sqlglot.tokens
1from __future__ import annotations 2 3import typing as t 4from enum import auto 5 6from sqlglot.errors import TokenError 7from sqlglot.helper import AutoName 8from sqlglot.trie import TrieResult, in_trie, new_trie 9 10 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 DQMARK = auto() 25 SEMICOLON = auto() 26 STAR = auto() 27 BACKSLASH = auto() 28 SLASH = auto() 29 LT = auto() 30 LTE = auto() 31 GT = auto() 32 GTE = auto() 33 NOT = auto() 34 EQ = auto() 35 NEQ = auto() 36 NULLSAFE_EQ = auto() 37 AND = auto() 38 OR = auto() 39 AMP = auto() 40 DPIPE = auto() 41 PIPE = auto() 42 CARET = auto() 43 TILDA = auto() 44 ARROW = auto() 45 DARROW = auto() 46 FARROW = auto() 47 HASH = auto() 48 HASH_ARROW = auto() 49 DHASH_ARROW = auto() 50 LR_ARROW = auto() 51 DAT = auto() 52 LT_AT = auto() 53 AT_GT = auto() 54 DOLLAR = auto() 55 PARAMETER = auto() 56 SESSION_PARAMETER = auto() 57 DAMP = auto() 58 XOR = auto() 59 60 BLOCK_START = auto() 61 BLOCK_END = auto() 62 63 SPACE = auto() 64 BREAK = auto() 65 66 STRING = auto() 67 NUMBER = auto() 68 IDENTIFIER = auto() 69 DATABASE = auto() 70 COLUMN = auto() 71 COLUMN_DEF = auto() 72 SCHEMA = auto() 73 TABLE = auto() 74 VAR = auto() 75 BIT_STRING = auto() 76 HEX_STRING = auto() 77 BYTE_STRING = auto() 78 NATIONAL_STRING = auto() 79 RAW_STRING = auto() 80 81 # types 82 BIT = auto() 83 BOOLEAN = auto() 84 TINYINT = auto() 85 UTINYINT = auto() 86 SMALLINT = auto() 87 USMALLINT = auto() 88 MEDIUMINT = auto() 89 INT = auto() 90 UINT = auto() 91 BIGINT = auto() 92 UBIGINT = auto() 93 INT128 = auto() 94 UINT128 = auto() 95 INT256 = auto() 96 UINT256 = auto() 97 FLOAT = auto() 98 DOUBLE = auto() 99 DECIMAL = auto() 100 BIGDECIMAL = auto() 101 CHAR = auto() 102 NCHAR = auto() 103 VARCHAR = auto() 104 NVARCHAR = auto() 105 TEXT = auto() 106 MEDIUMTEXT = auto() 107 LONGTEXT = auto() 108 MEDIUMBLOB = auto() 109 LONGBLOB = auto() 110 BINARY = auto() 111 VARBINARY = auto() 112 JSON = auto() 113 JSONB = auto() 114 TIME = auto() 115 TIMETZ = auto() 116 TIMESTAMP = auto() 117 TIMESTAMPTZ = auto() 118 TIMESTAMPLTZ = auto() 119 DATETIME = auto() 120 DATETIME64 = auto() 121 DATE = auto() 122 INT4RANGE = auto() 123 INT4MULTIRANGE = auto() 124 INT8RANGE = auto() 125 INT8MULTIRANGE = auto() 126 NUMRANGE = auto() 127 NUMMULTIRANGE = auto() 128 TSRANGE = auto() 129 TSMULTIRANGE = auto() 130 TSTZRANGE = auto() 131 TSTZMULTIRANGE = auto() 132 DATERANGE = auto() 133 DATEMULTIRANGE = auto() 134 UUID = auto() 135 GEOGRAPHY = auto() 136 NULLABLE = auto() 137 GEOMETRY = auto() 138 HLLSKETCH = auto() 139 HSTORE = auto() 140 SUPER = auto() 141 SERIAL = auto() 142 SMALLSERIAL = auto() 143 BIGSERIAL = auto() 144 XML = auto() 145 YEAR = auto() 146 UNIQUEIDENTIFIER = auto() 147 USERDEFINED = auto() 148 MONEY = auto() 149 SMALLMONEY = auto() 150 ROWVERSION = auto() 151 IMAGE = auto() 152 VARIANT = auto() 153 OBJECT = auto() 154 INET = auto() 155 IPADDRESS = auto() 156 IPPREFIX = auto() 157 ENUM = auto() 158 ENUM8 = auto() 159 ENUM16 = auto() 160 FIXEDSTRING = auto() 161 LOWCARDINALITY = auto() 162 NESTED = auto() 163 UNKNOWN = auto() 164 165 # keywords 166 ALIAS = auto() 167 ALTER = auto() 168 ALWAYS = auto() 169 ALL = auto() 170 ANTI = auto() 171 ANY = auto() 172 APPLY = auto() 173 ARRAY = auto() 174 ASC = auto() 175 ASOF = auto() 176 AUTO_INCREMENT = auto() 177 BEGIN = auto() 178 BETWEEN = auto() 179 CACHE = auto() 180 CASE = auto() 181 CHARACTER_SET = auto() 182 CLUSTER_BY = auto() 183 COLLATE = auto() 184 COMMAND = auto() 185 COMMENT = auto() 186 COMMIT = auto() 187 CONNECT_BY = auto() 188 CONSTRAINT = auto() 189 CREATE = auto() 190 CROSS = auto() 191 CUBE = auto() 192 CURRENT_DATE = auto() 193 CURRENT_DATETIME = auto() 194 CURRENT_TIME = auto() 195 CURRENT_TIMESTAMP = auto() 196 CURRENT_USER = auto() 197 DEFAULT = auto() 198 DELETE = auto() 199 DESC = auto() 200 DESCRIBE = auto() 201 DICTIONARY = auto() 202 DISTINCT = auto() 203 DISTRIBUTE_BY = auto() 204 DIV = auto() 205 DROP = auto() 206 ELSE = auto() 207 END = auto() 208 ESCAPE = auto() 209 EXCEPT = auto() 210 EXECUTE = auto() 211 EXISTS = auto() 212 FALSE = auto() 213 FETCH = auto() 214 FILTER = auto() 215 FINAL = auto() 216 FIRST = auto() 217 FOR = auto() 218 FORCE = auto() 219 FOREIGN_KEY = auto() 220 FORMAT = auto() 221 FROM = auto() 222 FULL = auto() 223 FUNCTION = auto() 224 GLOB = auto() 225 GLOBAL = auto() 226 GROUP_BY = auto() 227 GROUPING_SETS = auto() 228 HAVING = auto() 229 HINT = auto() 230 IGNORE = auto() 231 ILIKE = auto() 232 ILIKE_ANY = auto() 233 IN = auto() 234 INDEX = auto() 235 INNER = auto() 236 INSERT = auto() 237 INTERSECT = auto() 238 INTERVAL = auto() 239 INTO = auto() 240 INTRODUCER = auto() 241 IRLIKE = auto() 242 IS = auto() 243 ISNULL = auto() 244 JOIN = auto() 245 JOIN_MARKER = auto() 246 KEEP = auto() 247 LANGUAGE = auto() 248 LATERAL = auto() 249 LEFT = auto() 250 LIKE = auto() 251 LIKE_ANY = auto() 252 LIMIT = auto() 253 LOAD = auto() 254 LOCK = auto() 255 MAP = auto() 256 MATCH_RECOGNIZE = auto() 257 MEMBER_OF = auto() 258 MERGE = auto() 259 MOD = auto() 260 NATURAL = auto() 261 NEXT = auto() 262 NOTNULL = auto() 263 NULL = auto() 264 OFFSET = auto() 265 ON = auto() 266 ORDER_BY = auto() 267 ORDERED = auto() 268 ORDINALITY = auto() 269 OUTER = auto() 270 OVER = auto() 271 OVERLAPS = auto() 272 OVERWRITE = auto() 273 PARTITION = auto() 274 PARTITION_BY = auto() 275 PERCENT = auto() 276 PIVOT = auto() 277 PLACEHOLDER = auto() 278 PRAGMA = auto() 279 PRIMARY_KEY = auto() 280 PROCEDURE = auto() 281 PROPERTIES = auto() 282 PSEUDO_TYPE = auto() 283 QUALIFY = auto() 284 QUOTE = auto() 285 RANGE = auto() 286 RECURSIVE = auto() 287 REPLACE = auto() 288 RETURNING = auto() 289 REFERENCES = auto() 290 RIGHT = auto() 291 RLIKE = auto() 292 ROLLBACK = auto() 293 ROLLUP = auto() 294 ROW = auto() 295 ROWS = auto() 296 SELECT = auto() 297 SEMI = auto() 298 SEPARATOR = auto() 299 SERDE_PROPERTIES = auto() 300 SET = auto() 301 SETTINGS = auto() 302 SHOW = auto() 303 SIMILAR_TO = auto() 304 SOME = auto() 305 SORT_BY = auto() 306 START_WITH = auto() 307 STRUCT = auto() 308 TABLE_SAMPLE = auto() 309 TEMPORARY = auto() 310 TOP = auto() 311 THEN = auto() 312 TRUE = auto() 313 UNCACHE = auto() 314 UNION = auto() 315 UNNEST = auto() 316 UNPIVOT = auto() 317 UPDATE = auto() 318 USE = auto() 319 USING = auto() 320 VALUES = auto() 321 VIEW = auto() 322 VOLATILE = auto() 323 WHEN = auto() 324 WHERE = auto() 325 WINDOW = auto() 326 WITH = auto() 327 UNIQUE = auto() 328 329 330class Token: 331 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 332 333 @classmethod 334 def number(cls, number: int) -> Token: 335 """Returns a NUMBER token with `number` as its text.""" 336 return cls(TokenType.NUMBER, str(number)) 337 338 @classmethod 339 def string(cls, string: str) -> Token: 340 """Returns a STRING token with `string` as its text.""" 341 return cls(TokenType.STRING, string) 342 343 @classmethod 344 def identifier(cls, identifier: str) -> Token: 345 """Returns an IDENTIFIER token with `identifier` as its text.""" 346 return cls(TokenType.IDENTIFIER, identifier) 347 348 @classmethod 349 def var(cls, var: str) -> Token: 350 """Returns an VAR token with `var` as its text.""" 351 return cls(TokenType.VAR, var) 352 353 def __init__( 354 self, 355 token_type: TokenType, 356 text: str, 357 line: int = 1, 358 col: int = 1, 359 start: int = 0, 360 end: int = 0, 361 comments: t.List[str] = [], 362 ) -> None: 363 """Token initializer. 364 365 Args: 366 token_type: The TokenType Enum. 367 text: The text of the token. 368 line: The line that the token ends on. 369 col: The column that the token ends on. 370 start: The start index of the token. 371 end: The ending index of the token. 372 comments: The comments to attach to the token. 373 """ 374 self.token_type = token_type 375 self.text = text 376 self.line = line 377 self.col = col 378 self.start = start 379 self.end = end 380 self.comments = comments 381 382 def __repr__(self) -> str: 383 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 384 return f"<Token {attributes}>" 385 386 387class _Tokenizer(type): 388 def __new__(cls, clsname, bases, attrs): 389 klass = super().__new__(cls, clsname, bases, attrs) 390 391 def _convert_quotes(arr: t.List[str | t.Tuple[str, str]]) -> t.Dict[str, str]: 392 return dict( 393 (item, item) if isinstance(item, str) else (item[0], item[1]) for item in arr 394 ) 395 396 def _quotes_to_format( 397 token_type: TokenType, arr: t.List[str | t.Tuple[str, str]] 398 ) -> t.Dict[str, t.Tuple[str, TokenType]]: 399 return {k: (v, token_type) for k, v in _convert_quotes(arr).items()} 400 401 klass._QUOTES = _convert_quotes(klass.QUOTES) 402 klass._IDENTIFIERS = _convert_quotes(klass.IDENTIFIERS) 403 404 klass._FORMAT_STRINGS = { 405 **{ 406 p + s: (e, TokenType.NATIONAL_STRING) 407 for s, e in klass._QUOTES.items() 408 for p in ("n", "N") 409 }, 410 **_quotes_to_format(TokenType.BIT_STRING, klass.BIT_STRINGS), 411 **_quotes_to_format(TokenType.BYTE_STRING, klass.BYTE_STRINGS), 412 **_quotes_to_format(TokenType.HEX_STRING, klass.HEX_STRINGS), 413 **_quotes_to_format(TokenType.RAW_STRING, klass.RAW_STRINGS), 414 } 415 416 klass._STRING_ESCAPES = set(klass.STRING_ESCAPES) 417 klass._IDENTIFIER_ESCAPES = set(klass.IDENTIFIER_ESCAPES) 418 klass._COMMENTS = { 419 **dict( 420 (comment, None) if isinstance(comment, str) else (comment[0], comment[1]) 421 for comment in klass.COMMENTS 422 ), 423 "{#": "#}", # Ensure Jinja comments are tokenized correctly in all dialects 424 } 425 426 klass._KEYWORD_TRIE = new_trie( 427 key.upper() 428 for key in ( 429 *klass.KEYWORDS, 430 *klass._COMMENTS, 431 *klass._QUOTES, 432 *klass._FORMAT_STRINGS, 433 ) 434 if " " in key or any(single in key for single in klass.SINGLE_TOKENS) 435 ) 436 437 return klass 438 439 440class Tokenizer(metaclass=_Tokenizer): 441 SINGLE_TOKENS = { 442 "(": TokenType.L_PAREN, 443 ")": TokenType.R_PAREN, 444 "[": TokenType.L_BRACKET, 445 "]": TokenType.R_BRACKET, 446 "{": TokenType.L_BRACE, 447 "}": TokenType.R_BRACE, 448 "&": TokenType.AMP, 449 "^": TokenType.CARET, 450 ":": TokenType.COLON, 451 ",": TokenType.COMMA, 452 ".": TokenType.DOT, 453 "-": TokenType.DASH, 454 "=": TokenType.EQ, 455 ">": TokenType.GT, 456 "<": TokenType.LT, 457 "%": TokenType.MOD, 458 "!": TokenType.NOT, 459 "|": TokenType.PIPE, 460 "+": TokenType.PLUS, 461 ";": TokenType.SEMICOLON, 462 "/": TokenType.SLASH, 463 "\\": TokenType.BACKSLASH, 464 "*": TokenType.STAR, 465 "~": TokenType.TILDA, 466 "?": TokenType.PLACEHOLDER, 467 "@": TokenType.PARAMETER, 468 # used for breaking a var like x'y' but nothing else 469 # the token type doesn't matter 470 "'": TokenType.QUOTE, 471 "`": TokenType.IDENTIFIER, 472 '"': TokenType.IDENTIFIER, 473 "#": TokenType.HASH, 474 } 475 476 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 477 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 478 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 479 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 480 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 481 IDENTIFIER_ESCAPES = ['"'] 482 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 483 STRING_ESCAPES = ["'"] 484 VAR_SINGLE_TOKENS: t.Set[str] = set() 485 486 # Autofilled 487 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 488 489 _COMMENTS: t.Dict[str, str] = {} 490 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 491 _IDENTIFIERS: t.Dict[str, str] = {} 492 _IDENTIFIER_ESCAPES: t.Set[str] = set() 493 _QUOTES: t.Dict[str, str] = {} 494 _STRING_ESCAPES: t.Set[str] = set() 495 _KEYWORD_TRIE: t.Dict = {} 496 497 KEYWORDS: t.Dict[str, TokenType] = { 498 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 499 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 500 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 501 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 502 "/*+": TokenType.HINT, 503 "==": TokenType.EQ, 504 "::": TokenType.DCOLON, 505 "||": TokenType.DPIPE, 506 ">=": TokenType.GTE, 507 "<=": TokenType.LTE, 508 "<>": TokenType.NEQ, 509 "!=": TokenType.NEQ, 510 "<=>": TokenType.NULLSAFE_EQ, 511 "->": TokenType.ARROW, 512 "->>": TokenType.DARROW, 513 "=>": TokenType.FARROW, 514 "#>": TokenType.HASH_ARROW, 515 "#>>": TokenType.DHASH_ARROW, 516 "<->": TokenType.LR_ARROW, 517 "&&": TokenType.DAMP, 518 "??": TokenType.DQMARK, 519 "ALL": TokenType.ALL, 520 "ALWAYS": TokenType.ALWAYS, 521 "AND": TokenType.AND, 522 "ANTI": TokenType.ANTI, 523 "ANY": TokenType.ANY, 524 "ASC": TokenType.ASC, 525 "AS": TokenType.ALIAS, 526 "ASOF": TokenType.ASOF, 527 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 528 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 529 "BEGIN": TokenType.BEGIN, 530 "BETWEEN": TokenType.BETWEEN, 531 "CACHE": TokenType.CACHE, 532 "UNCACHE": TokenType.UNCACHE, 533 "CASE": TokenType.CASE, 534 "CHARACTER SET": TokenType.CHARACTER_SET, 535 "CLUSTER BY": TokenType.CLUSTER_BY, 536 "COLLATE": TokenType.COLLATE, 537 "COLUMN": TokenType.COLUMN, 538 "COMMIT": TokenType.COMMIT, 539 "CONNECT BY": TokenType.CONNECT_BY, 540 "CONSTRAINT": TokenType.CONSTRAINT, 541 "CREATE": TokenType.CREATE, 542 "CROSS": TokenType.CROSS, 543 "CUBE": TokenType.CUBE, 544 "CURRENT_DATE": TokenType.CURRENT_DATE, 545 "CURRENT_TIME": TokenType.CURRENT_TIME, 546 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 547 "CURRENT_USER": TokenType.CURRENT_USER, 548 "DATABASE": TokenType.DATABASE, 549 "DEFAULT": TokenType.DEFAULT, 550 "DELETE": TokenType.DELETE, 551 "DESC": TokenType.DESC, 552 "DESCRIBE": TokenType.DESCRIBE, 553 "DISTINCT": TokenType.DISTINCT, 554 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 555 "DIV": TokenType.DIV, 556 "DROP": TokenType.DROP, 557 "ELSE": TokenType.ELSE, 558 "END": TokenType.END, 559 "ESCAPE": TokenType.ESCAPE, 560 "EXCEPT": TokenType.EXCEPT, 561 "EXECUTE": TokenType.EXECUTE, 562 "EXISTS": TokenType.EXISTS, 563 "FALSE": TokenType.FALSE, 564 "FETCH": TokenType.FETCH, 565 "FILTER": TokenType.FILTER, 566 "FIRST": TokenType.FIRST, 567 "FULL": TokenType.FULL, 568 "FUNCTION": TokenType.FUNCTION, 569 "FOR": TokenType.FOR, 570 "FOREIGN KEY": TokenType.FOREIGN_KEY, 571 "FORMAT": TokenType.FORMAT, 572 "FROM": TokenType.FROM, 573 "GEOGRAPHY": TokenType.GEOGRAPHY, 574 "GEOMETRY": TokenType.GEOMETRY, 575 "GLOB": TokenType.GLOB, 576 "GROUP BY": TokenType.GROUP_BY, 577 "GROUPING SETS": TokenType.GROUPING_SETS, 578 "HAVING": TokenType.HAVING, 579 "ILIKE": TokenType.ILIKE, 580 "IN": TokenType.IN, 581 "INDEX": TokenType.INDEX, 582 "INET": TokenType.INET, 583 "INNER": TokenType.INNER, 584 "INSERT": TokenType.INSERT, 585 "INTERVAL": TokenType.INTERVAL, 586 "INTERSECT": TokenType.INTERSECT, 587 "INTO": TokenType.INTO, 588 "IS": TokenType.IS, 589 "ISNULL": TokenType.ISNULL, 590 "JOIN": TokenType.JOIN, 591 "KEEP": TokenType.KEEP, 592 "LATERAL": TokenType.LATERAL, 593 "LEFT": TokenType.LEFT, 594 "LIKE": TokenType.LIKE, 595 "LIMIT": TokenType.LIMIT, 596 "LOAD": TokenType.LOAD, 597 "LOCK": TokenType.LOCK, 598 "MERGE": TokenType.MERGE, 599 "NATURAL": TokenType.NATURAL, 600 "NEXT": TokenType.NEXT, 601 "NOT": TokenType.NOT, 602 "NOTNULL": TokenType.NOTNULL, 603 "NULL": TokenType.NULL, 604 "OBJECT": TokenType.OBJECT, 605 "OFFSET": TokenType.OFFSET, 606 "ON": TokenType.ON, 607 "OR": TokenType.OR, 608 "XOR": TokenType.XOR, 609 "ORDER BY": TokenType.ORDER_BY, 610 "ORDINALITY": TokenType.ORDINALITY, 611 "OUTER": TokenType.OUTER, 612 "OVER": TokenType.OVER, 613 "OVERLAPS": TokenType.OVERLAPS, 614 "OVERWRITE": TokenType.OVERWRITE, 615 "PARTITION": TokenType.PARTITION, 616 "PARTITION BY": TokenType.PARTITION_BY, 617 "PARTITIONED BY": TokenType.PARTITION_BY, 618 "PARTITIONED_BY": TokenType.PARTITION_BY, 619 "PERCENT": TokenType.PERCENT, 620 "PIVOT": TokenType.PIVOT, 621 "PRAGMA": TokenType.PRAGMA, 622 "PRIMARY KEY": TokenType.PRIMARY_KEY, 623 "PROCEDURE": TokenType.PROCEDURE, 624 "QUALIFY": TokenType.QUALIFY, 625 "RANGE": TokenType.RANGE, 626 "RECURSIVE": TokenType.RECURSIVE, 627 "REGEXP": TokenType.RLIKE, 628 "REPLACE": TokenType.REPLACE, 629 "RETURNING": TokenType.RETURNING, 630 "REFERENCES": TokenType.REFERENCES, 631 "RIGHT": TokenType.RIGHT, 632 "RLIKE": TokenType.RLIKE, 633 "ROLLBACK": TokenType.ROLLBACK, 634 "ROLLUP": TokenType.ROLLUP, 635 "ROW": TokenType.ROW, 636 "ROWS": TokenType.ROWS, 637 "SCHEMA": TokenType.SCHEMA, 638 "SELECT": TokenType.SELECT, 639 "SEMI": TokenType.SEMI, 640 "SET": TokenType.SET, 641 "SETTINGS": TokenType.SETTINGS, 642 "SHOW": TokenType.SHOW, 643 "SIMILAR TO": TokenType.SIMILAR_TO, 644 "SOME": TokenType.SOME, 645 "SORT BY": TokenType.SORT_BY, 646 "START WITH": TokenType.START_WITH, 647 "TABLE": TokenType.TABLE, 648 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 649 "TEMP": TokenType.TEMPORARY, 650 "TEMPORARY": TokenType.TEMPORARY, 651 "THEN": TokenType.THEN, 652 "TRUE": TokenType.TRUE, 653 "UNION": TokenType.UNION, 654 "UNKNOWN": TokenType.UNKNOWN, 655 "UNNEST": TokenType.UNNEST, 656 "UNPIVOT": TokenType.UNPIVOT, 657 "UPDATE": TokenType.UPDATE, 658 "USE": TokenType.USE, 659 "USING": TokenType.USING, 660 "UUID": TokenType.UUID, 661 "VALUES": TokenType.VALUES, 662 "VIEW": TokenType.VIEW, 663 "VOLATILE": TokenType.VOLATILE, 664 "WHEN": TokenType.WHEN, 665 "WHERE": TokenType.WHERE, 666 "WINDOW": TokenType.WINDOW, 667 "WITH": TokenType.WITH, 668 "APPLY": TokenType.APPLY, 669 "ARRAY": TokenType.ARRAY, 670 "BIT": TokenType.BIT, 671 "BOOL": TokenType.BOOLEAN, 672 "BOOLEAN": TokenType.BOOLEAN, 673 "BYTE": TokenType.TINYINT, 674 "TINYINT": TokenType.TINYINT, 675 "SHORT": TokenType.SMALLINT, 676 "SMALLINT": TokenType.SMALLINT, 677 "INT128": TokenType.INT128, 678 "INT2": TokenType.SMALLINT, 679 "INTEGER": TokenType.INT, 680 "INT": TokenType.INT, 681 "INT4": TokenType.INT, 682 "LONG": TokenType.BIGINT, 683 "BIGINT": TokenType.BIGINT, 684 "INT8": TokenType.BIGINT, 685 "DEC": TokenType.DECIMAL, 686 "DECIMAL": TokenType.DECIMAL, 687 "BIGDECIMAL": TokenType.BIGDECIMAL, 688 "BIGNUMERIC": TokenType.BIGDECIMAL, 689 "MAP": TokenType.MAP, 690 "NULLABLE": TokenType.NULLABLE, 691 "NUMBER": TokenType.DECIMAL, 692 "NUMERIC": TokenType.DECIMAL, 693 "FIXED": TokenType.DECIMAL, 694 "REAL": TokenType.FLOAT, 695 "FLOAT": TokenType.FLOAT, 696 "FLOAT4": TokenType.FLOAT, 697 "FLOAT8": TokenType.DOUBLE, 698 "DOUBLE": TokenType.DOUBLE, 699 "DOUBLE PRECISION": TokenType.DOUBLE, 700 "JSON": TokenType.JSON, 701 "CHAR": TokenType.CHAR, 702 "CHARACTER": TokenType.CHAR, 703 "NCHAR": TokenType.NCHAR, 704 "VARCHAR": TokenType.VARCHAR, 705 "VARCHAR2": TokenType.VARCHAR, 706 "NVARCHAR": TokenType.NVARCHAR, 707 "NVARCHAR2": TokenType.NVARCHAR, 708 "STR": TokenType.TEXT, 709 "STRING": TokenType.TEXT, 710 "TEXT": TokenType.TEXT, 711 "CLOB": TokenType.TEXT, 712 "LONGVARCHAR": TokenType.TEXT, 713 "BINARY": TokenType.BINARY, 714 "BLOB": TokenType.VARBINARY, 715 "BYTEA": TokenType.VARBINARY, 716 "VARBINARY": TokenType.VARBINARY, 717 "TIME": TokenType.TIME, 718 "TIMETZ": TokenType.TIMETZ, 719 "TIMESTAMP": TokenType.TIMESTAMP, 720 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 721 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 722 "DATE": TokenType.DATE, 723 "DATETIME": TokenType.DATETIME, 724 "INT4RANGE": TokenType.INT4RANGE, 725 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 726 "INT8RANGE": TokenType.INT8RANGE, 727 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 728 "NUMRANGE": TokenType.NUMRANGE, 729 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 730 "TSRANGE": TokenType.TSRANGE, 731 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 732 "TSTZRANGE": TokenType.TSTZRANGE, 733 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 734 "DATERANGE": TokenType.DATERANGE, 735 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 736 "UNIQUE": TokenType.UNIQUE, 737 "STRUCT": TokenType.STRUCT, 738 "VARIANT": TokenType.VARIANT, 739 "ALTER": TokenType.ALTER, 740 "ANALYZE": TokenType.COMMAND, 741 "CALL": TokenType.COMMAND, 742 "COMMENT": TokenType.COMMENT, 743 "COPY": TokenType.COMMAND, 744 "EXPLAIN": TokenType.COMMAND, 745 "GRANT": TokenType.COMMAND, 746 "OPTIMIZE": TokenType.COMMAND, 747 "PREPARE": TokenType.COMMAND, 748 "TRUNCATE": TokenType.COMMAND, 749 "VACUUM": TokenType.COMMAND, 750 "USER-DEFINED": TokenType.USERDEFINED, 751 } 752 753 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 754 " ": TokenType.SPACE, 755 "\t": TokenType.SPACE, 756 "\n": TokenType.BREAK, 757 "\r": TokenType.BREAK, 758 "\r\n": TokenType.BREAK, 759 } 760 761 COMMANDS = { 762 TokenType.COMMAND, 763 TokenType.EXECUTE, 764 TokenType.FETCH, 765 TokenType.SHOW, 766 } 767 768 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 769 770 # handle numeric literals like in hive (3L = BIGINT) 771 NUMERIC_LITERALS: t.Dict[str, str] = {} 772 ENCODE: t.Optional[str] = None 773 774 COMMENTS = ["--", ("/*", "*/")] 775 776 __slots__ = ( 777 "sql", 778 "size", 779 "tokens", 780 "_start", 781 "_current", 782 "_line", 783 "_col", 784 "_comments", 785 "_char", 786 "_end", 787 "_peek", 788 "_prev_token_line", 789 ) 790 791 def __init__(self) -> None: 792 self.reset() 793 794 def reset(self) -> None: 795 self.sql = "" 796 self.size = 0 797 self.tokens: t.List[Token] = [] 798 self._start = 0 799 self._current = 0 800 self._line = 1 801 self._col = 0 802 self._comments: t.List[str] = [] 803 804 self._char = "" 805 self._end = False 806 self._peek = "" 807 self._prev_token_line = -1 808 809 def tokenize(self, sql: str) -> t.List[Token]: 810 """Returns a list of tokens corresponding to the SQL string `sql`.""" 811 self.reset() 812 self.sql = sql 813 self.size = len(sql) 814 815 try: 816 self._scan() 817 except Exception as e: 818 start = max(self._current - 50, 0) 819 end = min(self._current + 50, self.size - 1) 820 context = self.sql[start:end] 821 raise TokenError(f"Error tokenizing '{context}'") from e 822 823 return self.tokens 824 825 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 826 while self.size and not self._end: 827 self._start = self._current 828 self._advance() 829 830 if self._char is None: 831 break 832 833 if self._char not in self.WHITE_SPACE: 834 if self._char.isdigit(): 835 self._scan_number() 836 elif self._char in self._IDENTIFIERS: 837 self._scan_identifier(self._IDENTIFIERS[self._char]) 838 else: 839 self._scan_keywords() 840 841 if until and until(): 842 break 843 844 if self.tokens and self._comments: 845 self.tokens[-1].comments.extend(self._comments) 846 847 def _chars(self, size: int) -> str: 848 if size == 1: 849 return self._char 850 851 start = self._current - 1 852 end = start + size 853 854 return self.sql[start:end] if end <= self.size else "" 855 856 def _advance(self, i: int = 1, alnum: bool = False) -> None: 857 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 858 self._col = 1 859 self._line += 1 860 else: 861 self._col += i 862 863 self._current += i 864 self._end = self._current >= self.size 865 self._char = self.sql[self._current - 1] 866 self._peek = "" if self._end else self.sql[self._current] 867 868 if alnum and self._char.isalnum(): 869 # Here we use local variables instead of attributes for better performance 870 _col = self._col 871 _current = self._current 872 _end = self._end 873 _peek = self._peek 874 875 while _peek.isalnum(): 876 _col += 1 877 _current += 1 878 _end = _current >= self.size 879 _peek = "" if _end else self.sql[_current] 880 881 self._col = _col 882 self._current = _current 883 self._end = _end 884 self._peek = _peek 885 self._char = self.sql[_current - 1] 886 887 @property 888 def _text(self) -> str: 889 return self.sql[self._start : self._current] 890 891 def peek(self, i: int = 0) -> str: 892 i = self._current + i 893 if i < self.size: 894 return self.sql[i] 895 return "" 896 897 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 898 self._prev_token_line = self._line 899 900 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 901 self.tokens[-1].comments.extend(self._comments) 902 self._comments = [] 903 904 self.tokens.append( 905 Token( 906 token_type, 907 text=self._text if text is None else text, 908 line=self._line, 909 col=self._col, 910 start=self._start, 911 end=self._current - 1, 912 comments=self._comments, 913 ) 914 ) 915 self._comments = [] 916 917 # If we have either a semicolon or a begin token before the command's token, we'll parse 918 # whatever follows the command's token as a string 919 if ( 920 token_type in self.COMMANDS 921 and self._peek != ";" 922 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 923 ): 924 start = self._current 925 tokens = len(self.tokens) 926 self._scan(lambda: self._peek == ";") 927 self.tokens = self.tokens[:tokens] 928 text = self.sql[start : self._current].strip() 929 if text: 930 self._add(TokenType.STRING, text) 931 932 def _scan_keywords(self) -> None: 933 size = 0 934 word = None 935 chars = self._text 936 char = chars 937 prev_space = False 938 skip = False 939 trie = self._KEYWORD_TRIE 940 single_token = char in self.SINGLE_TOKENS 941 942 while chars: 943 if skip: 944 result = TrieResult.PREFIX 945 else: 946 result, trie = in_trie(trie, char.upper()) 947 948 if result == TrieResult.FAILED: 949 break 950 if result == TrieResult.EXISTS: 951 word = chars 952 953 size += 1 954 end = self._current - 1 + size 955 956 if end < self.size: 957 char = self.sql[end] 958 single_token = single_token or char in self.SINGLE_TOKENS 959 is_space = char in self.WHITE_SPACE 960 961 if not is_space or not prev_space: 962 if is_space: 963 char = " " 964 chars += char 965 prev_space = is_space 966 skip = False 967 else: 968 skip = True 969 else: 970 char = "" 971 chars = " " 972 973 if not word: 974 if self._char in self.SINGLE_TOKENS: 975 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 976 return 977 self._scan_var() 978 return 979 980 if self._scan_string(word): 981 return 982 if self._scan_comment(word): 983 return 984 985 self._advance(size - 1) 986 word = word.upper() 987 self._add(self.KEYWORDS[word], text=word) 988 989 def _scan_comment(self, comment_start: str) -> bool: 990 if comment_start not in self._COMMENTS: 991 return False 992 993 comment_start_line = self._line 994 comment_start_size = len(comment_start) 995 comment_end = self._COMMENTS[comment_start] 996 997 if comment_end: 998 # Skip the comment's start delimiter 999 self._advance(comment_start_size) 1000 1001 comment_end_size = len(comment_end) 1002 while not self._end and self._chars(comment_end_size) != comment_end: 1003 self._advance(alnum=True) 1004 1005 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1006 self._advance(comment_end_size - 1) 1007 else: 1008 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1009 self._advance(alnum=True) 1010 self._comments.append(self._text[comment_start_size:]) 1011 1012 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1013 # Multiple consecutive comments are preserved by appending them to the current comments list. 1014 if comment_start_line == self._prev_token_line: 1015 self.tokens[-1].comments.extend(self._comments) 1016 self._comments = [] 1017 self._prev_token_line = self._line 1018 1019 return True 1020 1021 def _scan_number(self) -> None: 1022 if self._char == "0": 1023 peek = self._peek.upper() 1024 if peek == "B": 1025 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1026 elif peek == "X": 1027 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1028 1029 decimal = False 1030 scientific = 0 1031 1032 while True: 1033 if self._peek.isdigit(): 1034 self._advance() 1035 elif self._peek == "." and not decimal: 1036 after = self.peek(1) 1037 if after.isdigit() or not after.isalpha(): 1038 decimal = True 1039 self._advance() 1040 else: 1041 return self._add(TokenType.VAR) 1042 elif self._peek in ("-", "+") and scientific == 1: 1043 scientific += 1 1044 self._advance() 1045 elif self._peek.upper() == "E" and not scientific: 1046 scientific += 1 1047 self._advance() 1048 elif self._peek.isidentifier(): 1049 number_text = self._text 1050 literal = "" 1051 1052 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1053 literal += self._peek.upper() 1054 self._advance() 1055 1056 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1057 1058 if token_type: 1059 self._add(TokenType.NUMBER, number_text) 1060 self._add(TokenType.DCOLON, "::") 1061 return self._add(token_type, literal) 1062 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1063 return self._add(TokenType.VAR) 1064 1065 self._add(TokenType.NUMBER, number_text) 1066 return self._advance(-len(literal)) 1067 else: 1068 return self._add(TokenType.NUMBER) 1069 1070 def _scan_bits(self) -> None: 1071 self._advance() 1072 value = self._extract_value() 1073 try: 1074 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1075 int(value, 2) 1076 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1077 except ValueError: 1078 self._add(TokenType.IDENTIFIER) 1079 1080 def _scan_hex(self) -> None: 1081 self._advance() 1082 value = self._extract_value() 1083 try: 1084 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1085 int(value, 16) 1086 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1087 except ValueError: 1088 self._add(TokenType.IDENTIFIER) 1089 1090 def _extract_value(self) -> str: 1091 while True: 1092 char = self._peek.strip() 1093 if char and char not in self.SINGLE_TOKENS: 1094 self._advance(alnum=True) 1095 else: 1096 break 1097 1098 return self._text 1099 1100 def _scan_string(self, start: str) -> bool: 1101 base = None 1102 token_type = TokenType.STRING 1103 1104 if start in self._QUOTES: 1105 end = self._QUOTES[start] 1106 elif start in self._FORMAT_STRINGS: 1107 end, token_type = self._FORMAT_STRINGS[start] 1108 1109 if token_type == TokenType.HEX_STRING: 1110 base = 16 1111 elif token_type == TokenType.BIT_STRING: 1112 base = 2 1113 else: 1114 return False 1115 1116 self._advance(len(start)) 1117 text = self._extract_string(end) 1118 1119 if base: 1120 try: 1121 int(text, base) 1122 except: 1123 raise TokenError( 1124 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1125 ) 1126 else: 1127 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1128 1129 self._add(token_type, text) 1130 return True 1131 1132 def _scan_identifier(self, identifier_end: str) -> None: 1133 self._advance() 1134 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1135 self._add(TokenType.IDENTIFIER, text) 1136 1137 def _scan_var(self) -> None: 1138 while True: 1139 char = self._peek.strip() 1140 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1141 self._advance(alnum=True) 1142 else: 1143 break 1144 1145 self._add( 1146 TokenType.VAR 1147 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1148 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1149 ) 1150 1151 def _extract_string(self, delimiter: str, escapes=None) -> str: 1152 text = "" 1153 delim_size = len(delimiter) 1154 escapes = self._STRING_ESCAPES if escapes is None else escapes 1155 1156 while True: 1157 if self._char in escapes and (self._peek == delimiter or self._peek in escapes): 1158 if self._peek == delimiter: 1159 text += self._peek 1160 else: 1161 text += self._char + self._peek 1162 1163 if self._current + 1 < self.size: 1164 self._advance(2) 1165 else: 1166 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1167 else: 1168 if self._chars(delim_size) == delimiter: 1169 if delim_size > 1: 1170 self._advance(delim_size - 1) 1171 break 1172 1173 if self._end: 1174 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1175 1176 current = self._current - 1 1177 self._advance(alnum=True) 1178 text += self.sql[current : self._current - 1] 1179 1180 return text
12class TokenType(AutoName): 13 L_PAREN = auto() 14 R_PAREN = auto() 15 L_BRACKET = auto() 16 R_BRACKET = auto() 17 L_BRACE = auto() 18 R_BRACE = auto() 19 COMMA = auto() 20 DOT = auto() 21 DASH = auto() 22 PLUS = auto() 23 COLON = auto() 24 DCOLON = auto() 25 DQMARK = auto() 26 SEMICOLON = auto() 27 STAR = auto() 28 BACKSLASH = auto() 29 SLASH = auto() 30 LT = auto() 31 LTE = auto() 32 GT = auto() 33 GTE = auto() 34 NOT = auto() 35 EQ = auto() 36 NEQ = auto() 37 NULLSAFE_EQ = auto() 38 AND = auto() 39 OR = auto() 40 AMP = auto() 41 DPIPE = auto() 42 PIPE = auto() 43 CARET = auto() 44 TILDA = auto() 45 ARROW = auto() 46 DARROW = auto() 47 FARROW = auto() 48 HASH = auto() 49 HASH_ARROW = auto() 50 DHASH_ARROW = auto() 51 LR_ARROW = auto() 52 DAT = auto() 53 LT_AT = auto() 54 AT_GT = auto() 55 DOLLAR = auto() 56 PARAMETER = auto() 57 SESSION_PARAMETER = auto() 58 DAMP = auto() 59 XOR = auto() 60 61 BLOCK_START = auto() 62 BLOCK_END = auto() 63 64 SPACE = auto() 65 BREAK = auto() 66 67 STRING = auto() 68 NUMBER = auto() 69 IDENTIFIER = auto() 70 DATABASE = auto() 71 COLUMN = auto() 72 COLUMN_DEF = auto() 73 SCHEMA = auto() 74 TABLE = auto() 75 VAR = auto() 76 BIT_STRING = auto() 77 HEX_STRING = auto() 78 BYTE_STRING = auto() 79 NATIONAL_STRING = auto() 80 RAW_STRING = auto() 81 82 # types 83 BIT = auto() 84 BOOLEAN = auto() 85 TINYINT = auto() 86 UTINYINT = auto() 87 SMALLINT = auto() 88 USMALLINT = auto() 89 MEDIUMINT = auto() 90 INT = auto() 91 UINT = auto() 92 BIGINT = auto() 93 UBIGINT = auto() 94 INT128 = auto() 95 UINT128 = auto() 96 INT256 = auto() 97 UINT256 = auto() 98 FLOAT = auto() 99 DOUBLE = auto() 100 DECIMAL = auto() 101 BIGDECIMAL = auto() 102 CHAR = auto() 103 NCHAR = auto() 104 VARCHAR = auto() 105 NVARCHAR = auto() 106 TEXT = auto() 107 MEDIUMTEXT = auto() 108 LONGTEXT = auto() 109 MEDIUMBLOB = auto() 110 LONGBLOB = auto() 111 BINARY = auto() 112 VARBINARY = auto() 113 JSON = auto() 114 JSONB = auto() 115 TIME = auto() 116 TIMETZ = auto() 117 TIMESTAMP = auto() 118 TIMESTAMPTZ = auto() 119 TIMESTAMPLTZ = auto() 120 DATETIME = auto() 121 DATETIME64 = auto() 122 DATE = auto() 123 INT4RANGE = auto() 124 INT4MULTIRANGE = auto() 125 INT8RANGE = auto() 126 INT8MULTIRANGE = auto() 127 NUMRANGE = auto() 128 NUMMULTIRANGE = auto() 129 TSRANGE = auto() 130 TSMULTIRANGE = auto() 131 TSTZRANGE = auto() 132 TSTZMULTIRANGE = auto() 133 DATERANGE = auto() 134 DATEMULTIRANGE = auto() 135 UUID = auto() 136 GEOGRAPHY = auto() 137 NULLABLE = auto() 138 GEOMETRY = auto() 139 HLLSKETCH = auto() 140 HSTORE = auto() 141 SUPER = auto() 142 SERIAL = auto() 143 SMALLSERIAL = auto() 144 BIGSERIAL = auto() 145 XML = auto() 146 YEAR = auto() 147 UNIQUEIDENTIFIER = auto() 148 USERDEFINED = auto() 149 MONEY = auto() 150 SMALLMONEY = auto() 151 ROWVERSION = auto() 152 IMAGE = auto() 153 VARIANT = auto() 154 OBJECT = auto() 155 INET = auto() 156 IPADDRESS = auto() 157 IPPREFIX = auto() 158 ENUM = auto() 159 ENUM8 = auto() 160 ENUM16 = auto() 161 FIXEDSTRING = auto() 162 LOWCARDINALITY = auto() 163 NESTED = auto() 164 UNKNOWN = auto() 165 166 # keywords 167 ALIAS = auto() 168 ALTER = auto() 169 ALWAYS = auto() 170 ALL = auto() 171 ANTI = auto() 172 ANY = auto() 173 APPLY = auto() 174 ARRAY = auto() 175 ASC = auto() 176 ASOF = auto() 177 AUTO_INCREMENT = auto() 178 BEGIN = auto() 179 BETWEEN = auto() 180 CACHE = auto() 181 CASE = auto() 182 CHARACTER_SET = auto() 183 CLUSTER_BY = auto() 184 COLLATE = auto() 185 COMMAND = auto() 186 COMMENT = auto() 187 COMMIT = auto() 188 CONNECT_BY = auto() 189 CONSTRAINT = auto() 190 CREATE = auto() 191 CROSS = auto() 192 CUBE = auto() 193 CURRENT_DATE = auto() 194 CURRENT_DATETIME = auto() 195 CURRENT_TIME = auto() 196 CURRENT_TIMESTAMP = auto() 197 CURRENT_USER = auto() 198 DEFAULT = auto() 199 DELETE = auto() 200 DESC = auto() 201 DESCRIBE = auto() 202 DICTIONARY = auto() 203 DISTINCT = auto() 204 DISTRIBUTE_BY = auto() 205 DIV = auto() 206 DROP = auto() 207 ELSE = auto() 208 END = auto() 209 ESCAPE = auto() 210 EXCEPT = auto() 211 EXECUTE = auto() 212 EXISTS = auto() 213 FALSE = auto() 214 FETCH = auto() 215 FILTER = auto() 216 FINAL = auto() 217 FIRST = auto() 218 FOR = auto() 219 FORCE = auto() 220 FOREIGN_KEY = auto() 221 FORMAT = auto() 222 FROM = auto() 223 FULL = auto() 224 FUNCTION = auto() 225 GLOB = auto() 226 GLOBAL = auto() 227 GROUP_BY = auto() 228 GROUPING_SETS = auto() 229 HAVING = auto() 230 HINT = auto() 231 IGNORE = auto() 232 ILIKE = auto() 233 ILIKE_ANY = auto() 234 IN = auto() 235 INDEX = auto() 236 INNER = auto() 237 INSERT = auto() 238 INTERSECT = auto() 239 INTERVAL = auto() 240 INTO = auto() 241 INTRODUCER = auto() 242 IRLIKE = auto() 243 IS = auto() 244 ISNULL = auto() 245 JOIN = auto() 246 JOIN_MARKER = auto() 247 KEEP = auto() 248 LANGUAGE = auto() 249 LATERAL = auto() 250 LEFT = auto() 251 LIKE = auto() 252 LIKE_ANY = auto() 253 LIMIT = auto() 254 LOAD = auto() 255 LOCK = auto() 256 MAP = auto() 257 MATCH_RECOGNIZE = auto() 258 MEMBER_OF = auto() 259 MERGE = auto() 260 MOD = auto() 261 NATURAL = auto() 262 NEXT = auto() 263 NOTNULL = auto() 264 NULL = auto() 265 OFFSET = auto() 266 ON = auto() 267 ORDER_BY = auto() 268 ORDERED = auto() 269 ORDINALITY = auto() 270 OUTER = auto() 271 OVER = auto() 272 OVERLAPS = auto() 273 OVERWRITE = auto() 274 PARTITION = auto() 275 PARTITION_BY = auto() 276 PERCENT = auto() 277 PIVOT = auto() 278 PLACEHOLDER = auto() 279 PRAGMA = auto() 280 PRIMARY_KEY = auto() 281 PROCEDURE = auto() 282 PROPERTIES = auto() 283 PSEUDO_TYPE = auto() 284 QUALIFY = auto() 285 QUOTE = auto() 286 RANGE = auto() 287 RECURSIVE = auto() 288 REPLACE = auto() 289 RETURNING = auto() 290 REFERENCES = auto() 291 RIGHT = auto() 292 RLIKE = auto() 293 ROLLBACK = auto() 294 ROLLUP = auto() 295 ROW = auto() 296 ROWS = auto() 297 SELECT = auto() 298 SEMI = auto() 299 SEPARATOR = auto() 300 SERDE_PROPERTIES = auto() 301 SET = auto() 302 SETTINGS = auto() 303 SHOW = auto() 304 SIMILAR_TO = auto() 305 SOME = auto() 306 SORT_BY = auto() 307 START_WITH = auto() 308 STRUCT = auto() 309 TABLE_SAMPLE = auto() 310 TEMPORARY = auto() 311 TOP = auto() 312 THEN = auto() 313 TRUE = auto() 314 UNCACHE = auto() 315 UNION = auto() 316 UNNEST = auto() 317 UNPIVOT = auto() 318 UPDATE = auto() 319 USE = auto() 320 USING = auto() 321 VALUES = auto() 322 VIEW = auto() 323 VOLATILE = auto() 324 WHEN = auto() 325 WHERE = auto() 326 WINDOW = auto() 327 WITH = auto() 328 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'>
DQMARK =
<TokenType.DQMARK: 'DQMARK'>
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'>
DAT =
<TokenType.DAT: 'DAT'>
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'>
XOR =
<TokenType.XOR: 'XOR'>
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'>
MEDIUMINT =
<TokenType.MEDIUMINT: 'MEDIUMINT'>
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'>
TIMETZ =
<TokenType.TIMETZ: 'TIMETZ'>
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'>
YEAR =
<TokenType.YEAR: 'YEAR'>
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'>
IPADDRESS =
<TokenType.IPADDRESS: 'IPADDRESS'>
IPPREFIX =
<TokenType.IPPREFIX: 'IPPREFIX'>
ENUM =
<TokenType.ENUM: 'ENUM'>
ENUM8 =
<TokenType.ENUM8: 'ENUM8'>
ENUM16 =
<TokenType.ENUM16: 'ENUM16'>
FIXEDSTRING =
<TokenType.FIXEDSTRING: 'FIXEDSTRING'>
LOWCARDINALITY =
<TokenType.LOWCARDINALITY: 'LOWCARDINALITY'>
NESTED =
<TokenType.NESTED: 'NESTED'>
UNKNOWN =
<TokenType.UNKNOWN: 'UNKNOWN'>
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'>
CONNECT_BY =
<TokenType.CONNECT_BY: 'CONNECT_BY'>
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'>
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'>
MEMBER_OF =
<TokenType.MEMBER_OF: 'MEMBER_OF'>
MERGE =
<TokenType.MERGE: 'MERGE'>
MOD =
<TokenType.MOD: 'MOD'>
NATURAL =
<TokenType.NATURAL: 'NATURAL'>
NEXT =
<TokenType.NEXT: 'NEXT'>
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'>
START_WITH =
<TokenType.START_WITH: 'START_WITH'>
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:
331class Token: 332 __slots__ = ("token_type", "text", "line", "col", "start", "end", "comments") 333 334 @classmethod 335 def number(cls, number: int) -> Token: 336 """Returns a NUMBER token with `number` as its text.""" 337 return cls(TokenType.NUMBER, str(number)) 338 339 @classmethod 340 def string(cls, string: str) -> Token: 341 """Returns a STRING token with `string` as its text.""" 342 return cls(TokenType.STRING, string) 343 344 @classmethod 345 def identifier(cls, identifier: str) -> Token: 346 """Returns an IDENTIFIER token with `identifier` as its text.""" 347 return cls(TokenType.IDENTIFIER, identifier) 348 349 @classmethod 350 def var(cls, var: str) -> Token: 351 """Returns an VAR token with `var` as its text.""" 352 return cls(TokenType.VAR, var) 353 354 def __init__( 355 self, 356 token_type: TokenType, 357 text: str, 358 line: int = 1, 359 col: int = 1, 360 start: int = 0, 361 end: int = 0, 362 comments: t.List[str] = [], 363 ) -> None: 364 """Token initializer. 365 366 Args: 367 token_type: The TokenType Enum. 368 text: The text of the token. 369 line: The line that the token ends on. 370 col: The column that the token ends on. 371 start: The start index of the token. 372 end: The ending index of the token. 373 comments: The comments to attach to the token. 374 """ 375 self.token_type = token_type 376 self.text = text 377 self.line = line 378 self.col = col 379 self.start = start 380 self.end = end 381 self.comments = comments 382 383 def __repr__(self) -> str: 384 attributes = ", ".join(f"{k}: {getattr(self, k)}" for k in self.__slots__) 385 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] = [])
354 def __init__( 355 self, 356 token_type: TokenType, 357 text: str, 358 line: int = 1, 359 col: int = 1, 360 start: int = 0, 361 end: int = 0, 362 comments: t.List[str] = [], 363 ) -> None: 364 """Token initializer. 365 366 Args: 367 token_type: The TokenType Enum. 368 text: The text of the token. 369 line: The line that the token ends on. 370 col: The column that the token ends on. 371 start: The start index of the token. 372 end: The ending index of the token. 373 comments: The comments to attach to the token. 374 """ 375 self.token_type = token_type 376 self.text = text 377 self.line = line 378 self.col = col 379 self.start = start 380 self.end = end 381 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.
334 @classmethod 335 def number(cls, number: int) -> Token: 336 """Returns a NUMBER token with `number` as its text.""" 337 return cls(TokenType.NUMBER, str(number))
Returns a NUMBER token with number
as its text.
339 @classmethod 340 def string(cls, string: str) -> Token: 341 """Returns a STRING token with `string` as its text.""" 342 return cls(TokenType.STRING, string)
Returns a STRING token with string
as its text.
344 @classmethod 345 def identifier(cls, identifier: str) -> Token: 346 """Returns an IDENTIFIER token with `identifier` as its text.""" 347 return cls(TokenType.IDENTIFIER, identifier)
Returns an IDENTIFIER token with identifier
as its text.
class
Tokenizer:
441class Tokenizer(metaclass=_Tokenizer): 442 SINGLE_TOKENS = { 443 "(": TokenType.L_PAREN, 444 ")": TokenType.R_PAREN, 445 "[": TokenType.L_BRACKET, 446 "]": TokenType.R_BRACKET, 447 "{": TokenType.L_BRACE, 448 "}": TokenType.R_BRACE, 449 "&": TokenType.AMP, 450 "^": TokenType.CARET, 451 ":": TokenType.COLON, 452 ",": TokenType.COMMA, 453 ".": TokenType.DOT, 454 "-": TokenType.DASH, 455 "=": TokenType.EQ, 456 ">": TokenType.GT, 457 "<": TokenType.LT, 458 "%": TokenType.MOD, 459 "!": TokenType.NOT, 460 "|": TokenType.PIPE, 461 "+": TokenType.PLUS, 462 ";": TokenType.SEMICOLON, 463 "/": TokenType.SLASH, 464 "\\": TokenType.BACKSLASH, 465 "*": TokenType.STAR, 466 "~": TokenType.TILDA, 467 "?": TokenType.PLACEHOLDER, 468 "@": TokenType.PARAMETER, 469 # used for breaking a var like x'y' but nothing else 470 # the token type doesn't matter 471 "'": TokenType.QUOTE, 472 "`": TokenType.IDENTIFIER, 473 '"': TokenType.IDENTIFIER, 474 "#": TokenType.HASH, 475 } 476 477 BIT_STRINGS: t.List[str | t.Tuple[str, str]] = [] 478 BYTE_STRINGS: t.List[str | t.Tuple[str, str]] = [] 479 HEX_STRINGS: t.List[str | t.Tuple[str, str]] = [] 480 RAW_STRINGS: t.List[str | t.Tuple[str, str]] = [] 481 IDENTIFIERS: t.List[str | t.Tuple[str, str]] = ['"'] 482 IDENTIFIER_ESCAPES = ['"'] 483 QUOTES: t.List[t.Tuple[str, str] | str] = ["'"] 484 STRING_ESCAPES = ["'"] 485 VAR_SINGLE_TOKENS: t.Set[str] = set() 486 487 # Autofilled 488 IDENTIFIERS_CAN_START_WITH_DIGIT: bool = False 489 490 _COMMENTS: t.Dict[str, str] = {} 491 _FORMAT_STRINGS: t.Dict[str, t.Tuple[str, TokenType]] = {} 492 _IDENTIFIERS: t.Dict[str, str] = {} 493 _IDENTIFIER_ESCAPES: t.Set[str] = set() 494 _QUOTES: t.Dict[str, str] = {} 495 _STRING_ESCAPES: t.Set[str] = set() 496 _KEYWORD_TRIE: t.Dict = {} 497 498 KEYWORDS: t.Dict[str, TokenType] = { 499 **{f"{{%{postfix}": TokenType.BLOCK_START for postfix in ("", "+", "-")}, 500 **{f"{prefix}%}}": TokenType.BLOCK_END for prefix in ("", "+", "-")}, 501 **{f"{{{{{postfix}": TokenType.BLOCK_START for postfix in ("+", "-")}, 502 **{f"{prefix}}}}}": TokenType.BLOCK_END for prefix in ("+", "-")}, 503 "/*+": TokenType.HINT, 504 "==": TokenType.EQ, 505 "::": TokenType.DCOLON, 506 "||": TokenType.DPIPE, 507 ">=": TokenType.GTE, 508 "<=": TokenType.LTE, 509 "<>": TokenType.NEQ, 510 "!=": TokenType.NEQ, 511 "<=>": TokenType.NULLSAFE_EQ, 512 "->": TokenType.ARROW, 513 "->>": TokenType.DARROW, 514 "=>": TokenType.FARROW, 515 "#>": TokenType.HASH_ARROW, 516 "#>>": TokenType.DHASH_ARROW, 517 "<->": TokenType.LR_ARROW, 518 "&&": TokenType.DAMP, 519 "??": TokenType.DQMARK, 520 "ALL": TokenType.ALL, 521 "ALWAYS": TokenType.ALWAYS, 522 "AND": TokenType.AND, 523 "ANTI": TokenType.ANTI, 524 "ANY": TokenType.ANY, 525 "ASC": TokenType.ASC, 526 "AS": TokenType.ALIAS, 527 "ASOF": TokenType.ASOF, 528 "AUTOINCREMENT": TokenType.AUTO_INCREMENT, 529 "AUTO_INCREMENT": TokenType.AUTO_INCREMENT, 530 "BEGIN": TokenType.BEGIN, 531 "BETWEEN": TokenType.BETWEEN, 532 "CACHE": TokenType.CACHE, 533 "UNCACHE": TokenType.UNCACHE, 534 "CASE": TokenType.CASE, 535 "CHARACTER SET": TokenType.CHARACTER_SET, 536 "CLUSTER BY": TokenType.CLUSTER_BY, 537 "COLLATE": TokenType.COLLATE, 538 "COLUMN": TokenType.COLUMN, 539 "COMMIT": TokenType.COMMIT, 540 "CONNECT BY": TokenType.CONNECT_BY, 541 "CONSTRAINT": TokenType.CONSTRAINT, 542 "CREATE": TokenType.CREATE, 543 "CROSS": TokenType.CROSS, 544 "CUBE": TokenType.CUBE, 545 "CURRENT_DATE": TokenType.CURRENT_DATE, 546 "CURRENT_TIME": TokenType.CURRENT_TIME, 547 "CURRENT_TIMESTAMP": TokenType.CURRENT_TIMESTAMP, 548 "CURRENT_USER": TokenType.CURRENT_USER, 549 "DATABASE": TokenType.DATABASE, 550 "DEFAULT": TokenType.DEFAULT, 551 "DELETE": TokenType.DELETE, 552 "DESC": TokenType.DESC, 553 "DESCRIBE": TokenType.DESCRIBE, 554 "DISTINCT": TokenType.DISTINCT, 555 "DISTRIBUTE BY": TokenType.DISTRIBUTE_BY, 556 "DIV": TokenType.DIV, 557 "DROP": TokenType.DROP, 558 "ELSE": TokenType.ELSE, 559 "END": TokenType.END, 560 "ESCAPE": TokenType.ESCAPE, 561 "EXCEPT": TokenType.EXCEPT, 562 "EXECUTE": TokenType.EXECUTE, 563 "EXISTS": TokenType.EXISTS, 564 "FALSE": TokenType.FALSE, 565 "FETCH": TokenType.FETCH, 566 "FILTER": TokenType.FILTER, 567 "FIRST": TokenType.FIRST, 568 "FULL": TokenType.FULL, 569 "FUNCTION": TokenType.FUNCTION, 570 "FOR": TokenType.FOR, 571 "FOREIGN KEY": TokenType.FOREIGN_KEY, 572 "FORMAT": TokenType.FORMAT, 573 "FROM": TokenType.FROM, 574 "GEOGRAPHY": TokenType.GEOGRAPHY, 575 "GEOMETRY": TokenType.GEOMETRY, 576 "GLOB": TokenType.GLOB, 577 "GROUP BY": TokenType.GROUP_BY, 578 "GROUPING SETS": TokenType.GROUPING_SETS, 579 "HAVING": TokenType.HAVING, 580 "ILIKE": TokenType.ILIKE, 581 "IN": TokenType.IN, 582 "INDEX": TokenType.INDEX, 583 "INET": TokenType.INET, 584 "INNER": TokenType.INNER, 585 "INSERT": TokenType.INSERT, 586 "INTERVAL": TokenType.INTERVAL, 587 "INTERSECT": TokenType.INTERSECT, 588 "INTO": TokenType.INTO, 589 "IS": TokenType.IS, 590 "ISNULL": TokenType.ISNULL, 591 "JOIN": TokenType.JOIN, 592 "KEEP": TokenType.KEEP, 593 "LATERAL": TokenType.LATERAL, 594 "LEFT": TokenType.LEFT, 595 "LIKE": TokenType.LIKE, 596 "LIMIT": TokenType.LIMIT, 597 "LOAD": TokenType.LOAD, 598 "LOCK": TokenType.LOCK, 599 "MERGE": TokenType.MERGE, 600 "NATURAL": TokenType.NATURAL, 601 "NEXT": TokenType.NEXT, 602 "NOT": TokenType.NOT, 603 "NOTNULL": TokenType.NOTNULL, 604 "NULL": TokenType.NULL, 605 "OBJECT": TokenType.OBJECT, 606 "OFFSET": TokenType.OFFSET, 607 "ON": TokenType.ON, 608 "OR": TokenType.OR, 609 "XOR": TokenType.XOR, 610 "ORDER BY": TokenType.ORDER_BY, 611 "ORDINALITY": TokenType.ORDINALITY, 612 "OUTER": TokenType.OUTER, 613 "OVER": TokenType.OVER, 614 "OVERLAPS": TokenType.OVERLAPS, 615 "OVERWRITE": TokenType.OVERWRITE, 616 "PARTITION": TokenType.PARTITION, 617 "PARTITION BY": TokenType.PARTITION_BY, 618 "PARTITIONED BY": TokenType.PARTITION_BY, 619 "PARTITIONED_BY": TokenType.PARTITION_BY, 620 "PERCENT": TokenType.PERCENT, 621 "PIVOT": TokenType.PIVOT, 622 "PRAGMA": TokenType.PRAGMA, 623 "PRIMARY KEY": TokenType.PRIMARY_KEY, 624 "PROCEDURE": TokenType.PROCEDURE, 625 "QUALIFY": TokenType.QUALIFY, 626 "RANGE": TokenType.RANGE, 627 "RECURSIVE": TokenType.RECURSIVE, 628 "REGEXP": TokenType.RLIKE, 629 "REPLACE": TokenType.REPLACE, 630 "RETURNING": TokenType.RETURNING, 631 "REFERENCES": TokenType.REFERENCES, 632 "RIGHT": TokenType.RIGHT, 633 "RLIKE": TokenType.RLIKE, 634 "ROLLBACK": TokenType.ROLLBACK, 635 "ROLLUP": TokenType.ROLLUP, 636 "ROW": TokenType.ROW, 637 "ROWS": TokenType.ROWS, 638 "SCHEMA": TokenType.SCHEMA, 639 "SELECT": TokenType.SELECT, 640 "SEMI": TokenType.SEMI, 641 "SET": TokenType.SET, 642 "SETTINGS": TokenType.SETTINGS, 643 "SHOW": TokenType.SHOW, 644 "SIMILAR TO": TokenType.SIMILAR_TO, 645 "SOME": TokenType.SOME, 646 "SORT BY": TokenType.SORT_BY, 647 "START WITH": TokenType.START_WITH, 648 "TABLE": TokenType.TABLE, 649 "TABLESAMPLE": TokenType.TABLE_SAMPLE, 650 "TEMP": TokenType.TEMPORARY, 651 "TEMPORARY": TokenType.TEMPORARY, 652 "THEN": TokenType.THEN, 653 "TRUE": TokenType.TRUE, 654 "UNION": TokenType.UNION, 655 "UNKNOWN": TokenType.UNKNOWN, 656 "UNNEST": TokenType.UNNEST, 657 "UNPIVOT": TokenType.UNPIVOT, 658 "UPDATE": TokenType.UPDATE, 659 "USE": TokenType.USE, 660 "USING": TokenType.USING, 661 "UUID": TokenType.UUID, 662 "VALUES": TokenType.VALUES, 663 "VIEW": TokenType.VIEW, 664 "VOLATILE": TokenType.VOLATILE, 665 "WHEN": TokenType.WHEN, 666 "WHERE": TokenType.WHERE, 667 "WINDOW": TokenType.WINDOW, 668 "WITH": TokenType.WITH, 669 "APPLY": TokenType.APPLY, 670 "ARRAY": TokenType.ARRAY, 671 "BIT": TokenType.BIT, 672 "BOOL": TokenType.BOOLEAN, 673 "BOOLEAN": TokenType.BOOLEAN, 674 "BYTE": TokenType.TINYINT, 675 "TINYINT": TokenType.TINYINT, 676 "SHORT": TokenType.SMALLINT, 677 "SMALLINT": TokenType.SMALLINT, 678 "INT128": TokenType.INT128, 679 "INT2": TokenType.SMALLINT, 680 "INTEGER": TokenType.INT, 681 "INT": TokenType.INT, 682 "INT4": TokenType.INT, 683 "LONG": TokenType.BIGINT, 684 "BIGINT": TokenType.BIGINT, 685 "INT8": TokenType.BIGINT, 686 "DEC": TokenType.DECIMAL, 687 "DECIMAL": TokenType.DECIMAL, 688 "BIGDECIMAL": TokenType.BIGDECIMAL, 689 "BIGNUMERIC": TokenType.BIGDECIMAL, 690 "MAP": TokenType.MAP, 691 "NULLABLE": TokenType.NULLABLE, 692 "NUMBER": TokenType.DECIMAL, 693 "NUMERIC": TokenType.DECIMAL, 694 "FIXED": TokenType.DECIMAL, 695 "REAL": TokenType.FLOAT, 696 "FLOAT": TokenType.FLOAT, 697 "FLOAT4": TokenType.FLOAT, 698 "FLOAT8": TokenType.DOUBLE, 699 "DOUBLE": TokenType.DOUBLE, 700 "DOUBLE PRECISION": TokenType.DOUBLE, 701 "JSON": TokenType.JSON, 702 "CHAR": TokenType.CHAR, 703 "CHARACTER": TokenType.CHAR, 704 "NCHAR": TokenType.NCHAR, 705 "VARCHAR": TokenType.VARCHAR, 706 "VARCHAR2": TokenType.VARCHAR, 707 "NVARCHAR": TokenType.NVARCHAR, 708 "NVARCHAR2": TokenType.NVARCHAR, 709 "STR": TokenType.TEXT, 710 "STRING": TokenType.TEXT, 711 "TEXT": TokenType.TEXT, 712 "CLOB": TokenType.TEXT, 713 "LONGVARCHAR": TokenType.TEXT, 714 "BINARY": TokenType.BINARY, 715 "BLOB": TokenType.VARBINARY, 716 "BYTEA": TokenType.VARBINARY, 717 "VARBINARY": TokenType.VARBINARY, 718 "TIME": TokenType.TIME, 719 "TIMETZ": TokenType.TIMETZ, 720 "TIMESTAMP": TokenType.TIMESTAMP, 721 "TIMESTAMPTZ": TokenType.TIMESTAMPTZ, 722 "TIMESTAMPLTZ": TokenType.TIMESTAMPLTZ, 723 "DATE": TokenType.DATE, 724 "DATETIME": TokenType.DATETIME, 725 "INT4RANGE": TokenType.INT4RANGE, 726 "INT4MULTIRANGE": TokenType.INT4MULTIRANGE, 727 "INT8RANGE": TokenType.INT8RANGE, 728 "INT8MULTIRANGE": TokenType.INT8MULTIRANGE, 729 "NUMRANGE": TokenType.NUMRANGE, 730 "NUMMULTIRANGE": TokenType.NUMMULTIRANGE, 731 "TSRANGE": TokenType.TSRANGE, 732 "TSMULTIRANGE": TokenType.TSMULTIRANGE, 733 "TSTZRANGE": TokenType.TSTZRANGE, 734 "TSTZMULTIRANGE": TokenType.TSTZMULTIRANGE, 735 "DATERANGE": TokenType.DATERANGE, 736 "DATEMULTIRANGE": TokenType.DATEMULTIRANGE, 737 "UNIQUE": TokenType.UNIQUE, 738 "STRUCT": TokenType.STRUCT, 739 "VARIANT": TokenType.VARIANT, 740 "ALTER": TokenType.ALTER, 741 "ANALYZE": TokenType.COMMAND, 742 "CALL": TokenType.COMMAND, 743 "COMMENT": TokenType.COMMENT, 744 "COPY": TokenType.COMMAND, 745 "EXPLAIN": TokenType.COMMAND, 746 "GRANT": TokenType.COMMAND, 747 "OPTIMIZE": TokenType.COMMAND, 748 "PREPARE": TokenType.COMMAND, 749 "TRUNCATE": TokenType.COMMAND, 750 "VACUUM": TokenType.COMMAND, 751 "USER-DEFINED": TokenType.USERDEFINED, 752 } 753 754 WHITE_SPACE: t.Dict[t.Optional[str], TokenType] = { 755 " ": TokenType.SPACE, 756 "\t": TokenType.SPACE, 757 "\n": TokenType.BREAK, 758 "\r": TokenType.BREAK, 759 "\r\n": TokenType.BREAK, 760 } 761 762 COMMANDS = { 763 TokenType.COMMAND, 764 TokenType.EXECUTE, 765 TokenType.FETCH, 766 TokenType.SHOW, 767 } 768 769 COMMAND_PREFIX_TOKENS = {TokenType.SEMICOLON, TokenType.BEGIN} 770 771 # handle numeric literals like in hive (3L = BIGINT) 772 NUMERIC_LITERALS: t.Dict[str, str] = {} 773 ENCODE: t.Optional[str] = None 774 775 COMMENTS = ["--", ("/*", "*/")] 776 777 __slots__ = ( 778 "sql", 779 "size", 780 "tokens", 781 "_start", 782 "_current", 783 "_line", 784 "_col", 785 "_comments", 786 "_char", 787 "_end", 788 "_peek", 789 "_prev_token_line", 790 ) 791 792 def __init__(self) -> None: 793 self.reset() 794 795 def reset(self) -> None: 796 self.sql = "" 797 self.size = 0 798 self.tokens: t.List[Token] = [] 799 self._start = 0 800 self._current = 0 801 self._line = 1 802 self._col = 0 803 self._comments: t.List[str] = [] 804 805 self._char = "" 806 self._end = False 807 self._peek = "" 808 self._prev_token_line = -1 809 810 def tokenize(self, sql: str) -> t.List[Token]: 811 """Returns a list of tokens corresponding to the SQL string `sql`.""" 812 self.reset() 813 self.sql = sql 814 self.size = len(sql) 815 816 try: 817 self._scan() 818 except Exception as e: 819 start = max(self._current - 50, 0) 820 end = min(self._current + 50, self.size - 1) 821 context = self.sql[start:end] 822 raise TokenError(f"Error tokenizing '{context}'") from e 823 824 return self.tokens 825 826 def _scan(self, until: t.Optional[t.Callable] = None) -> None: 827 while self.size and not self._end: 828 self._start = self._current 829 self._advance() 830 831 if self._char is None: 832 break 833 834 if self._char not in self.WHITE_SPACE: 835 if self._char.isdigit(): 836 self._scan_number() 837 elif self._char in self._IDENTIFIERS: 838 self._scan_identifier(self._IDENTIFIERS[self._char]) 839 else: 840 self._scan_keywords() 841 842 if until and until(): 843 break 844 845 if self.tokens and self._comments: 846 self.tokens[-1].comments.extend(self._comments) 847 848 def _chars(self, size: int) -> str: 849 if size == 1: 850 return self._char 851 852 start = self._current - 1 853 end = start + size 854 855 return self.sql[start:end] if end <= self.size else "" 856 857 def _advance(self, i: int = 1, alnum: bool = False) -> None: 858 if self.WHITE_SPACE.get(self._char) is TokenType.BREAK: 859 self._col = 1 860 self._line += 1 861 else: 862 self._col += i 863 864 self._current += i 865 self._end = self._current >= self.size 866 self._char = self.sql[self._current - 1] 867 self._peek = "" if self._end else self.sql[self._current] 868 869 if alnum and self._char.isalnum(): 870 # Here we use local variables instead of attributes for better performance 871 _col = self._col 872 _current = self._current 873 _end = self._end 874 _peek = self._peek 875 876 while _peek.isalnum(): 877 _col += 1 878 _current += 1 879 _end = _current >= self.size 880 _peek = "" if _end else self.sql[_current] 881 882 self._col = _col 883 self._current = _current 884 self._end = _end 885 self._peek = _peek 886 self._char = self.sql[_current - 1] 887 888 @property 889 def _text(self) -> str: 890 return self.sql[self._start : self._current] 891 892 def peek(self, i: int = 0) -> str: 893 i = self._current + i 894 if i < self.size: 895 return self.sql[i] 896 return "" 897 898 def _add(self, token_type: TokenType, text: t.Optional[str] = None) -> None: 899 self._prev_token_line = self._line 900 901 if self._comments and token_type == TokenType.SEMICOLON and self.tokens: 902 self.tokens[-1].comments.extend(self._comments) 903 self._comments = [] 904 905 self.tokens.append( 906 Token( 907 token_type, 908 text=self._text if text is None else text, 909 line=self._line, 910 col=self._col, 911 start=self._start, 912 end=self._current - 1, 913 comments=self._comments, 914 ) 915 ) 916 self._comments = [] 917 918 # If we have either a semicolon or a begin token before the command's token, we'll parse 919 # whatever follows the command's token as a string 920 if ( 921 token_type in self.COMMANDS 922 and self._peek != ";" 923 and (len(self.tokens) == 1 or self.tokens[-2].token_type in self.COMMAND_PREFIX_TOKENS) 924 ): 925 start = self._current 926 tokens = len(self.tokens) 927 self._scan(lambda: self._peek == ";") 928 self.tokens = self.tokens[:tokens] 929 text = self.sql[start : self._current].strip() 930 if text: 931 self._add(TokenType.STRING, text) 932 933 def _scan_keywords(self) -> None: 934 size = 0 935 word = None 936 chars = self._text 937 char = chars 938 prev_space = False 939 skip = False 940 trie = self._KEYWORD_TRIE 941 single_token = char in self.SINGLE_TOKENS 942 943 while chars: 944 if skip: 945 result = TrieResult.PREFIX 946 else: 947 result, trie = in_trie(trie, char.upper()) 948 949 if result == TrieResult.FAILED: 950 break 951 if result == TrieResult.EXISTS: 952 word = chars 953 954 size += 1 955 end = self._current - 1 + size 956 957 if end < self.size: 958 char = self.sql[end] 959 single_token = single_token or char in self.SINGLE_TOKENS 960 is_space = char in self.WHITE_SPACE 961 962 if not is_space or not prev_space: 963 if is_space: 964 char = " " 965 chars += char 966 prev_space = is_space 967 skip = False 968 else: 969 skip = True 970 else: 971 char = "" 972 chars = " " 973 974 if not word: 975 if self._char in self.SINGLE_TOKENS: 976 self._add(self.SINGLE_TOKENS[self._char], text=self._char) 977 return 978 self._scan_var() 979 return 980 981 if self._scan_string(word): 982 return 983 if self._scan_comment(word): 984 return 985 986 self._advance(size - 1) 987 word = word.upper() 988 self._add(self.KEYWORDS[word], text=word) 989 990 def _scan_comment(self, comment_start: str) -> bool: 991 if comment_start not in self._COMMENTS: 992 return False 993 994 comment_start_line = self._line 995 comment_start_size = len(comment_start) 996 comment_end = self._COMMENTS[comment_start] 997 998 if comment_end: 999 # Skip the comment's start delimiter 1000 self._advance(comment_start_size) 1001 1002 comment_end_size = len(comment_end) 1003 while not self._end and self._chars(comment_end_size) != comment_end: 1004 self._advance(alnum=True) 1005 1006 self._comments.append(self._text[comment_start_size : -comment_end_size + 1]) 1007 self._advance(comment_end_size - 1) 1008 else: 1009 while not self._end and not self.WHITE_SPACE.get(self._peek) is TokenType.BREAK: 1010 self._advance(alnum=True) 1011 self._comments.append(self._text[comment_start_size:]) 1012 1013 # Leading comment is attached to the succeeding token, whilst trailing comment to the preceding. 1014 # Multiple consecutive comments are preserved by appending them to the current comments list. 1015 if comment_start_line == self._prev_token_line: 1016 self.tokens[-1].comments.extend(self._comments) 1017 self._comments = [] 1018 self._prev_token_line = self._line 1019 1020 return True 1021 1022 def _scan_number(self) -> None: 1023 if self._char == "0": 1024 peek = self._peek.upper() 1025 if peek == "B": 1026 return self._scan_bits() if self.BIT_STRINGS else self._add(TokenType.NUMBER) 1027 elif peek == "X": 1028 return self._scan_hex() if self.HEX_STRINGS else self._add(TokenType.NUMBER) 1029 1030 decimal = False 1031 scientific = 0 1032 1033 while True: 1034 if self._peek.isdigit(): 1035 self._advance() 1036 elif self._peek == "." and not decimal: 1037 after = self.peek(1) 1038 if after.isdigit() or not after.isalpha(): 1039 decimal = True 1040 self._advance() 1041 else: 1042 return self._add(TokenType.VAR) 1043 elif self._peek in ("-", "+") and scientific == 1: 1044 scientific += 1 1045 self._advance() 1046 elif self._peek.upper() == "E" and not scientific: 1047 scientific += 1 1048 self._advance() 1049 elif self._peek.isidentifier(): 1050 number_text = self._text 1051 literal = "" 1052 1053 while self._peek.strip() and self._peek not in self.SINGLE_TOKENS: 1054 literal += self._peek.upper() 1055 self._advance() 1056 1057 token_type = self.KEYWORDS.get(self.NUMERIC_LITERALS.get(literal, "")) 1058 1059 if token_type: 1060 self._add(TokenType.NUMBER, number_text) 1061 self._add(TokenType.DCOLON, "::") 1062 return self._add(token_type, literal) 1063 elif self.IDENTIFIERS_CAN_START_WITH_DIGIT: 1064 return self._add(TokenType.VAR) 1065 1066 self._add(TokenType.NUMBER, number_text) 1067 return self._advance(-len(literal)) 1068 else: 1069 return self._add(TokenType.NUMBER) 1070 1071 def _scan_bits(self) -> None: 1072 self._advance() 1073 value = self._extract_value() 1074 try: 1075 # If `value` can't be converted to a binary, fallback to tokenizing it as an identifier 1076 int(value, 2) 1077 self._add(TokenType.BIT_STRING, value[2:]) # Drop the 0b 1078 except ValueError: 1079 self._add(TokenType.IDENTIFIER) 1080 1081 def _scan_hex(self) -> None: 1082 self._advance() 1083 value = self._extract_value() 1084 try: 1085 # If `value` can't be converted to a hex, fallback to tokenizing it as an identifier 1086 int(value, 16) 1087 self._add(TokenType.HEX_STRING, value[2:]) # Drop the 0x 1088 except ValueError: 1089 self._add(TokenType.IDENTIFIER) 1090 1091 def _extract_value(self) -> str: 1092 while True: 1093 char = self._peek.strip() 1094 if char and char not in self.SINGLE_TOKENS: 1095 self._advance(alnum=True) 1096 else: 1097 break 1098 1099 return self._text 1100 1101 def _scan_string(self, start: str) -> bool: 1102 base = None 1103 token_type = TokenType.STRING 1104 1105 if start in self._QUOTES: 1106 end = self._QUOTES[start] 1107 elif start in self._FORMAT_STRINGS: 1108 end, token_type = self._FORMAT_STRINGS[start] 1109 1110 if token_type == TokenType.HEX_STRING: 1111 base = 16 1112 elif token_type == TokenType.BIT_STRING: 1113 base = 2 1114 else: 1115 return False 1116 1117 self._advance(len(start)) 1118 text = self._extract_string(end) 1119 1120 if base: 1121 try: 1122 int(text, base) 1123 except: 1124 raise TokenError( 1125 f"Numeric string contains invalid characters from {self._line}:{self._start}" 1126 ) 1127 else: 1128 text = text.encode(self.ENCODE).decode(self.ENCODE) if self.ENCODE else text 1129 1130 self._add(token_type, text) 1131 return True 1132 1133 def _scan_identifier(self, identifier_end: str) -> None: 1134 self._advance() 1135 text = self._extract_string(identifier_end, self._IDENTIFIER_ESCAPES) 1136 self._add(TokenType.IDENTIFIER, text) 1137 1138 def _scan_var(self) -> None: 1139 while True: 1140 char = self._peek.strip() 1141 if char and (char in self.VAR_SINGLE_TOKENS or char not in self.SINGLE_TOKENS): 1142 self._advance(alnum=True) 1143 else: 1144 break 1145 1146 self._add( 1147 TokenType.VAR 1148 if self.tokens and self.tokens[-1].token_type == TokenType.PARAMETER 1149 else self.KEYWORDS.get(self._text.upper(), TokenType.VAR) 1150 ) 1151 1152 def _extract_string(self, delimiter: str, escapes=None) -> str: 1153 text = "" 1154 delim_size = len(delimiter) 1155 escapes = self._STRING_ESCAPES if escapes is None else escapes 1156 1157 while True: 1158 if self._char in escapes and (self._peek == delimiter or self._peek in escapes): 1159 if self._peek == delimiter: 1160 text += self._peek 1161 else: 1162 text += self._char + self._peek 1163 1164 if self._current + 1 < self.size: 1165 self._advance(2) 1166 else: 1167 raise TokenError(f"Missing {delimiter} from {self._line}:{self._current}") 1168 else: 1169 if self._chars(delim_size) == delimiter: 1170 if delim_size > 1: 1171 self._advance(delim_size - 1) 1172 break 1173 1174 if self._end: 1175 raise TokenError(f"Missing {delimiter} from {self._line}:{self._start}") 1176 1177 current = self._current - 1 1178 self._advance(alnum=True) 1179 text += self.sql[current : self._current - 1] 1180 1181 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'>, '??': <TokenType.DQMARK: 'DQMARK'>, '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'>, 'CONNECT BY': <TokenType.CONNECT_BY: 'CONNECT_BY'>, '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'>, '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'>, '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'>, 'XOR': <TokenType.XOR: 'XOR'>, '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'>, 'START WITH': <TokenType.START_WITH: 'START_WITH'>, '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'>, 'UNKNOWN': <TokenType.UNKNOWN: 'UNKNOWN'>, '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'>, 'INT128': <TokenType.INT128: 'INT128'>, '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'>, 'TIMETZ': <TokenType.TIMETZ: 'TIMETZ'>, '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.COMMAND: 'COMMAND'>, <TokenType.SHOW: 'SHOW'>, <TokenType.FETCH: 'FETCH'>, <TokenType.EXECUTE: 'EXECUTE'>}
COMMAND_PREFIX_TOKENS =
{<TokenType.BEGIN: 'BEGIN'>, <TokenType.SEMICOLON: 'SEMICOLON'>}
def
reset(self) -> None:
795 def reset(self) -> None: 796 self.sql = "" 797 self.size = 0 798 self.tokens: t.List[Token] = [] 799 self._start = 0 800 self._current = 0 801 self._line = 1 802 self._col = 0 803 self._comments: t.List[str] = [] 804 805 self._char = "" 806 self._end = False 807 self._peek = "" 808 self._prev_token_line = -1
810 def tokenize(self, sql: str) -> t.List[Token]: 811 """Returns a list of tokens corresponding to the SQL string `sql`.""" 812 self.reset() 813 self.sql = sql 814 self.size = len(sql) 815 816 try: 817 self._scan() 818 except Exception as e: 819 start = max(self._current - 50, 0) 820 end = min(self._current + 50, self.size - 1) 821 context = self.sql[start:end] 822 raise TokenError(f"Error tokenizing '{context}'") from e 823 824 return self.tokens
Returns a list of tokens corresponding to the SQL string sql
.