sqlglot.dialects.duckdb

  1from __future__ import annotations
  2
  3import typing as t
  4
  5from sqlglot import exp, generator, parser, tokens, transforms
  6from sqlglot.expressions import DATA_TYPE
  7from sqlglot.dialects.dialect import (
  8    Dialect,
  9    JSON_EXTRACT_TYPE,
 10    NormalizationStrategy,
 11    approx_count_distinct_sql,
 12    arg_max_or_min_no_count,
 13    arrow_json_extract_sql,
 14    binary_from_function,
 15    bool_xor_sql,
 16    build_default_decimal_type,
 17    date_trunc_to_time,
 18    datestrtodate_sql,
 19    no_datetime_sql,
 20    encode_decode_sql,
 21    build_formatted_time,
 22    inline_array_unless_query,
 23    no_comment_column_constraint_sql,
 24    no_safe_divide_sql,
 25    no_time_sql,
 26    no_timestamp_sql,
 27    pivot_column_names,
 28    regexp_extract_sql,
 29    rename_func,
 30    str_position_sql,
 31    str_to_time_sql,
 32    timestamptrunc_sql,
 33    timestrtotime_sql,
 34    unit_to_var,
 35    unit_to_str,
 36    sha256_sql,
 37)
 38from sqlglot.helper import seq_get
 39from sqlglot.tokens import TokenType
 40
 41DATETIME_DELTA = t.Union[
 42    exp.DateAdd, exp.TimeAdd, exp.DatetimeAdd, exp.TsOrDsAdd, exp.DateSub, exp.DatetimeSub
 43]
 44
 45WINDOW_FUNCS_WITH_IGNORE_NULLS = (
 46    exp.FirstValue,
 47    exp.LastValue,
 48    exp.Lag,
 49    exp.Lead,
 50    exp.NthValue,
 51)
 52
 53
 54def _date_delta_sql(self: DuckDB.Generator, expression: DATETIME_DELTA) -> str:
 55    this = expression.this
 56    unit = unit_to_var(expression)
 57    op = (
 58        "+"
 59        if isinstance(expression, (exp.DateAdd, exp.TimeAdd, exp.DatetimeAdd, exp.TsOrDsAdd))
 60        else "-"
 61    )
 62
 63    to_type: t.Optional[DATA_TYPE] = None
 64    if isinstance(expression, exp.TsOrDsAdd):
 65        to_type = expression.return_type
 66    elif this.is_string:
 67        # Cast string literals (i.e function parameters) to the appropriate type for +/- interval to work
 68        to_type = (
 69            exp.DataType.Type.DATETIME
 70            if isinstance(expression, (exp.DatetimeAdd, exp.DatetimeSub))
 71            else exp.DataType.Type.DATE
 72        )
 73
 74    this = exp.cast(this, to_type) if to_type else this
 75
 76    return f"{self.sql(this)} {op} {self.sql(exp.Interval(this=expression.expression, unit=unit))}"
 77
 78
 79# BigQuery -> DuckDB conversion for the DATE function
 80def _date_sql(self: DuckDB.Generator, expression: exp.Date) -> str:
 81    result = f"CAST({self.sql(expression, 'this')} AS DATE)"
 82    zone = self.sql(expression, "zone")
 83
 84    if zone:
 85        date_str = self.func("STRFTIME", result, "'%d/%m/%Y'")
 86        date_str = f"{date_str} || ' ' || {zone}"
 87
 88        # This will create a TIMESTAMP with time zone information
 89        result = self.func("STRPTIME", date_str, "'%d/%m/%Y %Z'")
 90
 91    return result
 92
 93
 94# BigQuery -> DuckDB conversion for the TIME_DIFF function
 95def _timediff_sql(self: DuckDB.Generator, expression: exp.TimeDiff) -> str:
 96    this = exp.cast(expression.this, exp.DataType.Type.TIME)
 97    expr = exp.cast(expression.expression, exp.DataType.Type.TIME)
 98
 99    # Although the 2 dialects share similar signatures, BQ seems to inverse
100    # the sign of the result so the start/end time operands are flipped
101    return self.func("DATE_DIFF", unit_to_str(expression), expr, this)
102
103
104def _array_sort_sql(self: DuckDB.Generator, expression: exp.ArraySort) -> str:
105    if expression.expression:
106        self.unsupported("DuckDB ARRAY_SORT does not support a comparator")
107    return self.func("ARRAY_SORT", expression.this)
108
109
110def _sort_array_sql(self: DuckDB.Generator, expression: exp.SortArray) -> str:
111    name = "ARRAY_REVERSE_SORT" if expression.args.get("asc") == exp.false() else "ARRAY_SORT"
112    return self.func(name, expression.this)
113
114
115def _build_sort_array_desc(args: t.List) -> exp.Expression:
116    return exp.SortArray(this=seq_get(args, 0), asc=exp.false())
117
118
119def _build_date_diff(args: t.List) -> exp.Expression:
120    return exp.DateDiff(this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0))
121
122
123def _build_generate_series(end_exclusive: bool = False) -> t.Callable[[t.List], exp.GenerateSeries]:
124    def _builder(args: t.List) -> exp.GenerateSeries:
125        # Check https://duckdb.org/docs/sql/functions/nested.html#range-functions
126        if len(args) == 1:
127            # DuckDB uses 0 as a default for the series' start when it's omitted
128            args.insert(0, exp.Literal.number("0"))
129
130        gen_series = exp.GenerateSeries.from_arg_list(args)
131        gen_series.set("is_end_exclusive", end_exclusive)
132
133        return gen_series
134
135    return _builder
136
137
138def _build_make_timestamp(args: t.List) -> exp.Expression:
139    if len(args) == 1:
140        return exp.UnixToTime(this=seq_get(args, 0), scale=exp.UnixToTime.MICROS)
141
142    return exp.TimestampFromParts(
143        year=seq_get(args, 0),
144        month=seq_get(args, 1),
145        day=seq_get(args, 2),
146        hour=seq_get(args, 3),
147        min=seq_get(args, 4),
148        sec=seq_get(args, 5),
149    )
150
151
152def _struct_sql(self: DuckDB.Generator, expression: exp.Struct) -> str:
153    args: t.List[str] = []
154
155    # BigQuery allows inline construction such as "STRUCT<a STRING, b INTEGER>('str', 1)" which is
156    # canonicalized to "ROW('str', 1) AS STRUCT(a TEXT, b INT)" in DuckDB
157    # The transformation to ROW will take place if a cast to STRUCT / ARRAY of STRUCTs is found
158    ancestor_cast = expression.find_ancestor(exp.Cast)
159    is_struct_cast = ancestor_cast and any(
160        casted_type.is_type(exp.DataType.Type.STRUCT)
161        for casted_type in ancestor_cast.find_all(exp.DataType)
162    )
163
164    for i, expr in enumerate(expression.expressions):
165        is_property_eq = isinstance(expr, exp.PropertyEQ)
166        value = expr.expression if is_property_eq else expr
167
168        if is_struct_cast:
169            args.append(self.sql(value))
170        else:
171            key = expr.name if is_property_eq else f"_{i}"
172            args.append(f"{self.sql(exp.Literal.string(key))}: {self.sql(value)}")
173
174    csv_args = ", ".join(args)
175
176    return f"ROW({csv_args})" if is_struct_cast else f"{{{csv_args}}}"
177
178
179def _datatype_sql(self: DuckDB.Generator, expression: exp.DataType) -> str:
180    if expression.is_type("array"):
181        return f"{self.expressions(expression, flat=True)}[{self.expressions(expression, key='values', flat=True)}]"
182
183    # Modifiers are not supported for TIME, [TIME | TIMESTAMP] WITH TIME ZONE
184    if expression.is_type(
185        exp.DataType.Type.TIME, exp.DataType.Type.TIMETZ, exp.DataType.Type.TIMESTAMPTZ
186    ):
187        return expression.this.value
188
189    return self.datatype_sql(expression)
190
191
192def _json_format_sql(self: DuckDB.Generator, expression: exp.JSONFormat) -> str:
193    sql = self.func("TO_JSON", expression.this, expression.args.get("options"))
194    return f"CAST({sql} AS TEXT)"
195
196
197def _unix_to_time_sql(self: DuckDB.Generator, expression: exp.UnixToTime) -> str:
198    scale = expression.args.get("scale")
199    timestamp = expression.this
200
201    if scale in (None, exp.UnixToTime.SECONDS):
202        return self.func("TO_TIMESTAMP", timestamp)
203    if scale == exp.UnixToTime.MILLIS:
204        return self.func("EPOCH_MS", timestamp)
205    if scale == exp.UnixToTime.MICROS:
206        return self.func("MAKE_TIMESTAMP", timestamp)
207
208    return self.func("TO_TIMESTAMP", exp.Div(this=timestamp, expression=exp.func("POW", 10, scale)))
209
210
211WRAPPED_JSON_EXTRACT_EXPRESSIONS = (exp.Binary, exp.Bracket, exp.In)
212
213
214def _arrow_json_extract_sql(self: DuckDB.Generator, expression: JSON_EXTRACT_TYPE) -> str:
215    arrow_sql = arrow_json_extract_sql(self, expression)
216    if not expression.same_parent and isinstance(
217        expression.parent, WRAPPED_JSON_EXTRACT_EXPRESSIONS
218    ):
219        arrow_sql = self.wrap(arrow_sql)
220    return arrow_sql
221
222
223def _implicit_datetime_cast(
224    arg: t.Optional[exp.Expression], type: exp.DataType.Type = exp.DataType.Type.DATE
225) -> t.Optional[exp.Expression]:
226    return exp.cast(arg, type) if isinstance(arg, exp.Literal) else arg
227
228
229def _date_diff_sql(self: DuckDB.Generator, expression: exp.DateDiff) -> str:
230    this = _implicit_datetime_cast(expression.this)
231    expr = _implicit_datetime_cast(expression.expression)
232
233    return self.func("DATE_DIFF", unit_to_str(expression), expr, this)
234
235
236def _generate_datetime_array_sql(
237    self: DuckDB.Generator, expression: t.Union[exp.GenerateDateArray, exp.GenerateTimestampArray]
238) -> str:
239    is_generate_date_array = isinstance(expression, exp.GenerateDateArray)
240
241    type = exp.DataType.Type.DATE if is_generate_date_array else exp.DataType.Type.TIMESTAMP
242    start = _implicit_datetime_cast(expression.args.get("start"), type=type)
243    end = _implicit_datetime_cast(expression.args.get("end"), type=type)
244
245    # BQ's GENERATE_DATE_ARRAY & GENERATE_TIMESTAMP_ARRAY are transformed to DuckDB'S GENERATE_SERIES
246    gen_series: t.Union[exp.GenerateSeries, exp.Cast] = exp.GenerateSeries(
247        start=start, end=end, step=expression.args.get("step")
248    )
249
250    if is_generate_date_array:
251        # The GENERATE_SERIES result type is TIMESTAMP array, so to match BQ's semantics for
252        # GENERATE_DATE_ARRAY we must cast it back to DATE array
253        gen_series = exp.cast(gen_series, exp.DataType.build("ARRAY<DATE>"))
254
255    return self.sql(gen_series)
256
257
258class DuckDB(Dialect):
259    NULL_ORDERING = "nulls_are_last"
260    SUPPORTS_USER_DEFINED_TYPES = False
261    SAFE_DIVISION = True
262    INDEX_OFFSET = 1
263    CONCAT_COALESCE = True
264    SUPPORTS_ORDER_BY_ALL = True
265    SUPPORTS_FIXED_SIZE_ARRAYS = True
266    STRICT_JSON_PATH_SYNTAX = False
267
268    # https://duckdb.org/docs/sql/introduction.html#creating-a-new-table
269    NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE
270
271    def to_json_path(self, path: t.Optional[exp.Expression]) -> t.Optional[exp.Expression]:
272        if isinstance(path, exp.Literal):
273            # DuckDB also supports the JSON pointer syntax, where every path starts with a `/`.
274            # Additionally, it allows accessing the back of lists using the `[#-i]` syntax.
275            # This check ensures we'll avoid trying to parse these as JSON paths, which can
276            # either result in a noisy warning or in an invalid representation of the path.
277            path_text = path.name
278            if path_text.startswith("/") or "[#" in path_text:
279                return path
280
281        return super().to_json_path(path)
282
283    class Tokenizer(tokens.Tokenizer):
284        HEREDOC_STRINGS = ["$"]
285
286        HEREDOC_TAG_IS_IDENTIFIER = True
287        HEREDOC_STRING_ALTERNATIVE = TokenType.PARAMETER
288
289        KEYWORDS = {
290            **tokens.Tokenizer.KEYWORDS,
291            "//": TokenType.DIV,
292            "ATTACH": TokenType.COMMAND,
293            "BINARY": TokenType.VARBINARY,
294            "BITSTRING": TokenType.BIT,
295            "BPCHAR": TokenType.TEXT,
296            "CHAR": TokenType.TEXT,
297            "CHARACTER VARYING": TokenType.TEXT,
298            "EXCLUDE": TokenType.EXCEPT,
299            "LOGICAL": TokenType.BOOLEAN,
300            "ONLY": TokenType.ONLY,
301            "PIVOT_WIDER": TokenType.PIVOT,
302            "POSITIONAL": TokenType.POSITIONAL,
303            "SIGNED": TokenType.INT,
304            "STRING": TokenType.TEXT,
305            "SUMMARIZE": TokenType.SUMMARIZE,
306            "TIMESTAMP_S": TokenType.TIMESTAMP_S,
307            "TIMESTAMP_MS": TokenType.TIMESTAMP_MS,
308            "TIMESTAMP_NS": TokenType.TIMESTAMP_NS,
309            "TIMESTAMP_US": TokenType.TIMESTAMP,
310            "UBIGINT": TokenType.UBIGINT,
311            "UINTEGER": TokenType.UINT,
312            "USMALLINT": TokenType.USMALLINT,
313            "UTINYINT": TokenType.UTINYINT,
314            "VARCHAR": TokenType.TEXT,
315        }
316        KEYWORDS.pop("/*+")
317
318        SINGLE_TOKENS = {
319            **tokens.Tokenizer.SINGLE_TOKENS,
320            "$": TokenType.PARAMETER,
321        }
322
323    class Parser(parser.Parser):
324        BITWISE = {
325            **parser.Parser.BITWISE,
326            TokenType.TILDA: exp.RegexpLike,
327        }
328
329        FUNCTIONS_WITH_ALIASED_ARGS = {*parser.Parser.FUNCTIONS_WITH_ALIASED_ARGS, "STRUCT_PACK"}
330
331        FUNCTIONS = {
332            **parser.Parser.FUNCTIONS,
333            "ARRAY_REVERSE_SORT": _build_sort_array_desc,
334            "ARRAY_SORT": exp.SortArray.from_arg_list,
335            "DATEDIFF": _build_date_diff,
336            "DATE_DIFF": _build_date_diff,
337            "DATE_TRUNC": date_trunc_to_time,
338            "DATETRUNC": date_trunc_to_time,
339            "DECODE": lambda args: exp.Decode(
340                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
341            ),
342            "ENCODE": lambda args: exp.Encode(
343                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
344            ),
345            "EPOCH": exp.TimeToUnix.from_arg_list,
346            "EPOCH_MS": lambda args: exp.UnixToTime(
347                this=seq_get(args, 0), scale=exp.UnixToTime.MILLIS
348            ),
349            "JSON": exp.ParseJSON.from_arg_list,
350            "JSON_EXTRACT_PATH": parser.build_extract_json_with_path(exp.JSONExtract),
351            "JSON_EXTRACT_STRING": parser.build_extract_json_with_path(exp.JSONExtractScalar),
352            "LIST_HAS": exp.ArrayContains.from_arg_list,
353            "LIST_REVERSE_SORT": _build_sort_array_desc,
354            "LIST_SORT": exp.SortArray.from_arg_list,
355            "LIST_VALUE": lambda args: exp.Array(expressions=args),
356            "MAKE_TIME": exp.TimeFromParts.from_arg_list,
357            "MAKE_TIMESTAMP": _build_make_timestamp,
358            "MEDIAN": lambda args: exp.PercentileCont(
359                this=seq_get(args, 0), expression=exp.Literal.number(0.5)
360            ),
361            "QUANTILE_CONT": exp.PercentileCont.from_arg_list,
362            "QUANTILE_DISC": exp.PercentileDisc.from_arg_list,
363            "REGEXP_EXTRACT": lambda args: exp.RegexpExtract(
364                this=seq_get(args, 0), expression=seq_get(args, 1), group=seq_get(args, 2)
365            ),
366            "REGEXP_MATCHES": exp.RegexpLike.from_arg_list,
367            "REGEXP_REPLACE": lambda args: exp.RegexpReplace(
368                this=seq_get(args, 0),
369                expression=seq_get(args, 1),
370                replacement=seq_get(args, 2),
371                modifiers=seq_get(args, 3),
372            ),
373            "STRFTIME": build_formatted_time(exp.TimeToStr, "duckdb"),
374            "STRING_SPLIT": exp.Split.from_arg_list,
375            "STRING_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
376            "STRING_TO_ARRAY": exp.Split.from_arg_list,
377            "STRPTIME": build_formatted_time(exp.StrToTime, "duckdb"),
378            "STRUCT_PACK": exp.Struct.from_arg_list,
379            "STR_SPLIT": exp.Split.from_arg_list,
380            "STR_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
381            "TO_TIMESTAMP": exp.UnixToTime.from_arg_list,
382            "UNNEST": exp.Explode.from_arg_list,
383            "XOR": binary_from_function(exp.BitwiseXor),
384            "GENERATE_SERIES": _build_generate_series(),
385            "RANGE": _build_generate_series(end_exclusive=True),
386        }
387
388        FUNCTIONS.pop("DATE_SUB")
389        FUNCTIONS.pop("GLOB")
390
391        FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy()
392        FUNCTION_PARSERS.pop("DECODE")
393
394        NO_PAREN_FUNCTION_PARSERS = {
395            **parser.Parser.NO_PAREN_FUNCTION_PARSERS,
396            "MAP": lambda self: self._parse_map(),
397        }
398
399        TABLE_ALIAS_TOKENS = parser.Parser.TABLE_ALIAS_TOKENS - {
400            TokenType.SEMI,
401            TokenType.ANTI,
402        }
403
404        PLACEHOLDER_PARSERS = {
405            **parser.Parser.PLACEHOLDER_PARSERS,
406            TokenType.PARAMETER: lambda self: (
407                self.expression(exp.Placeholder, this=self._prev.text)
408                if self._match(TokenType.NUMBER) or self._match_set(self.ID_VAR_TOKENS)
409                else None
410            ),
411        }
412
413        TYPE_CONVERTERS = {
414            # https://duckdb.org/docs/sql/data_types/numeric
415            exp.DataType.Type.DECIMAL: build_default_decimal_type(precision=18, scale=3),
416            # https://duckdb.org/docs/sql/data_types/text
417            exp.DataType.Type.TEXT: lambda dtype: exp.DataType.build("TEXT"),
418        }
419
420        def _parse_table_sample(self, as_modifier: bool = False) -> t.Optional[exp.TableSample]:
421            # https://duckdb.org/docs/sql/samples.html
422            sample = super()._parse_table_sample(as_modifier=as_modifier)
423            if sample and not sample.args.get("method"):
424                if sample.args.get("size"):
425                    sample.set("method", exp.var("RESERVOIR"))
426                else:
427                    sample.set("method", exp.var("SYSTEM"))
428
429            return sample
430
431        def _parse_bracket(
432            self, this: t.Optional[exp.Expression] = None
433        ) -> t.Optional[exp.Expression]:
434            bracket = super()._parse_bracket(this)
435            if isinstance(bracket, exp.Bracket):
436                bracket.set("returns_list_for_maps", True)
437
438            return bracket
439
440        def _parse_map(self) -> exp.ToMap | exp.Map:
441            if self._match(TokenType.L_BRACE, advance=False):
442                return self.expression(exp.ToMap, this=self._parse_bracket())
443
444            args = self._parse_wrapped_csv(self._parse_assignment)
445            return self.expression(exp.Map, keys=seq_get(args, 0), values=seq_get(args, 1))
446
447        def _parse_struct_types(self, type_required: bool = False) -> t.Optional[exp.Expression]:
448            return self._parse_field_def()
449
450        def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]:
451            if len(aggregations) == 1:
452                return super()._pivot_column_names(aggregations)
453            return pivot_column_names(aggregations, dialect="duckdb")
454
455    class Generator(generator.Generator):
456        PARAMETER_TOKEN = "$"
457        NAMED_PLACEHOLDER_TOKEN = "$"
458        JOIN_HINTS = False
459        TABLE_HINTS = False
460        QUERY_HINTS = False
461        LIMIT_FETCH = "LIMIT"
462        STRUCT_DELIMITER = ("(", ")")
463        RENAME_TABLE_WITH_DB = False
464        NVL2_SUPPORTED = False
465        SEMI_ANTI_JOIN_WITH_SIDE = False
466        TABLESAMPLE_KEYWORDS = "USING SAMPLE"
467        TABLESAMPLE_SEED_KEYWORD = "REPEATABLE"
468        LAST_DAY_SUPPORTS_DATE_PART = False
469        JSON_KEY_VALUE_PAIR_SEP = ","
470        IGNORE_NULLS_IN_FUNC = True
471        JSON_PATH_BRACKETED_KEY_SUPPORTED = False
472        SUPPORTS_CREATE_TABLE_LIKE = False
473        MULTI_ARG_DISTINCT = False
474        CAN_IMPLEMENT_ARRAY_ANY = True
475        SUPPORTS_TO_NUMBER = False
476        COPY_HAS_INTO_KEYWORD = False
477        STAR_EXCEPT = "EXCLUDE"
478        PAD_FILL_PATTERN_IS_REQUIRED = True
479        ARRAY_CONCAT_IS_VAR_LEN = False
480
481        TRANSFORMS = {
482            **generator.Generator.TRANSFORMS,
483            exp.ApproxDistinct: approx_count_distinct_sql,
484            exp.Array: inline_array_unless_query,
485            exp.ArrayContainsAll: rename_func("ARRAY_HAS_ALL"),
486            exp.ArrayFilter: rename_func("LIST_FILTER"),
487            exp.ArraySize: rename_func("ARRAY_LENGTH"),
488            exp.ArgMax: arg_max_or_min_no_count("ARG_MAX"),
489            exp.ArgMin: arg_max_or_min_no_count("ARG_MIN"),
490            exp.ArraySort: _array_sort_sql,
491            exp.ArraySum: rename_func("LIST_SUM"),
492            exp.BitwiseXor: rename_func("XOR"),
493            exp.CommentColumnConstraint: no_comment_column_constraint_sql,
494            exp.CurrentDate: lambda *_: "CURRENT_DATE",
495            exp.CurrentTime: lambda *_: "CURRENT_TIME",
496            exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP",
497            exp.DayOfMonth: rename_func("DAYOFMONTH"),
498            exp.DayOfWeek: rename_func("DAYOFWEEK"),
499            exp.DayOfWeekIso: rename_func("ISODOW"),
500            exp.DayOfYear: rename_func("DAYOFYEAR"),
501            exp.DataType: _datatype_sql,
502            exp.Date: _date_sql,
503            exp.DateAdd: _date_delta_sql,
504            exp.DateFromParts: rename_func("MAKE_DATE"),
505            exp.DateSub: _date_delta_sql,
506            exp.DateDiff: _date_diff_sql,
507            exp.DateStrToDate: datestrtodate_sql,
508            exp.Datetime: no_datetime_sql,
509            exp.DatetimeSub: _date_delta_sql,
510            exp.DatetimeAdd: _date_delta_sql,
511            exp.DateToDi: lambda self,
512            e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.DATEINT_FORMAT}) AS INT)",
513            exp.Decode: lambda self, e: encode_decode_sql(self, e, "DECODE", replace=False),
514            exp.DiToDate: lambda self,
515            e: f"CAST(STRPTIME(CAST({self.sql(e, 'this')} AS TEXT), {DuckDB.DATEINT_FORMAT}) AS DATE)",
516            exp.Encode: lambda self, e: encode_decode_sql(self, e, "ENCODE", replace=False),
517            exp.GenerateDateArray: _generate_datetime_array_sql,
518            exp.GenerateTimestampArray: _generate_datetime_array_sql,
519            exp.Explode: rename_func("UNNEST"),
520            exp.IntDiv: lambda self, e: self.binary(e, "//"),
521            exp.IsInf: rename_func("ISINF"),
522            exp.IsNan: rename_func("ISNAN"),
523            exp.JSONExtract: _arrow_json_extract_sql,
524            exp.JSONExtractScalar: _arrow_json_extract_sql,
525            exp.JSONFormat: _json_format_sql,
526            exp.LogicalOr: rename_func("BOOL_OR"),
527            exp.LogicalAnd: rename_func("BOOL_AND"),
528            exp.MD5Digest: lambda self, e: self.func("UNHEX", self.func("MD5", e.this)),
529            exp.MonthsBetween: lambda self, e: self.func(
530                "DATEDIFF",
531                "'month'",
532                exp.cast(e.expression, exp.DataType.Type.TIMESTAMP, copy=True),
533                exp.cast(e.this, exp.DataType.Type.TIMESTAMP, copy=True),
534            ),
535            exp.PercentileCont: rename_func("QUANTILE_CONT"),
536            exp.PercentileDisc: rename_func("QUANTILE_DISC"),
537            # DuckDB doesn't allow qualified columns inside of PIVOT expressions.
538            # See: https://github.com/duckdb/duckdb/blob/671faf92411182f81dce42ac43de8bfb05d9909e/src/planner/binder/tableref/bind_pivot.cpp#L61-L62
539            exp.Pivot: transforms.preprocess([transforms.unqualify_columns]),
540            exp.RegexpExtract: regexp_extract_sql,
541            exp.RegexpReplace: lambda self, e: self.func(
542                "REGEXP_REPLACE",
543                e.this,
544                e.expression,
545                e.args.get("replacement"),
546                e.args.get("modifiers"),
547            ),
548            exp.RegexpLike: rename_func("REGEXP_MATCHES"),
549            exp.RegexpSplit: rename_func("STR_SPLIT_REGEX"),
550            exp.Return: lambda self, e: self.sql(e, "this"),
551            exp.ReturnsProperty: lambda self, e: "TABLE" if isinstance(e.this, exp.Schema) else "",
552            exp.Rand: rename_func("RANDOM"),
553            exp.SafeDivide: no_safe_divide_sql,
554            exp.SHA2: sha256_sql,
555            exp.Split: rename_func("STR_SPLIT"),
556            exp.SortArray: _sort_array_sql,
557            exp.StrPosition: str_position_sql,
558            exp.StrToUnix: lambda self, e: self.func(
559                "EPOCH", self.func("STRPTIME", e.this, self.format_time(e))
560            ),
561            exp.Struct: _struct_sql,
562            exp.Transform: rename_func("LIST_TRANSFORM"),
563            exp.TimeAdd: _date_delta_sql,
564            exp.Time: no_time_sql,
565            exp.TimeDiff: _timediff_sql,
566            exp.Timestamp: no_timestamp_sql,
567            exp.TimestampDiff: lambda self, e: self.func(
568                "DATE_DIFF", exp.Literal.string(e.unit), e.expression, e.this
569            ),
570            exp.TimestampTrunc: timestamptrunc_sql(),
571            exp.TimeStrToDate: lambda self, e: self.sql(exp.cast(e.this, exp.DataType.Type.DATE)),
572            exp.TimeStrToTime: timestrtotime_sql,
573            exp.TimeStrToUnix: lambda self, e: self.func(
574                "EPOCH", exp.cast(e.this, exp.DataType.Type.TIMESTAMP)
575            ),
576            exp.TimeToStr: lambda self, e: self.func("STRFTIME", e.this, self.format_time(e)),
577            exp.TimeToUnix: rename_func("EPOCH"),
578            exp.TsOrDiToDi: lambda self,
579            e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS TEXT), '-', ''), 1, 8) AS INT)",
580            exp.TsOrDsAdd: _date_delta_sql,
581            exp.TsOrDsDiff: lambda self, e: self.func(
582                "DATE_DIFF",
583                f"'{e.args.get('unit') or 'DAY'}'",
584                exp.cast(e.expression, exp.DataType.Type.TIMESTAMP),
585                exp.cast(e.this, exp.DataType.Type.TIMESTAMP),
586            ),
587            exp.UnixToStr: lambda self, e: self.func(
588                "STRFTIME", self.func("TO_TIMESTAMP", e.this), self.format_time(e)
589            ),
590            exp.DatetimeTrunc: lambda self, e: self.func(
591                "DATE_TRUNC", unit_to_str(e), exp.cast(e.this, exp.DataType.Type.DATETIME)
592            ),
593            exp.UnixToTime: _unix_to_time_sql,
594            exp.UnixToTimeStr: lambda self, e: f"CAST(TO_TIMESTAMP({self.sql(e, 'this')}) AS TEXT)",
595            exp.VariancePop: rename_func("VAR_POP"),
596            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
597            exp.Xor: bool_xor_sql,
598        }
599
600        SUPPORTED_JSON_PATH_PARTS = {
601            exp.JSONPathKey,
602            exp.JSONPathRoot,
603            exp.JSONPathSubscript,
604            exp.JSONPathWildcard,
605        }
606
607        TYPE_MAPPING = {
608            **generator.Generator.TYPE_MAPPING,
609            exp.DataType.Type.BINARY: "BLOB",
610            exp.DataType.Type.BPCHAR: "TEXT",
611            exp.DataType.Type.CHAR: "TEXT",
612            exp.DataType.Type.FLOAT: "REAL",
613            exp.DataType.Type.NCHAR: "TEXT",
614            exp.DataType.Type.NVARCHAR: "TEXT",
615            exp.DataType.Type.UINT: "UINTEGER",
616            exp.DataType.Type.VARBINARY: "BLOB",
617            exp.DataType.Type.ROWVERSION: "BLOB",
618            exp.DataType.Type.VARCHAR: "TEXT",
619            exp.DataType.Type.TIMESTAMPNTZ: "TIMESTAMP",
620            exp.DataType.Type.TIMESTAMP_S: "TIMESTAMP_S",
621            exp.DataType.Type.TIMESTAMP_MS: "TIMESTAMP_MS",
622            exp.DataType.Type.TIMESTAMP_NS: "TIMESTAMP_NS",
623        }
624
625        # https://github.com/duckdb/duckdb/blob/ff7f24fd8e3128d94371827523dae85ebaf58713/third_party/libpg_query/grammar/keywords/reserved_keywords.list#L1-L77
626        RESERVED_KEYWORDS = {
627            "array",
628            "analyse",
629            "union",
630            "all",
631            "when",
632            "in_p",
633            "default",
634            "create_p",
635            "window",
636            "asymmetric",
637            "to",
638            "else",
639            "localtime",
640            "from",
641            "end_p",
642            "select",
643            "current_date",
644            "foreign",
645            "with",
646            "grant",
647            "session_user",
648            "or",
649            "except",
650            "references",
651            "fetch",
652            "limit",
653            "group_p",
654            "leading",
655            "into",
656            "collate",
657            "offset",
658            "do",
659            "then",
660            "localtimestamp",
661            "check_p",
662            "lateral_p",
663            "current_role",
664            "where",
665            "asc_p",
666            "placing",
667            "desc_p",
668            "user",
669            "unique",
670            "initially",
671            "column",
672            "both",
673            "some",
674            "as",
675            "any",
676            "only",
677            "deferrable",
678            "null_p",
679            "current_time",
680            "true_p",
681            "table",
682            "case",
683            "trailing",
684            "variadic",
685            "for",
686            "on",
687            "distinct",
688            "false_p",
689            "not",
690            "constraint",
691            "current_timestamp",
692            "returning",
693            "primary",
694            "intersect",
695            "having",
696            "analyze",
697            "current_user",
698            "and",
699            "cast",
700            "symmetric",
701            "using",
702            "order",
703            "current_catalog",
704        }
705
706        UNWRAPPED_INTERVAL_VALUES = (exp.Literal, exp.Paren)
707
708        # DuckDB doesn't generally support CREATE TABLE .. properties
709        # https://duckdb.org/docs/sql/statements/create_table.html
710        PROPERTIES_LOCATION = {
711            prop: exp.Properties.Location.UNSUPPORTED
712            for prop in generator.Generator.PROPERTIES_LOCATION
713        }
714
715        # There are a few exceptions (e.g. temporary tables) which are supported or
716        # can be transpiled to DuckDB, so we explicitly override them accordingly
717        PROPERTIES_LOCATION[exp.LikeProperty] = exp.Properties.Location.POST_SCHEMA
718        PROPERTIES_LOCATION[exp.TemporaryProperty] = exp.Properties.Location.POST_CREATE
719        PROPERTIES_LOCATION[exp.ReturnsProperty] = exp.Properties.Location.POST_ALIAS
720
721        def fromiso8601timestamp_sql(self, expression: exp.FromISO8601Timestamp) -> str:
722            return self.sql(exp.cast(expression.this, exp.DataType.Type.TIMESTAMPTZ))
723
724        def strtotime_sql(self, expression: exp.StrToTime) -> str:
725            if expression.args.get("safe"):
726                formatted_time = self.format_time(expression)
727                return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS TIMESTAMP)"
728            return str_to_time_sql(self, expression)
729
730        def strtodate_sql(self, expression: exp.StrToDate) -> str:
731            if expression.args.get("safe"):
732                formatted_time = self.format_time(expression)
733                return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS DATE)"
734            return f"CAST({str_to_time_sql(self, expression)} AS DATE)"
735
736        def parsejson_sql(self, expression: exp.ParseJSON) -> str:
737            arg = expression.this
738            if expression.args.get("safe"):
739                return self.sql(exp.case().when(exp.func("json_valid", arg), arg).else_(exp.null()))
740            return self.func("JSON", arg)
741
742        def timefromparts_sql(self, expression: exp.TimeFromParts) -> str:
743            nano = expression.args.get("nano")
744            if nano is not None:
745                expression.set(
746                    "sec", expression.args["sec"] + nano.pop() / exp.Literal.number(1000000000.0)
747                )
748
749            return rename_func("MAKE_TIME")(self, expression)
750
751        def timestampfromparts_sql(self, expression: exp.TimestampFromParts) -> str:
752            sec = expression.args["sec"]
753
754            milli = expression.args.get("milli")
755            if milli is not None:
756                sec += milli.pop() / exp.Literal.number(1000.0)
757
758            nano = expression.args.get("nano")
759            if nano is not None:
760                sec += nano.pop() / exp.Literal.number(1000000000.0)
761
762            if milli or nano:
763                expression.set("sec", sec)
764
765            return rename_func("MAKE_TIMESTAMP")(self, expression)
766
767        def tablesample_sql(
768            self,
769            expression: exp.TableSample,
770            tablesample_keyword: t.Optional[str] = None,
771        ) -> str:
772            if not isinstance(expression.parent, exp.Select):
773                # This sample clause only applies to a single source, not the entire resulting relation
774                tablesample_keyword = "TABLESAMPLE"
775
776            if expression.args.get("size"):
777                method = expression.args.get("method")
778                if method and method.name.upper() != "RESERVOIR":
779                    self.unsupported(
780                        f"Sampling method {method} is not supported with a discrete sample count, "
781                        "defaulting to reservoir sampling"
782                    )
783                    expression.set("method", exp.var("RESERVOIR"))
784
785            return super().tablesample_sql(expression, tablesample_keyword=tablesample_keyword)
786
787        def interval_sql(self, expression: exp.Interval) -> str:
788            multiplier: t.Optional[int] = None
789            unit = expression.text("unit").lower()
790
791            if unit.startswith("week"):
792                multiplier = 7
793            if unit.startswith("quarter"):
794                multiplier = 90
795
796            if multiplier:
797                return f"({multiplier} * {super().interval_sql(exp.Interval(this=expression.this, unit=exp.var('DAY')))})"
798
799            return super().interval_sql(expression)
800
801        def columndef_sql(self, expression: exp.ColumnDef, sep: str = " ") -> str:
802            if isinstance(expression.parent, exp.UserDefinedFunction):
803                return self.sql(expression, "this")
804            return super().columndef_sql(expression, sep)
805
806        def join_sql(self, expression: exp.Join) -> str:
807            if (
808                expression.side == "LEFT"
809                and not expression.args.get("on")
810                and isinstance(expression.this, exp.Unnest)
811            ):
812                # Some dialects support `LEFT JOIN UNNEST(...)` without an explicit ON clause
813                # DuckDB doesn't, but we can just add a dummy ON clause that is always true
814                return super().join_sql(expression.on(exp.true()))
815
816            return super().join_sql(expression)
817
818        def generateseries_sql(self, expression: exp.GenerateSeries) -> str:
819            # GENERATE_SERIES(a, b) -> [a, b], RANGE(a, b) -> [a, b)
820            if expression.args.get("is_end_exclusive"):
821                return rename_func("RANGE")(self, expression)
822
823            return self.function_fallback_sql(expression)
824
825        def bracket_sql(self, expression: exp.Bracket) -> str:
826            this = expression.this
827            if isinstance(this, exp.Array):
828                this.replace(exp.paren(this))
829
830            bracket = super().bracket_sql(expression)
831
832            if not expression.args.get("returns_list_for_maps"):
833                if not this.type:
834                    from sqlglot.optimizer.annotate_types import annotate_types
835
836                    this = annotate_types(this)
837
838                if this.is_type(exp.DataType.Type.MAP):
839                    bracket = f"({bracket})[1]"
840
841            return bracket
842
843        def withingroup_sql(self, expression: exp.WithinGroup) -> str:
844            expression_sql = self.sql(expression, "expression")
845
846            func = expression.this
847            if isinstance(func, exp.PERCENTILES):
848                # Make the order key the first arg and slide the fraction to the right
849                # https://duckdb.org/docs/sql/aggregates#ordered-set-aggregate-functions
850                order_col = expression.find(exp.Ordered)
851                if order_col:
852                    func.set("expression", func.this)
853                    func.set("this", order_col.this)
854
855            this = self.sql(expression, "this").rstrip(")")
856
857            return f"{this}{expression_sql})"
858
859        def length_sql(self, expression: exp.Length) -> str:
860            arg = expression.this
861
862            # Dialects like BQ and Snowflake also accept binary values as args, so
863            # DDB will attempt to infer the type or resort to case/when resolution
864            if not expression.args.get("binary") or arg.is_string:
865                return self.func("LENGTH", arg)
866
867            if not arg.type:
868                from sqlglot.optimizer.annotate_types import annotate_types
869
870                arg = annotate_types(arg)
871
872            if arg.is_type(*exp.DataType.TEXT_TYPES):
873                return self.func("LENGTH", arg)
874
875            # We need these casts to make duckdb's static type checker happy
876            blob = exp.cast(arg, exp.DataType.Type.VARBINARY)
877            varchar = exp.cast(arg, exp.DataType.Type.VARCHAR)
878
879            case = (
880                exp.case(self.func("TYPEOF", arg))
881                .when(
882                    "'VARCHAR'", exp.Anonymous(this="LENGTH", expressions=[varchar])
883                )  # anonymous to break length_sql recursion
884                .when("'BLOB'", self.func("OCTET_LENGTH", blob))
885            )
886
887            return self.sql(case)
888
889        def objectinsert_sql(self, expression: exp.ObjectInsert) -> str:
890            this = expression.this
891            key = expression.args.get("key")
892            key_sql = key.name if isinstance(key, exp.Expression) else ""
893            value_sql = self.sql(expression, "value")
894
895            kv_sql = f"{key_sql} := {value_sql}"
896
897            # If the input struct is empty e.g. transpiling OBJECT_INSERT(OBJECT_CONSTRUCT(), key, value) from Snowflake
898            # then we can generate STRUCT_PACK which will build it since STRUCT_INSERT({}, key := value) is not valid DuckDB
899            if isinstance(this, exp.Struct) and not this.expressions:
900                return self.func("STRUCT_PACK", kv_sql)
901
902            return self.func("STRUCT_INSERT", this, kv_sql)
903
904        def unnest_sql(self, expression: exp.Unnest) -> str:
905            explode_array = expression.args.get("explode_array")
906            if explode_array:
907                # In BigQuery, UNNESTing a nested array leads to explosion of the top-level array & struct
908                # This is transpiled to DDB by transforming "FROM UNNEST(...)" to "FROM (SELECT UNNEST(..., max_depth => 2))"
909                expression.expressions.append(
910                    exp.Kwarg(this=exp.var("max_depth"), expression=exp.Literal.number(2))
911                )
912
913                # If BQ's UNNEST is aliased, we transform it from a column alias to a table alias in DDB
914                alias = expression.args.get("alias")
915                if alias:
916                    expression.set("alias", None)
917                    alias = exp.TableAlias(this=seq_get(alias.args.get("columns"), 0))
918
919                unnest_sql = super().unnest_sql(expression)
920                select = exp.Select(expressions=[unnest_sql]).subquery(alias)
921                return self.sql(select)
922
923            return super().unnest_sql(expression)
924
925        def ignorenulls_sql(self, expression: exp.IgnoreNulls) -> str:
926            if isinstance(expression.this, WINDOW_FUNCS_WITH_IGNORE_NULLS):
927                # DuckDB should render IGNORE NULLS only for the general-purpose
928                # window functions that accept it e.g. FIRST_VALUE(... IGNORE NULLS) OVER (...)
929                return super().ignorenulls_sql(expression)
930
931            return self.sql(expression, "this")
932
933        def arraytostring_sql(self, expression: exp.ArrayToString) -> str:
934            this = self.sql(expression, "this")
935            null_text = self.sql(expression, "null")
936
937            if null_text:
938                this = f"LIST_TRANSFORM({this}, x -> COALESCE(x, {null_text}))"
939
940            return self.func("ARRAY_TO_STRING", this, expression.expression)