sqlglot.dialects.snowflake

   1from __future__ import annotations
   2
   3import typing as t
   4
   5from sqlglot import exp, generator, parser, tokens, transforms
   6from sqlglot.dialects.dialect import (
   7    Dialect,
   8    NormalizationStrategy,
   9    binary_from_function,
  10    build_default_decimal_type,
  11    build_timestamp_from_parts,
  12    date_delta_sql,
  13    date_trunc_to_time,
  14    datestrtodate_sql,
  15    build_formatted_time,
  16    if_sql,
  17    inline_array_sql,
  18    max_or_greatest,
  19    min_or_least,
  20    rename_func,
  21    timestamptrunc_sql,
  22    timestrtotime_sql,
  23    var_map_sql,
  24    map_date_part,
  25    no_safe_divide_sql,
  26    no_timestamp_sql,
  27    timestampdiff_sql,
  28    no_make_interval_sql,
  29)
  30from sqlglot.generator import unsupported_args
  31from sqlglot.helper import flatten, is_float, is_int, seq_get
  32from sqlglot.tokens import TokenType
  33
  34if t.TYPE_CHECKING:
  35    from sqlglot._typing import E
  36
  37
  38# from https://docs.snowflake.com/en/sql-reference/functions/to_timestamp.html
  39def _build_datetime(
  40    name: str, kind: exp.DataType.Type, safe: bool = False
  41) -> t.Callable[[t.List], exp.Func]:
  42    def _builder(args: t.List) -> exp.Func:
  43        value = seq_get(args, 0)
  44        int_value = value is not None and is_int(value.name)
  45
  46        if isinstance(value, exp.Literal):
  47            # Converts calls like `TO_TIME('01:02:03')` into casts
  48            if len(args) == 1 and value.is_string and not int_value:
  49                return (
  50                    exp.TryCast(this=value, to=exp.DataType.build(kind))
  51                    if safe
  52                    else exp.cast(value, kind)
  53                )
  54
  55            # Handles `TO_TIMESTAMP(str, fmt)` and `TO_TIMESTAMP(num, scale)` as special
  56            # cases so we can transpile them, since they're relatively common
  57            if kind == exp.DataType.Type.TIMESTAMP:
  58                if int_value and not safe:
  59                    # TRY_TO_TIMESTAMP('integer') is not parsed into exp.UnixToTime as
  60                    # it's not easily transpilable
  61                    return exp.UnixToTime(this=value, scale=seq_get(args, 1))
  62                if not is_float(value.this):
  63                    expr = build_formatted_time(exp.StrToTime, "snowflake")(args)
  64                    expr.set("safe", safe)
  65                    return expr
  66
  67        if kind == exp.DataType.Type.DATE and not int_value:
  68            formatted_exp = build_formatted_time(exp.TsOrDsToDate, "snowflake")(args)
  69            formatted_exp.set("safe", safe)
  70            return formatted_exp
  71
  72        return exp.Anonymous(this=name, expressions=args)
  73
  74    return _builder
  75
  76
  77def _build_object_construct(args: t.List) -> t.Union[exp.StarMap, exp.Struct]:
  78    expression = parser.build_var_map(args)
  79
  80    if isinstance(expression, exp.StarMap):
  81        return expression
  82
  83    return exp.Struct(
  84        expressions=[
  85            exp.PropertyEQ(this=k, expression=v) for k, v in zip(expression.keys, expression.values)
  86        ]
  87    )
  88
  89
  90def _build_datediff(args: t.List) -> exp.DateDiff:
  91    return exp.DateDiff(
  92        this=seq_get(args, 2), expression=seq_get(args, 1), unit=map_date_part(seq_get(args, 0))
  93    )
  94
  95
  96def _build_date_time_add(expr_type: t.Type[E]) -> t.Callable[[t.List], E]:
  97    def _builder(args: t.List) -> E:
  98        return expr_type(
  99            this=seq_get(args, 2),
 100            expression=seq_get(args, 1),
 101            unit=map_date_part(seq_get(args, 0)),
 102        )
 103
 104    return _builder
 105
 106
 107# https://docs.snowflake.com/en/sql-reference/functions/div0
 108def _build_if_from_div0(args: t.List) -> exp.If:
 109    lhs = exp._wrap(seq_get(args, 0), exp.Binary)
 110    rhs = exp._wrap(seq_get(args, 1), exp.Binary)
 111
 112    cond = exp.EQ(this=rhs, expression=exp.Literal.number(0)).and_(
 113        exp.Is(this=lhs, expression=exp.null()).not_()
 114    )
 115    true = exp.Literal.number(0)
 116    false = exp.Div(this=lhs, expression=rhs)
 117    return exp.If(this=cond, true=true, false=false)
 118
 119
 120# https://docs.snowflake.com/en/sql-reference/functions/zeroifnull
 121def _build_if_from_zeroifnull(args: t.List) -> exp.If:
 122    cond = exp.Is(this=seq_get(args, 0), expression=exp.Null())
 123    return exp.If(this=cond, true=exp.Literal.number(0), false=seq_get(args, 0))
 124
 125
 126# https://docs.snowflake.com/en/sql-reference/functions/zeroifnull
 127def _build_if_from_nullifzero(args: t.List) -> exp.If:
 128    cond = exp.EQ(this=seq_get(args, 0), expression=exp.Literal.number(0))
 129    return exp.If(this=cond, true=exp.Null(), false=seq_get(args, 0))
 130
 131
 132def _regexpilike_sql(self: Snowflake.Generator, expression: exp.RegexpILike) -> str:
 133    flag = expression.text("flag")
 134
 135    if "i" not in flag:
 136        flag += "i"
 137
 138    return self.func(
 139        "REGEXP_LIKE", expression.this, expression.expression, exp.Literal.string(flag)
 140    )
 141
 142
 143def _build_regexp_replace(args: t.List) -> exp.RegexpReplace:
 144    regexp_replace = exp.RegexpReplace.from_arg_list(args)
 145
 146    if not regexp_replace.args.get("replacement"):
 147        regexp_replace.set("replacement", exp.Literal.string(""))
 148
 149    return regexp_replace
 150
 151
 152def _show_parser(*args: t.Any, **kwargs: t.Any) -> t.Callable[[Snowflake.Parser], exp.Show]:
 153    def _parse(self: Snowflake.Parser) -> exp.Show:
 154        return self._parse_show_snowflake(*args, **kwargs)
 155
 156    return _parse
 157
 158
 159def _date_trunc_to_time(args: t.List) -> exp.DateTrunc | exp.TimestampTrunc:
 160    trunc = date_trunc_to_time(args)
 161    trunc.set("unit", map_date_part(trunc.args["unit"]))
 162    return trunc
 163
 164
 165def _unqualify_unpivot_columns(expression: exp.Expression) -> exp.Expression:
 166    """
 167    Snowflake doesn't allow columns referenced in UNPIVOT to be qualified,
 168    so we need to unqualify them.
 169
 170    Example:
 171        >>> from sqlglot import parse_one
 172        >>> expr = parse_one("SELECT * FROM m_sales UNPIVOT(sales FOR month IN (m_sales.jan, feb, mar, april))")
 173        >>> print(_unqualify_unpivot_columns(expr).sql(dialect="snowflake"))
 174        SELECT * FROM m_sales UNPIVOT(sales FOR month IN (jan, feb, mar, april))
 175    """
 176    if isinstance(expression, exp.Pivot) and expression.unpivot:
 177        expression = transforms.unqualify_columns(expression)
 178
 179    return expression
 180
 181
 182def _flatten_structured_types_unless_iceberg(expression: exp.Expression) -> exp.Expression:
 183    assert isinstance(expression, exp.Create)
 184
 185    def _flatten_structured_type(expression: exp.DataType) -> exp.DataType:
 186        if expression.this in exp.DataType.NESTED_TYPES:
 187            expression.set("expressions", None)
 188        return expression
 189
 190    props = expression.args.get("properties")
 191    if isinstance(expression.this, exp.Schema) and not (props and props.find(exp.IcebergProperty)):
 192        for schema_expression in expression.this.expressions:
 193            if isinstance(schema_expression, exp.ColumnDef):
 194                column_type = schema_expression.kind
 195                if isinstance(column_type, exp.DataType):
 196                    column_type.transform(_flatten_structured_type, copy=False)
 197
 198    return expression
 199
 200
 201def _unnest_generate_date_array(unnest: exp.Unnest) -> None:
 202    generate_date_array = unnest.expressions[0]
 203    start = generate_date_array.args.get("start")
 204    end = generate_date_array.args.get("end")
 205    step = generate_date_array.args.get("step")
 206
 207    if not start or not end or not isinstance(step, exp.Interval) or step.name != "1":
 208        return
 209
 210    unit = step.args.get("unit")
 211
 212    unnest_alias = unnest.args.get("alias")
 213    if unnest_alias:
 214        unnest_alias = unnest_alias.copy()
 215        sequence_value_name = seq_get(unnest_alias.columns, 0) or "value"
 216    else:
 217        sequence_value_name = "value"
 218
 219    # We'll add the next sequence value to the starting date and project the result
 220    date_add = _build_date_time_add(exp.DateAdd)(
 221        [unit, exp.cast(sequence_value_name, "int"), exp.cast(start, "date")]
 222    ).as_(sequence_value_name)
 223
 224    # We use DATEDIFF to compute the number of sequence values needed
 225    number_sequence = Snowflake.Parser.FUNCTIONS["ARRAY_GENERATE_RANGE"](
 226        [exp.Literal.number(0), _build_datediff([unit, start, end]) + 1]
 227    )
 228
 229    unnest.set("expressions", [number_sequence])
 230    unnest.replace(exp.select(date_add).from_(unnest.copy()).subquery(unnest_alias))
 231
 232
 233def _transform_generate_date_array(expression: exp.Expression) -> exp.Expression:
 234    if isinstance(expression, exp.Select):
 235        for generate_date_array in expression.find_all(exp.GenerateDateArray):
 236            parent = generate_date_array.parent
 237
 238            # If GENERATE_DATE_ARRAY is used directly as an array (e.g passed into ARRAY_LENGTH), the transformed Snowflake
 239            # query is the following (it'll be unnested properly on the next iteration due to copy):
 240            # SELECT ref(GENERATE_DATE_ARRAY(...)) -> SELECT ref((SELECT ARRAY_AGG(*) FROM UNNEST(GENERATE_DATE_ARRAY(...))))
 241            if not isinstance(parent, exp.Unnest):
 242                unnest = exp.Unnest(expressions=[generate_date_array.copy()])
 243                generate_date_array.replace(
 244                    exp.select(exp.ArrayAgg(this=exp.Star())).from_(unnest).subquery()
 245                )
 246
 247            if (
 248                isinstance(parent, exp.Unnest)
 249                and isinstance(parent.parent, (exp.From, exp.Join))
 250                and len(parent.expressions) == 1
 251            ):
 252                _unnest_generate_date_array(parent)
 253
 254    return expression
 255
 256
 257def _build_regexp_extract(expr_type: t.Type[E]) -> t.Callable[[t.List], E]:
 258    def _builder(args: t.List) -> E:
 259        return expr_type(
 260            this=seq_get(args, 0),
 261            expression=seq_get(args, 1),
 262            position=seq_get(args, 2),
 263            occurrence=seq_get(args, 3),
 264            parameters=seq_get(args, 4),
 265            group=seq_get(args, 5) or exp.Literal.number(0),
 266        )
 267
 268    return _builder
 269
 270
 271def _regexpextract_sql(self, expression: exp.RegexpExtract | exp.RegexpExtractAll) -> str:
 272    # Other dialects don't support all of the following parameters, so we need to
 273    # generate default values as necessary to ensure the transpilation is correct
 274    group = expression.args.get("group")
 275
 276    # To avoid generating all these default values, we set group to None if
 277    # it's 0 (also default value) which doesn't trigger the following chain
 278    if group and group.name == "0":
 279        group = None
 280
 281    parameters = expression.args.get("parameters") or (group and exp.Literal.string("c"))
 282    occurrence = expression.args.get("occurrence") or (parameters and exp.Literal.number(1))
 283    position = expression.args.get("position") or (occurrence and exp.Literal.number(1))
 284
 285    return self.func(
 286        "REGEXP_SUBSTR" if isinstance(expression, exp.RegexpExtract) else "REGEXP_EXTRACT_ALL",
 287        expression.this,
 288        expression.expression,
 289        position,
 290        occurrence,
 291        parameters,
 292        group,
 293    )
 294
 295
 296def _json_extract_value_array_sql(
 297    self: Snowflake.Generator, expression: exp.JSONValueArray | exp.JSONExtractArray
 298) -> str:
 299    json_extract = exp.JSONExtract(this=expression.this, expression=expression.expression)
 300    ident = exp.to_identifier("x")
 301
 302    if isinstance(expression, exp.JSONValueArray):
 303        this: exp.Expression = exp.cast(ident, to=exp.DataType.Type.VARCHAR)
 304    else:
 305        this = exp.ParseJSON(this=f"TO_JSON({ident})")
 306
 307    transform_lambda = exp.Lambda(expressions=[ident], this=this)
 308
 309    return self.func("TRANSFORM", json_extract, transform_lambda)
 310
 311
 312class Snowflake(Dialect):
 313    # https://docs.snowflake.com/en/sql-reference/identifiers-syntax
 314    NORMALIZATION_STRATEGY = NormalizationStrategy.UPPERCASE
 315    NULL_ORDERING = "nulls_are_large"
 316    TIME_FORMAT = "'YYYY-MM-DD HH24:MI:SS'"
 317    SUPPORTS_USER_DEFINED_TYPES = False
 318    SUPPORTS_SEMI_ANTI_JOIN = False
 319    PREFER_CTE_ALIAS_COLUMN = True
 320    TABLESAMPLE_SIZE_IS_PERCENT = True
 321    COPY_PARAMS_ARE_CSV = False
 322    ARRAY_AGG_INCLUDES_NULLS = None
 323
 324    TIME_MAPPING = {
 325        "YYYY": "%Y",
 326        "yyyy": "%Y",
 327        "YY": "%y",
 328        "yy": "%y",
 329        "MMMM": "%B",
 330        "mmmm": "%B",
 331        "MON": "%b",
 332        "mon": "%b",
 333        "MM": "%m",
 334        "mm": "%m",
 335        "DD": "%d",
 336        "dd": "%-d",
 337        "DY": "%a",
 338        "dy": "%w",
 339        "HH24": "%H",
 340        "hh24": "%H",
 341        "HH12": "%I",
 342        "hh12": "%I",
 343        "MI": "%M",
 344        "mi": "%M",
 345        "SS": "%S",
 346        "ss": "%S",
 347        "FF": "%f",
 348        "ff": "%f",
 349        "FF6": "%f",
 350        "ff6": "%f",
 351    }
 352
 353    def quote_identifier(self, expression: E, identify: bool = True) -> E:
 354        # This disables quoting DUAL in SELECT ... FROM DUAL, because Snowflake treats an
 355        # unquoted DUAL keyword in a special way and does not map it to a user-defined table
 356        if (
 357            isinstance(expression, exp.Identifier)
 358            and isinstance(expression.parent, exp.Table)
 359            and expression.name.lower() == "dual"
 360        ):
 361            return expression  # type: ignore
 362
 363        return super().quote_identifier(expression, identify=identify)
 364
 365    class Parser(parser.Parser):
 366        IDENTIFY_PIVOT_STRINGS = True
 367        DEFAULT_SAMPLING_METHOD = "BERNOULLI"
 368        COLON_IS_VARIANT_EXTRACT = True
 369
 370        ID_VAR_TOKENS = {
 371            *parser.Parser.ID_VAR_TOKENS,
 372            TokenType.MATCH_CONDITION,
 373        }
 374
 375        TABLE_ALIAS_TOKENS = parser.Parser.TABLE_ALIAS_TOKENS | {TokenType.WINDOW}
 376        TABLE_ALIAS_TOKENS.discard(TokenType.MATCH_CONDITION)
 377
 378        FUNCTIONS = {
 379            **parser.Parser.FUNCTIONS,
 380            "APPROX_PERCENTILE": exp.ApproxQuantile.from_arg_list,
 381            "ARRAY_CONSTRUCT": lambda args: exp.Array(expressions=args),
 382            "ARRAY_CONTAINS": lambda args: exp.ArrayContains(
 383                this=seq_get(args, 1), expression=seq_get(args, 0)
 384            ),
 385            "ARRAY_GENERATE_RANGE": lambda args: exp.GenerateSeries(
 386                # ARRAY_GENERATE_RANGE has an exlusive end; we normalize it to be inclusive
 387                start=seq_get(args, 0),
 388                end=exp.Sub(this=seq_get(args, 1), expression=exp.Literal.number(1)),
 389                step=seq_get(args, 2),
 390            ),
 391            "BITXOR": binary_from_function(exp.BitwiseXor),
 392            "BIT_XOR": binary_from_function(exp.BitwiseXor),
 393            "BOOLXOR": binary_from_function(exp.Xor),
 394            "DATE": _build_datetime("DATE", exp.DataType.Type.DATE),
 395            "DATE_TRUNC": _date_trunc_to_time,
 396            "DATEADD": _build_date_time_add(exp.DateAdd),
 397            "DATEDIFF": _build_datediff,
 398            "DIV0": _build_if_from_div0,
 399            "EDITDISTANCE": lambda args: exp.Levenshtein(
 400                this=seq_get(args, 0), expression=seq_get(args, 1), max_dist=seq_get(args, 2)
 401            ),
 402            "FLATTEN": exp.Explode.from_arg_list,
 403            "GET_PATH": lambda args, dialect: exp.JSONExtract(
 404                this=seq_get(args, 0), expression=dialect.to_json_path(seq_get(args, 1))
 405            ),
 406            "IFF": exp.If.from_arg_list,
 407            "LAST_DAY": lambda args: exp.LastDay(
 408                this=seq_get(args, 0), unit=map_date_part(seq_get(args, 1))
 409            ),
 410            "LEN": lambda args: exp.Length(this=seq_get(args, 0), binary=True),
 411            "LENGTH": lambda args: exp.Length(this=seq_get(args, 0), binary=True),
 412            "LISTAGG": exp.GroupConcat.from_arg_list,
 413            "NULLIFZERO": _build_if_from_nullifzero,
 414            "OBJECT_CONSTRUCT": _build_object_construct,
 415            "REGEXP_EXTRACT_ALL": _build_regexp_extract(exp.RegexpExtractAll),
 416            "REGEXP_REPLACE": _build_regexp_replace,
 417            "REGEXP_SUBSTR": _build_regexp_extract(exp.RegexpExtract),
 418            "REGEXP_SUBSTR_ALL": _build_regexp_extract(exp.RegexpExtractAll),
 419            "RLIKE": exp.RegexpLike.from_arg_list,
 420            "SQUARE": lambda args: exp.Pow(this=seq_get(args, 0), expression=exp.Literal.number(2)),
 421            "TIMEADD": _build_date_time_add(exp.TimeAdd),
 422            "TIMEDIFF": _build_datediff,
 423            "TIMESTAMPADD": _build_date_time_add(exp.DateAdd),
 424            "TIMESTAMPDIFF": _build_datediff,
 425            "TIMESTAMPFROMPARTS": build_timestamp_from_parts,
 426            "TIMESTAMP_FROM_PARTS": build_timestamp_from_parts,
 427            "TIMESTAMPNTZFROMPARTS": build_timestamp_from_parts,
 428            "TIMESTAMP_NTZ_FROM_PARTS": build_timestamp_from_parts,
 429            "TRY_PARSE_JSON": lambda args: exp.ParseJSON(this=seq_get(args, 0), safe=True),
 430            "TRY_TO_DATE": _build_datetime("TRY_TO_DATE", exp.DataType.Type.DATE, safe=True),
 431            "TRY_TO_TIMESTAMP": _build_datetime(
 432                "TRY_TO_TIMESTAMP", exp.DataType.Type.TIMESTAMP, safe=True
 433            ),
 434            "TO_DATE": _build_datetime("TO_DATE", exp.DataType.Type.DATE),
 435            "TO_NUMBER": lambda args: exp.ToNumber(
 436                this=seq_get(args, 0),
 437                format=seq_get(args, 1),
 438                precision=seq_get(args, 2),
 439                scale=seq_get(args, 3),
 440            ),
 441            "TO_TIME": _build_datetime("TO_TIME", exp.DataType.Type.TIME),
 442            "TO_TIMESTAMP": _build_datetime("TO_TIMESTAMP", exp.DataType.Type.TIMESTAMP),
 443            "TO_TIMESTAMP_LTZ": _build_datetime("TO_TIMESTAMP_LTZ", exp.DataType.Type.TIMESTAMPLTZ),
 444            "TO_TIMESTAMP_NTZ": _build_datetime("TO_TIMESTAMP_NTZ", exp.DataType.Type.TIMESTAMP),
 445            "TO_TIMESTAMP_TZ": _build_datetime("TO_TIMESTAMP_TZ", exp.DataType.Type.TIMESTAMPTZ),
 446            "TO_VARCHAR": exp.ToChar.from_arg_list,
 447            "ZEROIFNULL": _build_if_from_zeroifnull,
 448        }
 449
 450        FUNCTION_PARSERS = {
 451            **parser.Parser.FUNCTION_PARSERS,
 452            "DATE_PART": lambda self: self._parse_date_part(),
 453            "OBJECT_CONSTRUCT_KEEP_NULL": lambda self: self._parse_json_object(),
 454        }
 455        FUNCTION_PARSERS.pop("TRIM")
 456
 457        TIMESTAMPS = parser.Parser.TIMESTAMPS - {TokenType.TIME}
 458
 459        RANGE_PARSERS = {
 460            **parser.Parser.RANGE_PARSERS,
 461            TokenType.LIKE_ANY: parser.binary_range_parser(exp.LikeAny),
 462            TokenType.ILIKE_ANY: parser.binary_range_parser(exp.ILikeAny),
 463        }
 464
 465        ALTER_PARSERS = {
 466            **parser.Parser.ALTER_PARSERS,
 467            "UNSET": lambda self: self.expression(
 468                exp.Set,
 469                tag=self._match_text_seq("TAG"),
 470                expressions=self._parse_csv(self._parse_id_var),
 471                unset=True,
 472            ),
 473        }
 474
 475        STATEMENT_PARSERS = {
 476            **parser.Parser.STATEMENT_PARSERS,
 477            TokenType.SHOW: lambda self: self._parse_show(),
 478        }
 479
 480        PROPERTY_PARSERS = {
 481            **parser.Parser.PROPERTY_PARSERS,
 482            "LOCATION": lambda self: self._parse_location_property(),
 483            "TAG": lambda self: self._parse_tag(),
 484        }
 485
 486        TYPE_CONVERTERS = {
 487            # https://docs.snowflake.com/en/sql-reference/data-types-numeric#number
 488            exp.DataType.Type.DECIMAL: build_default_decimal_type(precision=38, scale=0),
 489        }
 490
 491        SHOW_PARSERS = {
 492            "SCHEMAS": _show_parser("SCHEMAS"),
 493            "TERSE SCHEMAS": _show_parser("SCHEMAS"),
 494            "OBJECTS": _show_parser("OBJECTS"),
 495            "TERSE OBJECTS": _show_parser("OBJECTS"),
 496            "TABLES": _show_parser("TABLES"),
 497            "TERSE TABLES": _show_parser("TABLES"),
 498            "VIEWS": _show_parser("VIEWS"),
 499            "TERSE VIEWS": _show_parser("VIEWS"),
 500            "PRIMARY KEYS": _show_parser("PRIMARY KEYS"),
 501            "TERSE PRIMARY KEYS": _show_parser("PRIMARY KEYS"),
 502            "IMPORTED KEYS": _show_parser("IMPORTED KEYS"),
 503            "TERSE IMPORTED KEYS": _show_parser("IMPORTED KEYS"),
 504            "UNIQUE KEYS": _show_parser("UNIQUE KEYS"),
 505            "TERSE UNIQUE KEYS": _show_parser("UNIQUE KEYS"),
 506            "SEQUENCES": _show_parser("SEQUENCES"),
 507            "TERSE SEQUENCES": _show_parser("SEQUENCES"),
 508            "COLUMNS": _show_parser("COLUMNS"),
 509            "USERS": _show_parser("USERS"),
 510            "TERSE USERS": _show_parser("USERS"),
 511        }
 512
 513        CONSTRAINT_PARSERS = {
 514            **parser.Parser.CONSTRAINT_PARSERS,
 515            "WITH": lambda self: self._parse_with_constraint(),
 516            "MASKING": lambda self: self._parse_with_constraint(),
 517            "PROJECTION": lambda self: self._parse_with_constraint(),
 518            "TAG": lambda self: self._parse_with_constraint(),
 519        }
 520
 521        STAGED_FILE_SINGLE_TOKENS = {
 522            TokenType.DOT,
 523            TokenType.MOD,
 524            TokenType.SLASH,
 525        }
 526
 527        FLATTEN_COLUMNS = ["SEQ", "KEY", "PATH", "INDEX", "VALUE", "THIS"]
 528
 529        SCHEMA_KINDS = {"OBJECTS", "TABLES", "VIEWS", "SEQUENCES", "UNIQUE KEYS", "IMPORTED KEYS"}
 530
 531        NON_TABLE_CREATABLES = {"STORAGE INTEGRATION", "TAG", "WAREHOUSE", "STREAMLIT"}
 532
 533        LAMBDAS = {
 534            **parser.Parser.LAMBDAS,
 535            TokenType.ARROW: lambda self, expressions: self.expression(
 536                exp.Lambda,
 537                this=self._replace_lambda(
 538                    self._parse_assignment(),
 539                    expressions,
 540                ),
 541                expressions=[e.this if isinstance(e, exp.Cast) else e for e in expressions],
 542            ),
 543        }
 544
 545        def _negate_range(
 546            self, this: t.Optional[exp.Expression] = None
 547        ) -> t.Optional[exp.Expression]:
 548            if not this:
 549                return this
 550
 551            query = this.args.get("query")
 552            if isinstance(this, exp.In) and isinstance(query, exp.Query):
 553                # Snowflake treats `value NOT IN (subquery)` as `VALUE <> ALL (subquery)`, so
 554                # we do this conversion here to avoid parsing it into `NOT value IN (subquery)`
 555                # which can produce different results (most likely a SnowFlake bug).
 556                #
 557                # https://docs.snowflake.com/en/sql-reference/functions/in
 558                # Context: https://github.com/tobymao/sqlglot/issues/3890
 559                return self.expression(
 560                    exp.NEQ, this=this.this, expression=exp.All(this=query.unnest())
 561                )
 562
 563            return self.expression(exp.Not, this=this)
 564
 565        def _parse_tag(self) -> exp.Tags:
 566            return self.expression(
 567                exp.Tags,
 568                expressions=self._parse_wrapped_csv(self._parse_property),
 569            )
 570
 571        def _parse_with_constraint(self) -> t.Optional[exp.Expression]:
 572            if self._prev.token_type != TokenType.WITH:
 573                self._retreat(self._index - 1)
 574
 575            if self._match_text_seq("MASKING", "POLICY"):
 576                policy = self._parse_column()
 577                return self.expression(
 578                    exp.MaskingPolicyColumnConstraint,
 579                    this=policy.to_dot() if isinstance(policy, exp.Column) else policy,
 580                    expressions=self._match(TokenType.USING)
 581                    and self._parse_wrapped_csv(self._parse_id_var),
 582                )
 583            if self._match_text_seq("PROJECTION", "POLICY"):
 584                policy = self._parse_column()
 585                return self.expression(
 586                    exp.ProjectionPolicyColumnConstraint,
 587                    this=policy.to_dot() if isinstance(policy, exp.Column) else policy,
 588                )
 589            if self._match(TokenType.TAG):
 590                return self._parse_tag()
 591
 592            return None
 593
 594        def _parse_with_property(self) -> t.Optional[exp.Expression] | t.List[exp.Expression]:
 595            if self._match(TokenType.TAG):
 596                return self._parse_tag()
 597
 598            return super()._parse_with_property()
 599
 600        def _parse_create(self) -> exp.Create | exp.Command:
 601            expression = super()._parse_create()
 602            if isinstance(expression, exp.Create) and expression.kind in self.NON_TABLE_CREATABLES:
 603                # Replace the Table node with the enclosed Identifier
 604                expression.this.replace(expression.this.this)
 605
 606            return expression
 607
 608        # https://docs.snowflake.com/en/sql-reference/functions/date_part.html
 609        # https://docs.snowflake.com/en/sql-reference/functions-date-time.html#label-supported-date-time-parts
 610        def _parse_date_part(self: Snowflake.Parser) -> t.Optional[exp.Expression]:
 611            this = self._parse_var() or self._parse_type()
 612
 613            if not this:
 614                return None
 615
 616            self._match(TokenType.COMMA)
 617            expression = self._parse_bitwise()
 618            this = map_date_part(this)
 619            name = this.name.upper()
 620
 621            if name.startswith("EPOCH"):
 622                if name == "EPOCH_MILLISECOND":
 623                    scale = 10**3
 624                elif name == "EPOCH_MICROSECOND":
 625                    scale = 10**6
 626                elif name == "EPOCH_NANOSECOND":
 627                    scale = 10**9
 628                else:
 629                    scale = None
 630
 631                ts = self.expression(exp.Cast, this=expression, to=exp.DataType.build("TIMESTAMP"))
 632                to_unix: exp.Expression = self.expression(exp.TimeToUnix, this=ts)
 633
 634                if scale:
 635                    to_unix = exp.Mul(this=to_unix, expression=exp.Literal.number(scale))
 636
 637                return to_unix
 638
 639            return self.expression(exp.Extract, this=this, expression=expression)
 640
 641        def _parse_bracket_key_value(self, is_map: bool = False) -> t.Optional[exp.Expression]:
 642            if is_map:
 643                # Keys are strings in Snowflake's objects, see also:
 644                # - https://docs.snowflake.com/en/sql-reference/data-types-semistructured
 645                # - https://docs.snowflake.com/en/sql-reference/functions/object_construct
 646                return self._parse_slice(self._parse_string())
 647
 648            return self._parse_slice(self._parse_alias(self._parse_assignment(), explicit=True))
 649
 650        def _parse_lateral(self) -> t.Optional[exp.Lateral]:
 651            lateral = super()._parse_lateral()
 652            if not lateral:
 653                return lateral
 654
 655            if isinstance(lateral.this, exp.Explode):
 656                table_alias = lateral.args.get("alias")
 657                columns = [exp.to_identifier(col) for col in self.FLATTEN_COLUMNS]
 658                if table_alias and not table_alias.args.get("columns"):
 659                    table_alias.set("columns", columns)
 660                elif not table_alias:
 661                    exp.alias_(lateral, "_flattened", table=columns, copy=False)
 662
 663            return lateral
 664
 665        def _parse_table_parts(
 666            self, schema: bool = False, is_db_reference: bool = False, wildcard: bool = False
 667        ) -> exp.Table:
 668            # https://docs.snowflake.com/en/user-guide/querying-stage
 669            if self._match(TokenType.STRING, advance=False):
 670                table = self._parse_string()
 671            elif self._match_text_seq("@", advance=False):
 672                table = self._parse_location_path()
 673            else:
 674                table = None
 675
 676            if table:
 677                file_format = None
 678                pattern = None
 679
 680                wrapped = self._match(TokenType.L_PAREN)
 681                while self._curr and wrapped and not self._match(TokenType.R_PAREN):
 682                    if self._match_text_seq("FILE_FORMAT", "=>"):
 683                        file_format = self._parse_string() or super()._parse_table_parts(
 684                            is_db_reference=is_db_reference
 685                        )
 686                    elif self._match_text_seq("PATTERN", "=>"):
 687                        pattern = self._parse_string()
 688                    else:
 689                        break
 690
 691                    self._match(TokenType.COMMA)
 692
 693                table = self.expression(exp.Table, this=table, format=file_format, pattern=pattern)
 694            else:
 695                table = super()._parse_table_parts(schema=schema, is_db_reference=is_db_reference)
 696
 697            return table
 698
 699        def _parse_id_var(
 700            self,
 701            any_token: bool = True,
 702            tokens: t.Optional[t.Collection[TokenType]] = None,
 703        ) -> t.Optional[exp.Expression]:
 704            if self._match_text_seq("IDENTIFIER", "("):
 705                identifier = (
 706                    super()._parse_id_var(any_token=any_token, tokens=tokens)
 707                    or self._parse_string()
 708                )
 709                self._match_r_paren()
 710                return self.expression(exp.Anonymous, this="IDENTIFIER", expressions=[identifier])
 711
 712            return super()._parse_id_var(any_token=any_token, tokens=tokens)
 713
 714        def _parse_show_snowflake(self, this: str) -> exp.Show:
 715            scope = None
 716            scope_kind = None
 717
 718            # will identity SHOW TERSE SCHEMAS but not SHOW TERSE PRIMARY KEYS
 719            # which is syntactically valid but has no effect on the output
 720            terse = self._tokens[self._index - 2].text.upper() == "TERSE"
 721
 722            history = self._match_text_seq("HISTORY")
 723
 724            like = self._parse_string() if self._match(TokenType.LIKE) else None
 725
 726            if self._match(TokenType.IN):
 727                if self._match_text_seq("ACCOUNT"):
 728                    scope_kind = "ACCOUNT"
 729                elif self._match_set(self.DB_CREATABLES):
 730                    scope_kind = self._prev.text.upper()
 731                    if self._curr:
 732                        scope = self._parse_table_parts()
 733                elif self._curr:
 734                    scope_kind = "SCHEMA" if this in self.SCHEMA_KINDS else "TABLE"
 735                    scope = self._parse_table_parts()
 736
 737            return self.expression(
 738                exp.Show,
 739                **{
 740                    "terse": terse,
 741                    "this": this,
 742                    "history": history,
 743                    "like": like,
 744                    "scope": scope,
 745                    "scope_kind": scope_kind,
 746                    "starts_with": self._match_text_seq("STARTS", "WITH") and self._parse_string(),
 747                    "limit": self._parse_limit(),
 748                    "from": self._parse_string() if self._match(TokenType.FROM) else None,
 749                },
 750            )
 751
 752        def _parse_location_property(self) -> exp.LocationProperty:
 753            self._match(TokenType.EQ)
 754            return self.expression(exp.LocationProperty, this=self._parse_location_path())
 755
 756        def _parse_file_location(self) -> t.Optional[exp.Expression]:
 757            # Parse either a subquery or a staged file
 758            return (
 759                self._parse_select(table=True, parse_subquery_alias=False)
 760                if self._match(TokenType.L_PAREN, advance=False)
 761                else self._parse_table_parts()
 762            )
 763
 764        def _parse_location_path(self) -> exp.Var:
 765            parts = [self._advance_any(ignore_reserved=True)]
 766
 767            # We avoid consuming a comma token because external tables like @foo and @bar
 768            # can be joined in a query with a comma separator, as well as closing paren
 769            # in case of subqueries
 770            while self._is_connected() and not self._match_set(
 771                (TokenType.COMMA, TokenType.L_PAREN, TokenType.R_PAREN), advance=False
 772            ):
 773                parts.append(self._advance_any(ignore_reserved=True))
 774
 775            return exp.var("".join(part.text for part in parts if part))
 776
 777        def _parse_lambda_arg(self) -> t.Optional[exp.Expression]:
 778            this = super()._parse_lambda_arg()
 779
 780            if not this:
 781                return this
 782
 783            typ = self._parse_types()
 784
 785            if typ:
 786                return self.expression(exp.Cast, this=this, to=typ)
 787
 788            return this
 789
 790    class Tokenizer(tokens.Tokenizer):
 791        STRING_ESCAPES = ["\\", "'"]
 792        HEX_STRINGS = [("x'", "'"), ("X'", "'")]
 793        RAW_STRINGS = ["$$"]
 794        COMMENTS = ["--", "//", ("/*", "*/")]
 795        NESTED_COMMENTS = False
 796
 797        KEYWORDS = {
 798            **tokens.Tokenizer.KEYWORDS,
 799            "BYTEINT": TokenType.INT,
 800            "CHAR VARYING": TokenType.VARCHAR,
 801            "CHARACTER VARYING": TokenType.VARCHAR,
 802            "EXCLUDE": TokenType.EXCEPT,
 803            "ILIKE ANY": TokenType.ILIKE_ANY,
 804            "LIKE ANY": TokenType.LIKE_ANY,
 805            "MATCH_CONDITION": TokenType.MATCH_CONDITION,
 806            "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE,
 807            "MINUS": TokenType.EXCEPT,
 808            "NCHAR VARYING": TokenType.VARCHAR,
 809            "PUT": TokenType.COMMAND,
 810            "REMOVE": TokenType.COMMAND,
 811            "RM": TokenType.COMMAND,
 812            "SAMPLE": TokenType.TABLE_SAMPLE,
 813            "SQL_DOUBLE": TokenType.DOUBLE,
 814            "SQL_VARCHAR": TokenType.VARCHAR,
 815            "STORAGE INTEGRATION": TokenType.STORAGE_INTEGRATION,
 816            "TAG": TokenType.TAG,
 817            "TIMESTAMP_TZ": TokenType.TIMESTAMPTZ,
 818            "TOP": TokenType.TOP,
 819            "WAREHOUSE": TokenType.WAREHOUSE,
 820            "STREAMLIT": TokenType.STREAMLIT,
 821        }
 822        KEYWORDS.pop("/*+")
 823
 824        SINGLE_TOKENS = {
 825            **tokens.Tokenizer.SINGLE_TOKENS,
 826            "$": TokenType.PARAMETER,
 827        }
 828
 829        VAR_SINGLE_TOKENS = {"$"}
 830
 831        COMMANDS = tokens.Tokenizer.COMMANDS - {TokenType.SHOW}
 832
 833    class Generator(generator.Generator):
 834        PARAMETER_TOKEN = "$"
 835        MATCHED_BY_SOURCE = False
 836        SINGLE_STRING_INTERVAL = True
 837        JOIN_HINTS = False
 838        TABLE_HINTS = False
 839        QUERY_HINTS = False
 840        AGGREGATE_FILTER_SUPPORTED = False
 841        SUPPORTS_TABLE_COPY = False
 842        COLLATE_IS_FUNC = True
 843        LIMIT_ONLY_LITERALS = True
 844        JSON_KEY_VALUE_PAIR_SEP = ","
 845        INSERT_OVERWRITE = " OVERWRITE INTO"
 846        STRUCT_DELIMITER = ("(", ")")
 847        COPY_PARAMS_ARE_WRAPPED = False
 848        COPY_PARAMS_EQ_REQUIRED = True
 849        STAR_EXCEPT = "EXCLUDE"
 850        SUPPORTS_EXPLODING_PROJECTIONS = False
 851        ARRAY_CONCAT_IS_VAR_LEN = False
 852        SUPPORTS_CONVERT_TIMEZONE = True
 853        EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False
 854        SUPPORTS_MEDIAN = True
 855        ARRAY_SIZE_NAME = "ARRAY_SIZE"
 856
 857        TRANSFORMS = {
 858            **generator.Generator.TRANSFORMS,
 859            exp.ApproxDistinct: rename_func("APPROX_COUNT_DISTINCT"),
 860            exp.ArgMax: rename_func("MAX_BY"),
 861            exp.ArgMin: rename_func("MIN_BY"),
 862            exp.Array: inline_array_sql,
 863            exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CAT"),
 864            exp.ArrayContains: lambda self, e: self.func("ARRAY_CONTAINS", e.expression, e.this),
 865            exp.AtTimeZone: lambda self, e: self.func(
 866                "CONVERT_TIMEZONE", e.args.get("zone"), e.this
 867            ),
 868            exp.BitwiseXor: rename_func("BITXOR"),
 869            exp.Create: transforms.preprocess([_flatten_structured_types_unless_iceberg]),
 870            exp.DateAdd: date_delta_sql("DATEADD"),
 871            exp.DateDiff: date_delta_sql("DATEDIFF"),
 872            exp.DatetimeAdd: date_delta_sql("TIMESTAMPADD"),
 873            exp.DatetimeDiff: timestampdiff_sql,
 874            exp.DateStrToDate: datestrtodate_sql,
 875            exp.DayOfMonth: rename_func("DAYOFMONTH"),
 876            exp.DayOfWeek: rename_func("DAYOFWEEK"),
 877            exp.DayOfYear: rename_func("DAYOFYEAR"),
 878            exp.Explode: rename_func("FLATTEN"),
 879            exp.Extract: rename_func("DATE_PART"),
 880            exp.FromTimeZone: lambda self, e: self.func(
 881                "CONVERT_TIMEZONE", e.args.get("zone"), "'UTC'", e.this
 882            ),
 883            exp.GenerateSeries: lambda self, e: self.func(
 884                "ARRAY_GENERATE_RANGE", e.args["start"], e.args["end"] + 1, e.args.get("step")
 885            ),
 886            exp.GroupConcat: rename_func("LISTAGG"),
 887            exp.If: if_sql(name="IFF", false_value="NULL"),
 888            exp.JSONExtractArray: _json_extract_value_array_sql,
 889            exp.JSONExtractScalar: lambda self, e: self.func(
 890                "JSON_EXTRACT_PATH_TEXT", e.this, e.expression
 891            ),
 892            exp.JSONObject: lambda self, e: self.func("OBJECT_CONSTRUCT_KEEP_NULL", *e.expressions),
 893            exp.JSONPathRoot: lambda *_: "",
 894            exp.JSONValueArray: _json_extract_value_array_sql,
 895            exp.LogicalAnd: rename_func("BOOLAND_AGG"),
 896            exp.LogicalOr: rename_func("BOOLOR_AGG"),
 897            exp.Map: lambda self, e: var_map_sql(self, e, "OBJECT_CONSTRUCT"),
 898            exp.MakeInterval: no_make_interval_sql,
 899            exp.Max: max_or_greatest,
 900            exp.Min: min_or_least,
 901            exp.ParseJSON: lambda self, e: self.func(
 902                "TRY_PARSE_JSON" if e.args.get("safe") else "PARSE_JSON", e.this
 903            ),
 904            exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}",
 905            exp.PercentileCont: transforms.preprocess(
 906                [transforms.add_within_group_for_percentiles]
 907            ),
 908            exp.PercentileDisc: transforms.preprocess(
 909                [transforms.add_within_group_for_percentiles]
 910            ),
 911            exp.Pivot: transforms.preprocess([_unqualify_unpivot_columns]),
 912            exp.RegexpExtract: _regexpextract_sql,
 913            exp.RegexpExtractAll: _regexpextract_sql,
 914            exp.RegexpILike: _regexpilike_sql,
 915            exp.Rand: rename_func("RANDOM"),
 916            exp.Select: transforms.preprocess(
 917                [
 918                    transforms.eliminate_distinct_on,
 919                    transforms.explode_to_unnest(),
 920                    transforms.eliminate_semi_and_anti_joins,
 921                    _transform_generate_date_array,
 922                ]
 923            ),
 924            exp.SafeDivide: lambda self, e: no_safe_divide_sql(self, e, "IFF"),
 925            exp.SHA: rename_func("SHA1"),
 926            exp.StarMap: rename_func("OBJECT_CONSTRUCT"),
 927            exp.StartsWith: rename_func("STARTSWITH"),
 928            exp.StrPosition: lambda self, e: self.func(
 929                "POSITION", e.args.get("substr"), e.this, e.args.get("position")
 930            ),
 931            exp.StrToDate: lambda self, e: self.func("DATE", e.this, self.format_time(e)),
 932            exp.Stuff: rename_func("INSERT"),
 933            exp.TimeAdd: date_delta_sql("TIMEADD"),
 934            exp.Timestamp: no_timestamp_sql,
 935            exp.TimestampAdd: date_delta_sql("TIMESTAMPADD"),
 936            exp.TimestampDiff: lambda self, e: self.func(
 937                "TIMESTAMPDIFF", e.unit, e.expression, e.this
 938            ),
 939            exp.TimestampTrunc: timestamptrunc_sql(),
 940            exp.TimeStrToTime: timestrtotime_sql,
 941            exp.TimeToUnix: lambda self, e: f"EXTRACT(epoch_second FROM {self.sql(e, 'this')})",
 942            exp.ToArray: rename_func("TO_ARRAY"),
 943            exp.ToChar: lambda self, e: self.function_fallback_sql(e),
 944            exp.ToDouble: rename_func("TO_DOUBLE"),
 945            exp.TsOrDsAdd: date_delta_sql("DATEADD", cast=True),
 946            exp.TsOrDsDiff: date_delta_sql("DATEDIFF"),
 947            exp.TsOrDsToDate: lambda self, e: self.func(
 948                "TRY_TO_DATE" if e.args.get("safe") else "TO_DATE", e.this, self.format_time(e)
 949            ),
 950            exp.UnixToTime: rename_func("TO_TIMESTAMP"),
 951            exp.Uuid: rename_func("UUID_STRING"),
 952            exp.VarMap: lambda self, e: var_map_sql(self, e, "OBJECT_CONSTRUCT"),
 953            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
 954            exp.Xor: rename_func("BOOLXOR"),
 955            exp.Levenshtein: unsupported_args("ins_cost", "del_cost", "sub_cost")(
 956                rename_func("EDITDISTANCE")
 957            ),
 958        }
 959
 960        SUPPORTED_JSON_PATH_PARTS = {
 961            exp.JSONPathKey,
 962            exp.JSONPathRoot,
 963            exp.JSONPathSubscript,
 964        }
 965
 966        TYPE_MAPPING = {
 967            **generator.Generator.TYPE_MAPPING,
 968            exp.DataType.Type.NESTED: "OBJECT",
 969            exp.DataType.Type.STRUCT: "OBJECT",
 970        }
 971
 972        PROPERTIES_LOCATION = {
 973            **generator.Generator.PROPERTIES_LOCATION,
 974            exp.SetProperty: exp.Properties.Location.UNSUPPORTED,
 975            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
 976        }
 977
 978        UNSUPPORTED_VALUES_EXPRESSIONS = {
 979            exp.Map,
 980            exp.StarMap,
 981            exp.Struct,
 982            exp.VarMap,
 983        }
 984
 985        def with_properties(self, properties: exp.Properties) -> str:
 986            return self.properties(properties, wrapped=False, prefix=self.sep(""), sep=" ")
 987
 988        def values_sql(self, expression: exp.Values, values_as_table: bool = True) -> str:
 989            if expression.find(*self.UNSUPPORTED_VALUES_EXPRESSIONS):
 990                values_as_table = False
 991
 992            return super().values_sql(expression, values_as_table=values_as_table)
 993
 994        def datatype_sql(self, expression: exp.DataType) -> str:
 995            expressions = expression.expressions
 996            if (
 997                expressions
 998                and expression.is_type(*exp.DataType.STRUCT_TYPES)
 999                and any(isinstance(field_type, exp.DataType) for field_type in expressions)
1000            ):
1001                # The correct syntax is OBJECT [ (<key> <value_type [NOT NULL] [, ...]) ]
1002                return "OBJECT"
1003
1004            return super().datatype_sql(expression)
1005
1006        def tonumber_sql(self, expression: exp.ToNumber) -> str:
1007            return self.func(
1008                "TO_NUMBER",
1009                expression.this,
1010                expression.args.get("format"),
1011                expression.args.get("precision"),
1012                expression.args.get("scale"),
1013            )
1014
1015        def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str:
1016            milli = expression.args.get("milli")
1017            if milli is not None:
1018                milli_to_nano = milli.pop() * exp.Literal.number(1000000)
1019                expression.set("nano", milli_to_nano)
1020
1021            return rename_func("TIMESTAMP_FROM_PARTS")(self, expression)
1022
1023        def cast_sql(self, expression: exp.Cast, safe_prefix: t.Optional[str] = None) -> str:
1024            if expression.is_type(exp.DataType.Type.GEOGRAPHY):
1025                return self.func("TO_GEOGRAPHY", expression.this)
1026            if expression.is_type(exp.DataType.Type.GEOMETRY):
1027                return self.func("TO_GEOMETRY", expression.this)
1028
1029            return super().cast_sql(expression, safe_prefix=safe_prefix)
1030
1031        def trycast_sql(self, expression: exp.TryCast) -> str:
1032            value = expression.this
1033
1034            if value.type is None:
1035                from sqlglot.optimizer.annotate_types import annotate_types
1036
1037                value = annotate_types(value)
1038
1039            if value.is_type(*exp.DataType.TEXT_TYPES, exp.DataType.Type.UNKNOWN):
1040                return super().trycast_sql(expression)
1041
1042            # TRY_CAST only works for string values in Snowflake
1043            return self.cast_sql(expression)
1044
1045        def log_sql(self, expression: exp.Log) -> str:
1046            if not expression.expression:
1047                return self.func("LN", expression.this)
1048
1049            return super().log_sql(expression)
1050
1051        def unnest_sql(self, expression: exp.Unnest) -> str:
1052            unnest_alias = expression.args.get("alias")
1053            offset = expression.args.get("offset")
1054
1055            columns = [
1056                exp.to_identifier("seq"),
1057                exp.to_identifier("key"),
1058                exp.to_identifier("path"),
1059                offset.pop() if isinstance(offset, exp.Expression) else exp.to_identifier("index"),
1060                seq_get(unnest_alias.columns if unnest_alias else [], 0)
1061                or exp.to_identifier("value"),
1062                exp.to_identifier("this"),
1063            ]
1064
1065            if unnest_alias:
1066                unnest_alias.set("columns", columns)
1067            else:
1068                unnest_alias = exp.TableAlias(this="_u", columns=columns)
1069
1070            explode = f"TABLE(FLATTEN(INPUT => {self.sql(expression.expressions[0])}))"
1071            alias = self.sql(unnest_alias)
1072            alias = f" AS {alias}" if alias else ""
1073            return f"{explode}{alias}"
1074
1075        def show_sql(self, expression: exp.Show) -> str:
1076            terse = "TERSE " if expression.args.get("terse") else ""
1077            history = " HISTORY" if expression.args.get("history") else ""
1078            like = self.sql(expression, "like")
1079            like = f" LIKE {like}" if like else ""
1080
1081            scope = self.sql(expression, "scope")
1082            scope = f" {scope}" if scope else ""
1083
1084            scope_kind = self.sql(expression, "scope_kind")
1085            if scope_kind:
1086                scope_kind = f" IN {scope_kind}"
1087
1088            starts_with = self.sql(expression, "starts_with")
1089            if starts_with:
1090                starts_with = f" STARTS WITH {starts_with}"
1091
1092            limit = self.sql(expression, "limit")
1093
1094            from_ = self.sql(expression, "from")
1095            if from_:
1096                from_ = f" FROM {from_}"
1097
1098            return f"SHOW {terse}{expression.name}{history}{like}{scope_kind}{scope}{starts_with}{limit}{from_}"
1099
1100        def describe_sql(self, expression: exp.Describe) -> str:
1101            # Default to table if kind is unknown
1102            kind_value = expression.args.get("kind") or "TABLE"
1103            kind = f" {kind_value}" if kind_value else ""
1104            this = f" {self.sql(expression, 'this')}"
1105            expressions = self.expressions(expression, flat=True)
1106            expressions = f" {expressions}" if expressions else ""
1107            return f"DESCRIBE{kind}{this}{expressions}"
1108
1109        def generatedasidentitycolumnconstraint_sql(
1110            self, expression: exp.GeneratedAsIdentityColumnConstraint
1111        ) -> str:
1112            start = expression.args.get("start")
1113            start = f" START {start}" if start else ""
1114            increment = expression.args.get("increment")
1115            increment = f" INCREMENT {increment}" if increment else ""
1116            return f"AUTOINCREMENT{start}{increment}"
1117
1118        def cluster_sql(self, expression: exp.Cluster) -> str:
1119            return f"CLUSTER BY ({self.expressions(expression, flat=True)})"
1120
1121        def struct_sql(self, expression: exp.Struct) -> str:
1122            keys = []
1123            values = []
1124
1125            for i, e in enumerate(expression.expressions):
1126                if isinstance(e, exp.PropertyEQ):
1127                    keys.append(
1128                        exp.Literal.string(e.name) if isinstance(e.this, exp.Identifier) else e.this
1129                    )
1130                    values.append(e.expression)
1131                else:
1132                    keys.append(exp.Literal.string(f"_{i}"))
1133                    values.append(e)
1134
1135            return self.func("OBJECT_CONSTRUCT", *flatten(zip(keys, values)))
1136
1137        @unsupported_args("weight", "accuracy")
1138        def approxquantile_sql(self, expression: exp.ApproxQuantile) -> str:
1139            return self.func("APPROX_PERCENTILE", expression.this, expression.args.get("quantile"))
1140
1141        def alterset_sql(self, expression: exp.AlterSet) -> str:
1142            exprs = self.expressions(expression, flat=True)
1143            exprs = f" {exprs}" if exprs else ""
1144            file_format = self.expressions(expression, key="file_format", flat=True, sep=" ")
1145            file_format = f" STAGE_FILE_FORMAT = ({file_format})" if file_format else ""
1146            copy_options = self.expressions(expression, key="copy_options", flat=True, sep=" ")
1147            copy_options = f" STAGE_COPY_OPTIONS = ({copy_options})" if copy_options else ""
1148            tag = self.expressions(expression, key="tag", flat=True)
1149            tag = f" TAG {tag}" if tag else ""
1150
1151            return f"SET{exprs}{file_format}{copy_options}{tag}"
1152
1153        def strtotime_sql(self, expression: exp.StrToTime):
1154            safe_prefix = "TRY_" if expression.args.get("safe") else ""
1155            return self.func(
1156                f"{safe_prefix}TO_TIMESTAMP", expression.this, self.format_time(expression)
1157            )
1158
1159        def timestampsub_sql(self, expression: exp.TimestampSub):
1160            return self.sql(
1161                exp.TimestampAdd(
1162                    this=expression.this,
1163                    expression=expression.expression * -1,
1164                    unit=expression.unit,
1165                )
1166            )
1167
1168        def jsonextract_sql(self, expression: exp.JSONExtract):
1169            this = expression.this
1170
1171            # JSON strings are valid coming from other dialects such as BQ
1172            return self.func(
1173                "GET_PATH",
1174                exp.ParseJSON(this=this) if this.is_string else this,
1175                expression.expression,
1176            )
1177
1178        def timetostr_sql(self, expression: exp.TimeToStr) -> str:
1179            this = expression.this
1180            if not isinstance(this, exp.TsOrDsToTimestamp):
1181                this = exp.cast(this, exp.DataType.Type.TIMESTAMP)
1182
1183            return self.func("TO_CHAR", this, self.format_time(expression))