 1100 raise RuntimeError( 1101 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1102 ) 1103 else: 1104 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1105 1106 self._add(token_type, text) 1107 return True 1108 1109 def _scan_identifier(self, identifier_end: str) -> None: 1110 self._advance() 1111 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1112 self._add(TokenType.IDENTIFIER, text) 1113 1114 def _scan_var(self) -> None: 1115 while True: 1116 char = self._peek.strip() 1117 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1118 self._advance(alnum=True) 1119 else: 1120 break 1121 1122 self._add( 1123 TokenType.VAR 1124 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1125 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1126 ) 1127 1128 def _extract_string(self, delimiter: str, escapes=None) -> str: 1129 text = "" 1130 delim_size = len(delimiter) 1131 escapes = self._STRING_ESCAPES if escapes is None else escapes 1132 1133 while True: 1134 if self._char in escapes and (self._peek == delimiter or self._peek in escapes): 1135 if self._peek == delimiter: 1136 text += self._peek 1137 else: 1138 text += self._char + self._peek 1139 1140 if self._current + 1 < self.size: 1141 self._advance(2) 1142 else: 1143 raise RuntimeError(f"Missing {delimiter} from {self._line}:{self._current}") 1144 else: 1145 if self._chars(delim_size) == delimiter: 1146 if delim_size > 1: 1147 self._advance(delim_size - 1) 1148 break 1149 1150 if self._end: 1151 raise RuntimeError(f"Missing {delimiter} from {self._line}:{self._start}") 1152 1153 current = self._current - 1 1154 self._advance(alnum=True) 1155 text += self.sql[current : self._current - 1] 1156 1157 return text
SINGLE_TOKENS =
{'(': <TokenType.L_PAREN: 'L_PAREN'>, ')': <TokenType.R_PAREN: 'R_PAREN'>, '[': <TokenType.L_BRACKET: 'L_BRACKET'>, ']': <TokenType.R_BRACKET: 'R_BRACKET'>, '{': <TokenType.L_BRACE: 'L_BRACE'>, '}': <TokenType.R_BRACE: 'R_BRACE'>, '&': <TokenType.AMP: 'AMP'>, '^': <TokenType.CARET: 'CARET'>, ':': <TokenType.COLON: 'COLON'>, ',': <TokenType.COMMA: 'COMMA'>, '.': <TokenType.DOT: 'DOT'>, '-': <TokenType.DASH: 'DASH'>, '=': <TokenType.EQ: 'EQ'>, '>': <TokenType.GT: 'GT'>, '<': <TokenType.LT: 'LT'>, '%': <TokenType.MOD: 'MOD'>, '!': <TokenType.NOT: 'NOT'>, '|': <TokenType.PIPE: 'PIPE'>, '+': <TokenType.PLUS: 'PLUS'>, ';': <TokenType.SEMICOLON: 'SEMICOLON'>, '/': <TokenType.SLASH: 'SLASH'>, '\\': <TokenType.BACKSLASH: 'BACKSLASH'>, '*': <TokenType.STAR: 'STAR'>, '~': <TokenType.TILDA: 'TILDA'>, '?': <TokenType.PLACEHOLDER: 'PLACEHOLDER'>, '@': <TokenType.PARAMETER: 'PARAMETER'>, "'": <TokenType.QUOTE: 'QUOTE'>, '`': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '"': <TokenType.IDENTIFIER: 'IDENTIFIER'>, '#': <TokenType.HASH: 'HASH'>}
KEYWORDS: Dict[str, sqlglot.tokens.TokenType] =
{'{%': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{%-': <TokenType.BLOCK_START: 'BLOCK_START'>, '%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '+%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-%}': <TokenType.BLOCK_END: 'BLOCK_END'>, '{{+': <TokenType.BLOCK_START: 'BLOCK_START'>, '{{-': <TokenType.BLOCK_START: 'BLOCK_START'>, '+}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '-}}': <TokenType.BLOCK_END: 'BLOCK_END'>, '/*+': <TokenType.HINT: 'HINT'>, '==': <TokenType.EQ: 'EQ'>, '::': <TokenType.DCOLON: 'DCOLON'>, '||': <TokenType.DPIPE: 'DPIPE'>, '>=': <TokenType.GTE: 'GTE'>, '<=': <TokenType.LTE: 'LTE'>, '<>': <TokenType.NEQ: 'NEQ'>, '!=': <TokenType.NEQ: 'NEQ'>, '<=>': <TokenType.NULLSAFE_EQ: 'NULLSAFE_EQ'>, '->': <TokenType.ARROW: 'ARROW'>, '->>': <TokenType.DARROW: 'DARROW'>, '=>': <TokenType.FARROW: 'FARROW'>, '#>': <TokenType.HASH_ARROW: 'HASH_ARROW'>, '#>>': <TokenType.DHASH_ARROW: 'DHASH_ARROW'>, '<->': <TokenType.LR_ARROW: 'LR_ARROW'>, '&&': <TokenType.DAMP: 'DAMP'>, 'ALL': <TokenType.ALL: 'ALL'>, 'ALWAYS': <TokenType.ALWAYS: 'ALWAYS'>, 'AND': <TokenType.AND: 'AND'>, 'ANTI': <TokenType.ANTI: 'ANTI'>, 'ANY': <TokenType.ANY: 'ANY'>, 'ASC': <TokenType.ASC: 'ASC'>, 'AS': <TokenType.ALIAS: 'ALIAS'>, 'ASOF': <TokenType.ASOF: 'ASOF'>, 'AUTOINCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'AUTO_INCREMENT': <TokenType.AUTO_INCREMENT: 'AUTO_INCREMENT'>, 'BEGIN': <TokenType.BEGIN: 'BEGIN'>, 'BETWEEN': <TokenType.BETWEEN: 'BETWEEN'>, 'CACHE': <TokenType.CACHE: 'CACHE'>, 'UNCACHE': <TokenType.UNCACHE: 'UNCACHE'>, 'CASE': <TokenType.CASE: 'CASE'>, 'CHARACTER SET': <TokenType.CHARACTER_SET: 'CHARACTER_SET'>, 'CLUSTER BY': <TokenType.CLUSTER_BY: 'CLUSTER_BY'>, 'COLLATE': <TokenType.COLLATE: 'COLLATE'>, 'COLUMN': <TokenType.COLUMN: 'COLUMN'>, 'COMMIT': <TokenType.COMMIT: 'COMMIT'>, 'CONSTRAINT': <TokenType.CONSTRAINT: 'CONSTRAINT'>, 'CREATE': <TokenType.CREATE: 'CREATE'>, 'CROSS': <TokenType.CROSS: 'CROSS'>, 'CUBE': <TokenType.CUBE: 'CUBE'>, 'CURRENT_DATE': <TokenType.CURRENT_DATE: 'CURRENT_DATE'>, 'CURRENT_TIME': <TokenType.CURRENT_TIME: 'CURRENT_TIME'>, 'CURRENT_TIMESTAMP': <TokenType.CURRENT_TIMESTAMP: 'CURRENT_TIMESTAMP'>, 'CURRENT_USER': <TokenType.CURRENT_USER: 'CURRENT_USER'>, 'DATABASE': <TokenType.DATABASE: 'DATABASE'>, 'DEFAULT': <TokenType.DEFAULT: 'DEFAULT'>, 'DELETE': <TokenType.DELETE: 'DELETE'>, 'DESC': <TokenType.DESC: 'DESC'>, 'DESCRIBE': <TokenType.DESCRIBE: 'DESCRIBE'>, 'DISTINCT': <TokenType.DISTINCT: 'DISTINCT'>, 'DISTRIBUTE BY': <TokenType.DISTRIBUTE_BY: 'DISTRIBUTE_BY'>, 'DIV': <TokenType.DIV: 'DIV'>, 'DROP': <TokenType.DROP: 'DROP'>, 'ELSE': <TokenType.ELSE: 'ELSE'>, 'END': <TokenType.END: 'END'>, 'ESCAPE': <TokenType.ESCAPE: 'ESCAPE'>, 'EXCEPT': <TokenType.EXCEPT: 'EXCEPT'>, 'EXECUTE': <TokenType.EXECUTE: 'EXECUTE'>, 'EXISTS': <TokenType.EXISTS: 'EXISTS'>, 'FALSE': <TokenType.FALSE: 'FALSE'>, 'FETCH': <TokenType.FETCH: 'FETCH'>, 'FILTER': <TokenType.FILTER: 'FILTER'>, 'FIRST': <TokenType.FIRST: 'FIRST'>, 'FULL': <TokenType.FULL: 'FULL'>, 'FUNCTION': <TokenType.FUNCTION: 'FUNCTION'>, 'FOR': <TokenType.FOR: 'FOR'>, 'FOREIGN KEY': <TokenType.FOREIGN_KEY: 'FOREIGN_KEY'>, 'FORMAT': <TokenType.FORMAT: 'FORMAT'>, 'FROM': <TokenType.FROM: 'FROM'>, 'GEOGRAPHY': <TokenType.GEOGRAPHY: 'GEOGRAPHY'>, 'GEOMETRY': <TokenType.GEOMETRY: 'GEOMETRY'>, 'GLOB': <TokenType.GLOB: 'GLOB'>, 'GROUP BY': <TokenType.GROUP_BY: 'GROUP_BY'>, 'GROUPING SETS': <TokenType.GROUPING_SETS: 'GROUPING_SETS'>, 'HAVING': <TokenType.HAVING: 'HAVING'>, 'IF': <TokenType.IF: 'IF'>, 'ILIKE': <TokenType.ILIKE: 'ILIKE'>, 'IN': <TokenType.IN: 'IN'>, 'INDEX': <TokenType.INDEX: 'INDEX'>, 'INET': <TokenType.INET: 'INET'>, 'INNER': <TokenType.INNER: 'INNER'>, 'INSERT': <TokenType.INSERT: 'INSERT'>, 'INTERVAL': <TokenType.INTERVAL: 'INTERVAL'>, 'INTERSECT': <TokenType.INTERSECT: 'INTERSECT'>, 'INTO': <TokenType.INTO: 'INTO'>, 'IS': <TokenType.IS: 'IS'>, 'ISNULL': <TokenType.ISNULL: 'ISNULL'>, 'JOIN': <TokenType.JOIN: 'JOIN'>, 'KEEP': <TokenType.KEEP: 'KEEP'>, 'LATERAL': <TokenType.LATERAL: 'LATERAL'>, 'LEFT': <TokenType.LEFT: 'LEFT'>, 'LIKE': <TokenType.LIKE: 'LIKE'>, 'LIMIT': <TokenType.LIMIT: 'LIMIT'>, 'LOAD': <TokenType.LOAD: 'LOAD'>, 'LOCK': <TokenType.LOCK: 'LOCK'>, 'MERGE': <TokenType.MERGE: 'MERGE'>, 'NATURAL': <TokenType.NATURAL: 'NATURAL'>, 'NEXT': <TokenType.NEXT: 'NEXT'>, 'NEXT VALUE FOR': <TokenType.NEXT_VALUE_FOR: 'NEXT_VALUE_FOR'>, 'NOT': <TokenType.NOT: 'NOT'>, 'NOTNULL': <TokenType.NOTNULL: 'NOTNULL'>, 'NULL': <TokenType.NULL: 'NULL'>, 'OBJECT': <TokenType.OBJECT: 'OBJECT'>, 'OFFSET': <TokenType.OFFSET: 'OFFSET'>, 'ON': <TokenType.ON: 'ON'>, 'OR': <TokenType.OR: 'OR'>, 'ORDER BY': <TokenType.ORDER_BY: 'ORDER_BY'>, 'ORDINALITY': <TokenType.ORDINALITY: 'ORDINALITY'>, 'OUTER': <TokenType.OUTER: 'OUTER'>, 'OVER': <TokenType.OVER: 'OVER'>, 'OVERLAPS': <TokenType.OVERLAPS: 'OVERLAPS'>, 'OVERWRITE': <TokenType.OVERWRITE: 'OVERWRITE'>, 'PARTITION': <TokenType.PARTITION: 'PARTITION'>, 'PARTITION BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PARTITIONED_BY': <TokenType.PARTITION_BY: 'PARTITION_BY'>, 'PERCENT': <TokenType.PERCENT: 'PERCENT'>, 'PIVOT': <TokenType.PIVOT: 'PIVOT'>, 'PRAGMA': <TokenType.PRAGMA: 'PRAGMA'>, 'PRIMARY KEY': <TokenType.PRIMARY_KEY: 'PRIMARY_KEY'>, 'PROCEDURE': <TokenType.PROCEDURE: 'PROCEDURE'>, 'QUALIFY': <TokenType.QUALIFY: 'QUALIFY'>, 'RANGE': <TokenType.RANGE: 'RANGE'>, 'RECURSIVE': <TokenType.RECURSIVE: 'RECURSIVE'>, 'REGEXP': <TokenType.RLIKE: 'RLIKE'>, 'REPLACE': <TokenType.REPLACE: 'REPLACE'>, 'RETURNING': <TokenType.RETURNING: 'RETURNING'>, 'REFERENCES': <TokenType.REFERENCES: 'REFERENCES'>, 'RIGHT': <TokenType.RIGHT: 'RIGHT'>, 'RLIKE': <TokenType.RLIKE: 'RLIKE'>, 'ROLLBACK': <TokenType.ROLLBACK: 'ROLLBACK'>, 'ROLLUP': <TokenType.ROLLUP: 'ROLLUP'>, 'ROW': <TokenType.ROW: 'ROW'>, 'ROWS': <TokenType.ROWS: 'ROWS'>, 'SCHEMA': <TokenType.SCHEMA: 'SCHEMA'>, 'SELECT': <TokenType.SELECT: 'SELECT'>, 'SEMI': <TokenType.SEMI: 'SEMI'>, 'SET': <TokenType.SET: 'SET'>, 'SETTINGS': <TokenType.SETTINGS: 'SETTINGS'>, 'SHOW': <TokenType.SHOW: 'SHOW'>, 'SIMILAR TO': <TokenType.SIMILAR_TO: 'SIMILAR_TO'>, 'SOME': <TokenType.SOME: 'SOME'>, 'SORT BY': <TokenType.SORT_BY: 'SORT_BY'>, 'TABLE': <TokenType.TABLE: 'TABLE'>, 'TABLESAMPLE': <TokenType.TABLE_SAMPLE: 'TABLE_SAMPLE'>, 'TEMP': <TokenType.TEMPORARY: 'TEMPORARY'>, 'TEMPORARY': <TokenType.TEMPORARY: 'TEMPORARY'>, 'THEN': <TokenType.THEN: 'THEN'>, 'TRUE': <TokenType.TRUE: 'TRUE'>, 'UNION': <TokenType.UNION: 'UNION'>, 'UNNEST': <TokenType.UNNEST: 'UNNEST'>, 'UNPIVOT': <TokenType.UNPIVOT: 'UNPIVOT'>, 'UPDATE': <TokenType.UPDATE: 'UPDATE'>, 'USE': <TokenType.USE: 'USE'>, 'USING': <TokenType.USING: 'USING'>, 'UUID': <TokenType.UUID: 'UUID'>, 'VALUES': <TokenType.VALUES: 'VALUES'>, 'VIEW': <TokenType.VIEW: 'VIEW'>, 'VOLATILE': <TokenType.VOLATILE: 'VOLATILE'>, 'WHEN': <TokenType.WHEN: 'WHEN'>, 'WHERE': <TokenType.WHERE: 'WHERE'>, 'WINDOW': <TokenType.WINDOW: 'WINDOW'>, 'WITH': <TokenType.WITH: 'WITH'>, 'APPLY': <TokenType.APPLY: 'APPLY'>, 'ARRAY': <TokenType.ARRAY: 'ARRAY'>, 'BIT': <TokenType.BIT: 'BIT'>, 'BOOL': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BOOLEAN': <TokenType.BOOLEAN: 'BOOLEAN'>, 'BYTE': <TokenType.TINYINT: 'TINYINT'>, 'TINYINT': <TokenType.TINYINT: 'TINYINT'>, 'SHORT': <TokenType.SMALLINT: 'SMALLINT'>, 'SMALLINT': <TokenType.SMALLINT: 'SMALLINT'>, 'INT2': <TokenType.SMALLINT: 'SMALLINT'>, 'INTEGER': <TokenType.INT: 'INT'>, 'INT': <TokenType.INT: 'INT'>, 'INT4': <TokenType.INT: 'INT'>, 'LONG': <TokenType.BIGINT: 'BIGINT'>, 'BIGINT': <TokenType.BIGINT: 'BIGINT'>, 'INT8': <TokenType.BIGINT: 'BIGINT'>, 'DEC': <TokenType.DECIMAL: 'DECIMAL'>, 'DECIMAL': <TokenType.DECIMAL: 'DECIMAL'>, 'BIGDECIMAL': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'BIGNUMERIC': <TokenType.BIGDECIMAL: 'BIGDECIMAL'>, 'MAP': <TokenType.MAP: 'MAP'>, 'NULLABLE': <TokenType.NULLABLE: 'NULLABLE'>, 'NUMBER': <TokenType.DECIMAL: 'DECIMAL'>, 'NUMERIC': <TokenType.DECIMAL: 'DECIMAL'>, 'FIXED': <TokenType.DECIMAL: 'DECIMAL'>, 'REAL': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT4': <TokenType.FLOAT: 'FLOAT'>, 'FLOAT8': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE': <TokenType.DOUBLE: 'DOUBLE'>, 'DOUBLE PRECISION': <TokenType.DOUBLE: 'DOUBLE'>, 'JSON': <TokenType.JSON: 'JSON'>, 'CHAR': <TokenType.CHAR: 'CHAR'>, 'CHARACTER': <TokenType.CHAR: 'CHAR'>, 'NCHAR': <TokenType.NCHAR: 'NCHAR'>, 'VARCHAR': <TokenType.VARCHAR: 'VARCHAR'>, 'VARCHAR2': <TokenType.VARCHAR: 'VARCHAR'>, 'NVARCHAR': <TokenType.NVARCHAR: 'NVARCHAR'>, 'NVARCHAR2': <TokenType.NVARCHAR: 'NVARCHAR'>, 'STR': <TokenType.TEXT: 'TEXT'>, 'STRING': <TokenType.TEXT: 'TEXT'>, 'TEXT': <TokenType.TEXT: 'TEXT'>, 'CLOB': <TokenType.TEXT: 'TEXT'>, 'LONGVARCHAR': <TokenType.TEXT: 'TEXT'>, 'BINARY': <TokenType.BINARY: 'BINARY'>, 'BLOB': <TokenType.VARBINARY: 'VARBINARY'>, 'BYTEA': <TokenType.VARBINARY: 'VARBINARY'>, 'VARBINARY': <TokenType.VARBINARY: 'VARBINARY'>, 'TIME': <TokenType.TIME: 'TIME'>, 'TIMESTAMP': <TokenType.TIMESTAMP: 'TIMESTAMP'>, 'TIMESTAMPTZ': <TokenType.TIMESTAMPTZ: 'TIMESTAMPTZ'>, 'TIMESTAMPLTZ': <TokenType.TIMESTAMPLTZ: 'TIMESTAMPLTZ'>, 'DATE': <TokenType.DATE: 'DATE'>, 'DATETIME': <TokenType.DATETIME: 'DATETIME'>, 'INT4RANGE': <TokenType.INT4RANGE: 'INT4RANGE'>, 'INT4MULTIRANGE': <TokenType.INT4MULTIRANGE: 'INT4MULTIRANGE'>, 'INT8RANGE': <TokenType.INT8RANGE: 'INT8RANGE'>, 'INT8MULTIRANGE': <TokenType.INT8MULTIRANGE: 'INT8MULTIRANGE'>, 'NUMRANGE': <TokenType.NUMRANGE: 'NUMRANGE'>, 'NUMMULTIRANGE': <TokenType.NUMMULTIRANGE: 'NUMMULTIRANGE'>, 'TSRANGE': <TokenType.TSRANGE: 'TSRANGE'>, 'TSMULTIRANGE': <TokenType.TSMULTIRANGE: 'TSMULTIRANGE'>, 'TSTZRANGE': <TokenType.TSTZRANGE: 'TSTZRANGE'>, 'TSTZMULTIRANGE': <TokenType.TSTZMULTIRANGE: 'TSTZMULTIRANGE'>, 'DATERANGE': <TokenType.DATERANGE: 'DATERANGE'>, 'DATEMULTIRANGE': <TokenType.DATEMULTIRANGE: 'DATEMULTIRANGE'>, 'UNIQUE': <TokenType.UNIQUE: 'UNIQUE'>, 'STRUCT': <TokenType.STRUCT: 'STRUCT'>, 'VARIANT': <TokenType.VARIANT: 'VARIANT'>, 'ALTER': <TokenType.ALTER: 'ALTER'>, 'ANALYZE': <TokenType.COMMAND: 'COMMAND'>, 'CALL': <TokenType.COMMAND: 'COMMAND'>, 'COMMENT': <TokenType.COMMENT: 'COMMENT'>, 'COPY': <TokenType.COMMAND: 'COMMAND'>, 'EXPLAIN': <TokenType.COMMAND: 'COMMAND'>, 'GRANT': <TokenType.COMMAND: 'COMMAND'>, 'OPTIMIZE': <TokenType.COMMAND: 'COMMAND'>, 'PREPARE': <TokenType.COMMAND: 'COMMAND'>, 'TRUNCATE': <TokenType.COMMAND: 'COMMAND'>, 'VACUUM': <TokenType.COMMAND: 'COMMAND'>, 'USER-DEFINED': <TokenType.USERDEFINED: 'USERDEFINED'>}
WHITE_SPACE: Dict[Optional[str], sqlglot.tokens.TokenType] =
{' ': <TokenType.SPACE: 'SPACE'>, '\t': <TokenType.SPACE: 'SPACE'>, '\n': <TokenType.BREAK: 'BREAK'>, '\r': <TokenType.BREAK: 'BREAK'>, '\r\n': <TokenType.BREAK: 'BREAK'>}
COMMANDS =
{<TokenType.FETCH: 'FETCH'>, <TokenType.SHOW: 'SHOW'>, <TokenType.EXECUTE: 'EXECUTE'>, <TokenType.COMMAND: 'COMMAND'>}
COMMAND_PREFIX_TOKENS =
{<TokenType.BEGIN: 'BEGIN'>, <TokenType.SEMICOLON: 'SEMICOLON'>}
def
reset(self) -> None:
774 def reset(self) -> None: 775 self.sql = "" 776 self.size = 0 777 self.tokens: t.List[Token] = [] 778 self._start = 0 779 self._current = 0 780 self._line = 1 781 self._col = 0 782 self._comments: t.List[str] = [] 783 784 self._char = "" 785 self._end = False 786 self._peek = "" 787 self._prev_token_line = -1
789 def tokenize(self, sql: str) -> t.List[Token]: 790 """Returns a list of tokens corresponding to the SQL string `sql`.""" 791 self.reset() 792 self.sql = sql 793 self.size = len(sql) 794 795 try: 796 self._scan() 797 except Exception as e: 798 start = max(self._current - 50, 0) 799 end = min(self._current + 50, self.size - 1) 800 context = self.sql[start:end] 801 raise ValueError(f"Error tokenizing '{context}'") from e 802 803 return self.tokens
Returns a list of tokens corresponding to the SQL string sql
.