sqlglot.dialects.clickhouse
1from __future__ import annotations 2 3import typing as t 4 5from sqlglot import exp, generator, parser, tokens 6from sqlglot.dialects.dialect import Dialect, inline_array_sql, var_map_sql 7from sqlglot.errors import ParseError 8from sqlglot.parser import parse_var_map 9from sqlglot.tokens import TokenType 10 11 12def _lower_func(sql: str) -> str: 13 index = sql.index("(") 14 return sql[:index].lower() + sql[index:] 15 16 17class ClickHouse(Dialect): 18 normalize_functions = None 19 null_ordering = "nulls_are_last" 20 21 class Tokenizer(tokens.Tokenizer): 22 COMMENTS = ["--", "#", "#!", ("/*", "*/")] 23 IDENTIFIERS = ['"', "`"] 24 25 KEYWORDS = { 26 **tokens.Tokenizer.KEYWORDS, 27 "ASOF": TokenType.ASOF, 28 "GLOBAL": TokenType.GLOBAL, 29 "DATETIME64": TokenType.DATETIME, 30 "FINAL": TokenType.FINAL, 31 "FLOAT32": TokenType.FLOAT, 32 "FLOAT64": TokenType.DOUBLE, 33 "INT16": TokenType.SMALLINT, 34 "INT32": TokenType.INT, 35 "INT64": TokenType.BIGINT, 36 "INT8": TokenType.TINYINT, 37 "TUPLE": TokenType.STRUCT, 38 } 39 40 class Parser(parser.Parser): 41 FUNCTIONS = { 42 **parser.Parser.FUNCTIONS, # type: ignore 43 "MAP": parse_var_map, 44 "QUANTILE": lambda params, args: exp.Quantile(this=args, quantile=params), 45 "QUANTILES": lambda params, args: exp.Quantiles(parameters=params, expressions=args), 46 "QUANTILEIF": lambda params, args: exp.QuantileIf(parameters=params, expressions=args), 47 } 48 49 RANGE_PARSERS = { 50 **parser.Parser.RANGE_PARSERS, 51 TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN) 52 and self._parse_in(this, is_global=True), 53 } 54 55 JOIN_KINDS = {*parser.Parser.JOIN_KINDS, TokenType.ANY, TokenType.ASOF} # type: ignore 56 57 TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {TokenType.ANY} # type: ignore 58 59 def _parse_in( 60 self, this: t.Optional[exp.Expression], is_global: bool = False 61 ) -> exp.Expression: 62 this = super()._parse_in(this) 63 this.set("is_global", is_global) 64 return this 65 66 def _parse_table( 67 self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None 68 ) -> t.Optional[exp.Expression]: 69 this = super()._parse_table(schema=schema, alias_tokens=alias_tokens) 70 71 if self._match(TokenType.FINAL): 72 this = self.expression(exp.Final, this=this) 73 74 return this 75 76 def _parse_position(self, haystack_first: bool = False) -> exp.Expression: 77 return super()._parse_position(haystack_first=True) 78 79 # https://clickhouse.com/docs/en/sql-reference/statements/select/with/ 80 def _parse_cte(self) -> exp.Expression: 81 index = self._index 82 try: 83 # WITH <identifier> AS <subquery expression> 84 return super()._parse_cte() 85 except ParseError: 86 # WITH <expression> AS <identifier> 87 self._retreat(index) 88 statement = self._parse_statement() 89 90 if statement and isinstance(statement.this, exp.Alias): 91 self.raise_error("Expected CTE to have alias") 92 93 return self.expression(exp.CTE, this=statement, alias=statement and statement.this) 94 95 class Generator(generator.Generator): 96 STRUCT_DELIMITER = ("(", ")") 97 98 TYPE_MAPPING = { 99 **generator.Generator.TYPE_MAPPING, # type: ignore 100 exp.DataType.Type.NULLABLE: "Nullable", 101 exp.DataType.Type.DATETIME: "DateTime64", 102 exp.DataType.Type.MAP: "Map", 103 exp.DataType.Type.ARRAY: "Array", 104 exp.DataType.Type.STRUCT: "Tuple", 105 exp.DataType.Type.TINYINT: "Int8", 106 exp.DataType.Type.SMALLINT: "Int16", 107 exp.DataType.Type.INT: "Int32", 108 exp.DataType.Type.BIGINT: "Int64", 109 exp.DataType.Type.FLOAT: "Float32", 110 exp.DataType.Type.DOUBLE: "Float64", 111 } 112 113 TRANSFORMS = { 114 **generator.Generator.TRANSFORMS, # type: ignore 115 exp.Array: inline_array_sql, 116 exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})", 117 exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL", 118 exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)), 119 exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)), 120 exp.Quantile: lambda self, e: f"quantile{self._param_args_sql(e, 'quantile', 'this')}", 121 exp.Quantiles: lambda self, e: f"quantiles{self._param_args_sql(e, 'parameters', 'expressions')}", 122 exp.QuantileIf: lambda self, e: f"quantileIf{self._param_args_sql(e, 'parameters', 'expressions')}", 123 } 124 125 EXPLICIT_UNION = True 126 127 def _param_args_sql( 128 self, expression: exp.Expression, params_name: str, args_name: str 129 ) -> str: 130 params = self.format_args(self.expressions(expression, params_name)) 131 args = self.format_args(self.expressions(expression, args_name)) 132 return f"({params})({args})" 133 134 def cte_sql(self, expression: exp.CTE) -> str: 135 if isinstance(expression.this, exp.Alias): 136 return self.sql(expression, "this") 137 138 return super().cte_sql(expression)
18class ClickHouse(Dialect): 19 normalize_functions = None 20 null_ordering = "nulls_are_last" 21 22 class Tokenizer(tokens.Tokenizer): 23 COMMENTS = ["--", "#", "#!", ("/*", "*/")] 24 IDENTIFIERS = ['"', "`"] 25 26 KEYWORDS = { 27 **tokens.Tokenizer.KEYWORDS, 28 "ASOF": TokenType.ASOF, 29 "GLOBAL": TokenType.GLOBAL, 30 "DATETIME64": TokenType.DATETIME, 31 "FINAL": TokenType.FINAL, 32 "FLOAT32": TokenType.FLOAT, 33 "FLOAT64": TokenType.DOUBLE, 34 "INT16": TokenType.SMALLINT, 35 "INT32": TokenType.INT, 36 "INT64": TokenType.BIGINT, 37 "INT8": TokenType.TINYINT, 38 "TUPLE": TokenType.STRUCT, 39 } 40 41 class Parser(parser.Parser): 42 FUNCTIONS = { 43 **parser.Parser.FUNCTIONS, # type: ignore 44 "MAP": parse_var_map, 45 "QUANTILE": lambda params, args: exp.Quantile(this=args, quantile=params), 46 "QUANTILES": lambda params, args: exp.Quantiles(parameters=params, expressions=args), 47 "QUANTILEIF": lambda params, args: exp.QuantileIf(parameters=params, expressions=args), 48 } 49 50 RANGE_PARSERS = { 51 **parser.Parser.RANGE_PARSERS, 52 TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN) 53 and self._parse_in(this, is_global=True), 54 } 55 56 JOIN_KINDS = {*parser.Parser.JOIN_KINDS, TokenType.ANY, TokenType.ASOF} # type: ignore 57 58 TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {TokenType.ANY} # type: ignore 59 60 def _parse_in( 61 self, this: t.Optional[exp.Expression], is_global: bool = False 62 ) -> exp.Expression: 63 this = super()._parse_in(this) 64 this.set("is_global", is_global) 65 return this 66 67 def _parse_table( 68 self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None 69 ) -> t.Optional[exp.Expression]: 70 this = super()._parse_table(schema=schema, alias_tokens=alias_tokens) 71 72 if self._match(TokenType.FINAL): 73 this = self.expression(exp.Final, this=this) 74 75 return this 76 77 def _parse_position(self, haystack_first: bool = False) -> exp.Expression: 78 return super()._parse_position(haystack_first=True) 79 80 # https://clickhouse.com/docs/en/sql-reference/statements/select/with/ 81 def _parse_cte(self) -> exp.Expression: 82 index = self._index 83 try: 84 # WITH <identifier> AS <subquery expression> 85 return super()._parse_cte() 86 except ParseError: 87 # WITH <expression> AS <identifier> 88 self._retreat(index) 89 statement = self._parse_statement() 90 91 if statement and isinstance(statement.this, exp.Alias): 92 self.raise_error("Expected CTE to have alias") 93 94 return self.expression(exp.CTE, this=statement, alias=statement and statement.this) 95 96 class Generator(generator.Generator): 97 STRUCT_DELIMITER = ("(", ")") 98 99 TYPE_MAPPING = { 100 **generator.Generator.TYPE_MAPPING, # type: ignore 101 exp.DataType.Type.NULLABLE: "Nullable", 102 exp.DataType.Type.DATETIME: "DateTime64", 103 exp.DataType.Type.MAP: "Map", 104 exp.DataType.Type.ARRAY: "Array", 105 exp.DataType.Type.STRUCT: "Tuple", 106 exp.DataType.Type.TINYINT: "Int8", 107 exp.DataType.Type.SMALLINT: "Int16", 108 exp.DataType.Type.INT: "Int32", 109 exp.DataType.Type.BIGINT: "Int64", 110 exp.DataType.Type.FLOAT: "Float32", 111 exp.DataType.Type.DOUBLE: "Float64", 112 } 113 114 TRANSFORMS = { 115 **generator.Generator.TRANSFORMS, # type: ignore 116 exp.Array: inline_array_sql, 117 exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})", 118 exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL", 119 exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)), 120 exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)), 121 exp.Quantile: lambda self, e: f"quantile{self._param_args_sql(e, 'quantile', 'this')}", 122 exp.Quantiles: lambda self, e: f"quantiles{self._param_args_sql(e, 'parameters', 'expressions')}", 123 exp.QuantileIf: lambda self, e: f"quantileIf{self._param_args_sql(e, 'parameters', 'expressions')}", 124 } 125 126 EXPLICIT_UNION = True 127 128 def _param_args_sql( 129 self, expression: exp.Expression, params_name: str, args_name: str 130 ) -> str: 131 params = self.format_args(self.expressions(expression, params_name)) 132 args = self.format_args(self.expressions(expression, args_name)) 133 return f"({params})({args})" 134 135 def cte_sql(self, expression: exp.CTE) -> str: 136 if isinstance(expression.this, exp.Alias): 137 return self.sql(expression, "this") 138 139 return super().cte_sql(expression)
22 class Tokenizer(tokens.Tokenizer): 23 COMMENTS = ["--", "#", "#!", ("/*", "*/")] 24 IDENTIFIERS = ['"', "`"] 25 26 KEYWORDS = { 27 **tokens.Tokenizer.KEYWORDS, 28 "ASOF": TokenType.ASOF, 29 "GLOBAL": TokenType.GLOBAL, 30 "DATETIME64": TokenType.DATETIME, 31 "FINAL": TokenType.FINAL, 32 "FLOAT32": TokenType.FLOAT, 33 "FLOAT64": TokenType.DOUBLE, 34 "INT16": TokenType.SMALLINT, 35 "INT32": TokenType.INT, 36 "INT64": TokenType.BIGINT, 37 "INT8": TokenType.TINYINT, 38 "TUPLE": TokenType.STRUCT, 39 }
Inherited Members
41 class Parser(parser.Parser): 42 FUNCTIONS = { 43 **parser.Parser.FUNCTIONS, # type: ignore 44 "MAP": parse_var_map, 45 "QUANTILE": lambda params, args: exp.Quantile(this=args, quantile=params), 46 "QUANTILES": lambda params, args: exp.Quantiles(parameters=params, expressions=args), 47 "QUANTILEIF": lambda params, args: exp.QuantileIf(parameters=params, expressions=args), 48 } 49 50 RANGE_PARSERS = { 51 **parser.Parser.RANGE_PARSERS, 52 TokenType.GLOBAL: lambda self, this: self._match(TokenType.IN) 53 and self._parse_in(this, is_global=True), 54 } 55 56 JOIN_KINDS = {*parser.Parser.JOIN_KINDS, TokenType.ANY, TokenType.ASOF} # type: ignore 57 58 TABLE_ALIAS_TOKENS = {*parser.Parser.TABLE_ALIAS_TOKENS} - {TokenType.ANY} # type: ignore 59 60 def _parse_in( 61 self, this: t.Optional[exp.Expression], is_global: bool = False 62 ) -> exp.Expression: 63 this = super()._parse_in(this) 64 this.set("is_global", is_global) 65 return this 66 67 def _parse_table( 68 self, schema: bool = False, alias_tokens: t.Optional[t.Collection[TokenType]] = None 69 ) -> t.Optional[exp.Expression]: 70 this = super()._parse_table(schema=schema, alias_tokens=alias_tokens) 71 72 if self._match(TokenType.FINAL): 73 this = self.expression(exp.Final, this=this) 74 75 return this 76 77 def _parse_position(self, haystack_first: bool = False) -> exp.Expression: 78 return super()._parse_position(haystack_first=True) 79 80 # https://clickhouse.com/docs/en/sql-reference/statements/select/with/ 81 def _parse_cte(self) -> exp.Expression: 82 index = self._index 83 try: 84 # WITH <identifier> AS <subquery expression> 85 return super()._parse_cte() 86 except ParseError: 87 # WITH <expression> AS <identifier> 88 self._retreat(index) 89 statement = self._parse_statement() 90 91 if statement and isinstance(statement.this, exp.Alias): 92 self.raise_error("Expected CTE to have alias") 93 94 return self.expression(exp.CTE, this=statement, alias=statement and statement.this)
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.RAISE
- 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
96 class Generator(generator.Generator): 97 STRUCT_DELIMITER = ("(", ")") 98 99 TYPE_MAPPING = { 100 **generator.Generator.TYPE_MAPPING, # type: ignore 101 exp.DataType.Type.NULLABLE: "Nullable", 102 exp.DataType.Type.DATETIME: "DateTime64", 103 exp.DataType.Type.MAP: "Map", 104 exp.DataType.Type.ARRAY: "Array", 105 exp.DataType.Type.STRUCT: "Tuple", 106 exp.DataType.Type.TINYINT: "Int8", 107 exp.DataType.Type.SMALLINT: "Int16", 108 exp.DataType.Type.INT: "Int32", 109 exp.DataType.Type.BIGINT: "Int64", 110 exp.DataType.Type.FLOAT: "Float32", 111 exp.DataType.Type.DOUBLE: "Float64", 112 } 113 114 TRANSFORMS = { 115 **generator.Generator.TRANSFORMS, # type: ignore 116 exp.Array: inline_array_sql, 117 exp.StrPosition: lambda self, e: f"position({self.format_args(e.this, e.args.get('substr'), e.args.get('position'))})", 118 exp.Final: lambda self, e: f"{self.sql(e, 'this')} FINAL", 119 exp.Map: lambda self, e: _lower_func(var_map_sql(self, e)), 120 exp.VarMap: lambda self, e: _lower_func(var_map_sql(self, e)), 121 exp.Quantile: lambda self, e: f"quantile{self._param_args_sql(e, 'quantile', 'this')}", 122 exp.Quantiles: lambda self, e: f"quantiles{self._param_args_sql(e, 'parameters', 'expressions')}", 123 exp.QuantileIf: lambda self, e: f"quantileIf{self._param_args_sql(e, 'parameters', 'expressions')}", 124 } 125 126 EXPLICIT_UNION = True 127 128 def _param_args_sql( 129 self, expression: exp.Expression, params_name: str, args_name: str 130 ) -> str: 131 params = self.format_args(self.expressions(expression, params_name)) 132 args = self.format_args(self.expressions(expression, args_name)) 133 return f"({params})({args})" 134 135 def cte_sql(self, expression: exp.CTE) -> str: 136 if isinstance(expression.this, exp.Alias): 137 return self.sql(expression, "this") 138 139 return super().cte_sql(expression)
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: ".
- 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
- 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
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
- columndef_sql
- columnconstraint_sql
- autoincrementcolumnconstraint_sql
- compresscolumnconstraint_sql
- generatedasidentitycolumnconstraint_sql
- notnullcolumnconstraint_sql
- primarykeycolumnconstraint_sql
- uniquecolumnconstraint_sql
- create_sql
- describe_sql
- prepend_ctes
- with_sql
- tablealias_sql
- bitstring_sql
- hexstring_sql
- datatype_sql
- directory_sql
- delete_sql
- drop_sql
- except_sql
- except_op
- fetch_sql
- filter_sql
- hint_sql
- index_sql
- identifier_sql
- national_sql
- partition_sql
- properties_sql
- root_properties
- properties
- with_properties
- locate_properties
- property_sql
- likeproperty_sql
- fallbackproperty_sql
- journalproperty_sql
- freespaceproperty_sql
- afterjournalproperty_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
- 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
- lock_sql
- literal_sql
- loaddata_sql
- null_sql
- boolean_sql
- order_sql
- cluster_sql
- distribute_sql
- sort_sql
- ordered_sql
- matchrecognize_sql
- query_modifiers
- select_sql
- schema_sql
- star_sql
- structkwarg_sql
- parameter_sql
- sessionparameter_sql
- placeholder_sql
- subquery_sql
- qualify_sql
- union_sql
- union_op
- unnest_sql
- where_sql
- window_sql
- partition_by_sql
- window_spec_sql
- withingroup_sql
- between_sql
- bracket_sql
- all_sql
- any_sql
- exists_sql
- case_sql
- constraint_sql
- extract_sql
- trim_sql
- concat_sql
- check_sql
- foreignkey_sql
- primarykey_sql
- unique_sql
- if_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
- 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
- is_sql
- like_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