Edit on GitHub

sqlglot.dialects.clickhouse

View Source

  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    inline_array_sql,
  9    no_pivot_sql,
 10    rename_func,
 11    var_map_sql,
 12)
 13from sqlglot.errors import ParseError
 14from sqlglot.parser import parse_var_map
 15from sqlglot.tokens import Token, TokenType
 16
 17
 18def _lower_func(sql: str) -> str:
 19    index = sql.index("(")
 20    return sql[:index].lower() + sql[index:]
 21
 22
 23class ClickHouse(Dialect):
 24    normalize_functions = None
 25    null_ordering = "nulls_are_last"
 26
 27    class Tokenizer(tokens.Tokenizer):
 28        COMMENTS = ["--", "#", "#!", ("/*", "*/")]
 29        IDENTIFIERS = ['"', "`"]
 30        STRING_ESCAPES = ["'", "\\"]
 31        BIT_STRINGS = [("0b", "")]
 32        HEX_STRINGS = [("0x", ""), ("0X", "")]
 33
 34        KEYWORDS = {
 35            **tokens.Tokenizer.KEYWORDS,
 36            "ATTACH": TokenType.COMMAND,
 37            "DATETIME64": TokenType.DATETIME64,
 38            "DICTIONARY": TokenType.DICTIONARY,
 39            "FINAL": TokenType.FINAL,
 40            "FLOAT32": TokenType.FLOAT,
 41            "FLOAT64": TokenType.DOUBLE,
 42            "GLOBAL": TokenType.GLOBAL,
 43            "INT128": TokenType.INT128,
 44            "INT16": TokenType.SMALLINT,
 45            "INT256": TokenType.INT256,
 46            "INT32": TokenType.INT,
 47            "INT64": TokenType.BIGINT,
 48            "INT8": TokenType.TINYINT,
 49            "MAP": TokenType.MAP,
 50            "TUPLE": TokenType.STRUCT,
 51            "UINT128": TokenType.UINT128,
 52            "UINT16": TokenType.USMALLINT,
 53            "UINT256": TokenType.UINT256,
 54            "UINT32": TokenType.UINT,
 55            "UINT64": TokenType.UBIGINT,
 56            "UINT8": TokenType.UTINYINT,
 57        }
 58
 59    class Parser(parser.Parser):
 60        FUNCTIONS = {
 61            **parser.Parser.FUNCTIONS,
 62            "ANY": exp.AnyValue.from_arg_list,
 63            "MAP": parse_var_map,
 64            "MATCH": exp.RegexpLike.from_arg_list,
 65            "UNIQ": exp.ApproxDistinct.from_arg_list,
 66        }
 67
 68        FUNCTIONS_WITH_ALIASED_ARGS = {*parser.Parser.FUNCTIONS_WITH_ALIASED_ARGS, "TUPLE"}
 69
 70        FUNCTION_PARSERS = {
 71            **parser.Parser.FUNCTION_PARSERS,
 72            "QUANTILE": lambda self: self._parse_quantile(),
 73        }
 74
 75        FUNCTION_PARSERS.pop("MATCH")
 76
 77        NO_PAREN_FUNCTION_PARSERS = parser.Parser.NO_PAREN_FUNCTION_PARSERS.copy()
 78        NO_PAREN_FUNCTION_PARSERS.pop(TokenType.ANY)
 79
 80        RANGE_PARSERS = {
 81            **parser.Parser.RANGE_PARSERS,
 82            TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN)
 83            and self._parse_in(this, is_global=True),
 84        }
 85
 86        # The PLACEHOLDER entry is popped because 1) it doesn't affect Clickhouse (it corresponds to
 87        # the postgres-specific JSONBContains parser) and 2) it makes parsing the ternary op simpler.
 88        COLUMN_OPERATORS = parser.Parser.COLUMN_OPERATORS.copy()
 89        COLUMN_OPERATORS.pop(TokenType.PLACEHOLDER)
 90
 91        JOIN_KINDS = {
 92            *parser.Parser.JOIN_KINDS,
 93            TokenType.ANY,
 94            TokenType.ASOF,
 95            TokenType.ANTI,
 96            TokenType.SEMI,
 97        }
 98
 99        TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {
100            TokenType.ANY,
101            TokenType.SEMI,
102            TokenType.ANTI,
103            TokenType.SETTINGS,
104            TokenType.FORMAT,
105        }
106
107        LOG_DEFAULTS_TO_LN = True
108
109        QUERY_MODIFIER_PARSERS = {
110            **parser.Parser.QUERY_MODIFIER_PARSERS,
111            "settings": lambda self: self._parse_csv(self._parse_conjunction)
112            if self._match(TokenType.SETTINGS)
113            else None,
114            "format": lambda self: self._parse_id_var() if self._match(TokenType.FORMAT) else None,
115        }
116
117        def _parse_conjunction(self) -> t.Optional[exp.Expression]:
118            this = super()._parse_conjunction()
119
120            if self._match(TokenType.PLACEHOLDER):
121                return self.expression(
122                    exp.If,
123                    this=this,
124                    true=self._parse_conjunction(),
125                    false=self._match(TokenType.COLON) and self._parse_conjunction(),
126                )
127
128            return this
129
130        def _parse_placeholder(self) -> t.Optional[exp.Expression]:
131            """
132            Parse a placeholder expression like SELECT {abc: UInt32} or FROM {table: Identifier}
133            https://clickhouse.com/docs/en/sql-reference/syntax#defining-and-using-query-parameters
134            """
135            if not self._match(TokenType.L_BRACE):
136                return None
137
138            this = self._parse_id_var()
139            self._match(TokenType.COLON)
140            kind = self._parse_types(check_func=False) or (
141                self._match_text_seq("IDENTIFIER") and "Identifier"
142            )
143
144            if not kind:
145                self.raise_error("Expecting a placeholder type or 'Identifier' for tables")
146            elif not self._match(TokenType.R_BRACE):
147                self.raise_error("Expecting }")
148
149            return self.expression(exp.Placeholder, this=this, kind=kind)
150
151        def _parse_in(self, this: t.Optional[exp.Expression], is_global: bool = False) -> exp.In:
152            this = super()._parse_in(this)
153            this.set("is_global", is_global)
154            return this
155
156        def _parse_table(
157            self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None
158        ) -> t.Optional[exp.Expression]:
159            this = super()._parse_table(schema=schema, alias_tokens=alias_tokens)
160
161            if self._match(TokenType.FINAL):
162                this = self.expression(exp.Final, this=this)
163
164            return this
165
166        def _parse_position(self, haystack_first: bool = False) -> exp.Expression:
167            return super()._parse_position(haystack_first=True)
168
169        # https://clickhouse.com/docs/en/sql-reference/statements/select/with/
170        def _parse_cte(self) -> exp.Expression:
171            index = self._index
172            try:
173                # WITH <identifier> AS <subquery expression>
174                return super()._parse_cte()
175            except ParseError:
176                # WITH <expression> AS <identifier>
177                self._retreat(index)
178                statement = self._parse_statement()
179
180                if statement and isinstance(statement.this, exp.Alias):
181                    self.raise_error("Expected CTE to have alias")
182
183                return self.expression(exp.CTE, this=statement, alias=statement and statement.this)
184
185        def _parse_join_parts(
186            self,
187        ) -> t.Tuple[t.Optional[Token], t.Optional[Token], t.Optional[Token]]:
188            is_global = self._match(TokenType.GLOBAL) and self._prev
189            kind_pre = self._match_set(self.JOIN_KINDS, advance=False) and self._prev
190            if kind_pre:
191                kind = self._match_set(self.JOIN_KINDS) and self._prev
192                side = self._match_set(self.JOIN_SIDES) and self._prev
193                return is_global, side, kind
194            return (
195                is_global,
196                self._match_set(self.JOIN_SIDES) and self._prev,
197                self._match_set(self.JOIN_KINDS) and self._prev,
198            )
199
200        def _parse_join(self, skip_join_token: bool = False) -> t.Optional[exp.Expression]:
201            join = super()._parse_join(skip_join_token)
202
203            if join:
204                join.set("global", join.args.pop("method", None))
205            return join
206
207        def _parse_function(
208            self, functions: t.Optional[t.Dict[str, t.Callable]] = None, anonymous: bool = False
209        ) -> t.Optional[exp.Expression]:
210            func = super()._parse_function(functions, anonymous)
211
212            if isinstance(func, exp.Anonymous):
213                params = self._parse_func_params(func)
214
215                if params:
216                    return self.expression(
217                        exp.ParameterizedAgg,
218                        this=func.this,
219                        expressions=func.expressions,
220                        params=params,
221                    )
222
223            return func
224
225        def _parse_func_params(
226            self, this: t.Optional[exp.Func] = None
227        ) -> t.Optional[t.List[t.Optional[exp.Expression]]]:
228            if self._match_pair(TokenType.R_PAREN, TokenType.L_PAREN):
229                return self._parse_csv(self._parse_lambda)
230            if self._match(TokenType.L_PAREN):
231                params = self._parse_csv(self._parse_lambda)
232                self._match_r_paren(this)
233                return params
234            return None
235
236        def _parse_quantile(self) -> exp.Quantile:
237            this = self._parse_lambda()
238            params = self._parse_func_params()
239            if params:
240                return self.expression(exp.Quantile, this=params[0], quantile=this)
241            return self.expression(exp.Quantile, this=this, quantile=exp.Literal.number(0.5))
242
243        def _parse_wrapped_id_vars(
244            self, optional: bool = False
245        ) -> t.List[t.Optional[exp.Expression]]:
246            return super()._parse_wrapped_id_vars(optional=True)
247
248        def _parse_primary_key(
249            self, wrapped_optional: bool = False, in_props: bool = False
250        ) -> exp.Expression:
251            return super()._parse_primary_key(
252                wrapped_optional=wrapped_optional or in_props, in_props=in_props
253            )
254
255    class Generator(generator.Generator):
256        STRUCT_DELIMITER = ("(", ")")
257
258        TYPE_MAPPING = {
259            **generator.Generator.TYPE_MAPPING,
260            exp.DataType.Type.ARRAY: "Array",
261            exp.DataType.Type.BIGINT: "Int64",
262            exp.DataType.Type.DATETIME64: "DateTime64",
263            exp.DataType.Type.DOUBLE: "Float64",
264            exp.DataType.Type.FLOAT: "Float32",
265            exp.DataType.Type.INT: "Int32",
266            exp.DataType.Type.INT128: "Int128",
267            exp.DataType.Type.INT256: "Int256",
268            exp.DataType.Type.MAP: "Map",
269            exp.DataType.Type.NULLABLE: "Nullable",
270            exp.DataType.Type.SMALLINT: "Int16",
271            exp.DataType.Type.STRUCT: "Tuple",
272            exp.DataType.Type.TINYINT: "Int8",
273            exp.DataType.Type.UBIGINT: "UInt64",
274            exp.DataType.Type.UINT: "UInt32",
275            exp.DataType.Type.UINT128: "UInt128",
276            exp.DataType.Type.UINT256: "UInt256",
277            exp.DataType.Type.USMALLINT: "UInt16",
278            exp.DataType.Type.UTINYINT: "UInt8",
279        }
280
281        TRANSFORMS = {
282            **generator.Generator.TRANSFORMS,
283            exp.AnyValue: rename_func("any"),
284            exp.ApproxDistinct: rename_func("uniq"),
285            exp.Array: inline_array_sql,
286            exp.CastToStrType: rename_func("CAST"),
287            exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL",
288            exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)),
289            exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}",
290            exp.Pivot: no_pivot_sql,
291            exp.Quantile: lambda self, e: self.func("quantile", e.args.get("quantile"))
292            + f"({self.sql(e, 'this')})",
293            exp.RegexpLike: lambda self, e: f"match({self.format_args(e.this, e.expression)})",
294            exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})",
295            exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)),
296        }
297
298        PROPERTIES_LOCATION = {
299            **generator.Generator.PROPERTIES_LOCATION,
300            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
301            exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA,
302        }
303
304        JOIN_HINTS = False
305        TABLE_HINTS = False
306        EXPLICIT_UNION = True
307        GROUPINGS_SEP = ""
308
309        def cte_sql(self, expression: exp.CTE) -> str:
310            if isinstance(expression.this, exp.Alias):
311                return self.sql(expression, "this")
312
313            return super().cte_sql(expression)
314
315        def after_limit_modifiers(self, expression: exp.Expression) -> t.List[str]:
316            return super().after_limit_modifiers(expression) + [
317                self.seg("SETTINGS ") + self.expressions(expression, key="settings", flat=True)
318                if expression.args.get("settings")
319                else "",
320                self.seg("FORMAT ") + self.sql(expression, "format")
321                if expression.args.get("format")
322                else "",
323            ]
324
325        def parameterizedagg_sql(self, expression: exp.Anonymous) -> str:
326            params = self.expressions(expression, "params", flat=True)
327            return self.func(expression.name, *expression.expressions) + f"({params})"
328
329        def placeholder_sql(self, expression: exp.Placeholder) -> str:
330            return f"{{{expression.name}: {self.sql(expression, 'kind')}}}"

class ClickHouse(sqlglot.dialects.dialect.Dialect): View Source

 24class ClickHouse(Dialect):
 25    normalize_functions = None
 26    null_ordering = "nulls_are_last"
 27
 28    class Tokenizer(tokens.Tokenizer):
 29        COMMENTS = ["--", "#", "#!", ("/*", "*/")]
 30        IDENTIFIERS = ['"', "`"]
 31        STRING_ESCAPES = ["'", "\\"]
 32        BIT_STRINGS = [("0b", "")]
 33        HEX_STRINGS = [("0x", ""), ("0X", "")]
 34
 35        KEYWORDS = {
 36            **tokens.Tokenizer.KEYWORDS,
 37            "ATTACH": TokenType.COMMAND,
 38            "DATETIME64": TokenType.DATETIME64,
 39            "DICTIONARY": TokenType.DICTIONARY,
 40            "FINAL": TokenType.FINAL,
 41            "FLOAT32": TokenType.FLOAT,
 42            "FLOAT64": TokenType.DOUBLE,
 43            "GLOBAL": TokenType.GLOBAL,
 44            "INT128": TokenType.INT128,
 45            "INT16": TokenType.SMALLINT,
 46            "INT256": TokenType.INT256,
 47            "INT32": TokenType.INT,
 48            "INT64": TokenType.BIGINT,
 49            "INT8": TokenType.TINYINT,
 50            "MAP": TokenType.MAP,
 51            "TUPLE": TokenType.STRUCT,
 52            "UINT128": TokenType.UINT128,
 53            "UINT16": TokenType.USMALLINT,
 54            "UINT256": TokenType.UINT256,
 55            "UINT32": TokenType.UINT,
 56            "UINT64": TokenType.UBIGINT,
 57            "UINT8": TokenType.UTINYINT,
 58        }
 59
 60    class Parser(parser.Parser):
 61        FUNCTIONS = {
 62            **parser.Parser.FUNCTIONS,
 63            "ANY": exp.AnyValue.from_arg_list,
 64            "MAP": parse_var_map,
 65            "MATCH": exp.RegexpLike.from_arg_list,
 66            "UNIQ": exp.ApproxDistinct.from_arg_list,
 67        }
 68
 69        FUNCTIONS_WITH_ALIASED_ARGS = {*parser.Parser.FUNCTIONS_WITH_ALIASED_ARGS, "TUPLE"}
 70
 71        FUNCTION_PARSERS = {
 72            **parser.Parser.FUNCTION_PARSERS,
 73            "QUANTILE": lambda self: self._parse_quantile(),
 74        }
 75
 76        FUNCTION_PARSERS.pop("MATCH")
 77
 78        NO_PAREN_FUNCTION_PARSERS = parser.Parser.NO_PAREN_FUNCTION_PARSERS.copy()
 79        NO_PAREN_FUNCTION_PARSERS.pop(TokenType.ANY)
 80
 81        RANGE_PARSERS = {
 82            **parser.Parser.RANGE_PARSERS,
 83            TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN)
 84            and self._parse_in(this, is_global=True),
 85        }
 86
 87        # The PLACEHOLDER entry is popped because 1) it doesn't affect Clickhouse (it corresponds to
 88        # the postgres-specific JSONBContains parser) and 2) it makes parsing the ternary op simpler.
 89        COLUMN_OPERATORS = parser.Parser.COLUMN_OPERATORS.copy()
 90        COLUMN_OPERATORS.pop(TokenType.PLACEHOLDER)
 91
 92        JOIN_KINDS = {
 93            *parser.Parser.JOIN_KINDS,
 94            TokenType.ANY,
 95            TokenType.ASOF,
 96            TokenType.ANTI,
 97            TokenType.SEMI,
 98        }
 99
100        TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {
101            TokenType.ANY,
102            TokenType.SEMI,
103            TokenType.ANTI,
104            TokenType.SETTINGS,
105            TokenType.FORMAT,
106        }
107
108        LOG_DEFAULTS_TO_LN = True
109
110        QUERY_MODIFIER_PARSERS = {
111            **parser.Parser.QUERY_MODIFIER_PARSERS,
112            "settings": lambda self: self._parse_csv(self._parse_conjunction)
113            if self._match(TokenType.SETTINGS)
114            else None,
115            "format": lambda self: self._parse_id_var() if self._match(TokenType.FORMAT) else None,
116        }
117
118        def _parse_conjunction(self) -> t.Optional[exp.Expression]:
119            this = super()._parse_conjunction()
120
121            if self._match(TokenType.PLACEHOLDER):
122                return self.expression(
123                    exp.If,
124                    this=this,
125                    true=self._parse_conjunction(),
126                    false=self._match(TokenType.COLON) and self._parse_conjunction(),
127                )
128
129            return this
130
131        def _parse_placeholder(self) -> t.Optional[exp.Expression]:
132            """
133            Parse a placeholder expression like SELECT {abc: UInt32} or FROM {table: Identifier}
134            https://clickhouse.com/docs/en/sql-reference/syntax#defining-and-using-query-parameters
135            """
136            if not self._match(TokenType.L_BRACE):
137                return None
138
139            this = self._parse_id_var()
140            self._match(TokenType.COLON)
141            kind = self._parse_types(check_func=False) or (
142                self._match_text_seq("IDENTIFIER") and "Identifier"
143            )
144
145            if not kind:
146                self.raise_error("Expecting a placeholder type or 'Identifier' for tables")
147            elif not self._match(TokenType.R_BRACE):
148                self.raise_error("Expecting }")
149
150            return self.expression(exp.Placeholder, this=this, kind=kind)
151
152        def _parse_in(self, this: t.Optional[exp.Expression], is_global: bool = False) -> exp.In:
153            this = super()._parse_in(this)
154            this.set("is_global", is_global)
155            return this
156
157        def _parse_table(
158            self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None
159        ) -> t.Optional[exp.Expression]:
160            this = super()._parse_table(schema=schema, alias_tokens=alias_tokens)
161
162            if self._match(TokenType.FINAL):
163                this = self.expression(exp.Final, this=this)
164
165            return this
166
167        def _parse_position(self, haystack_first: bool = False) -> exp.Expression:
168            return super()._parse_position(haystack_first=True)
169
170        # https://clickhouse.com/docs/en/sql-reference/statements/select/with/
171        def _parse_cte(self) -> exp.Expression:
172            index = self._index
173            try:
174                # WITH <identifier> AS <subquery expression>
175                return super()._parse_cte()
176            except ParseError:
177                # WITH <expression> AS <identifier>
178                self._retreat(index)
179                statement = self._parse_statement()
180
181                if statement and isinstance(statement.this, exp.Alias):
182                    self.raise_error("Expected CTE to have alias")
183
184                return self.expression(exp.CTE, this=statement, alias=statement and statement.this)
185
186        def _parse_join_parts(
187            self,
188        ) -> t.Tuple[t.Optional[Token], t.Optional[Token], t.Optional[Token]]:
189            is_global = self._match(TokenType.GLOBAL) and self._prev
190            kind_pre = self._match_set(self.JOIN_KINDS, advance=False) and self._prev
191            if kind_pre:
192                kind = self._match_set(self.JOIN_KINDS) and self._prev
193                side = self._match_set(self.JOIN_SIDES) and self._prev
194                return is_global, side, kind
195            return (
196                is_global,
197                self._match_set(self.JOIN_SIDES) and self._prev,
198                self._match_set(self.JOIN_KINDS) and self._prev,
199            )
200
201        def _parse_join(self, skip_join_token: bool = False) -> t.Optional[exp.Expression]:
202            join = super()._parse_join(skip_join_token)
203
204            if join:
205                join.set("global", join.args.pop("method", None))
206            return join
207
208        def _parse_function(
209            self, functions: t.Optional[t.Dict[str, t.Callable]] = None, anonymous: bool = False
210        ) -> t.Optional[exp.Expression]:
211            func = super()._parse_function(functions, anonymous)
212
213            if isinstance(func, exp.Anonymous):
214                params = self._parse_func_params(func)
215
216                if params:
217                    return self.expression(
218                        exp.ParameterizedAgg,
219                        this=func.this,
220                        expressions=func.expressions,
221                        params=params,
222                    )
223
224            return func
225
226        def _parse_func_params(
227            self, this: t.Optional[exp.Func] = None
228        ) -> t.Optional[t.List[t.Optional[exp.Expression]]]:
229            if self._match_pair(TokenType.R_PAREN, TokenType.L_PAREN):
230                return self._parse_csv(self._parse_lambda)
231            if self._match(TokenType.L_PAREN):
232                params = self._parse_csv(self._parse_lambda)
233                self._match_r_paren(this)
234                return params
235            return None
236
237        def _parse_quantile(self) -> exp.Quantile:
238            this = self._parse_lambda()
239            params = self._parse_func_params()
240            if params:
241                return self.expression(exp.Quantile, this=params[0], quantile=this)
242            return self.expression(exp.Quantile, this=this, quantile=exp.Literal.number(0.5))
243
244        def _parse_wrapped_id_vars(
245            self, optional: bool = False
246        ) -> t.List[t.Optional[exp.Expression]]:
247            return super()._parse_wrapped_id_vars(optional=True)
248
249        def _parse_primary_key(
250            self, wrapped_optional: bool = False, in_props: bool = False
251        ) -> exp.Expression:
252            return super()._parse_primary_key(
253                wrapped_optional=wrapped_optional or in_props, in_props=in_props
254            )
255
256    class Generator(generator.Generator):
257        STRUCT_DELIMITER = ("(", ")")
258
259        TYPE_MAPPING = {
260            **generator.Generator.TYPE_MAPPING,
261            exp.DataType.Type.ARRAY: "Array",
262            exp.DataType.Type.BIGINT: "Int64",
263            exp.DataType.Type.DATETIME64: "DateTime64",
264            exp.DataType.Type.DOUBLE: "Float64",
265            exp.DataType.Type.FLOAT: "Float32",
266            exp.DataType.Type.INT: "Int32",
267            exp.DataType.Type.INT128: "Int128",
268            exp.DataType.Type.INT256: "Int256",
269            exp.DataType.Type.MAP: "Map",
270            exp.DataType.Type.NULLABLE: "Nullable",
271            exp.DataType.Type.SMALLINT: "Int16",
272            exp.DataType.Type.STRUCT: "Tuple",
273            exp.DataType.Type.TINYINT: "Int8",
274            exp.DataType.Type.UBIGINT: "UInt64",
275            exp.DataType.Type.UINT: "UInt32",
276            exp.DataType.Type.UINT128: "UInt128",
277            exp.DataType.Type.UINT256: "UInt256",
278            exp.DataType.Type.USMALLINT: "UInt16",
279            exp.DataType.Type.UTINYINT: "UInt8",
280        }
281
282        TRANSFORMS = {
283            **generator.Generator.TRANSFORMS,
284            exp.AnyValue: rename_func("any"),
285            exp.ApproxDistinct: rename_func("uniq"),
286            exp.Array: inline_array_sql,
287            exp.CastToStrType: rename_func("CAST"),
288            exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL",
289            exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)),
290            exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}",
291            exp.Pivot: no_pivot_sql,
292            exp.Quantile: lambda self, e: self.func("quantile", e.args.get("quantile"))
293            + f"({self.sql(e, 'this')})",
294            exp.RegexpLike: lambda self, e: f"match({self.format_args(e.this, e.expression)})",
295            exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})",
296            exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)),
297        }
298
299        PROPERTIES_LOCATION = {
300            **generator.Generator.PROPERTIES_LOCATION,
301            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
302            exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA,
303        }
304
305        JOIN_HINTS = False
306        TABLE_HINTS = False
307        EXPLICIT_UNION = True
308        GROUPINGS_SEP = ""
309
310        def cte_sql(self, expression: exp.CTE) -> str:
311            if isinstance(expression.this, exp.Alias):
312                return self.sql(expression, "this")
313
314            return super().cte_sql(expression)
315
316        def after_limit_modifiers(self, expression: exp.Expression) -> t.List[str]:
317            return super().after_limit_modifiers(expression) + [
318                self.seg("SETTINGS ") + self.expressions(expression, key="settings", flat=True)
319                if expression.args.get("settings")
320                else "",
321                self.seg("FORMAT ") + self.sql(expression, "format")
322                if expression.args.get("format")
323                else "",
324            ]
325
326        def parameterizedagg_sql(self, expression: exp.Anonymous) -> str:
327            params = self.expressions(expression, "params", flat=True)
328            return self.func(expression.name, *expression.expressions) + f"({params})"
329
330        def placeholder_sql(self, expression: exp.Placeholder) -> str:
331            return f"{{{expression.name}: {self.sql(expression, 'kind')}}}"

Inherited Members

sqlglot.dialects.dialect.Dialect: get_or_raise; format_time; parse; parse_into; generate; transpile; tokenize; parser; generator

class ClickHouse.Tokenizer(sqlglot.tokens.Tokenizer): View Source

28    class Tokenizer(tokens.Tokenizer):
29        COMMENTS = ["--", "#", "#!", ("/*", "*/")]
30        IDENTIFIERS = ['"', "`"]
31        STRING_ESCAPES = ["'", "\\"]
32        BIT_STRINGS = [("0b", "")]
33        HEX_STRINGS = [("0x", ""), ("0X", "")]
34
35        KEYWORDS = {
36            **tokens.Tokenizer.KEYWORDS,
37            "ATTACH": TokenType.COMMAND,
38            "DATETIME64": TokenType.DATETIME64,
39            "DICTIONARY": TokenType.DICTIONARY,
40            "FINAL": TokenType.FINAL,
41            "FLOAT32": TokenType.FLOAT,
42            "FLOAT64": TokenType.DOUBLE,
43            "GLOBAL": TokenType.GLOBAL,
44            "INT128": TokenType.INT128,
45            "INT16": TokenType.SMALLINT,
46            "INT256": TokenType.INT256,
47            "INT32": TokenType.INT,
48            "INT64": TokenType.BIGINT,
49            "INT8": TokenType.TINYINT,
50            "MAP": TokenType.MAP,
51            "TUPLE": TokenType.STRUCT,
52            "UINT128": TokenType.UINT128,
53            "UINT16": TokenType.USMALLINT,
54            "UINT256": TokenType.UINT256,
55            "UINT32": TokenType.UINT,
56            "UINT64": TokenType.UBIGINT,
57            "UINT8": TokenType.UTINYINT,
58        }

Inherited Members

sqlglot.tokens.Tokenizer: reset; tokenize

class ClickHouse.Parser(sqlglot.parser.Parser): View Source

 60    class Parser(parser.Parser):
 61        FUNCTIONS = {
 62            **parser.Parser.FUNCTIONS,
 63            "ANY": exp.AnyValue.from_arg_list,
 64            "MAP": parse_var_map,
 65            "MATCH": exp.RegexpLike.from_arg_list,
 66            "UNIQ": exp.ApproxDistinct.from_arg_list,
 67        }
 68
 69        FUNCTIONS_WITH_ALIASED_ARGS = {*parser.Parser.FUNCTIONS_WITH_ALIASED_ARGS, "TUPLE"}
 70
 71        FUNCTION_PARSERS = {
 72            **parser.Parser.FUNCTION_PARSERS,
 73            "QUANTILE": lambda self: self._parse_quantile(),
 74        }
 75
 76        FUNCTION_PARSERS.pop("MATCH")
 77
 78        NO_PAREN_FUNCTION_PARSERS = parser.Parser.NO_PAREN_FUNCTION_PARSERS.copy()
 79        NO_PAREN_FUNCTION_PARSERS.pop(TokenType.ANY)
 80
 81        RANGE_PARSERS = {
 82            **parser.Parser.RANGE_PARSERS,
 83            TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN)
 84            and self._parse_in(this, is_global=True),
 85        }
 86
 87        # The PLACEHOLDER entry is popped because 1) it doesn't affect Clickhouse (it corresponds to
 88        # the postgres-specific JSONBContains parser) and 2) it makes parsing the ternary op simpler.
 89        COLUMN_OPERATORS = parser.Parser.COLUMN_OPERATORS.copy()
 90        COLUMN_OPERATORS.pop(TokenType.PLACEHOLDER)
 91
 92        JOIN_KINDS = {
 93            *parser.Parser.JOIN_KINDS,
 94            TokenType.ANY,
 95            TokenType.ASOF,
 96            TokenType.ANTI,
 97            TokenType.SEMI,
 98        }
 99
100        TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {
101            TokenType.ANY,
102            TokenType.SEMI,
103            TokenType.ANTI,
104            TokenType.SETTINGS,
105            TokenType.FORMAT,
106        }
107
108        LOG_DEFAULTS_TO_LN = True
109
110        QUERY_MODIFIER_PARSERS = {
111            **parser.Parser.QUERY_MODIFIER_PARSERS,
112            "settings": lambda self: self._parse_csv(self._parse_conjunction)
113            if self._match(TokenType.SETTINGS)
114            else None,
115            "format": lambda self: self._parse_id_var() if self._match(TokenType.FORMAT) else None,
116        }
117
118        def _parse_conjunction(self) -> t.Optional[exp.Expression]:
119            this = super()._parse_conjunction()
120
121            if self._match(TokenType.PLACEHOLDER):
122                return self.expression(
123                    exp.If,
124                    this=this,
125                    true=self._parse_conjunction(),
126                    false=self._match(TokenType.COLON) and self._parse_conjunction(),
127                )
128
129            return this
130
131        def _parse_placeholder(self) -> t.Optional[exp.Expression]:
132            """
133            Parse a placeholder expression like SELECT {abc: UInt32} or FROM {table: Identifier}
134            https://clickhouse.com/docs/en/sql-reference/syntax#defining-and-using-query-parameters
135            """
136            if not self._match(TokenType.L_BRACE):
137                return None
138
139            this = self._parse_id_var()
140            self._match(TokenType.COLON)
141            kind = self._parse_types(check_func=False) or (
142                self._match_text_seq("IDENTIFIER") and "Identifier"
143            )
144
145            if not kind:
146                self.raise_error("Expecting a placeholder type or 'Identifier' for tables")
147            elif not self._match(TokenType.R_BRACE):
148                self.raise_error("Expecting }")
149
150            return self.expression(exp.Placeholder, this=this, kind=kind)
151
152        def _parse_in(self, this: t.Optional[exp.Expression], is_global: bool = False) -> exp.In:
153            this = super()._parse_in(this)
154            this.set("is_global", is_global)
155            return this
156
157        def _parse_table(
158            self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None
159        ) -> t.Optional[exp.Expression]:
160            this = super()._parse_table(schema=schema, alias_tokens=alias_tokens)
161
162            if self._match(TokenType.FINAL):
163                this = self.expression(exp.Final, this=this)
164
165            return this
166
167        def _parse_position(self, haystack_first: bool = False) -> exp.Expression:
168            return super()._parse_position(haystack_first=True)
169
170        # https://clickhouse.com/docs/en/sql-reference/statements/select/with/
171        def _parse_cte(self) -> exp.Expression:
172            index = self._index
173            try:
174                # WITH <identifier> AS <subquery expression>
175                return super()._parse_cte()
176            except ParseError:
177                # WITH <expression> AS <identifier>
178                self._retreat(index)
179                statement = self._parse_statement()
180
181                if statement and isinstance(statement.this, exp.Alias):
182                    self.raise_error("Expected CTE to have alias")
183
184                return self.expression(exp.CTE, this=statement, alias=statement and statement.this)
185
186        def _parse_join_parts(
187            self,
188        ) -> t.Tuple[t.Optional[Token], t.Optional[Token], t.Optional[Token]]:
189            is_global = self._match(TokenType.GLOBAL) and self._prev
190            kind_pre = self._match_set(self.JOIN_KINDS, advance=False) and self._prev
191            if kind_pre:
192                kind = self._match_set(self.JOIN_KINDS) and self._prev
193                side = self._match_set(self.JOIN_SIDES) and self._prev
194                return is_global, side, kind
195            return (
196                is_global,
197                self._match_set(self.JOIN_SIDES) and self._prev,
198                self._match_set(self.JOIN_KINDS) and self._prev,
199            )
200
201        def _parse_join(self, skip_join_token: bool = False) -> t.Optional[exp.Expression]:
202            join = super()._parse_join(skip_join_token)
203
204            if join:
205                join.set("global", join.args.pop("method", None))
206            return join
207
208        def _parse_function(
209            self, functions: t.Optional[t.Dict[str, t.Callable]] = None, anonymous: bool = False
210        ) -> t.Optional[exp.Expression]:
211            func = super()._parse_function(functions, anonymous)
212
213            if isinstance(func, exp.Anonymous):
214                params = self._parse_func_params(func)
215
216                if params:
217                    return self.expression(
218                        exp.ParameterizedAgg,
219                        this=func.this,
220                        expressions=func.expressions,
221                        params=params,
222                    )
223
224            return func
225
226        def _parse_func_params(
227            self, this: t.Optional[exp.Func] = None
228        ) -> t.Optional[t.List[t.Optional[exp.Expression]]]:
229            if self._match_pair(TokenType.R_PAREN, TokenType.L_PAREN):
230                return self._parse_csv(self._parse_lambda)
231            if self._match(TokenType.L_PAREN):
232                params = self._parse_csv(self._parse_lambda)
233                self._match_r_paren(this)
234                return params
235            return None
236
237        def _parse_quantile(self) -> exp.Quantile:
238            this = self._parse_lambda()
239            params = self._parse_func_params()
240            if params:
241                return self.expression(exp.Quantile, this=params[0], quantile=this)
242            return self.expression(exp.Quantile, this=this, quantile=exp.Literal.number(0.5))
243
244        def _parse_wrapped_id_vars(
245            self, optional: bool = False
246        ) -> t.List[t.Optional[exp.Expression]]:
247            return super()._parse_wrapped_id_vars(optional=True)
248
249        def _parse_primary_key(
250            self, wrapped_optional: bool = False, in_props: bool = False
251        ) -> exp.Expression:
252            return super()._parse_primary_key(
253                wrapped_optional=wrapped_optional or in_props, in_props=in_props
254            )

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"

Inherited Members

sqlglot.parser.Parser: Parser; reset; parse; parse_into; check_errors; raise_error; expression; validate_expression

class ClickHouse.Generator(sqlglot.generator.Generator): View Source

256    class Generator(generator.Generator):
257        STRUCT_DELIMITER = ("(", ")")
258
259        TYPE_MAPPING = {
260            **generator.Generator.TYPE_MAPPING,
261            exp.DataType.Type.ARRAY: "Array",
262            exp.DataType.Type.BIGINT: "Int64",
263            exp.DataType.Type.DATETIME64: "DateTime64",
264            exp.DataType.Type.DOUBLE: "Float64",
265            exp.DataType.Type.FLOAT: "Float32",
266            exp.DataType.Type.INT: "Int32",
267            exp.DataType.Type.INT128: "Int128",
268            exp.DataType.Type.INT256: "Int256",
269            exp.DataType.Type.MAP: "Map",
270            exp.DataType.Type.NULLABLE: "Nullable",
271            exp.DataType.Type.SMALLINT: "Int16",
272            exp.DataType.Type.STRUCT: "Tuple",
273            exp.DataType.Type.TINYINT: "Int8",
274            exp.DataType.Type.UBIGINT: "UInt64",
275            exp.DataType.Type.UINT: "UInt32",
276            exp.DataType.Type.UINT128: "UInt128",
277            exp.DataType.Type.UINT256: "UInt256",
278            exp.DataType.Type.USMALLINT: "UInt16",
279            exp.DataType.Type.UTINYINT: "UInt8",
280        }
281
282        TRANSFORMS = {
283            **generator.Generator.TRANSFORMS,
284            exp.AnyValue: rename_func("any"),
285            exp.ApproxDistinct: rename_func("uniq"),
286            exp.Array: inline_array_sql,
287            exp.CastToStrType: rename_func("CAST"),
288            exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL",
289            exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)),
290            exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}",
291            exp.Pivot: no_pivot_sql,
292            exp.Quantile: lambda self, e: self.func("quantile", e.args.get("quantile"))
293            + f"({self.sql(e, 'this')})",
294            exp.RegexpLike: lambda self, e: f"match({self.format_args(e.this, e.expression)})",
295            exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})",
296            exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)),
297        }
298
299        PROPERTIES_LOCATION = {
300            **generator.Generator.PROPERTIES_LOCATION,
301            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
302            exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA,
303        }
304
305        JOIN_HINTS = False
306        TABLE_HINTS = False
307        EXPLICIT_UNION = True
308        GROUPINGS_SEP = ""
309
310        def cte_sql(self, expression: exp.CTE) -> str:
311            if isinstance(expression.this, exp.Alias):
312                return self.sql(expression, "this")
313
314            return super().cte_sql(expression)
315
316        def after_limit_modifiers(self, expression: exp.Expression) -> t.List[str]:
317            return super().after_limit_modifiers(expression) + [
318                self.seg("SETTINGS ") + self.expressions(expression, key="settings", flat=True)
319                if expression.args.get("settings")
320                else "",
321                self.seg("FORMAT ") + self.sql(expression, "format")
322                if expression.args.get("format")
323                else "",
324            ]
325
326        def parameterizedagg_sql(self, expression: exp.Anonymous) -> str:
327            params = self.expressions(expression, "params", flat=True)
328            return self.func(expression.name, *expression.expressions) + f"({params})"
329
330        def placeholder_sql(self, expression: exp.Placeholder) -> str:
331            return f"{{{expression.name}: {self.sql(expression, 'kind')}}}"

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 cte_sql(self, expression: sqlglot.expressions.CTE) -> str: View Source

310        def cte_sql(self, expression: exp.CTE) -> str:
311            if isinstance(expression.this, exp.Alias):
312                return self.sql(expression, "this")
313
314            return super().cte_sql(expression)

def after_limit_modifiers(self, expression: sqlglot.expressions.Expression) -> List[str]: View Source

316        def after_limit_modifiers(self, expression: exp.Expression) -> t.List[str]:
317            return super().after_limit_modifiers(expression) + [
318                self.seg("SETTINGS ") + self.expressions(expression, key="settings", flat=True)
319                if expression.args.get("settings")
320                else "",
321                self.seg("FORMAT ") + self.sql(expression, "format")
322                if expression.args.get("format")
323                else "",
324            ]

def parameterizedagg_sql(self, expression: sqlglot.expressions.Anonymous) -> str: View Source

326        def parameterizedagg_sql(self, expression: exp.Anonymous) -> str:
327            params = self.expressions(expression, "params", flat=True)
328            return self.func(expression.name, *expression.expressions) + f"({params})"

def placeholder_sql(self, expression: sqlglot.expressions.Placeholder) -> str: View Source

330        def placeholder_sql(self, expression: exp.Placeholder) -> str:
331            return f"{{{expression.name}: {self.sql(expression, 'kind')}}}"

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; 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; tablesample_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; select_sql; schema_sql; star_sql; parameter_sql; sessionparameter_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