Edit on GitHub

sqlglot.dialects.duckdb

  1from __future__ import annotations
  2
  3import typing as t
  4
  5from sqlglot import exp, generator, parser, tokens
  6from sqlglot.dialects.dialect import (
  7    Dialect,
  8    approx_count_distinct_sql,
  9    arrow_json_extract_scalar_sql,
 10    arrow_json_extract_sql,
 11    datestrtodate_sql,
 12    format_time_lambda,
 13    no_comment_column_constraint_sql,
 14    no_properties_sql,
 15    no_safe_divide_sql,
 16    pivot_column_names,
 17    rename_func,
 18    str_position_sql,
 19    str_to_time_sql,
 20    timestamptrunc_sql,
 21    timestrtotime_sql,
 22    ts_or_ds_to_date_sql,
 23)
 24from sqlglot.helper import seq_get
 25from sqlglot.tokens import TokenType
 26
 27
 28def _ts_or_ds_add_sql(self: generator.Generator, expression: exp.TsOrDsAdd) -> str:
 29    this = self.sql(expression, "this")
 30    unit = self.sql(expression, "unit").strip("'") or "DAY"
 31    return f"CAST({this} AS DATE) + {self.sql(exp.Interval(this=expression.expression, unit=unit))}"
 32
 33
 34def _date_delta_sql(self: generator.Generator, expression: exp.DateAdd | exp.DateSub) -> str:
 35    this = self.sql(expression, "this")
 36    unit = self.sql(expression, "unit").strip("'") or "DAY"
 37    op = "+" if isinstance(expression, exp.DateAdd) else "-"
 38    return f"{this} {op} {self.sql(exp.Interval(this=expression.expression, unit=unit))}"
 39
 40
 41def _array_sort_sql(self: generator.Generator, expression: exp.ArraySort) -> str:
 42    if expression.expression:
 43        self.unsupported("DUCKDB ARRAY_SORT does not support a comparator")
 44    return f"ARRAY_SORT({self.sql(expression, 'this')})"
 45
 46
 47def _sort_array_sql(self: generator.Generator, expression: exp.SortArray) -> str:
 48    this = self.sql(expression, "this")
 49    if expression.args.get("asc") == exp.false():
 50        return f"ARRAY_REVERSE_SORT({this})"
 51    return f"ARRAY_SORT({this})"
 52
 53
 54def _sort_array_reverse(args: t.List) -> exp.Expression:
 55    return exp.SortArray(this=seq_get(args, 0), asc=exp.false())
 56
 57
 58def _parse_date_diff(args: t.List) -> exp.Expression:
 59    return exp.DateDiff(
 60        this=seq_get(args, 2),
 61        expression=seq_get(args, 1),
 62        unit=seq_get(args, 0),
 63    )
 64
 65
 66def _struct_sql(self: generator.Generator, expression: exp.Struct) -> str:
 67    args = [
 68        f"'{e.name or e.this.name}': {self.sql(e, 'expression')}" for e in expression.expressions
 69    ]
 70    return f"{{{', '.join(args)}}}"
 71
 72
 73def _datatype_sql(self: generator.Generator, expression: exp.DataType) -> str:
 74    if expression.is_type("array"):
 75        return f"{self.expressions(expression, flat=True)}[]"
 76    return self.datatype_sql(expression)
 77
 78
 79def _regexp_extract_sql(self: generator.Generator, expression: exp.RegexpExtract) -> str:
 80    bad_args = list(filter(expression.args.get, ("position", "occurrence")))
 81    if bad_args:
 82        self.unsupported(f"REGEXP_EXTRACT does not support arg(s) {bad_args}")
 83
 84    return self.func(
 85        "REGEXP_EXTRACT",
 86        expression.args.get("this"),
 87        expression.args.get("expression"),
 88        expression.args.get("group"),
 89    )
 90
 91
 92class DuckDB(Dialect):
 93    null_ordering = "nulls_are_last"
 94
 95    class Tokenizer(tokens.Tokenizer):
 96        KEYWORDS = {
 97            **tokens.Tokenizer.KEYWORDS,
 98            "~": TokenType.RLIKE,
 99            ":=": TokenType.EQ,
100            "//": TokenType.DIV,
101            "ATTACH": TokenType.COMMAND,
102            "BINARY": TokenType.VARBINARY,
103            "BPCHAR": TokenType.TEXT,
104            "BITSTRING": TokenType.BIT,
105            "CHAR": TokenType.TEXT,
106            "CHARACTER VARYING": TokenType.TEXT,
107            "EXCLUDE": TokenType.EXCEPT,
108            "INT1": TokenType.TINYINT,
109            "LOGICAL": TokenType.BOOLEAN,
110            "NUMERIC": TokenType.DOUBLE,
111            "PIVOT_WIDER": TokenType.PIVOT,
112            "SIGNED": TokenType.INT,
113            "STRING": TokenType.VARCHAR,
114            "UBIGINT": TokenType.UBIGINT,
115            "UINTEGER": TokenType.UINT,
116            "USMALLINT": TokenType.USMALLINT,
117            "UTINYINT": TokenType.UTINYINT,
118        }
119
120    class Parser(parser.Parser):
121        FUNCTIONS = {
122            **parser.Parser.FUNCTIONS,
123            "ARRAY_LENGTH": exp.ArraySize.from_arg_list,
124            "ARRAY_SORT": exp.SortArray.from_arg_list,
125            "ARRAY_REVERSE_SORT": _sort_array_reverse,
126            "DATEDIFF": _parse_date_diff,
127            "DATE_DIFF": _parse_date_diff,
128            "EPOCH": exp.TimeToUnix.from_arg_list,
129            "EPOCH_MS": lambda args: exp.UnixToTime(
130                this=exp.Div(
131                    this=seq_get(args, 0),
132                    expression=exp.Literal.number(1000),
133                )
134            ),
135            "LIST_REVERSE_SORT": _sort_array_reverse,
136            "LIST_SORT": exp.SortArray.from_arg_list,
137            "LIST_VALUE": exp.Array.from_arg_list,
138            "REGEXP_MATCHES": exp.RegexpLike.from_arg_list,
139            "STRFTIME": format_time_lambda(exp.TimeToStr, "duckdb"),
140            "STRING_SPLIT": exp.Split.from_arg_list,
141            "STRING_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
142            "STRING_TO_ARRAY": exp.Split.from_arg_list,
143            "STRPTIME": format_time_lambda(exp.StrToTime, "duckdb"),
144            "STRUCT_PACK": exp.Struct.from_arg_list,
145            "STR_SPLIT": exp.Split.from_arg_list,
146            "STR_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
147            "TO_TIMESTAMP": exp.UnixToTime.from_arg_list,
148            "UNNEST": exp.Explode.from_arg_list,
149        }
150
151        TYPE_TOKENS = {
152            *parser.Parser.TYPE_TOKENS,
153            TokenType.UBIGINT,
154            TokenType.UINT,
155            TokenType.USMALLINT,
156            TokenType.UTINYINT,
157        }
158
159        def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]:
160            if len(aggregations) == 1:
161                return super()._pivot_column_names(aggregations)
162            return pivot_column_names(aggregations, dialect="duckdb")
163
164    class Generator(generator.Generator):
165        JOIN_HINTS = False
166        TABLE_HINTS = False
167        LIMIT_FETCH = "LIMIT"
168        STRUCT_DELIMITER = ("(", ")")
169        RENAME_TABLE_WITH_DB = False
170
171        TRANSFORMS = {
172            **generator.Generator.TRANSFORMS,
173            exp.ApproxDistinct: approx_count_distinct_sql,
174            exp.Array: lambda self, e: self.func("ARRAY", e.expressions[0])
175            if isinstance(seq_get(e.expressions, 0), exp.Select)
176            else rename_func("LIST_VALUE")(self, e),
177            exp.ArraySize: rename_func("ARRAY_LENGTH"),
178            exp.ArraySort: _array_sort_sql,
179            exp.ArraySum: rename_func("LIST_SUM"),
180            exp.CommentColumnConstraint: no_comment_column_constraint_sql,
181            exp.CurrentDate: lambda self, e: "CURRENT_DATE",
182            exp.CurrentTime: lambda self, e: "CURRENT_TIME",
183            exp.CurrentTimestamp: lambda self, e: "CURRENT_TIMESTAMP",
184            exp.DayOfMonth: rename_func("DAYOFMONTH"),
185            exp.DayOfWeek: rename_func("DAYOFWEEK"),
186            exp.DayOfYear: rename_func("DAYOFYEAR"),
187            exp.DataType: _datatype_sql,
188            exp.DateAdd: _date_delta_sql,
189            exp.DateSub: _date_delta_sql,
190            exp.DateDiff: lambda self, e: self.func(
191                "DATE_DIFF", f"'{e.args.get('unit', 'day')}'", e.expression, e.this
192            ),
193            exp.DateStrToDate: datestrtodate_sql,
194            exp.DateToDi: lambda self, e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.dateint_format}) AS INT)",
195            exp.DiToDate: lambda self, e: f"CAST(STRPTIME(CAST({self.sql(e, 'this')} AS TEXT), {DuckDB.dateint_format}) AS DATE)",
196            exp.Explode: rename_func("UNNEST"),
197            exp.IntDiv: lambda self, e: self.binary(e, "//"),
198            exp.JSONExtract: arrow_json_extract_sql,
199            exp.JSONExtractScalar: arrow_json_extract_scalar_sql,
200            exp.JSONBExtract: arrow_json_extract_sql,
201            exp.JSONBExtractScalar: arrow_json_extract_scalar_sql,
202            exp.LogicalOr: rename_func("BOOL_OR"),
203            exp.LogicalAnd: rename_func("BOOL_AND"),
204            exp.Properties: no_properties_sql,
205            exp.RegexpExtract: _regexp_extract_sql,
206            exp.RegexpLike: rename_func("REGEXP_MATCHES"),
207            exp.RegexpSplit: rename_func("STR_SPLIT_REGEX"),
208            exp.SafeDivide: no_safe_divide_sql,
209            exp.Split: rename_func("STR_SPLIT"),
210            exp.SortArray: _sort_array_sql,
211            exp.StrPosition: str_position_sql,
212            exp.StrToDate: lambda self, e: f"CAST({str_to_time_sql(self, e)} AS DATE)",
213            exp.StrToTime: str_to_time_sql,
214            exp.StrToUnix: lambda self, e: f"EPOCH(STRPTIME({self.sql(e, 'this')}, {self.format_time(e)}))",
215            exp.Struct: _struct_sql,
216            exp.TimestampTrunc: timestamptrunc_sql,
217            exp.TimeStrToDate: lambda self, e: f"CAST({self.sql(e, 'this')} AS DATE)",
218            exp.TimeStrToTime: timestrtotime_sql,
219            exp.TimeStrToUnix: lambda self, e: f"EPOCH(CAST({self.sql(e, 'this')} AS TIMESTAMP))",
220            exp.TimeToStr: lambda self, e: f"STRFTIME({self.sql(e, 'this')}, {self.format_time(e)})",
221            exp.TimeToUnix: rename_func("EPOCH"),
222            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS TEXT), '-', ''), 1, 8) AS INT)",
223            exp.TsOrDsAdd: _ts_or_ds_add_sql,
224            exp.TsOrDsToDate: ts_or_ds_to_date_sql("duckdb"),
225            exp.UnixToStr: lambda self, e: f"STRFTIME(TO_TIMESTAMP({self.sql(e, 'this')}), {self.format_time(e)})",
226            exp.UnixToTime: rename_func("TO_TIMESTAMP"),
227            exp.UnixToTimeStr: lambda self, e: f"CAST(TO_TIMESTAMP({self.sql(e, 'this')}) AS TEXT)",
228            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
229        }
230
231        TYPE_MAPPING = {
232            **generator.Generator.TYPE_MAPPING,
233            exp.DataType.Type.BINARY: "BLOB",
234            exp.DataType.Type.CHAR: "TEXT",
235            exp.DataType.Type.FLOAT: "REAL",
236            exp.DataType.Type.NCHAR: "TEXT",
237            exp.DataType.Type.NVARCHAR: "TEXT",
238            exp.DataType.Type.UINT: "UINTEGER",
239            exp.DataType.Type.VARBINARY: "BLOB",
240            exp.DataType.Type.VARCHAR: "TEXT",
241        }
242
243        STAR_MAPPING = {**generator.Generator.STAR_MAPPING, "except": "EXCLUDE"}
244
245        PROPERTIES_LOCATION = {
246            **generator.Generator.PROPERTIES_LOCATION,
247            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
248        }
249
250        def tablesample_sql(
251            self, expression: exp.TableSample, seed_prefix: str = "SEED", sep: str = " AS "
252        ) -> str:
253            return super().tablesample_sql(expression, seed_prefix="REPEATABLE", sep=sep)
class DuckDB(sqlglot.dialects.dialect.Dialect):
 93class DuckDB(Dialect):
 94    null_ordering = "nulls_are_last"
 95
 96    class Tokenizer(tokens.Tokenizer):
 97        KEYWORDS = {
 98            **tokens.Tokenizer.KEYWORDS,
 99            "~": TokenType.RLIKE,
100            ":=": TokenType.EQ,
101            "//": TokenType.DIV,
102            "ATTACH": TokenType.COMMAND,
103            "BINARY": TokenType.VARBINARY,
104            "BPCHAR": TokenType.TEXT,
105            "BITSTRING": TokenType.BIT,
106            "CHAR": TokenType.TEXT,
107            "CHARACTER VARYING": TokenType.TEXT,
108            "EXCLUDE": TokenType.EXCEPT,
109            "INT1": TokenType.TINYINT,
110            "LOGICAL": TokenType.BOOLEAN,
111            "NUMERIC": TokenType.DOUBLE,
112            "PIVOT_WIDER": TokenType.PIVOT,
113            "SIGNED": TokenType.INT,
114            "STRING": TokenType.VARCHAR,
115            "UBIGINT": TokenType.UBIGINT,
116            "UINTEGER": TokenType.UINT,
117            "USMALLINT": TokenType.USMALLINT,
118            "UTINYINT": TokenType.UTINYINT,
119        }
120
121    class Parser(parser.Parser):
122        FUNCTIONS = {
123            **parser.Parser.FUNCTIONS,
124            "ARRAY_LENGTH": exp.ArraySize.from_arg_list,
125            "ARRAY_SORT": exp.SortArray.from_arg_list,
126            "ARRAY_REVERSE_SORT": _sort_array_reverse,
127            "DATEDIFF": _parse_date_diff,
128            "DATE_DIFF": _parse_date_diff,
129            "EPOCH": exp.TimeToUnix.from_arg_list,
130            "EPOCH_MS": lambda args: exp.UnixToTime(
131                this=exp.Div(
132                    this=seq_get(args, 0),
133                    expression=exp.Literal.number(1000),
134                )
135            ),
136            "LIST_REVERSE_SORT": _sort_array_reverse,
137            "LIST_SORT": exp.SortArray.from_arg_list,
138            "LIST_VALUE": exp.Array.from_arg_list,
139            "REGEXP_MATCHES": exp.RegexpLike.from_arg_list,
140            "STRFTIME": format_time_lambda(exp.TimeToStr, "duckdb"),
141            "STRING_SPLIT": exp.Split.from_arg_list,
142            "STRING_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
143            "STRING_TO_ARRAY": exp.Split.from_arg_list,
144            "STRPTIME": format_time_lambda(exp.StrToTime, "duckdb"),
145            "STRUCT_PACK": exp.Struct.from_arg_list,
146            "STR_SPLIT": exp.Split.from_arg_list,
147            "STR_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
148            "TO_TIMESTAMP": exp.UnixToTime.from_arg_list,
149            "UNNEST": exp.Explode.from_arg_list,
150        }
151
152        TYPE_TOKENS = {
153            *parser.Parser.TYPE_TOKENS,
154            TokenType.UBIGINT,
155            TokenType.UINT,
156            TokenType.USMALLINT,
157            TokenType.UTINYINT,
158        }
159
160        def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]:
161            if len(aggregations) == 1:
162                return super()._pivot_column_names(aggregations)
163            return pivot_column_names(aggregations, dialect="duckdb")
164
165    class Generator(generator.Generator):
166        JOIN_HINTS = False
167        TABLE_HINTS = False
168        LIMIT_FETCH = "LIMIT"
169        STRUCT_DELIMITER = ("(", ")")
170        RENAME_TABLE_WITH_DB = False
171
172        TRANSFORMS = {
173            **generator.Generator.TRANSFORMS,
174            exp.ApproxDistinct: approx_count_distinct_sql,
175            exp.Array: lambda self, e: self.func("ARRAY", e.expressions[0])
176            if isinstance(seq_get(e.expressions, 0), exp.Select)
177            else rename_func("LIST_VALUE")(self, e),
178            exp.ArraySize: rename_func("ARRAY_LENGTH"),
179            exp.ArraySort: _array_sort_sql,
180            exp.ArraySum: rename_func("LIST_SUM"),
181            exp.CommentColumnConstraint: no_comment_column_constraint_sql,
182            exp.CurrentDate: lambda self, e: "CURRENT_DATE",
183            exp.CurrentTime: lambda self, e: "CURRENT_TIME",
184            exp.CurrentTimestamp: lambda self, e: "CURRENT_TIMESTAMP",
185            exp.DayOfMonth: rename_func("DAYOFMONTH"),
186            exp.DayOfWeek: rename_func("DAYOFWEEK"),
187            exp.DayOfYear: rename_func("DAYOFYEAR"),
188            exp.DataType: _datatype_sql,
189            exp.DateAdd: _date_delta_sql,
190            exp.DateSub: _date_delta_sql,
191            exp.DateDiff: lambda self, e: self.func(
192                "DATE_DIFF", f"'{e.args.get('unit', 'day')}'", e.expression, e.this
193            ),
194            exp.DateStrToDate: datestrtodate_sql,
195            exp.DateToDi: lambda self, e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.dateint_format}) AS INT)",
196            exp.DiToDate: lambda self, e: f"CAST(STRPTIME(CAST({self.sql(e, 'this')} AS TEXT), {DuckDB.dateint_format}) AS DATE)",
197            exp.Explode: rename_func("UNNEST"),
198            exp.IntDiv: lambda self, e: self.binary(e, "//"),
199            exp.JSONExtract: arrow_json_extract_sql,
200            exp.JSONExtractScalar: arrow_json_extract_scalar_sql,
201            exp.JSONBExtract: arrow_json_extract_sql,
202            exp.JSONBExtractScalar: arrow_json_extract_scalar_sql,
203            exp.LogicalOr: rename_func("BOOL_OR"),
204            exp.LogicalAnd: rename_func("BOOL_AND"),
205            exp.Properties: no_properties_sql,
206            exp.RegexpExtract: _regexp_extract_sql,
207            exp.RegexpLike: rename_func("REGEXP_MATCHES"),
208            exp.RegexpSplit: rename_func("STR_SPLIT_REGEX"),
209            exp.SafeDivide: no_safe_divide_sql,
210            exp.Split: rename_func("STR_SPLIT"),
211            exp.SortArray: _sort_array_sql,
212            exp.StrPosition: str_position_sql,
213            exp.StrToDate: lambda self, e: f"CAST({str_to_time_sql(self, e)} AS DATE)",
214            exp.StrToTime: str_to_time_sql,
215            exp.StrToUnix: lambda self, e: f"EPOCH(STRPTIME({self.sql(e, 'this')}, {self.format_time(e)}))",
216            exp.Struct: _struct_sql,
217            exp.TimestampTrunc: timestamptrunc_sql,
218            exp.TimeStrToDate: lambda self, e: f"CAST({self.sql(e, 'this')} AS DATE)",
219            exp.TimeStrToTime: timestrtotime_sql,
220            exp.TimeStrToUnix: lambda self, e: f"EPOCH(CAST({self.sql(e, 'this')} AS TIMESTAMP))",
221            exp.TimeToStr: lambda self, e: f"STRFTIME({self.sql(e, 'this')}, {self.format_time(e)})",
222            exp.TimeToUnix: rename_func("EPOCH"),
223            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS TEXT), '-', ''), 1, 8) AS INT)",
224            exp.TsOrDsAdd: _ts_or_ds_add_sql,
225            exp.TsOrDsToDate: ts_or_ds_to_date_sql("duckdb"),
226            exp.UnixToStr: lambda self, e: f"STRFTIME(TO_TIMESTAMP({self.sql(e, 'this')}), {self.format_time(e)})",
227            exp.UnixToTime: rename_func("TO_TIMESTAMP"),
228            exp.UnixToTimeStr: lambda self, e: f"CAST(TO_TIMESTAMP({self.sql(e, 'this')}) AS TEXT)",
229            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
230        }
231
232        TYPE_MAPPING = {
233            **generator.Generator.TYPE_MAPPING,
234            exp.DataType.Type.BINARY: "BLOB",
235            exp.DataType.Type.CHAR: "TEXT",
236            exp.DataType.Type.FLOAT: "REAL",
237            exp.DataType.Type.NCHAR: "TEXT",
238            exp.DataType.Type.NVARCHAR: "TEXT",
239            exp.DataType.Type.UINT: "UINTEGER",
240            exp.DataType.Type.VARBINARY: "BLOB",
241            exp.DataType.Type.VARCHAR: "TEXT",
242        }
243
244        STAR_MAPPING = {**generator.Generator.STAR_MAPPING, "except": "EXCLUDE"}
245
246        PROPERTIES_LOCATION = {
247            **generator.Generator.PROPERTIES_LOCATION,
248            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
249        }
250
251        def tablesample_sql(
252            self, expression: exp.TableSample, seed_prefix: str = "SEED", sep: str = " AS "
253        ) -> str:
254            return super().tablesample_sql(expression, seed_prefix="REPEATABLE", sep=sep)
class DuckDB.Tokenizer(sqlglot.tokens.Tokenizer):
 96    class Tokenizer(tokens.Tokenizer):
 97        KEYWORDS = {
 98            **tokens.Tokenizer.KEYWORDS,
 99            "~": TokenType.RLIKE,
100            ":=": TokenType.EQ,
101            "//": TokenType.DIV,
102            "ATTACH": TokenType.COMMAND,
103            "BINARY": TokenType.VARBINARY,
104            "BPCHAR": TokenType.TEXT,
105            "BITSTRING": TokenType.BIT,
106            "CHAR": TokenType.TEXT,
107            "CHARACTER VARYING": TokenType.TEXT,
108            "EXCLUDE": TokenType.EXCEPT,
109            "INT1": TokenType.TINYINT,
110            "LOGICAL": TokenType.BOOLEAN,
111            "NUMERIC": TokenType.DOUBLE,
112            "PIVOT_WIDER": TokenType.PIVOT,
113            "SIGNED": TokenType.INT,
114            "STRING": TokenType.VARCHAR,
115            "UBIGINT": TokenType.UBIGINT,
116            "UINTEGER": TokenType.UINT,
117            "USMALLINT": TokenType.USMALLINT,
118            "UTINYINT": TokenType.UTINYINT,
119        }
class DuckDB.Parser(sqlglot.parser.Parser):
121    class Parser(parser.Parser):
122        FUNCTIONS = {
123            **parser.Parser.FUNCTIONS,
124            "ARRAY_LENGTH": exp.ArraySize.from_arg_list,
125            "ARRAY_SORT": exp.SortArray.from_arg_list,
126            "ARRAY_REVERSE_SORT": _sort_array_reverse,
127            "DATEDIFF": _parse_date_diff,
128            "DATE_DIFF": _parse_date_diff,
129            "EPOCH": exp.TimeToUnix.from_arg_list,
130            "EPOCH_MS": lambda args: exp.UnixToTime(
131                this=exp.Div(
132                    this=seq_get(args, 0),
133                    expression=exp.Literal.number(1000),
134                )
135            ),
136            "LIST_REVERSE_SORT": _sort_array_reverse,
137            "LIST_SORT": exp.SortArray.from_arg_list,
138            "LIST_VALUE": exp.Array.from_arg_list,
139            "REGEXP_MATCHES": exp.RegexpLike.from_arg_list,
140            "STRFTIME": format_time_lambda(exp.TimeToStr, "duckdb"),
141            "STRING_SPLIT": exp.Split.from_arg_list,
142            "STRING_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
143            "STRING_TO_ARRAY": exp.Split.from_arg_list,
144            "STRPTIME": format_time_lambda(exp.StrToTime, "duckdb"),
145            "STRUCT_PACK": exp.Struct.from_arg_list,
146            "STR_SPLIT": exp.Split.from_arg_list,
147            "STR_SPLIT_REGEX": exp.RegexpSplit.from_arg_list,
148            "TO_TIMESTAMP": exp.UnixToTime.from_arg_list,
149            "UNNEST": exp.Explode.from_arg_list,
150        }
151
152        TYPE_TOKENS = {
153            *parser.Parser.TYPE_TOKENS,
154            TokenType.UBIGINT,
155            TokenType.UINT,
156            TokenType.USMALLINT,
157            TokenType.UTINYINT,
158        }
159
160        def _pivot_column_names(self, aggregations: t.List[exp.Expression]) -> t.List[str]:
161            if len(aggregations) == 1:
162                return super()._pivot_column_names(aggregations)
163            return pivot_column_names(aggregations, dialect="duckdb")

Parser consumes a list of tokens produced by the sqlglot.tokens.Tokenizer and produces a parsed syntax tree.

Arguments:
  • error_level: the desired error level. Default: ErrorLevel.IMMEDIATE
  • error_message_context: determines the amount of context to capture from a query string when displaying the error message (in number of characters). Default: 50.
  • index_offset: Index offset for arrays eg ARRAY[0] vs ARRAY[1] as the head of a list. Default: 0
  • alias_post_tablesample: If the table alias comes after tablesample. Default: False
  • max_errors: Maximum number of error messages to include in a raised ParseError. This is only relevant if error_level is ErrorLevel.RAISE. Default: 3
  • null_ordering: Indicates the default null ordering method to use if not explicitly set. Options are "nulls_are_small", "nulls_are_large", "nulls_are_last". Default: "nulls_are_small"
class DuckDB.Generator(sqlglot.generator.Generator):
165    class Generator(generator.Generator):
166        JOIN_HINTS = False
167        TABLE_HINTS = False
168        LIMIT_FETCH = "LIMIT"
169        STRUCT_DELIMITER = ("(", ")")
170        RENAME_TABLE_WITH_DB = False
171
172        TRANSFORMS = {
173            **generator.Generator.TRANSFORMS,
174            exp.ApproxDistinct: approx_count_distinct_sql,
175            exp.Array: lambda self, e: self.func("ARRAY", e.expressions[0])
176            if isinstance(seq_get(e.expressions, 0), exp.Select)
177            else rename_func("LIST_VALUE")(self, e),
178            exp.ArraySize: rename_func("ARRAY_LENGTH"),
179            exp.ArraySort: _array_sort_sql,
180            exp.ArraySum: rename_func("LIST_SUM"),
181            exp.CommentColumnConstraint: no_comment_column_constraint_sql,
182            exp.CurrentDate: lambda self, e: "CURRENT_DATE",
183            exp.CurrentTime: lambda self, e: "CURRENT_TIME",
184            exp.CurrentTimestamp: lambda self, e: "CURRENT_TIMESTAMP",
185            exp.DayOfMonth: rename_func("DAYOFMONTH"),
186            exp.DayOfWeek: rename_func("DAYOFWEEK"),
187            exp.DayOfYear: rename_func("DAYOFYEAR"),
188            exp.DataType: _datatype_sql,
189            exp.DateAdd: _date_delta_sql,
190            exp.DateSub: _date_delta_sql,
191            exp.DateDiff: lambda self, e: self.func(
192                "DATE_DIFF", f"'{e.args.get('unit', 'day')}'", e.expression, e.this
193            ),
194            exp.DateStrToDate: datestrtodate_sql,
195            exp.DateToDi: lambda self, e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.dateint_format}) AS INT)",
196            exp.DiToDate: lambda self, e: f"CAST(STRPTIME(CAST({self.sql(e, 'this')} AS TEXT), {DuckDB.dateint_format}) AS DATE)",
197            exp.Explode: rename_func("UNNEST"),
198            exp.IntDiv: lambda self, e: self.binary(e, "//"),
199            exp.JSONExtract: arrow_json_extract_sql,
200            exp.JSONExtractScalar: arrow_json_extract_scalar_sql,
201            exp.JSONBExtract: arrow_json_extract_sql,
202            exp.JSONBExtractScalar: arrow_json_extract_scalar_sql,
203            exp.LogicalOr: rename_func("BOOL_OR"),
204            exp.LogicalAnd: rename_func("BOOL_AND"),
205            exp.Properties: no_properties_sql,
206            exp.RegexpExtract: _regexp_extract_sql,
207            exp.RegexpLike: rename_func("REGEXP_MATCHES"),
208            exp.RegexpSplit: rename_func("STR_SPLIT_REGEX"),
209            exp.SafeDivide: no_safe_divide_sql,
210            exp.Split: rename_func("STR_SPLIT"),
211            exp.SortArray: _sort_array_sql,
212            exp.StrPosition: str_position_sql,
213            exp.StrToDate: lambda self, e: f"CAST({str_to_time_sql(self, e)} AS DATE)",
214            exp.StrToTime: str_to_time_sql,
215            exp.StrToUnix: lambda self, e: f"EPOCH(STRPTIME({self.sql(e, 'this')}, {self.format_time(e)}))",
216            exp.Struct: _struct_sql,
217            exp.TimestampTrunc: timestamptrunc_sql,
218            exp.TimeStrToDate: lambda self, e: f"CAST({self.sql(e, 'this')} AS DATE)",
219            exp.TimeStrToTime: timestrtotime_sql,
220            exp.TimeStrToUnix: lambda self, e: f"EPOCH(CAST({self.sql(e, 'this')} AS TIMESTAMP))",
221            exp.TimeToStr: lambda self, e: f"STRFTIME({self.sql(e, 'this')}, {self.format_time(e)})",
222            exp.TimeToUnix: rename_func("EPOCH"),
223            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS TEXT), '-', ''), 1, 8) AS INT)",
224            exp.TsOrDsAdd: _ts_or_ds_add_sql,
225            exp.TsOrDsToDate: ts_or_ds_to_date_sql("duckdb"),
226            exp.UnixToStr: lambda self, e: f"STRFTIME(TO_TIMESTAMP({self.sql(e, 'this')}), {self.format_time(e)})",
227            exp.UnixToTime: rename_func("TO_TIMESTAMP"),
228            exp.UnixToTimeStr: lambda self, e: f"CAST(TO_TIMESTAMP({self.sql(e, 'this')}) AS TEXT)",
229            exp.WeekOfYear: rename_func("WEEKOFYEAR"),
230        }
231
232        TYPE_MAPPING = {
233            **generator.Generator.TYPE_MAPPING,
234            exp.DataType.Type.BINARY: "BLOB",
235            exp.DataType.Type.CHAR: "TEXT",
236            exp.DataType.Type.FLOAT: "REAL",
237            exp.DataType.Type.NCHAR: "TEXT",
238            exp.DataType.Type.NVARCHAR: "TEXT",
239            exp.DataType.Type.UINT: "UINTEGER",
240            exp.DataType.Type.VARBINARY: "BLOB",
241            exp.DataType.Type.VARCHAR: "TEXT",
242        }
243
244        STAR_MAPPING = {**generator.Generator.STAR_MAPPING, "except": "EXCLUDE"}
245
246        PROPERTIES_LOCATION = {
247            **generator.Generator.PROPERTIES_LOCATION,
248            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
249        }
250
251        def tablesample_sql(
252            self, expression: exp.TableSample, seed_prefix: str = "SEED", sep: str = " AS "
253        ) -> str:
254            return super().tablesample_sql(expression, seed_prefix="REPEATABLE", sep=sep)

Generator interprets the given syntax tree and produces a SQL string as an output.

Arguments:
  • time_mapping (dict): the dictionary of custom time mappings in which the key represents a python time format and the output the target time format
  • time_trie (trie): a trie of the time_mapping keys
  • pretty (bool): if set to True the returned string will be formatted. Default: False.
  • quote_start (str): specifies which starting character to use to delimit quotes. Default: '.
  • quote_end (str): specifies which ending character to use to delimit quotes. Default: '.
  • identifier_start (str): specifies which starting character to use to delimit identifiers. Default: ".
  • identifier_end (str): specifies which ending character to use to delimit identifiers. Default: ".
  • bit_start (str): specifies which starting character to use to delimit bit literals. Default: None.
  • bit_end (str): specifies which ending character to use to delimit bit literals. Default: None.
  • hex_start (str): specifies which starting character to use to delimit hex literals. Default: None.
  • hex_end (str): specifies which ending character to use to delimit hex literals. Default: None.
  • byte_start (str): specifies which starting character to use to delimit byte literals. Default: None.
  • byte_end (str): specifies which ending character to use to delimit byte literals. Default: None.
  • raw_start (str): specifies which starting character to use to delimit raw literals. Default: None.
  • raw_end (str): specifies which ending character to use to delimit raw literals. Default: None.
  • identify (bool | str): 'always': always quote, 'safe': quote identifiers if they don't contain an upcase, True defaults to always.
  • normalize (bool): if set to True all identifiers will lower cased
  • string_escape (str): specifies a string escape character. Default: '.
  • identifier_escape (str): specifies an identifier escape character. Default: ".
  • pad (int): determines padding in a formatted string. Default: 2.
  • indent (int): determines the size of indentation in a formatted string. Default: 4.
  • unnest_column_only (bool): if true unnest table aliases are considered only as column aliases
  • normalize_functions (str): normalize function names, "upper", "lower", or None Default: "upper"
  • alias_post_tablesample (bool): if the table alias comes after tablesample Default: False
  • identifiers_can_start_with_digit (bool): if an unquoted identifier can start with digit Default: False
  • unsupported_level (ErrorLevel): determines the generator's behavior when it encounters unsupported expressions. Default ErrorLevel.WARN.
  • null_ordering (str): Indicates the default null ordering method to use if not explicitly set. Options are "nulls_are_small", "nulls_are_large", "nulls_are_last". Default: "nulls_are_small"
  • max_unsupported (int): Maximum number of unsupported messages to include in a raised UnsupportedError. This is only relevant if unsupported_level is ErrorLevel.RAISE. Default: 3
  • leading_comma (bool): if the the comma is leading or trailing in select statements Default: False
  • max_text_width: The max number of characters in a segment before creating new lines in pretty mode. The default is on the smaller end because the length only represents a segment and not the true line length. Default: 80
  • comments: Whether or not to preserve comments in the output SQL code. Default: True
def tablesample_sql( self, expression: sqlglot.expressions.TableSample, seed_prefix: str = 'SEED', sep: str = ' AS ') -> str:
251        def tablesample_sql(
252            self, expression: exp.TableSample, seed_prefix: str = "SEED", sep: str = " AS "
253        ) -> str:
254            return super().tablesample_sql(expression, seed_prefix="REPEATABLE", sep=sep)
Inherited Members
sqlglot.generator.Generator
Generator
generate
unsupported
sep
seg
pad_comment
maybe_comment
wrap
no_identify
normalize_func
indent
sql
uncache_sql
cache_sql
characterset_sql
column_sql
columnposition_sql
columndef_sql
columnconstraint_sql
autoincrementcolumnconstraint_sql
compresscolumnconstraint_sql
generatedasidentitycolumnconstraint_sql
notnullcolumnconstraint_sql
primarykeycolumnconstraint_sql
uniquecolumnconstraint_sql
create_sql
clone_sql
describe_sql
prepend_ctes
with_sql
cte_sql
tablealias_sql
bitstring_sql
hexstring_sql
bytestring_sql
rawstring_sql
datatypesize_sql
datatype_sql
directory_sql
delete_sql
drop_sql
except_sql
except_op
fetch_sql
filter_sql
hint_sql
index_sql
identifier_sql
inputoutputformat_sql
national_sql
partition_sql
properties_sql
root_properties
properties
with_properties
locate_properties
property_sql
likeproperty_sql
fallbackproperty_sql
journalproperty_sql
freespaceproperty_sql
checksumproperty_sql
mergeblockratioproperty_sql
datablocksizeproperty_sql
blockcompressionproperty_sql
isolatedloadingproperty_sql
lockingproperty_sql
withdataproperty_sql
insert_sql
intersect_sql
intersect_op
introducer_sql
pseudotype_sql
onconflict_sql
returning_sql
rowformatdelimitedproperty_sql
table_sql
pivot_sql
tuple_sql
update_sql
values_sql
var_sql
into_sql
from_sql
group_sql
having_sql
join_sql
lambda_sql
lateral_sql
limit_sql
offset_sql
setitem_sql
set_sql
pragma_sql
lock_sql
literal_sql
loaddata_sql
null_sql
boolean_sql
order_sql
cluster_sql
distribute_sql
sort_sql
ordered_sql
matchrecognize_sql
query_modifiers
after_having_modifiers
after_limit_modifiers
select_sql
schema_sql
star_sql
parameter_sql
sessionparameter_sql
placeholder_sql
subquery_sql
qualify_sql
union_sql
union_op
unnest_sql
where_sql
window_sql
partition_by_sql
windowspec_sql
withingroup_sql
between_sql
bracket_sql
all_sql
any_sql
exists_sql
case_sql
constraint_sql
nextvaluefor_sql
extract_sql
trim_sql
concat_sql
check_sql
foreignkey_sql
primarykey_sql
if_sql
matchagainst_sql
jsonkeyvalue_sql
jsonobject_sql
openjsoncolumndef_sql
openjson_sql
in_sql
in_unnest_op
interval_sql
return_sql
reference_sql
anonymous_sql
paren_sql
neg_sql
not_sql
alias_sql
aliases_sql
attimezone_sql
add_sql
and_sql
connector_sql
bitwiseand_sql
bitwiseleftshift_sql
bitwisenot_sql
bitwiseor_sql
bitwiserightshift_sql
bitwisexor_sql
cast_sql
currentdate_sql
collate_sql
command_sql
comment_sql
mergetreettlaction_sql
mergetreettl_sql
transaction_sql
commit_sql
rollback_sql
altercolumn_sql
renametable_sql
altertable_sql
droppartition_sql
addconstraint_sql
distinct_sql
ignorenulls_sql
respectnulls_sql
intdiv_sql
dpipe_sql
div_sql
overlaps_sql
distance_sql
dot_sql
eq_sql
escape_sql
glob_sql
gt_sql
gte_sql
ilike_sql
ilikeany_sql
is_sql
like_sql
likeany_sql
similarto_sql
lt_sql
lte_sql
mod_sql
mul_sql
neq_sql
nullsafeeq_sql
nullsafeneq_sql
or_sql
slice_sql
sub_sql
trycast_sql
use_sql
binary
function_fallback_sql
func
format_args
text_width
format_time
expressions
op_expressions
naked_property
set_operation
tag_sql
token_sql
userdefinedfunction_sql
joinhint_sql
kwarg_sql
when_sql
merge_sql
tochar_sql
dictproperty_sql
dictrange_sql
dictsubproperty_sql