Supports BigQuery Standard SQL.
1"""Supports BigQuery Standard SQL.""" 2 3from __future__ import annotations 4 5import re 6import typing as t 7 8from sqlglot import exp, generator, parser, tokens, transforms 9from sqlglot.dialects.dialect import ( 10 Dialect, 11 datestrtodate_sql, 12 inline_array_sql, 13 no_ilike_sql, 14 rename_func, 15 timestrtotime_sql, 16) 17from sqlglot.helper import seq_get 18from sqlglot.tokens import TokenType 19 20E = t.TypeVar("E", bound=exp.Expression) 21 22 23def _date_add(expression_class: t.Type[E]) -> t.Callable[[t.Sequence], E]: 24 def func(args): 25 interval = seq_get(args, 1) 26 return expression_class( 27 this=seq_get(args, 0), 28 expression=interval.this, 29 unit=interval.args.get("unit"), 30 ) 31 32 return func 33 34 35def _date_add_sql( 36 data_type: str, kind: str 37) -> t.Callable[[generator.Generator, exp.Expression], str]: 38 def func(self, expression): 39 this = self.sql(expression, "this") 40 return f"{data_type}_{kind}({this}, {self.sql(exp.Interval(this=expression.expression, unit=expression.args.get('unit') or exp.Literal.string('day')))})" 41 42 return func 43 44 45def _derived_table_values_to_unnest(self: generator.Generator, expression: exp.Values) -> str: 46 if not isinstance(expression.unnest().parent, exp.From): 47 expression = t.cast(exp.Values, transforms.remove_precision_parameterized_types(expression)) 48 return self.values_sql(expression) 49 rows = [tuple_exp.expressions for tuple_exp in expression.find_all(exp.Tuple)] 50 structs = [] 51 for row in rows: 52 aliases = [ 53 exp.alias_(value, column_name) 54 for value, column_name in zip(row, expression.args["alias"].args["columns"]) 55 ] 56 structs.append(exp.Struct(expressions=aliases)) 57 unnest_exp = exp.Unnest(expressions=[exp.Array(expressions=structs)]) 58 return self.unnest_sql(unnest_exp) 59 60 61def _returnsproperty_sql(self: generator.Generator, expression: exp.ReturnsProperty) -> str: 62 this = expression.this 63 if isinstance(this, exp.Schema): 64 this = f"{this.this} <{self.expressions(this)}>" 65 else: 66 this = self.sql(this) 67 return f"RETURNS {this}" 68 69 70def _create_sql(self: generator.Generator, expression: exp.Create) -> str: 71 kind = expression.args["kind"] 72 returns = expression.find(exp.ReturnsProperty) 73 if kind.upper() == "FUNCTION" and returns and returns.args.get("is_table"): 74 expression = expression.copy() 75 expression.set("kind", "TABLE FUNCTION") 76 if isinstance( 77 expression.expression, 78 ( 79 exp.Subquery, 80 exp.Literal, 81 ), 82 ): 83 expression.set("expression", expression.expression.this) 84 85 return self.create_sql(expression) 86 87 return self.create_sql(expression) 88 89 90def _unqualify_unnest(expression: exp.Expression) -> exp.Expression: 91 """Remove references to unnest table aliases since bigquery doesn't allow them. 92 93 These are added by the optimizer's qualify_column step. 94 """ 95 if isinstance(expression, exp.Select): 96 unnests = { 97 unnest.alias 98 for unnest in expression.args.get("from", exp.From(expressions=[])).expressions 99 if isinstance(unnest, exp.Unnest) and unnest.alias 100 } 101 102 if unnests: 103 expression = expression.copy() 104 105 for select in expression.expressions: 106 for column in select.find_all(exp.Column): 107 if column.table in unnests: 108 column.set("table", None) 109 110 return expression 111 112 113class BigQuery(Dialect): 114 unnest_column_only = True 115 time_mapping = { 116 "%M": "%-M", 117 "%d": "%-d", 118 "%m": "%-m", 119 "%y": "%-y", 120 "%H": "%-H", 121 "%I": "%-I", 122 "%S": "%-S", 123 "%j": "%-j", 124 } 125 126 class Tokenizer(tokens.Tokenizer): 127 QUOTES = [ 128 (prefix + quote, quote) if prefix else quote 129 for quote in ["'", '"', '"""', "'''"] 130 for prefix in ["", "r", "R"] 131 ] 132 COMMENTS = ["--", "#", ("/*", "*/")] 133 IDENTIFIERS = ["`"] 134 STRING_ESCAPES = ["\\"] 135 HEX_STRINGS = [("0x", ""), ("0X", "")] 136 137 KEYWORDS = { 138 **tokens.Tokenizer.KEYWORDS, 139 "BEGIN": TokenType.COMMAND, 140 "BEGIN TRANSACTION": TokenType.BEGIN, 141 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 142 "CURRENT_TIME": TokenType.CURRENT_TIME, 143 "DECLARE": TokenType.COMMAND, 144 "GEOGRAPHY": TokenType.GEOGRAPHY, 145 "FLOAT64": TokenType.DOUBLE, 146 "INT64": TokenType.BIGINT, 147 "NOT DETERMINISTIC": TokenType.VOLATILE, 148 "UNKNOWN": TokenType.NULL, 149 } 150 KEYWORDS.pop("DIV") 151 152 class Parser(parser.Parser): 153 FUNCTIONS = { 154 **parser.Parser.FUNCTIONS, # type: ignore 155 "DATE_TRUNC": lambda args: exp.DateTrunc( 156 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 157 this=seq_get(args, 0), 158 ), 159 "DATE_ADD": _date_add(exp.DateAdd), 160 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 161 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 162 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 163 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 164 this=seq_get(args, 0), 165 expression=seq_get(args, 1), 166 position=seq_get(args, 2), 167 occurrence=seq_get(args, 3), 168 group=exp.Literal.number(1) 169 if re.compile(str(seq_get(args, 1))).groups == 1 170 else None, 171 ), 172 "TIME_ADD": _date_add(exp.TimeAdd), 173 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 174 "DATE_SUB": _date_add(exp.DateSub), 175 "DATETIME_SUB": _date_add(exp.DatetimeSub), 176 "TIME_SUB": _date_add(exp.TimeSub), 177 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 178 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 179 this=seq_get(args, 1), format=seq_get(args, 0) 180 ), 181 } 182 183 FUNCTION_PARSERS = { 184 **parser.Parser.FUNCTION_PARSERS, # type: ignore 185 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 186 } 187 FUNCTION_PARSERS.pop("TRIM") 188 189 NO_PAREN_FUNCTIONS = { 190 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 191 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 192 TokenType.CURRENT_TIME: exp.CurrentTime, 193 } 194 195 NESTED_TYPE_TOKENS = { 196 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 197 TokenType.TABLE, 198 } 199 200 ID_VAR_TOKENS = { 201 *parser.Parser.ID_VAR_TOKENS, # type: ignore 202 TokenType.VALUES, 203 } 204 205 PROPERTY_PARSERS = { 206 **parser.Parser.PROPERTY_PARSERS, # type: ignore 207 "NOT DETERMINISTIC": lambda self: self.expression( 208 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 209 ), 210 } 211 212 class Generator(generator.Generator): 213 TRANSFORMS = { 214 **generator.Generator.TRANSFORMS, # type: ignore 215 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 216 exp.ArraySize: rename_func("ARRAY_LENGTH"), 217 exp.DateAdd: _date_add_sql("DATE", "ADD"), 218 exp.DateSub: _date_add_sql("DATE", "SUB"), 219 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 220 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 221 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 222 exp.DateStrToDate: datestrtodate_sql, 223 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 224 exp.GroupConcat: rename_func("STRING_AGG"), 225 exp.ILike: no_ilike_sql, 226 exp.IntDiv: rename_func("DIV"), 227 exp.Select: transforms.preprocess( 228 [_unqualify_unnest], transforms.delegate("select_sql") 229 ), 230 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 231 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 232 exp.TimeSub: _date_add_sql("TIME", "SUB"), 233 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 234 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 235 exp.TimeStrToTime: timestrtotime_sql, 236 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 237 exp.VariancePop: rename_func("VAR_POP"), 238 exp.Values: _derived_table_values_to_unnest, 239 exp.ReturnsProperty: _returnsproperty_sql, 240 exp.Create: _create_sql, 241 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 242 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 243 if e.name == "IMMUTABLE" 244 else "NOT DETERMINISTIC", 245 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 246 } 247 248 TYPE_MAPPING = { 249 **generator.Generator.TYPE_MAPPING, # type: ignore 250 exp.DataType.Type.TINYINT: "INT64", 251 exp.DataType.Type.SMALLINT: "INT64", 252 exp.DataType.Type.INT: "INT64", 253 exp.DataType.Type.BIGINT: "INT64", 254 exp.DataType.Type.DECIMAL: "NUMERIC", 255 exp.DataType.Type.FLOAT: "FLOAT64", 256 exp.DataType.Type.DOUBLE: "FLOAT64", 257 exp.DataType.Type.BOOLEAN: "BOOL", 258 exp.DataType.Type.TEXT: "STRING", 259 exp.DataType.Type.VARCHAR: "STRING", 260 exp.DataType.Type.NVARCHAR: "STRING", 261 } 262 PROPERTIES_LOCATION = { 263 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 264 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 265 } 266 267 EXPLICIT_UNION = True 268 269 def array_sql(self, expression: exp.Array) -> str: 270 first_arg = seq_get(expression.expressions, 0) 271 if isinstance(first_arg, exp.Subqueryable): 272 return f"ARRAY{self.wrap(self.sql(first_arg))}" 273 274 return inline_array_sql(self, expression) 275 276 def transaction_sql(self, *_) -> str: 277 return "BEGIN TRANSACTION" 278 279 def commit_sql(self, *_) -> str: 280 return "COMMIT TRANSACTION" 281 282 def rollback_sql(self, *_) -> str: 283 return "ROLLBACK TRANSACTION" 284 285 def in_unnest_op(self, expression: exp.Unnest) -> str: 286 return self.sql(expression) 287 288 def except_op(self, expression: exp.Except) -> str: 289 if not expression.args.get("distinct", False): 290 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 291 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 292 293 def intersect_op(self, expression: exp.Intersect) -> str: 294 if not expression.args.get("distinct", False): 295 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 296 return f"INTERSECT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}"
114class BigQuery(Dialect): 115 unnest_column_only = True 116 time_mapping = { 117 "%M": "%-M", 118 "%d": "%-d", 119 "%m": "%-m", 120 "%y": "%-y", 121 "%H": "%-H", 122 "%I": "%-I", 123 "%S": "%-S", 124 "%j": "%-j", 125 } 126 127 class Tokenizer(tokens.Tokenizer): 128 QUOTES = [ 129 (prefix + quote, quote) if prefix else quote 130 for quote in ["'", '"', '"""', "'''"] 131 for prefix in ["", "r", "R"] 132 ] 133 COMMENTS = ["--", "#", ("/*", "*/")] 134 IDENTIFIERS = ["`"] 135 STRING_ESCAPES = ["\\"] 136 HEX_STRINGS = [("0x", ""), ("0X", "")] 137 138 KEYWORDS = { 139 **tokens.Tokenizer.KEYWORDS, 140 "BEGIN": TokenType.COMMAND, 141 "BEGIN TRANSACTION": TokenType.BEGIN, 142 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 143 "CURRENT_TIME": TokenType.CURRENT_TIME, 144 "DECLARE": TokenType.COMMAND, 145 "GEOGRAPHY": TokenType.GEOGRAPHY, 146 "FLOAT64": TokenType.DOUBLE, 147 "INT64": TokenType.BIGINT, 148 "NOT DETERMINISTIC": TokenType.VOLATILE, 149 "UNKNOWN": TokenType.NULL, 150 } 151 KEYWORDS.pop("DIV") 152 153 class Parser(parser.Parser): 154 FUNCTIONS = { 155 **parser.Parser.FUNCTIONS, # type: ignore 156 "DATE_TRUNC": lambda args: exp.DateTrunc( 157 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 158 this=seq_get(args, 0), 159 ), 160 "DATE_ADD": _date_add(exp.DateAdd), 161 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 162 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 163 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 164 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 165 this=seq_get(args, 0), 166 expression=seq_get(args, 1), 167 position=seq_get(args, 2), 168 occurrence=seq_get(args, 3), 169 group=exp.Literal.number(1) 170 if re.compile(str(seq_get(args, 1))).groups == 1 171 else None, 172 ), 173 "TIME_ADD": _date_add(exp.TimeAdd), 174 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 175 "DATE_SUB": _date_add(exp.DateSub), 176 "DATETIME_SUB": _date_add(exp.DatetimeSub), 177 "TIME_SUB": _date_add(exp.TimeSub), 178 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 179 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 180 this=seq_get(args, 1), format=seq_get(args, 0) 181 ), 182 } 183 184 FUNCTION_PARSERS = { 185 **parser.Parser.FUNCTION_PARSERS, # type: ignore 186 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 187 } 188 FUNCTION_PARSERS.pop("TRIM") 189 190 NO_PAREN_FUNCTIONS = { 191 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 192 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 193 TokenType.CURRENT_TIME: exp.CurrentTime, 194 } 195 196 NESTED_TYPE_TOKENS = { 197 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 198 TokenType.TABLE, 199 } 200 201 ID_VAR_TOKENS = { 202 *parser.Parser.ID_VAR_TOKENS, # type: ignore 203 TokenType.VALUES, 204 } 205 206 PROPERTY_PARSERS = { 207 **parser.Parser.PROPERTY_PARSERS, # type: ignore 208 "NOT DETERMINISTIC": lambda self: self.expression( 209 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 210 ), 211 } 212 213 class Generator(generator.Generator): 214 TRANSFORMS = { 215 **generator.Generator.TRANSFORMS, # type: ignore 216 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 217 exp.ArraySize: rename_func("ARRAY_LENGTH"), 218 exp.DateAdd: _date_add_sql("DATE", "ADD"), 219 exp.DateSub: _date_add_sql("DATE", "SUB"), 220 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 221 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 222 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 223 exp.DateStrToDate: datestrtodate_sql, 224 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 225 exp.GroupConcat: rename_func("STRING_AGG"), 226 exp.ILike: no_ilike_sql, 227 exp.IntDiv: rename_func("DIV"), 228 exp.Select: transforms.preprocess( 229 [_unqualify_unnest], transforms.delegate("select_sql") 230 ), 231 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 232 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 233 exp.TimeSub: _date_add_sql("TIME", "SUB"), 234 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 235 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 236 exp.TimeStrToTime: timestrtotime_sql, 237 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 238 exp.VariancePop: rename_func("VAR_POP"), 239 exp.Values: _derived_table_values_to_unnest, 240 exp.ReturnsProperty: _returnsproperty_sql, 241 exp.Create: _create_sql, 242 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 243 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 244 if e.name == "IMMUTABLE" 245 else "NOT DETERMINISTIC", 246 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 247 } 248 249 TYPE_MAPPING = { 250 **generator.Generator.TYPE_MAPPING, # type: ignore 251 exp.DataType.Type.TINYINT: "INT64", 252 exp.DataType.Type.SMALLINT: "INT64", 253 exp.DataType.Type.INT: "INT64", 254 exp.DataType.Type.BIGINT: "INT64", 255 exp.DataType.Type.DECIMAL: "NUMERIC", 256 exp.DataType.Type.FLOAT: "FLOAT64", 257 exp.DataType.Type.DOUBLE: "FLOAT64", 258 exp.DataType.Type.BOOLEAN: "BOOL", 259 exp.DataType.Type.TEXT: "STRING", 260 exp.DataType.Type.VARCHAR: "STRING", 261 exp.DataType.Type.NVARCHAR: "STRING", 262 } 263 PROPERTIES_LOCATION = { 264 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 265 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 266 } 267 268 EXPLICIT_UNION = True 269 270 def array_sql(self, expression: exp.Array) -> str: 271 first_arg = seq_get(expression.expressions, 0) 272 if isinstance(first_arg, exp.Subqueryable): 273 return f"ARRAY{self.wrap(self.sql(first_arg))}" 274 275 return inline_array_sql(self, expression) 276 277 def transaction_sql(self, *_) -> str: 278 return "BEGIN TRANSACTION" 279 280 def commit_sql(self, *_) -> str: 281 return "COMMIT TRANSACTION" 282 283 def rollback_sql(self, *_) -> str: 284 return "ROLLBACK TRANSACTION" 285 286 def in_unnest_op(self, expression: exp.Unnest) -> str: 287 return self.sql(expression) 288 289 def except_op(self, expression: exp.Except) -> str: 290 if not expression.args.get("distinct", False): 291 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 292 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 293 294 def intersect_op(self, expression: exp.Intersect) -> str: 295 if not expression.args.get("distinct", False): 296 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 297 return f"INTERSECT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}"
Inherited Members
127 class Tokenizer(tokens.Tokenizer): 128 QUOTES = [ 129 (prefix + quote, quote) if prefix else quote 130 for quote in ["'", '"', '"""', "'''"] 131 for prefix in ["", "r", "R"] 132 ] 133 COMMENTS = ["--", "#", ("/*", "*/")] 134 IDENTIFIERS = ["`"] 135 STRING_ESCAPES = ["\\"] 136 HEX_STRINGS = [("0x", ""), ("0X", "")] 137 138 KEYWORDS = { 139 **tokens.Tokenizer.KEYWORDS, 140 "BEGIN": TokenType.COMMAND, 141 "BEGIN TRANSACTION": TokenType.BEGIN, 142 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 143 "CURRENT_TIME": TokenType.CURRENT_TIME, 144 "DECLARE": TokenType.COMMAND, 145 "GEOGRAPHY": TokenType.GEOGRAPHY, 146 "FLOAT64": TokenType.DOUBLE, 147 "INT64": TokenType.BIGINT, 148 "NOT DETERMINISTIC": TokenType.VOLATILE, 149 "UNKNOWN": TokenType.NULL, 150 } 151 KEYWORDS.pop("DIV")
Inherited Members
153 class Parser(parser.Parser): 154 FUNCTIONS = { 155 **parser.Parser.FUNCTIONS, # type: ignore 156 "DATE_TRUNC": lambda args: exp.DateTrunc( 157 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 158 this=seq_get(args, 0), 159 ), 160 "DATE_ADD": _date_add(exp.DateAdd), 161 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 162 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 163 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 164 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 165 this=seq_get(args, 0), 166 expression=seq_get(args, 1), 167 position=seq_get(args, 2), 168 occurrence=seq_get(args, 3), 169 group=exp.Literal.number(1) 170 if re.compile(str(seq_get(args, 1))).groups == 1 171 else None, 172 ), 173 "TIME_ADD": _date_add(exp.TimeAdd), 174 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 175 "DATE_SUB": _date_add(exp.DateSub), 176 "DATETIME_SUB": _date_add(exp.DatetimeSub), 177 "TIME_SUB": _date_add(exp.TimeSub), 178 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 179 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 180 this=seq_get(args, 1), format=seq_get(args, 0) 181 ), 182 } 183 184 FUNCTION_PARSERS = { 185 **parser.Parser.FUNCTION_PARSERS, # type: ignore 186 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 187 } 188 FUNCTION_PARSERS.pop("TRIM") 189 190 NO_PAREN_FUNCTIONS = { 191 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 192 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 193 TokenType.CURRENT_TIME: exp.CurrentTime, 194 } 195 196 NESTED_TYPE_TOKENS = { 197 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 198 TokenType.TABLE, 199 } 200 201 ID_VAR_TOKENS = { 202 *parser.Parser.ID_VAR_TOKENS, # type: ignore 203 TokenType.VALUES, 204 } 205 206 PROPERTY_PARSERS = { 207 **parser.Parser.PROPERTY_PARSERS, # type: ignore 208 "NOT DETERMINISTIC": lambda self: self.expression( 209 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 210 ), 211 }
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
213 class Generator(generator.Generator): 214 TRANSFORMS = { 215 **generator.Generator.TRANSFORMS, # type: ignore 216 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 217 exp.ArraySize: rename_func("ARRAY_LENGTH"), 218 exp.DateAdd: _date_add_sql("DATE", "ADD"), 219 exp.DateSub: _date_add_sql("DATE", "SUB"), 220 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 221 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 222 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 223 exp.DateStrToDate: datestrtodate_sql, 224 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 225 exp.GroupConcat: rename_func("STRING_AGG"), 226 exp.ILike: no_ilike_sql, 227 exp.IntDiv: rename_func("DIV"), 228 exp.Select: transforms.preprocess( 229 [_unqualify_unnest], transforms.delegate("select_sql") 230 ), 231 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 232 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 233 exp.TimeSub: _date_add_sql("TIME", "SUB"), 234 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 235 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 236 exp.TimeStrToTime: timestrtotime_sql, 237 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 238 exp.VariancePop: rename_func("VAR_POP"), 239 exp.Values: _derived_table_values_to_unnest, 240 exp.ReturnsProperty: _returnsproperty_sql, 241 exp.Create: _create_sql, 242 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 243 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 244 if e.name == "IMMUTABLE" 245 else "NOT DETERMINISTIC", 246 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 247 } 248 249 TYPE_MAPPING = { 250 **generator.Generator.TYPE_MAPPING, # type: ignore 251 exp.DataType.Type.TINYINT: "INT64", 252 exp.DataType.Type.SMALLINT: "INT64", 253 exp.DataType.Type.INT: "INT64", 254 exp.DataType.Type.BIGINT: "INT64", 255 exp.DataType.Type.DECIMAL: "NUMERIC", 256 exp.DataType.Type.FLOAT: "FLOAT64", 257 exp.DataType.Type.DOUBLE: "FLOAT64", 258 exp.DataType.Type.BOOLEAN: "BOOL", 259 exp.DataType.Type.TEXT: "STRING", 260 exp.DataType.Type.VARCHAR: "STRING", 261 exp.DataType.Type.NVARCHAR: "STRING", 262 } 263 PROPERTIES_LOCATION = { 264 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 265 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 266 } 267 268 EXPLICIT_UNION = True 269 270 def array_sql(self, expression: exp.Array) -> str: 271 first_arg = seq_get(expression.expressions, 0) 272 if isinstance(first_arg, exp.Subqueryable): 273 return f"ARRAY{self.wrap(self.sql(first_arg))}" 274 275 return inline_array_sql(self, expression) 276 277 def transaction_sql(self, *_) -> str: 278 return "BEGIN TRANSACTION" 279 280 def commit_sql(self, *_) -> str: 281 return "COMMIT TRANSACTION" 282 283 def rollback_sql(self, *_) -> str: 284 return "ROLLBACK TRANSACTION" 285 286 def in_unnest_op(self, expression: exp.Unnest) -> str: 287 return self.sql(expression) 288 289 def except_op(self, expression: exp.Except) -> str: 290 if not expression.args.get("distinct", False): 291 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 292 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 293 294 def intersect_op(self, expression: exp.Intersect) -> str: 295 if not expression.args.get("distinct", False): 296 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 297 return f"INTERSECT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}"
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): if set to True all identifiers will be delimited by the corresponding character.
- 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
- generatedasidentitycolumnconstraint_sql
- notnullcolumnconstraint_sql
- primarykeycolumnconstraint_sql
- uniquecolumnconstraint_sql
- create_sql
- describe_sql
- prepend_ctes
- with_sql
- cte_sql
- tablealias_sql
- bitstring_sql
- hexstring_sql
- datatype_sql
- directory_sql
- delete_sql
- drop_sql
- except_sql
- 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
- insert_sql
- intersect_sql
- introducer_sql
- pseudotype_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
- 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
- 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
- altercolumn_sql
- renametable_sql
- altertable_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- intdiv_sql
- dpipe_sql
- div_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