sqlglot.dialects.duckdb

   1from __future__ import annotations
   2
   3import typing as t
   4
   5from sqlglot import exp, generator, parser, tokens, transforms
   6
   7from sqlglot.expressions import DATA_TYPE
   8from sqlglot.dialects.dialect import (
   9    Dialect,
  10    JSON_EXTRACT_TYPE,
  11    NormalizationStrategy,
  12    Version,
  13    approx_count_distinct_sql,
  14    arrow_json_extract_sql,
  15    binary_from_function,
  16    bool_xor_sql,
  17    build_default_decimal_type,
  18    count_if_to_sum,
  19    date_trunc_to_time,
  20    datestrtodate_sql,
  21    no_datetime_sql,
  22    encode_decode_sql,
  23    build_formatted_time,
  24    inline_array_unless_query,
  25    no_comment_column_constraint_sql,
  26    no_time_sql,
  27    no_timestamp_sql,
  28    pivot_column_names,
  29    rename_func,
  30    remove_from_array_using_filter,
  31    strposition_sql,
  32    str_to_time_sql,
  33    timestamptrunc_sql,
  34    timestrtotime_sql,
  35    unit_to_var,
  36    unit_to_str,
  37    sha256_sql,
  38    build_regexp_extract,
  39    explode_to_unnest_sql,
  40    no_make_interval_sql,
  41    groupconcat_sql,
  42)
  43from sqlglot.generator import unsupported_args
  44from sqlglot.helper import seq_get
  45from sqlglot.tokens import TokenType
  46from sqlglot.parser import binary_range_parser
  47
  48DATETIME_DELTA = t.Union[
  49    exp.DateAdd, exp.TimeAdd, exp.DatetimeAdd, exp.TsOrDsAdd, exp.DateSub, exp.DatetimeSub
  50]
  51
  52
  53def _date_delta_sql(self: DuckDB.Generator, expression: DATETIME_DELTA) -> str:
  54    this = expression.this
  55    unit = unit_to_var(expression)
  56    op = (
  57        "+"
  58        if isinstance(expression, (exp.DateAdd, exp.TimeAdd, exp.DatetimeAdd, exp.TsOrDsAdd))
  59        else "-"
  60    )
  61
  62    to_type: t.Optional[DATA_TYPE] = None
  63    if isinstance(expression, exp.TsOrDsAdd):
  64        to_type = expression.return_type
  65    elif this.is_string:
  66        # Cast string literals (i.e function parameters) to the appropriate type for +/- interval to work
  67        to_type = (
  68            exp.DataType.Type.DATETIME
  69            if isinstance(expression, (exp.DatetimeAdd, exp.DatetimeSub))
  70            else exp.DataType.Type.DATE
  71        )
  72
  73    this = exp.cast(this, to_type) if to_type else this
  74
  75    expr = expression.expression
  76    interval = expr if isinstance(expr, exp.Interval) else exp.Interval(this=expr, unit=unit)
  77
  78    return f"{self.sql(this)} {op} {self.sql(interval)}"
  79
  80
  81# BigQuery -> DuckDB conversion for the DATE function
  82def _date_sql(self: DuckDB.Generator, expression: exp.Date) -> str:
  83    result = f"CAST({self.sql(expression, 'this')} AS DATE)"
  84    zone = self.sql(expression, "zone")
  85
  86    if zone:
  87        date_str = self.func("STRFTIME", result, "'%d/%m/%Y'")
  88        date_str = f"{date_str} || ' ' || {zone}"
  89
  90        # This will create a TIMESTAMP with time zone information
  91        result = self.func("STRPTIME", date_str, "'%d/%m/%Y %Z'")
  92
  93    return result
  94
  95
  96# BigQuery -> DuckDB conversion for the TIME_DIFF function
  97def _timediff_sql(self: DuckDB.Generator, expression: exp.TimeDiff) -> str:
  98    this = exp.cast(expression.this, exp.DataType.Type.TIME)
  99    expr = exp.cast(expression.expression, exp.DataType.Type.TIME)
 100
 101    # Although the 2 dialects share similar signatures, BQ seems to inverse
 102    # the sign of the result so the start/end time operands are flipped
 103    return self.func("DATE_DIFF", unit_to_str(expression), expr, this)
 104
 105
 106@unsupported_args(("expression", "DuckDB's ARRAY_SORT does not support a comparator."))
 107def _array_sort_sql(self: DuckDB.Generator, expression: exp.ArraySort) -> str:
 108    return self.func("ARRAY_SORT", expression.this)
 109
 110
 111def _sort_array_sql(self: DuckDB.Generator, expression: exp.SortArray) -> str:
 112    name = "ARRAY_REVERSE_SORT" if expression.args.get("asc") == exp.false() else "ARRAY_SORT"
 113    return self.func(name, expression.this)
 114
 115
 116def _build_sort_array_desc(args: t.List) -> exp.Expression:
 117    return exp.SortArray(this=seq_get(args, 0), asc=exp.false())
 118
 119
 120def _build_date_diff(args: t.List) -> exp.Expression:
 121    return exp.DateDiff(this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0))
 122
 123
 124def _build_generate_series(end_exclusive: bool = False) -> t.Callable[[t.List], exp.GenerateSeries]:
 125    def _builder(args: t.List) -> exp.GenerateSeries:
 126        # Check https://duckdb.org/docs/sql/functions/nested.html#range-functions
 127        if len(args) == 1:
 128            # DuckDB uses 0 as a default for the series' start when it's omitted
 129            args.insert(0, exp.Literal.number("0"))
 130
 131        gen_series = exp.GenerateSeries.from_arg_list(args)
 132        gen_series.set("is_end_exclusive", end_exclusive)
 133
 134        return gen_series
 135
 136    return _builder
 137
 138
 139def _build_make_timestamp(args: t.List) -> exp.Expression:
 140    if len(args) == 1:
 141        return exp.UnixToTime(this=seq_get(args, 0), scale=exp.UnixToTime.MICROS)
 142
 143    return exp.TimestampFromParts(
 144        year=seq_get(args, 0),
 145        month=seq_get(args, 1),
 146        day=seq_get(args, 2),
 147        hour=seq_get(args, 3),
 148        min=seq_get(args, 4),
 149        sec=seq_get(args, 5),
 150    )
 151
 152
 153def _show_parser(*args: t.Any, **kwargs: t.Any) -> t.Callable[[DuckDB.Parser], exp.Show]:
 154    def _parse(self: DuckDB.Parser) -> exp.Show:
 155        return self._parse_show_duckdb(*args, **kwargs)
 156
 157    return _parse
 158
 159
 160def _struct_sql(self: DuckDB.Generator, expression: exp.Struct) -> str:
 161    args: t.List[str] = []
 162
 163    # BigQuery allows inline construction such as "STRUCT<a STRING, b INTEGER>('str', 1)" which is
 164    # canonicalized to "ROW('str', 1) AS STRUCT(a TEXT, b INT)" in DuckDB
 165    # The transformation to ROW will take place if:
 166    #  1. The STRUCT itself does not have proper fields (key := value) as a "proper" STRUCT would
 167    #  2. A cast to STRUCT / ARRAY of STRUCTs is found
 168    ancestor_cast = expression.find_ancestor(exp.Cast)
 169    is_bq_inline_struct = (
 170        (expression.find(exp.PropertyEQ) is None)
 171        and ancestor_cast
 172        and any(
 173            casted_type.is_type(exp.DataType.Type.STRUCT)
 174            for casted_type in ancestor_cast.find_all(exp.DataType)
 175        )
 176    )
 177
 178    for i, expr in enumerate(expression.expressions):
 179        is_property_eq = isinstance(expr, exp.PropertyEQ)
 180        value = expr.expression if is_property_eq else expr
 181
 182        if is_bq_inline_struct:
 183            args.append(self.sql(value))
 184        else:
 185            if is_property_eq:
 186                if isinstance(expr.this, exp.Identifier):
 187                    key = self.sql(exp.Literal.string(expr.name))
 188                else:
 189                    key = self.sql(expr.this)
 190            else:
 191                key = self.sql(exp.Literal.string(f"_{i}"))
 192
 193            args.append(f"{key}: {self.sql(value)}")
 194
 195    csv_args = ", ".join(args)
 196
 197    return f"ROW({csv_args})" if is_bq_inline_struct else f"{{{csv_args}}}"
 198
 199
 200def _datatype_sql(self: DuckDB.Generator, expression: exp.DataType) -> str:
 201    if expression.is_type("array"):
 202        return f"{self.expressions(expression, flat=True)}[{self.expressions(expression, key='values', flat=True)}]"
 203
 204    # Modifiers are not supported for TIME, [TIME | TIMESTAMP] WITH TIME ZONE
 205    if expression.is_type(
 206        exp.DataType.Type.TIME, exp.DataType.Type.TIMETZ, exp.DataType.Type.TIMESTAMPTZ
 207    ):
 208        return expression.this.value
 209
 210    return self.datatype_sql(expression)
 211
 212
 213def _json_format_sql(self: DuckDB.Generator, expression: exp.JSONFormat) -> str:
 214    sql = self.func("TO_JSON", expression.this, expression.args.get("options"))
 215    return f"CAST({sql} AS TEXT)"
 216
 217
 218def _unix_to_time_sql(self: DuckDB.Generator, expression: exp.UnixToTime) -> str:
 219    scale = expression.args.get("scale")
 220    timestamp = expression.this
 221
 222    if scale in (None, exp.UnixToTime.SECONDS):
 223        return self.func("TO_TIMESTAMP", timestamp)
 224    if scale == exp.UnixToTime.MILLIS:
 225        return self.func("EPOCH_MS", timestamp)
 226    if scale == exp.UnixToTime.MICROS:
 227        return self.func("MAKE_TIMESTAMP", timestamp)
 228
 229    return self.func("TO_TIMESTAMP", exp.Div(this=timestamp, expression=exp.func("POW", 10, scale)))
 230
 231
 232WRAPPED_JSON_EXTRACT_EXPRESSIONS = (exp.Binary, exp.Bracket, exp.In)
 233
 234
 235def _arrow_json_extract_sql(self: DuckDB.Generator, expression: JSON_EXTRACT_TYPE) -> str:
 236    arrow_sql = arrow_json_extract_sql(self, expression)
 237    if not expression.same_parent and isinstance(
 238        expression.parent, WRAPPED_JSON_EXTRACT_EXPRESSIONS
 239    ):
 240        arrow_sql = self.wrap(arrow_sql)
 241    return arrow_sql
 242
 243
 244def _implicit_datetime_cast(
 245    arg: t.Optional[exp.Expression], type: exp.DataType.Type = exp.DataType.Type.DATE
 246) -> t.Optional[exp.Expression]:
 247    return exp.cast(arg, type) if isinstance(arg, exp.Literal) else arg
 248
 249
 250def _date_diff_sql(self: DuckDB.Generator, expression: exp.DateDiff) -> str:
 251    this = _implicit_datetime_cast(expression.this)
 252    expr = _implicit_datetime_cast(expression.expression)
 253
 254    return self.func("DATE_DIFF", unit_to_str(expression), expr, this)
 255
 256
 257def _generate_datetime_array_sql(
 258    self: DuckDB.Generator, expression: t.Union[exp.GenerateDateArray, exp.GenerateTimestampArray]
 259) -> str:
 260    is_generate_date_array = isinstance(expression, exp.GenerateDateArray)
 261
 262    type = exp.DataType.Type.DATE if is_generate_date_array else exp.DataType.Type.TIMESTAMP
 263    start = _implicit_datetime_cast(expression.args.get("start"), type=type)
 264    end = _implicit_datetime_cast(expression.args.get("end"), type=type)
 265
 266    # BQ's GENERATE_DATE_ARRAY & GENERATE_TIMESTAMP_ARRAY are transformed to DuckDB'S GENERATE_SERIES
 267    gen_series: t.Union[exp.GenerateSeries, exp.Cast] = exp.GenerateSeries(
 268        start=start, end=end, step=expression.args.get("step")
 269    )
 270
 271    if is_generate_date_array:
 272        # The GENERATE_SERIES result type is TIMESTAMP array, so to match BQ's semantics for
 273        # GENERATE_DATE_ARRAY we must cast it back to DATE array
 274        gen_series = exp.cast(gen_series, exp.DataType.build("ARRAY<DATE>"))
 275
 276    return self.sql(gen_series)
 277
 278
 279def _json_extract_value_array_sql(
 280    self: DuckDB.Generator, expression: exp.JSONValueArray | exp.JSONExtractArray
 281) -> str:
 282    json_extract = exp.JSONExtract(this=expression.this, expression=expression.expression)
 283    data_type = "ARRAY<STRING>" if isinstance(expression, exp.JSONValueArray) else "ARRAY<JSON>"
 284    return self.sql(exp.cast(json_extract, to=exp.DataType.build(data_type)))
 285
 286
 287class DuckDB(Dialect):
 288    NULL_ORDERING = "nulls_are_last"
 289    SUPPORTS_USER_DEFINED_TYPES = True
 290    SAFE_DIVISION = True
 291    INDEX_OFFSET = 1
 292    CONCAT_COALESCE = True
 293    SUPPORTS_ORDER_BY_ALL = True
 294    SUPPORTS_FIXED_SIZE_ARRAYS = True
 295    STRICT_JSON_PATH_SYNTAX = False
 296    NUMBERS_CAN_BE_UNDERSCORE_SEPARATED = True
 297
 298    # https://duckdb.org/docs/sql/introduction.html#creating-a-new-table
 299    NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE
 300
 301    DATE_PART_MAPPING = {
 302        **Dialect.DATE_PART_MAPPING,
 303        "DAYOFWEEKISO": "ISODOW",
 304    }
 305    DATE_PART_MAPPING.pop("WEEKDAY")
 306
 307    def to_json_path(self, path: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]:
 308        if isinstance(path, exp.Literal):
 309            # DuckDB also supports the JSON pointer syntax, where every path starts with a `/`.
 310            # Additionally, it allows accessing the back of lists using the `[#-i]` syntax.
 311            # This check ensures we'll avoid trying to parse these as JSON paths, which can
 312            # either result in a noisy warning or in an invalid representation of the path.
 313            path_text = path.name
 314            if path_text.startswith("/") or "[#" in path_text:
 315                return path
 316
 317        return super().to_json_path(path)
 318
 319    class Tokenizer(tokens.Tokenizer):
 320        BYTE_STRINGS = [("e'", "'"), ("E'", "'")]
 321        HEREDOC_STRINGS = ["$"]
 322
 323        HEREDOC_TAG_IS_IDENTIFIER = True
 324        HEREDOC_STRING_ALTERNATIVE = TokenType.PARAMETER
 325
 326        KEYWORDS = {
 327            **tokens.Tokenizer.KEYWORDS,
 328            "//": TokenType.DIV,
 329            "**": TokenType.DSTAR,
 330            "^@": TokenType.CARET_AT,
 331            "@>": TokenType.AT_GT,
 332            "<@": TokenType.LT_AT,
 333            "ATTACH": TokenType.ATTACH,
 334            "BINARY": TokenType.VARBINARY,
 335            "BITSTRING": TokenType.BIT,
 336            "BPCHAR": TokenType.TEXT,
 337            "CHAR": TokenType.TEXT,
 338            "DATETIME": TokenType.TIMESTAMPNTZ,
 339            "DETACH": TokenType.DETACH,
 340            "EXCLUDE": TokenType.EXCEPT,
 341            "LOGICAL": TokenType.BOOLEAN,
 342            "ONLY": TokenType.ONLY,
 343            "PIVOT_WIDER": TokenType.PIVOT,
 344            "POSITIONAL": TokenType.POSITIONAL,
 345            "RESET": TokenType.COMMAND,
 346            "SIGNED": TokenType.INT,
 347            "STRING": TokenType.TEXT,
 348            "SUMMARIZE": TokenType.SUMMARIZE,
 349            "TIMESTAMP": TokenType.TIMESTAMPNTZ,
 350            "TIMESTAMP_S": TokenType.TIMESTAMP_S,
 351            "TIMESTAMP_MS": TokenType.TIMESTAMP_MS,
 352            "TIMESTAMP_NS": TokenType.TIMESTAMP_NS,
 353            "TIMESTAMP_US": TokenType.TIMESTAMP,
 354            "UBIGINT": TokenType.UBIGINT,
 355            "UINTEGER": TokenType.UINT,
 356            "USMALLINT": TokenType.USMALLINT,
 357            "UTINYINT": TokenType.UTINYINT,
 358            "VARCHAR": TokenType.TEXT,
 359        }
 360        KEYWORDS.pop("/*+")
 361
 362        SINGLE_TOKENS = {
 363            **tokens.Tokenizer.SINGLE_TOKENS,
 364            "$": TokenType.PARAMETER,
 365        }
 366
 367        COMMANDS = tokens.Tokenizer.COMMANDS - {TokenType.SHOW}
 368
 369    class Parser(parser.Parser):
 370        MAP_KEYS_ARE_ARBITRARY_EXPRESSIONS = True
 371
 372        BITWISE = {
 373            **parser.Parser.BITWISE,
 374            TokenType.TILDA: exp.RegexpLike,
 375        }
 376        BITWISE.pop(TokenType.CARET)
 377
 378        RANGE_PARSERS = {
 379            **parser.Parser.RANGE_PARSERS,
 380            TokenType.DAMP: binary_range_parser(exp.ArrayOverlaps),
 381            TokenType.CARET_AT: binary_range_parser(exp.StartsWith),
 382        }
 383
 384        EXPONENT = {
 385            **parser.Parser.EXPONENT,
 386            TokenType.CARET: exp.Pow,
 387            TokenType.DSTAR: exp.Pow,
 388        }
 389
 390        FUNCTIONS_WITH_ALIASED_ARGS = {*parser.Parser.FUNCTIONS_WITH_ALIASED_ARGS, "STRUCT_PACK"}
 391
 392        SHOW_PARSERS = {
 393            "TABLES": _show_parser("TABLES"),
 394            "ALL TABLES": _show_parser("ALL TABLES"),
 395        }
 396
 397        FUNCTIONS = {
 398            **parser.Parser.FUNCTIONS,
 399            "ARRAY_REVERSE_SORT": _build_sort_array_desc,
 400            "ARRAY_SORT": exp.SortArray.from_arg_list,
 401            "DATEDIFF": _build_date_diff,
 402            "DATE_DIFF": _build_date_diff,
 403            "DATE_TRUNC": date_trunc_to_time,
 404            "DATETRUNC": date_trunc_to_time,
 405            "DECODE": lambda args: exp.Decode(
 406                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
 407            ),
 408            "EDITDIST3": exp.Levenshtein.from_arg_list,
 409            "ENCODE": lambda args: exp.Encode(
 410                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
 411            ),
 412            "EPOCH": exp.TimeToUnix.from_arg_list,
 413            "EPOCH_MS": lambda args: exp.UnixToTime(
 414                this=seq_get(args, 0), scale=exp.UnixToTime.MILLIS
 415            ),
 416            "GENERATE_SERIES": _build_generate_series(),
 417            "JSON": exp.ParseJSON.from_arg_list,
 418            "JSON_EXTRACT_PATH": parser.build_extract_json_with_path(exp.JSONExtract),
 419            "JSON_EXTRACT_STRING": parser.build_extract_json_with_path(exp.JSONExtractScalar),
 420            "LIST_CONTAINS": exp.ArrayContains.from_arg_list,
 421            "LIST_HAS": exp.ArrayContains.from_arg_list,
 422            "LIST_HAS_ANY": exp.ArrayOverlaps.from_arg_list,
 423            "LIST_REVERSE_SORT": _build_sort_array_desc,
 424            "LIST_SORT": exp.SortArray.from_arg_list,
 425            "LIST_VALUE": lambda args: exp.Array(expressions=args),
 426            "MAKE_TIME": exp.TimeFromParts.from_arg_list,
 427            "MAKE_TIMESTAMP": _build_make_timestamp,
 428            "QUANTILE_CONT": exp.PercentileCont.from_arg_list,
 429            "QUANTILE_DISC": exp.PercentileDisc.from_arg_list,
 430            "RANGE": _build_generate_series(end_exclusive=True),
 431            "REGEXP_EXTRACT": build_regexp_extract(exp.RegexpExtract),
 432            "REGEXP_EXTRACT_ALL": build_regexp_extract(exp.RegexpExtractAll),
 433            "REGEXP_MATCHES": exp.RegexpLike.from_arg_list,
 434            "REGEXP_REPLACE": lambda args: exp.RegexpReplace(
 435                this=seq_get(args, 0),
 436                expression=seq_get(args, 1),
 437                replacement=seq_get(args, 2),
 438                modifiers=seq_get(args, 3),
 439            ),
 440            "SHA256": lambda args: exp.SHA2(this=seq_get(args, 0), length=exp.Literal.number(256)),
 441            "STRFTIME": build_formatted_time(exp.TimeToStr, "duckdb"),
 442            "STRING_SPLIT": exp.Split.from_arg_list,
 443            "STRING_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
 444            "STRING_TO_ARRAY": exp.Split.from_arg_list,
 445            "STRPTIME": build_formatted_time(exp.StrToTime, "duckdb"),
 446            "STRUCT_PACK": exp.Struct.from_arg_list,
 447            "STR_SPLIT": exp.Split.from_arg_list,
 448            "STR_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
 449            "TIME_BUCKET": exp.DateBin.from_arg_list,
 450            "TO_TIMESTAMP": exp.UnixToTime.from_arg_list,
 451            "UNNEST": exp.Explode.from_arg_list,
 452            "XOR": binary_from_function(exp.BitwiseXor),
 453        }
 454
 455        FUNCTIONS.pop("DATE_SUB")
 456        FUNCTIONS.pop("GLOB")
 457
 458        FUNCTION_PARSERS = {
 459            **parser.Parser.FUNCTION_PARSERS,
 460            **dict.fromkeys(
 461                ("GROUP_CONCAT", "LISTAGG", "STRINGAGG"), lambda self: self._parse_string_agg()
 462            ),
 463        }
 464        FUNCTION_PARSERS.pop("DECODE")
 465
 466        NO_PAREN_FUNCTION_PARSERS = {
 467            **parser.Parser.NO_PAREN_FUNCTION_PARSERS,
 468            "MAP": lambda self: self._parse_map(),
 469            "@": lambda self: exp.Abs(this=self._parse_bitwise()),
 470        }
 471
 472        TABLE_ALIAS_TOKENS = parser.Parser.TABLE_ALIAS_TOKENS - {
 473            TokenType.SEMI,
 474            TokenType.ANTI,
 475        }
 476
 477        PLACEHOLDER_PARSERS = {
 478            **parser.Parser.PLACEHOLDER_PARSERS,
 479            TokenType.PARAMETER: lambda self: (
 480                self.expression(exp.Placeholder, this=self._prev.text)
 481                if self._match(TokenType.NUMBER) or self._match_set(self.ID_VAR_TOKENS)
 482                else None
 483            ),
 484        }
 485
 486        TYPE_CONVERTERS = {
 487            # https://duckdb.org/docs/sql/data_types/numeric
 488            exp.DataType.Type.DECIMAL: build_default_decimal_type(precision=18, scale=3),
 489            # https://duckdb.org/docs/sql/data_types/text
 490            exp.DataType.Type.TEXT: lambda dtype: exp.DataType.build("TEXT"),
 491        }
 492
 493        STATEMENT_PARSERS = {
 494            **parser.Parser.STATEMENT_PARSERS,
 495            TokenType.ATTACH: lambda self: self._parse_attach_detach(),
 496            TokenType.DETACH: lambda self: self._parse_attach_detach(is_attach=False),
 497            TokenType.SHOW: lambda self: self._parse_show(),
 498        }
 499
 500        SET_PARSERS = {
 501            **parser.Parser.SET_PARSERS,
 502            "VARIABLE": lambda self: self._parse_set_item_assignment("VARIABLE"),
 503        }
 504
 505        def _parse_lambda(self, alias: bool = False) -> t.Optional[exp.Expression]:
 506            index = self._index
 507            if not self._match_text_seq("LAMBDA"):
 508                return super()._parse_lambda(alias=alias)
 509
 510            expressions = self._parse_csv(self._parse_lambda_arg)
 511            if not self._match(TokenType.COLON):
 512                self._retreat(index)
 513                return None
 514
 515            this = self._replace_lambda(self._parse_assignment(), expressions)
 516            return self.expression(exp.Lambda, this=this, expressions=expressions, colon=True)
 517
 518        def _parse_expression(self) -> t.Optional[exp.Expression]:
 519            # DuckDB supports prefix aliases, e.g. foo: 1
 520            if self._next and self._next.token_type == TokenType.COLON:
 521                alias = self._parse_id_var(tokens=self.ALIAS_TOKENS)
 522                self._match(TokenType.COLON)
 523                comments = self._prev_comments or []
 524
 525                this = self._parse_assignment()
 526                if isinstance(this, exp.Expression):
 527                    # Moves the comment next to the alias in `alias: expr /* comment */`
 528                    comments += this.pop_comments() or []
 529
 530                return self.expression(exp.Alias, comments=comments, this=this, alias=alias)
 531
 532            return super()._parse_expression()
 533
 534        def _parse_table(
 535            self,
 536            schema: bool = False,
 537            joins: bool = False,
 538            alias_tokens: t.Optional[t.Collection[TokenType]] = None,
 539            parse_bracket: bool = False,
 540            is_db_reference: bool = False,
 541            parse_partition: bool = False,
 542            consume_pipe: bool = False,
 543        ) -> t.Optional[exp.Expression]:
 544            # DuckDB supports prefix aliases, e.g. FROM foo: bar
 545            if self._next and self._next.token_type == TokenType.COLON:
 546                alias = self._parse_table_alias(
 547                    alias_tokens=alias_tokens or self.TABLE_ALIAS_TOKENS
 548                )
 549                self._match(TokenType.COLON)
 550                comments = self._prev_comments or []
 551            else:
 552                alias = None
 553                comments = []
 554
 555            table = super()._parse_table(
 556                schema=schema,
 557                joins=joins,
 558                alias_tokens=alias_tokens,
 559                parse_bracket=parse_bracket,
 560                is_db_reference=is_db_reference,
 561                parse_partition=parse_partition,
 562            )
 563            if isinstance(table, exp.Expression) and isinstance(alias, exp.TableAlias):
 564                # Moves the comment next to the alias in `alias: table /* comment */`
 565                comments += table.pop_comments() or []
 566                alias.comments = alias.pop_comments() + comments
 567                table.set("alias", alias)
 568
 569            return table
 570
 571        def _parse_table_sample(self, as_modifier: bool = False) -> t.Optional[exp.TableSample]:
 572            # https://duckdb.org/docs/sql/samples.html
 573            sample = super()._parse_table_sample(as_modifier=as_modifier)
 574            if sample and not sample.args.get("method"):
 575                if sample.args.get("size"):
 576                    sample.set("method", exp.var("RESERVOIR"))
 577                else:
 578                    sample.set("method", exp.var("SYSTEM"))
 579
 580            return sample
 581
 582        def _parse_bracket(
 583            self, this: t.Optional[exp.Expression] = None
 584        ) -> t.Optional[exp.Expression]:
 585            bracket = super()._parse_bracket(this)
 586
 587            if self.dialect.version < Version("1.2.0") and isinstance(bracket, exp.Bracket):
 588                # https://duckdb.org/2025/02/05/announcing-duckdb-120.html#breaking-changes
 589                bracket.set("returns_list_for_maps", True)
 590
 591            return bracket
 592
 593        def _parse_map(self) -> exp.ToMap | exp.Map:
 594            if self._match(TokenType.L_BRACE, advance=False):
 595                return self.expression(exp.ToMap, this=self._parse_bracket())
 596
 597            args = self._parse_wrapped_csv(self._parse_assignment)
 598            return self.expression(exp.Map, keys=seq_get(args, 0), values=seq_get(args, 1))
 599
 600        def _parse_struct_types(self, type_required: bool = False) -> t.Optional[exp.Expression]:
 601            return self._parse_field_def()
 602
 603        def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]:
 604            if len(aggregations) == 1:
 605                return super()._pivot_column_names(aggregations)
 606            return pivot_column_names(aggregations, dialect="duckdb")
 607
 608        def _parse_attach_detach(self, is_attach=True) -> exp.Attach | exp.Detach:
 609            def _parse_attach_option() -> exp.AttachOption:
 610                return self.expression(
 611                    exp.AttachOption,
 612                    this=self._parse_var(any_token=True),
 613                    expression=self._parse_field(any_token=True),
 614                )
 615
 616            self._match(TokenType.DATABASE)
 617            exists = self._parse_exists(not_=is_attach)
 618            this = self._parse_alias(self._parse_primary_or_var(), explicit=True)
 619
 620            if self._match(TokenType.L_PAREN, advance=False):
 621                expressions = self._parse_wrapped_csv(_parse_attach_option)
 622            else:
 623                expressions = None
 624
 625            return (
 626                self.expression(exp.Attach, this=this, exists=exists, expressions=expressions)
 627                if is_attach
 628                else self.expression(exp.Detach, this=this, exists=exists)
 629            )
 630
 631        def _parse_show_duckdb(self, this: str) -> exp.Show:
 632            return self.expression(exp.Show, this=this)
 633
 634        def _parse_primary(self) -> t.Optional[exp.Expression]:
 635            if self._match_pair(TokenType.HASH, TokenType.NUMBER):
 636                return exp.PositionalColumn(this=exp.Literal.number(self._prev.text))
 637
 638            return super()._parse_primary()
 639
 640    class Generator(generator.Generator):
 641        PARAMETER_TOKEN = "$"
 642        NAMED_PLACEHOLDER_TOKEN = "$"
 643        JOIN_HINTS = False
 644        TABLE_HINTS = False
 645        QUERY_HINTS = False
 646        LIMIT_FETCH = "LIMIT"
 647        STRUCT_DELIMITER = ("(", ")")
 648        RENAME_TABLE_WITH_DB = False
 649        NVL2_SUPPORTED = False
 650        SEMI_ANTI_JOIN_WITH_SIDE = False
 651        TABLESAMPLE_KEYWORDS = "USING SAMPLE"
 652        TABLESAMPLE_SEED_KEYWORD = "REPEATABLE"
 653        LAST_DAY_SUPPORTS_DATE_PART = False
 654        JSON_KEY_VALUE_PAIR_SEP = ","
 655        IGNORE_NULLS_IN_FUNC = True
 656        JSON_PATH_BRACKETED_KEY_SUPPORTED = False
 657        SUPPORTS_CREATE_TABLE_LIKE = False
 658        MULTI_ARG_DISTINCT = False
 659        CAN_IMPLEMENT_ARRAY_ANY = True
 660        SUPPORTS_TO_NUMBER = False
 661        SUPPORTS_WINDOW_EXCLUDE = True
 662        COPY_HAS_INTO_KEYWORD = False
 663        STAR_EXCEPT = "EXCLUDE"
 664        PAD_FILL_PATTERN_IS_REQUIRED = True
 665        ARRAY_CONCAT_IS_VAR_LEN = False
 666        ARRAY_SIZE_DIM_REQUIRED = False
 667        NORMALIZE_EXTRACT_DATE_PARTS = True
 668        SUPPORTS_LIKE_QUANTIFIERS = False
 669
 670        TRANSFORMS = {
 671            **generator.Generator.TRANSFORMS,
 672            exp.ApproxDistinct: approx_count_distinct_sql,
 673            exp.Array: inline_array_unless_query,
 674            exp.ArrayFilter: rename_func("LIST_FILTER"),
 675            exp.ArrayRemove: remove_from_array_using_filter,
 676            exp.ArraySort: _array_sort_sql,
 677            exp.ArraySum: rename_func("LIST_SUM"),
 678            exp.BitwiseXor: rename_func("XOR"),
 679            exp.CommentColumnConstraint: no_comment_column_constraint_sql,
 680            exp.CurrentDate: lambda *_: "CURRENT_DATE",
 681            exp.CurrentTime: lambda *_: "CURRENT_TIME",
 682            exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP",
 683            exp.DayOfMonth: rename_func("DAYOFMONTH"),
 684            exp.DayOfWeek: rename_func("DAYOFWEEK"),
 685            exp.DayOfWeekIso: rename_func("ISODOW"),
 686            exp.DayOfYear: rename_func("DAYOFYEAR"),
 687            exp.DataType: _datatype_sql,
 688            exp.Date: _date_sql,
 689            exp.DateAdd: _date_delta_sql,
 690            exp.DateFromParts: rename_func("MAKE_DATE"),
 691            exp.DateSub: _date_delta_sql,
 692            exp.DateDiff: _date_diff_sql,
 693            exp.DateStrToDate: datestrtodate_sql,
 694            exp.Datetime: no_datetime_sql,
 695            exp.DatetimeSub: _date_delta_sql,
 696            exp.DatetimeAdd: _date_delta_sql,
 697            exp.DateToDi: lambda self,
 698            e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.DATEINT_FORMAT}) AS INT)",
 699            exp.Decode: lambda self, e: encode_decode_sql(self, e, "DECODE", replace=False),
 700            exp.DiToDate: lambda self,
 701            e: f"CAST(STRPTIME(CAST({self.sql(e, 'this')} AS TEXT), {DuckDB.DATEINT_FORMAT}) AS DATE)",
 702            exp.Encode: lambda self, e: encode_decode_sql(self, e, "ENCODE", replace=False),
 703            exp.GenerateDateArray: _generate_datetime_array_sql,
 704            exp.GenerateTimestampArray: _generate_datetime_array_sql,
 705            exp.GroupConcat: lambda self, e: groupconcat_sql(self, e, within_group=False),
 706            exp.HexString: lambda self, e: self.hexstring_sql(e, binary_function_repr="FROM_HEX"),
 707            exp.Explode: rename_func("UNNEST"),
 708            exp.IntDiv: lambda self, e: self.binary(e, "//"),
 709            exp.IsInf: rename_func("ISINF"),
 710            exp.IsNan: rename_func("ISNAN"),
 711            exp.JSONBExists: rename_func("JSON_EXISTS"),
 712            exp.JSONExtract: _arrow_json_extract_sql,
 713            exp.JSONExtractArray: _json_extract_value_array_sql,
 714            exp.JSONExtractScalar: _arrow_json_extract_sql,
 715            exp.JSONFormat: _json_format_sql,
 716            exp.JSONValueArray: _json_extract_value_array_sql,
 717            exp.Lateral: explode_to_unnest_sql,
 718            exp.LogicalOr: rename_func("BOOL_OR"),
 719            exp.LogicalAnd: rename_func("BOOL_AND"),
 720            exp.MakeInterval: lambda self, e: no_make_interval_sql(self, e, sep=" "),
 721            exp.MD5Digest: lambda self, e: self.func("UNHEX", self.func("MD5", e.this)),
 722            exp.MonthsBetween: lambda self, e: self.func(
 723                "DATEDIFF",
 724                "'month'",
 725                exp.cast(e.expression, exp.DataType.Type.TIMESTAMP, copy=True),
 726                exp.cast(e.this, exp.DataType.Type.TIMESTAMP, copy=True),
 727            ),
 728            exp.PercentileCont: rename_func("QUANTILE_CONT"),
 729            exp.PercentileDisc: rename_func("QUANTILE_DISC"),
 730            # DuckDB doesn't allow qualified columns inside of PIVOT expressions.
 731            # See: https://github.com/duckdb/duckdb/blob/671faf92411182f81dce42ac43de8bfb05d9909e/src/planner/binder/tableref/bind_pivot.cpp#L61-L62
 732            exp.Pivot: transforms.preprocess([transforms.unqualify_columns]),
 733            exp.RegexpReplace: lambda self, e: self.func(
 734                "REGEXP_REPLACE",
 735                e.this,
 736                e.expression,
 737                e.args.get("replacement"),
 738                e.args.get("modifiers"),
 739            ),
 740            exp.RegexpLike: rename_func("REGEXP_MATCHES"),
 741            exp.RegexpILike: lambda self, e: self.func(
 742                "REGEXP_MATCHES", e.this, e.expression, exp.Literal.string("i")
 743            ),
 744            exp.RegexpSplit: rename_func("STR_SPLIT_REGEX"),
 745            exp.Return: lambda self, e: self.sql(e, "this"),
 746            exp.ReturnsProperty: lambda self, e: "TABLE" if isinstance(e.this, exp.Schema) else "",
 747            exp.Rand: rename_func("RANDOM"),
 748            exp.SHA: rename_func("SHA1"),
 749            exp.SHA2: sha256_sql,
 750            exp.Split: rename_func("STR_SPLIT"),
 751            exp.SortArray: _sort_array_sql,
 752            exp.StrPosition: strposition_sql,
 753            exp.StrToUnix: lambda self, e: self.func(
 754                "EPOCH", self.func("STRPTIME", e.this, self.format_time(e))
 755            ),
 756            exp.Struct: _struct_sql,
 757            exp.Transform: rename_func("LIST_TRANSFORM"),
 758            exp.TimeAdd: _date_delta_sql,
 759            exp.Time: no_time_sql,
 760            exp.TimeDiff: _timediff_sql,
 761            exp.Timestamp: no_timestamp_sql,
 762            exp.TimestampDiff: lambda self, e: self.func(
 763                "DATE_DIFF", exp.Literal.string(e.unit), e.expression, e.this
 764            ),
 765            exp.TimestampTrunc: timestamptrunc_sql(),
 766            exp.TimeStrToDate: lambda self, e: self.sql(exp.cast(e.this, exp.DataType.Type.DATE)),
 767            exp.TimeStrToTime: timestrtotime_sql,
 768            exp.TimeStrToUnix: lambda self, e: self.func(
 769                "EPOCH", exp.cast(e.this, exp.DataType.Type.TIMESTAMP)
 770            ),
 771            exp.TimeToStr: lambda self, e: self.func("STRFTIME", e.this, self.format_time(e)),
 772            exp.TimeToUnix: rename_func("EPOCH"),
 773            exp.TsOrDiToDi: lambda self,
 774            e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS TEXT), '-', ''), 1, 8) AS INT)",
 775            exp.TsOrDsAdd: _date_delta_sql,
 776            exp.TsOrDsDiff: lambda self, e: self.func(
 777                "DATE_DIFF",
 778                f"'{e.args.get('unit') or 'DAY'}'",
 779                exp.cast(e.expression, exp.DataType.Type.TIMESTAMP),
 780                exp.cast(e.this, exp.DataType.Type.TIMESTAMP),
 781            ),
 782            exp.UnixToStr: lambda self, e: self.func(
 783                "STRFTIME", self.func("TO_TIMESTAMP", e.this), self.format_time(e)
 784            ),
 785            exp.DatetimeTrunc: lambda self, e: self.func(
 786                "DATE_TRUNC", unit_to_str(e), exp.cast(e.this, exp.DataType.Type.DATETIME)
 787            ),
 788            exp.UnixToTime: _unix_to_time_sql,
 789            exp.UnixToTimeStr: lambda self, e: f"CAST(TO_TIMESTAMP({self.sql(e, 'this')}) AS TEXT)",
 790            exp.VariancePop: rename_func("VAR_POP"),
 791            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
 792            exp.Xor: bool_xor_sql,
 793            exp.Levenshtein: unsupported_args("ins_cost", "del_cost", "sub_cost", "max_dist")(
 794                rename_func("LEVENSHTEIN")
 795            ),
 796            exp.JSONObjectAgg: rename_func("JSON_GROUP_OBJECT"),
 797            exp.JSONBObjectAgg: rename_func("JSON_GROUP_OBJECT"),
 798            exp.DateBin: rename_func("TIME_BUCKET"),
 799        }
 800
 801        SUPPORTED_JSON_PATH_PARTS = {
 802            exp.JSONPathKey,
 803            exp.JSONPathRoot,
 804            exp.JSONPathSubscript,
 805            exp.JSONPathWildcard,
 806        }
 807
 808        TYPE_MAPPING = {
 809            **generator.Generator.TYPE_MAPPING,
 810            exp.DataType.Type.BINARY: "BLOB",
 811            exp.DataType.Type.BPCHAR: "TEXT",
 812            exp.DataType.Type.CHAR: "TEXT",
 813            exp.DataType.Type.DATETIME: "TIMESTAMP",
 814            exp.DataType.Type.FLOAT: "REAL",
 815            exp.DataType.Type.JSONB: "JSON",
 816            exp.DataType.Type.NCHAR: "TEXT",
 817            exp.DataType.Type.NVARCHAR: "TEXT",
 818            exp.DataType.Type.UINT: "UINTEGER",
 819            exp.DataType.Type.VARBINARY: "BLOB",
 820            exp.DataType.Type.ROWVERSION: "BLOB",
 821            exp.DataType.Type.VARCHAR: "TEXT",
 822            exp.DataType.Type.TIMESTAMPNTZ: "TIMESTAMP",
 823            exp.DataType.Type.TIMESTAMP_S: "TIMESTAMP_S",
 824            exp.DataType.Type.TIMESTAMP_MS: "TIMESTAMP_MS",
 825            exp.DataType.Type.TIMESTAMP_NS: "TIMESTAMP_NS",
 826        }
 827
 828        # https://github.com/duckdb/duckdb/blob/ff7f24fd8e3128d94371827523dae85ebaf58713/third_party/libpg_query/grammar/keywords/reserved_keywords.list#L1-L77
 829        RESERVED_KEYWORDS = {
 830            "array",
 831            "analyse",
 832            "union",
 833            "all",
 834            "when",
 835            "in_p",
 836            "default",
 837            "create_p",
 838            "window",
 839            "asymmetric",
 840            "to",
 841            "else",
 842            "localtime",
 843            "from",
 844            "end_p",
 845            "select",
 846            "current_date",
 847            "foreign",
 848            "with",
 849            "grant",
 850            "session_user",
 851            "or",
 852            "except",
 853            "references",
 854            "fetch",
 855            "limit",
 856            "group_p",
 857            "leading",
 858            "into",
 859            "collate",
 860            "offset",
 861            "do",
 862            "then",
 863            "localtimestamp",
 864            "check_p",
 865            "lateral_p",
 866            "current_role",
 867            "where",
 868            "asc_p",
 869            "placing",
 870            "desc_p",
 871            "user",
 872            "unique",
 873            "initially",
 874            "column",
 875            "both",
 876            "some",
 877            "as",
 878            "any",
 879            "only",
 880            "deferrable",
 881            "null_p",
 882            "current_time",
 883            "true_p",
 884            "table",
 885            "case",
 886            "trailing",
 887            "variadic",
 888            "for",
 889            "on",
 890            "distinct",
 891            "false_p",
 892            "not",
 893            "constraint",
 894            "current_timestamp",
 895            "returning",
 896            "primary",
 897            "intersect",
 898            "having",
 899            "analyze",
 900            "current_user",
 901            "and",
 902            "cast",
 903            "symmetric",
 904            "using",
 905            "order",
 906            "current_catalog",
 907        }
 908
 909        UNWRAPPED_INTERVAL_VALUES = (exp.Literal, exp.Paren)
 910
 911        # DuckDB doesn't generally support CREATE TABLE .. properties
 912        # https://duckdb.org/docs/sql/statements/create_table.html
 913        PROPERTIES_LOCATION = {
 914            prop: exp.Properties.Location.UNSUPPORTED
 915            for prop in generator.Generator.PROPERTIES_LOCATION
 916        }
 917
 918        # There are a few exceptions (e.g. temporary tables) which are supported or
 919        # can be transpiled to DuckDB, so we explicitly override them accordingly
 920        PROPERTIES_LOCATION[exp.LikeProperty] = exp.Properties.Location.POST_SCHEMA
 921        PROPERTIES_LOCATION[exp.TemporaryProperty] = exp.Properties.Location.POST_CREATE
 922        PROPERTIES_LOCATION[exp.ReturnsProperty] = exp.Properties.Location.POST_ALIAS
 923
 924        IGNORE_RESPECT_NULLS_WINDOW_FUNCTIONS = (
 925            exp.FirstValue,
 926            exp.Lag,
 927            exp.LastValue,
 928            exp.Lead,
 929            exp.NthValue,
 930        )
 931
 932        def lambda_sql(
 933            self, expression: exp.Lambda, arrow_sep: str = "->", wrap: bool = True
 934        ) -> str:
 935            if expression.args.get("colon"):
 936                prefix = "LAMBDA "
 937                arrow_sep = ":"
 938                wrap = False
 939            else:
 940                prefix = ""
 941
 942            lambda_sql = super().lambda_sql(expression, arrow_sep=arrow_sep, wrap=wrap)
 943            return f"{prefix}{lambda_sql}"
 944
 945        def show_sql(self, expression: exp.Show) -> str:
 946            return f"SHOW {expression.name}"
 947
 948        def fromiso8601timestamp_sql(self, expression: exp.FromISO8601Timestamp) -> str:
 949            return self.sql(exp.cast(expression.this, exp.DataType.Type.TIMESTAMPTZ))
 950
 951        def strtotime_sql(self, expression: exp.StrToTime) -> str:
 952            if expression.args.get("safe"):
 953                formatted_time = self.format_time(expression)
 954                return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS TIMESTAMP)"
 955            return str_to_time_sql(self, expression)
 956
 957        def strtodate_sql(self, expression: exp.StrToDate) -> str:
 958            if expression.args.get("safe"):
 959                formatted_time = self.format_time(expression)
 960                return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS DATE)"
 961            return f"CAST({str_to_time_sql(self, expression)} AS DATE)"
 962
 963        def parsejson_sql(self, expression: exp.ParseJSON) -> str:
 964            arg = expression.this
 965            if expression.args.get("safe"):
 966                return self.sql(exp.case().when(exp.func("json_valid", arg), arg).else_(exp.null()))
 967            return self.func("JSON", arg)
 968
 969        def timefromparts_sql(self, expression: exp.TimeFromParts) -> str:
 970            nano = expression.args.get("nano")
 971            if nano is not None:
 972                expression.set(
 973                    "sec", expression.args["sec"] + nano.pop() / exp.Literal.number(1000000000.0)
 974                )
 975
 976            return rename_func("MAKE_TIME")(self, expression)
 977
 978        def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str:
 979            sec = expression.args["sec"]
 980
 981            milli = expression.args.get("milli")
 982            if milli is not None:
 983                sec += milli.pop() / exp.Literal.number(1000.0)
 984
 985            nano = expression.args.get("nano")
 986            if nano is not None:
 987                sec += nano.pop() / exp.Literal.number(1000000000.0)
 988
 989            if milli or nano:
 990                expression.set("sec", sec)
 991
 992            return rename_func("MAKE_TIMESTAMP")(self, expression)
 993
 994        def tablesample_sql(
 995            self,
 996            expression: exp.TableSample,
 997            tablesample_keyword: t.Optional[str] = None,
 998        ) -> str:
 999            if not isinstance(expression.parent, exp.Select):
1000                # This sample clause only applies to a single source, not the entire resulting relation
1001                tablesample_keyword = "TABLESAMPLE"
1002
1003            if expression.args.get("size"):
1004                method = expression.args.get("method")
1005                if method and method.name.upper() != "RESERVOIR":
1006                    self.unsupported(
1007                        f"Sampling method {method} is not supported with a discrete sample count, "
1008                        "defaulting to reservoir sampling"
1009                    )
1010                    expression.set("method", exp.var("RESERVOIR"))
1011
1012            return super().tablesample_sql(expression, tablesample_keyword=tablesample_keyword)
1013
1014        def columndef_sql(self, expression: exp.ColumnDef, sep: str = " ") -> str:
1015            if isinstance(expression.parent, exp.UserDefinedFunction):
1016                return self.sql(expression, "this")
1017            return super().columndef_sql(expression, sep)
1018
1019        def join_sql(self, expression: exp.Join) -> str:
1020            if (
1021                expression.side == "LEFT"
1022                and not expression.args.get("on")
1023                and isinstance(expression.this, exp.Unnest)
1024            ):
1025                # Some dialects support `LEFT JOIN UNNEST(...)` without an explicit ON clause
1026                # DuckDB doesn't, but we can just add a dummy ON clause that is always true
1027                return super().join_sql(expression.on(exp.true()))
1028
1029            return super().join_sql(expression)
1030
1031        def generateseries_sql(self, expression: exp.GenerateSeries) -> str:
1032            # GENERATE_SERIES(a, b) -> [a, b], RANGE(a, b) -> [a, b)
1033            if expression.args.get("is_end_exclusive"):
1034                return rename_func("RANGE")(self, expression)
1035
1036            return self.function_fallback_sql(expression)
1037
1038        def countif_sql(self, expression: exp.CountIf) -> str:
1039            if self.dialect.version >= Version("1.2"):
1040                return self.function_fallback_sql(expression)
1041
1042            # https://github.com/tobymao/sqlglot/pull/4749
1043            return count_if_to_sum(self, expression)
1044
1045        def bracket_sql(self, expression: exp.Bracket) -> str:
1046            if self.dialect.version >= Version("1.2"):
1047                return super().bracket_sql(expression)
1048
1049            # https://duckdb.org/2025/02/05/announcing-duckdb-120.html#breaking-changes
1050            this = expression.this
1051            if isinstance(this, exp.Array):
1052                this.replace(exp.paren(this))
1053
1054            bracket = super().bracket_sql(expression)
1055
1056            if not expression.args.get("returns_list_for_maps"):
1057                if not this.type:
1058                    from sqlglot.optimizer.annotate_types import annotate_types
1059
1060                    this = annotate_types(this, dialect=self.dialect)
1061
1062                if this.is_type(exp.DataType.Type.MAP):
1063                    bracket = f"({bracket})[1]"
1064
1065            return bracket
1066
1067        def withingroup_sql(self, expression: exp.WithinGroup) -> str:
1068            expression_sql = self.sql(expression, "expression")
1069
1070            func = expression.this
1071            if isinstance(func, exp.PERCENTILES):
1072                # Make the order key the first arg and slide the fraction to the right
1073                # https://duckdb.org/docs/sql/aggregates#ordered-set-aggregate-functions
1074                order_col = expression.find(exp.Ordered)
1075                if order_col:
1076                    func.set("expression", func.this)
1077                    func.set("this", order_col.this)
1078
1079            this = self.sql(expression, "this").rstrip(")")
1080
1081            return f"{this}{expression_sql})"
1082
1083        def length_sql(self, expression: exp.Length) -> str:
1084            arg = expression.this
1085
1086            # Dialects like BQ and Snowflake also accept binary values as args, so
1087            # DDB will attempt to infer the type or resort to case/when resolution
1088            if not expression.args.get("binary") or arg.is_string:
1089                return self.func("LENGTH", arg)
1090
1091            if not arg.type:
1092                from sqlglot.optimizer.annotate_types import annotate_types
1093
1094                arg = annotate_types(arg, dialect=self.dialect)
1095
1096            if arg.is_type(*exp.DataType.TEXT_TYPES):
1097                return self.func("LENGTH", arg)
1098
1099            # We need these casts to make duckdb's static type checker happy
1100            blob = exp.cast(arg, exp.DataType.Type.VARBINARY)
1101            varchar = exp.cast(arg, exp.DataType.Type.VARCHAR)
1102
1103            case = (
1104                exp.case(self.func("TYPEOF", arg))
1105                .when("'BLOB'", self.func("OCTET_LENGTH", blob))
1106                .else_(
1107                    exp.Anonymous(this="LENGTH", expressions=[varchar])
1108                )  # anonymous to break length_sql recursion
1109            )
1110
1111            return self.sql(case)
1112
1113        def objectinsert_sql(self, expression: exp.ObjectInsert) -> str:
1114            this = expression.this
1115            key = expression.args.get("key")
1116            key_sql = key.name if isinstance(key, exp.Expression) else ""
1117            value_sql = self.sql(expression, "value")
1118
1119            kv_sql = f"{key_sql} := {value_sql}"
1120
1121            # If the input struct is empty e.g. transpiling OBJECT_INSERT(OBJECT_CONSTRUCT(), key, value) from Snowflake
1122            # then we can generate STRUCT_PACK which will build it since STRUCT_INSERT({}, key := value) is not valid DuckDB
1123            if isinstance(this, exp.Struct) and not this.expressions:
1124                return self.func("STRUCT_PACK", kv_sql)
1125
1126            return self.func("STRUCT_INSERT", this, kv_sql)
1127
1128        def unnest_sql(self, expression: exp.Unnest) -> str:
1129            explode_array = expression.args.get("explode_array")
1130            if explode_array:
1131                # In BigQuery, UNNESTing a nested array leads to explosion of the top-level array & struct
1132                # This is transpiled to DDB by transforming "FROM UNNEST(...)" to "FROM (SELECT UNNEST(..., max_depth => 2))"
1133                expression.expressions.append(
1134                    exp.Kwarg(this=exp.var("max_depth"), expression=exp.Literal.number(2))
1135                )
1136
1137                # If BQ's UNNEST is aliased, we transform it from a column alias to a table alias in DDB
1138                alias = expression.args.get("alias")
1139                if alias:
1140                    expression.set("alias", None)
1141                    alias = exp.TableAlias(this=seq_get(alias.args.get("columns"), 0))
1142
1143                unnest_sql = super().unnest_sql(expression)
1144                select = exp.Select(expressions=[unnest_sql]).subquery(alias)
1145                return self.sql(select)
1146
1147            return super().unnest_sql(expression)
1148
1149        def ignorenulls_sql(self, expression: exp.IgnoreNulls) -> str:
1150            if isinstance(expression.this, self.IGNORE_RESPECT_NULLS_WINDOW_FUNCTIONS):
1151                # DuckDB should render IGNORE NULLS only for the general-purpose
1152                # window functions that accept it e.g. FIRST_VALUE(... IGNORE NULLS) OVER (...)
1153                return super().ignorenulls_sql(expression)
1154
1155            self.unsupported("IGNORE NULLS is not supported for non-window functions.")
1156            return self.sql(expression, "this")
1157
1158        def respectnulls_sql(self, expression: exp.RespectNulls) -> str:
1159            if isinstance(expression.this, self.IGNORE_RESPECT_NULLS_WINDOW_FUNCTIONS):
1160                # DuckDB should render RESPECT NULLS only for the general-purpose
1161                # window functions that accept it e.g. FIRST_VALUE(... RESPECT NULLS) OVER (...)
1162                return super().respectnulls_sql(expression)
1163
1164            self.unsupported("RESPECT NULLS is not supported for non-window functions.")
1165            return self.sql(expression, "this")
1166
1167        def arraytostring_sql(self, expression: exp.ArrayToString) -> str:
1168            this = self.sql(expression, "this")
1169            null_text = self.sql(expression, "null")
1170
1171            if null_text:
1172                this = f"LIST_TRANSFORM({this}, x -> COALESCE(x, {null_text}))"
1173
1174            return self.func("ARRAY_TO_STRING", this, expression.expression)
1175
1176        @unsupported_args("position", "occurrence")
1177        def regexpextract_sql(self, expression: exp.RegexpExtract) -> str:
1178            group = expression.args.get("group")
1179            params = expression.args.get("parameters")
1180
1181            # Do not render group if there is no following argument,
1182            # and it's the default value for this dialect
1183            if (
1184                not params
1185                and group
1186                and group.name == str(self.dialect.REGEXP_EXTRACT_DEFAULT_GROUP)
1187            ):
1188                group = None
1189            return self.func(
1190                "REGEXP_EXTRACT", expression.this, expression.expression, group, params
1191            )
1192
1193        @unsupported_args("culture")
1194        def numbertostr_sql(self, expression: exp.NumberToStr) -> str:
1195            fmt = expression.args.get("format")
1196            if fmt and fmt.is_int:
1197                return self.func("FORMAT", f"'{{:,.{fmt.name}f}}'", expression.this)
1198
1199            self.unsupported("Only integer formats are supported by NumberToStr")
1200            return self.function_fallback_sql(expression)
1201
1202        def autoincrementcolumnconstraint_sql(self, _) -> str:
1203            self.unsupported("The AUTOINCREMENT column constraint is not supported by DuckDB")
1204            return ""
1205
1206        def aliases_sql(self, expression: exp.Aliases) -> str:
1207            this = expression.this
1208            if isinstance(this, exp.Posexplode):
1209                return self.posexplode_sql(this)
1210
1211            return super().aliases_sql(expression)
1212
1213        def posexplode_sql(self, expression: exp.Posexplode) -> str:
1214            this = expression.this
1215            parent = expression.parent
1216
1217            # The default Spark aliases are "pos" and "col", unless specified otherwise
1218            pos, col = exp.to_identifier("pos"), exp.to_identifier("col")
1219
1220            if isinstance(parent, exp.Aliases):
1221                # Column case: SELECT POSEXPLODE(col) [AS (a, b)]
1222                pos, col = parent.expressions
1223            elif isinstance(parent, exp.Table):
1224                # Table case: SELECT * FROM POSEXPLODE(col) [AS (a, b)]
1225                alias = parent.args.get("alias")
1226                if alias:
1227                    pos, col = alias.columns or [pos, col]
1228                    alias.pop()
1229
1230            # Translate POSEXPLODE to UNNEST + GENERATE_SUBSCRIPTS
1231            # Note: In Spark pos is 0-indexed, but in DuckDB it's 1-indexed, so we subtract 1 from GENERATE_SUBSCRIPTS
1232            unnest_sql = self.sql(exp.Unnest(expressions=[this], alias=col))
1233            gen_subscripts = self.sql(
1234                exp.Alias(
1235                    this=exp.Anonymous(
1236                        this="GENERATE_SUBSCRIPTS", expressions=[this, exp.Literal.number(1)]
1237                    )
1238                    - exp.Literal.number(1),
1239                    alias=pos,
1240                )
1241            )
1242
1243            posexplode_sql = self.format_args(gen_subscripts, unnest_sql)
1244
1245            if isinstance(parent, exp.From) or (parent and isinstance(parent.parent, exp.From)):
1246                # SELECT * FROM POSEXPLODE(col) -> SELECT * FROM (SELECT GENERATE_SUBSCRIPTS(...), UNNEST(...))
1247                return self.sql(exp.Subquery(this=exp.Select(expressions=[posexplode_sql])))
1248
1249            return posexplode_sql