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 min_or_least, 14 no_ilike_sql, 15 rename_func, 16 timestrtotime_sql, 17 ts_or_ds_to_date_sql, 18) 19from sqlglot.helper import seq_get 20from sqlglot.tokens import TokenType 21 22E = t.TypeVar("E", bound=exp.Expression) 23 24 25def _date_add(expression_class: t.Type[E]) -> t.Callable[[t.Sequence], E]: 26 def func(args): 27 interval = seq_get(args, 1) 28 return expression_class( 29 this=seq_get(args, 0), 30 expression=interval.this, 31 unit=interval.args.get("unit"), 32 ) 33 34 return func 35 36 37def _date_add_sql( 38 data_type: str, kind: str 39) -> t.Callable[[generator.Generator, exp.Expression], str]: 40 def func(self, expression): 41 this = self.sql(expression, "this") 42 unit = expression.args.get("unit") 43 unit = exp.var(unit.name.upper() if unit else "DAY") 44 interval = exp.Interval(this=expression.expression, unit=unit) 45 return f"{data_type}_{kind}({this}, {self.sql(interval)})" 46 47 return func 48 49 50def _derived_table_values_to_unnest(self: generator.Generator, expression: exp.Values) -> str: 51 if not isinstance(expression.unnest().parent, exp.From): 52 expression = t.cast(exp.Values, transforms.remove_precision_parameterized_types(expression)) 53 return self.values_sql(expression) 54 rows = [tuple_exp.expressions for tuple_exp in expression.find_all(exp.Tuple)] 55 structs = [] 56 for row in rows: 57 aliases = [ 58 exp.alias_(value, column_name) 59 for value, column_name in zip(row, expression.args["alias"].args["columns"]) 60 ] 61 structs.append(exp.Struct(expressions=aliases)) 62 unnest_exp = exp.Unnest(expressions=[exp.Array(expressions=structs)]) 63 return self.unnest_sql(unnest_exp) 64 65 66def _returnsproperty_sql(self: generator.Generator, expression: exp.ReturnsProperty) -> str: 67 this = expression.this 68 if isinstance(this, exp.Schema): 69 this = f"{this.this} <{self.expressions(this)}>" 70 else: 71 this = self.sql(this) 72 return f"RETURNS {this}" 73 74 75def _create_sql(self: generator.Generator, expression: exp.Create) -> str: 76 kind = expression.args["kind"] 77 returns = expression.find(exp.ReturnsProperty) 78 if kind.upper() == "FUNCTION" and returns and returns.args.get("is_table"): 79 expression = expression.copy() 80 expression.set("kind", "TABLE FUNCTION") 81 if isinstance( 82 expression.expression, 83 ( 84 exp.Subquery, 85 exp.Literal, 86 ), 87 ): 88 expression.set("expression", expression.expression.this) 89 90 return self.create_sql(expression) 91 92 return self.create_sql(expression) 93 94 95def _unqualify_unnest(expression: exp.Expression) -> exp.Expression: 96 """Remove references to unnest table aliases since bigquery doesn't allow them. 97 98 These are added by the optimizer's qualify_column step. 99 """ 100 if isinstance(expression, exp.Select): 101 unnests = { 102 unnest.alias 103 for unnest in expression.args.get("from", exp.From(expressions=[])).expressions 104 if isinstance(unnest, exp.Unnest) and unnest.alias 105 } 106 107 if unnests: 108 expression = expression.copy() 109 110 for select in expression.expressions: 111 for column in select.find_all(exp.Column): 112 if column.table in unnests: 113 column.set("table", None) 114 115 return expression 116 117 118class BigQuery(Dialect): 119 unnest_column_only = True 120 time_mapping = { 121 "%M": "%-M", 122 "%d": "%-d", 123 "%m": "%-m", 124 "%y": "%-y", 125 "%H": "%-H", 126 "%I": "%-I", 127 "%S": "%-S", 128 "%j": "%-j", 129 } 130 131 class Tokenizer(tokens.Tokenizer): 132 QUOTES = [ 133 (prefix + quote, quote) if prefix else quote 134 for quote in ["'", '"', '"""', "'''"] 135 for prefix in ["", "r", "R"] 136 ] 137 COMMENTS = ["--", "#", ("/*", "*/")] 138 IDENTIFIERS = ["`"] 139 STRING_ESCAPES = ["\\"] 140 HEX_STRINGS = [("0x", ""), ("0X", "")] 141 142 KEYWORDS = { 143 **tokens.Tokenizer.KEYWORDS, 144 "BEGIN": TokenType.COMMAND, 145 "BEGIN TRANSACTION": TokenType.BEGIN, 146 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 147 "CURRENT_TIME": TokenType.CURRENT_TIME, 148 "DECLARE": TokenType.COMMAND, 149 "GEOGRAPHY": TokenType.GEOGRAPHY, 150 "FLOAT64": TokenType.DOUBLE, 151 "INT64": TokenType.BIGINT, 152 "NOT DETERMINISTIC": TokenType.VOLATILE, 153 "UNKNOWN": TokenType.NULL, 154 } 155 KEYWORDS.pop("DIV") 156 157 class Parser(parser.Parser): 158 FUNCTIONS = { 159 **parser.Parser.FUNCTIONS, # type: ignore 160 "DATE_TRUNC": lambda args: exp.DateTrunc( 161 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 162 this=seq_get(args, 0), 163 ), 164 "DATE_ADD": _date_add(exp.DateAdd), 165 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 166 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 167 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 168 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 169 this=seq_get(args, 0), 170 expression=seq_get(args, 1), 171 position=seq_get(args, 2), 172 occurrence=seq_get(args, 3), 173 group=exp.Literal.number(1) 174 if re.compile(str(seq_get(args, 1))).groups == 1 175 else None, 176 ), 177 "TIME_ADD": _date_add(exp.TimeAdd), 178 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 179 "DATE_SUB": _date_add(exp.DateSub), 180 "DATETIME_SUB": _date_add(exp.DatetimeSub), 181 "TIME_SUB": _date_add(exp.TimeSub), 182 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 183 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 184 this=seq_get(args, 1), format=seq_get(args, 0) 185 ), 186 } 187 188 FUNCTION_PARSERS = { 189 **parser.Parser.FUNCTION_PARSERS, # type: ignore 190 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 191 } 192 FUNCTION_PARSERS.pop("TRIM") 193 194 NO_PAREN_FUNCTIONS = { 195 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 196 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 197 TokenType.CURRENT_TIME: exp.CurrentTime, 198 } 199 200 NESTED_TYPE_TOKENS = { 201 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 202 TokenType.TABLE, 203 } 204 205 ID_VAR_TOKENS = { 206 *parser.Parser.ID_VAR_TOKENS, # type: ignore 207 TokenType.VALUES, 208 } 209 210 PROPERTY_PARSERS = { 211 **parser.Parser.PROPERTY_PARSERS, # type: ignore 212 "NOT DETERMINISTIC": lambda self: self.expression( 213 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 214 ), 215 } 216 217 INTEGER_DIVISION = False 218 219 class Generator(generator.Generator): 220 INTEGER_DIVISION = False 221 222 TRANSFORMS = { 223 **generator.Generator.TRANSFORMS, # type: ignore 224 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 225 exp.ArraySize: rename_func("ARRAY_LENGTH"), 226 exp.DateAdd: _date_add_sql("DATE", "ADD"), 227 exp.DateSub: _date_add_sql("DATE", "SUB"), 228 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 229 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 230 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 231 exp.DateStrToDate: datestrtodate_sql, 232 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 233 exp.GroupConcat: rename_func("STRING_AGG"), 234 exp.ILike: no_ilike_sql, 235 exp.IntDiv: rename_func("DIV"), 236 exp.Min: min_or_least, 237 exp.Select: transforms.preprocess( 238 [_unqualify_unnest], transforms.delegate("select_sql") 239 ), 240 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 241 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 242 exp.TimeSub: _date_add_sql("TIME", "SUB"), 243 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 244 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 245 exp.TimeStrToTime: timestrtotime_sql, 246 exp.TsOrDsToDate: ts_or_ds_to_date_sql("bigquery"), 247 exp.TsOrDsAdd: _date_add_sql("DATE", "ADD"), 248 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 249 exp.VariancePop: rename_func("VAR_POP"), 250 exp.Values: _derived_table_values_to_unnest, 251 exp.ReturnsProperty: _returnsproperty_sql, 252 exp.Create: _create_sql, 253 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 254 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 255 if e.name == "IMMUTABLE" 256 else "NOT DETERMINISTIC", 257 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 258 } 259 260 TYPE_MAPPING = { 261 **generator.Generator.TYPE_MAPPING, # type: ignore 262 exp.DataType.Type.TINYINT: "INT64", 263 exp.DataType.Type.SMALLINT: "INT64", 264 exp.DataType.Type.INT: "INT64", 265 exp.DataType.Type.BIGINT: "INT64", 266 exp.DataType.Type.DECIMAL: "NUMERIC", 267 exp.DataType.Type.FLOAT: "FLOAT64", 268 exp.DataType.Type.DOUBLE: "FLOAT64", 269 exp.DataType.Type.BOOLEAN: "BOOL", 270 exp.DataType.Type.TEXT: "STRING", 271 exp.DataType.Type.VARCHAR: "STRING", 272 exp.DataType.Type.NVARCHAR: "STRING", 273 } 274 PROPERTIES_LOCATION = { 275 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 276 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 277 } 278 279 EXPLICIT_UNION = True 280 281 def array_sql(self, expression: exp.Array) -> str: 282 first_arg = seq_get(expression.expressions, 0) 283 if isinstance(first_arg, exp.Subqueryable): 284 return f"ARRAY{self.wrap(self.sql(first_arg))}" 285 286 return inline_array_sql(self, expression) 287 288 def transaction_sql(self, *_) -> str: 289 return "BEGIN TRANSACTION" 290 291 def commit_sql(self, *_) -> str: 292 return "COMMIT TRANSACTION" 293 294 def rollback_sql(self, *_) -> str: 295 return "ROLLBACK TRANSACTION" 296 297 def in_unnest_op(self, expression: exp.Unnest) -> str: 298 return self.sql(expression) 299 300 def except_op(self, expression: exp.Except) -> str: 301 if not expression.args.get("distinct", False): 302 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 303 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 304 305 def intersect_op(self, expression: exp.Intersect) -> str: 306 if not expression.args.get("distinct", False): 307 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 308 return f"INTERSECT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}"
119class BigQuery(Dialect): 120 unnest_column_only = True 121 time_mapping = { 122 "%M": "%-M", 123 "%d": "%-d", 124 "%m": "%-m", 125 "%y": "%-y", 126 "%H": "%-H", 127 "%I": "%-I", 128 "%S": "%-S", 129 "%j": "%-j", 130 } 131 132 class Tokenizer(tokens.Tokenizer): 133 QUOTES = [ 134 (prefix + quote, quote) if prefix else quote 135 for quote in ["'", '"', '"""', "'''"] 136 for prefix in ["", "r", "R"] 137 ] 138 COMMENTS = ["--", "#", ("/*", "*/")] 139 IDENTIFIERS = ["`"] 140 STRING_ESCAPES = ["\\"] 141 HEX_STRINGS = [("0x", ""), ("0X", "")] 142 143 KEYWORDS = { 144 **tokens.Tokenizer.KEYWORDS, 145 "BEGIN": TokenType.COMMAND, 146 "BEGIN TRANSACTION": TokenType.BEGIN, 147 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 148 "CURRENT_TIME": TokenType.CURRENT_TIME, 149 "DECLARE": TokenType.COMMAND, 150 "GEOGRAPHY": TokenType.GEOGRAPHY, 151 "FLOAT64": TokenType.DOUBLE, 152 "INT64": TokenType.BIGINT, 153 "NOT DETERMINISTIC": TokenType.VOLATILE, 154 "UNKNOWN": TokenType.NULL, 155 } 156 KEYWORDS.pop("DIV") 157 158 class Parser(parser.Parser): 159 FUNCTIONS = { 160 **parser.Parser.FUNCTIONS, # type: ignore 161 "DATE_TRUNC": lambda args: exp.DateTrunc( 162 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 163 this=seq_get(args, 0), 164 ), 165 "DATE_ADD": _date_add(exp.DateAdd), 166 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 167 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 168 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 169 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 170 this=seq_get(args, 0), 171 expression=seq_get(args, 1), 172 position=seq_get(args, 2), 173 occurrence=seq_get(args, 3), 174 group=exp.Literal.number(1) 175 if re.compile(str(seq_get(args, 1))).groups == 1 176 else None, 177 ), 178 "TIME_ADD": _date_add(exp.TimeAdd), 179 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 180 "DATE_SUB": _date_add(exp.DateSub), 181 "DATETIME_SUB": _date_add(exp.DatetimeSub), 182 "TIME_SUB": _date_add(exp.TimeSub), 183 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 184 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 185 this=seq_get(args, 1), format=seq_get(args, 0) 186 ), 187 } 188 189 FUNCTION_PARSERS = { 190 **parser.Parser.FUNCTION_PARSERS, # type: ignore 191 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 192 } 193 FUNCTION_PARSERS.pop("TRIM") 194 195 NO_PAREN_FUNCTIONS = { 196 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 197 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 198 TokenType.CURRENT_TIME: exp.CurrentTime, 199 } 200 201 NESTED_TYPE_TOKENS = { 202 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 203 TokenType.TABLE, 204 } 205 206 ID_VAR_TOKENS = { 207 *parser.Parser.ID_VAR_TOKENS, # type: ignore 208 TokenType.VALUES, 209 } 210 211 PROPERTY_PARSERS = { 212 **parser.Parser.PROPERTY_PARSERS, # type: ignore 213 "NOT DETERMINISTIC": lambda self: self.expression( 214 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 215 ), 216 } 217 218 INTEGER_DIVISION = False 219 220 class Generator(generator.Generator): 221 INTEGER_DIVISION = False 222 223 TRANSFORMS = { 224 **generator.Generator.TRANSFORMS, # type: ignore 225 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 226 exp.ArraySize: rename_func("ARRAY_LENGTH"), 227 exp.DateAdd: _date_add_sql("DATE", "ADD"), 228 exp.DateSub: _date_add_sql("DATE", "SUB"), 229 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 230 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 231 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 232 exp.DateStrToDate: datestrtodate_sql, 233 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 234 exp.GroupConcat: rename_func("STRING_AGG"), 235 exp.ILike: no_ilike_sql, 236 exp.IntDiv: rename_func("DIV"), 237 exp.Min: min_or_least, 238 exp.Select: transforms.preprocess( 239 [_unqualify_unnest], transforms.delegate("select_sql") 240 ), 241 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 242 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 243 exp.TimeSub: _date_add_sql("TIME", "SUB"), 244 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 245 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 246 exp.TimeStrToTime: timestrtotime_sql, 247 exp.TsOrDsToDate: ts_or_ds_to_date_sql("bigquery"), 248 exp.TsOrDsAdd: _date_add_sql("DATE", "ADD"), 249 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 250 exp.VariancePop: rename_func("VAR_POP"), 251 exp.Values: _derived_table_values_to_unnest, 252 exp.ReturnsProperty: _returnsproperty_sql, 253 exp.Create: _create_sql, 254 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 255 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 256 if e.name == "IMMUTABLE" 257 else "NOT DETERMINISTIC", 258 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 259 } 260 261 TYPE_MAPPING = { 262 **generator.Generator.TYPE_MAPPING, # type: ignore 263 exp.DataType.Type.TINYINT: "INT64", 264 exp.DataType.Type.SMALLINT: "INT64", 265 exp.DataType.Type.INT: "INT64", 266 exp.DataType.Type.BIGINT: "INT64", 267 exp.DataType.Type.DECIMAL: "NUMERIC", 268 exp.DataType.Type.FLOAT: "FLOAT64", 269 exp.DataType.Type.DOUBLE: "FLOAT64", 270 exp.DataType.Type.BOOLEAN: "BOOL", 271 exp.DataType.Type.TEXT: "STRING", 272 exp.DataType.Type.VARCHAR: "STRING", 273 exp.DataType.Type.NVARCHAR: "STRING", 274 } 275 PROPERTIES_LOCATION = { 276 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 277 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 278 } 279 280 EXPLICIT_UNION = True 281 282 def array_sql(self, expression: exp.Array) -> str: 283 first_arg = seq_get(expression.expressions, 0) 284 if isinstance(first_arg, exp.Subqueryable): 285 return f"ARRAY{self.wrap(self.sql(first_arg))}" 286 287 return inline_array_sql(self, expression) 288 289 def transaction_sql(self, *_) -> str: 290 return "BEGIN TRANSACTION" 291 292 def commit_sql(self, *_) -> str: 293 return "COMMIT TRANSACTION" 294 295 def rollback_sql(self, *_) -> str: 296 return "ROLLBACK TRANSACTION" 297 298 def in_unnest_op(self, expression: exp.Unnest) -> str: 299 return self.sql(expression) 300 301 def except_op(self, expression: exp.Except) -> str: 302 if not expression.args.get("distinct", False): 303 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 304 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 305 306 def intersect_op(self, expression: exp.Intersect) -> str: 307 if not expression.args.get("distinct", False): 308 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 309 return f"INTERSECT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}"
Inherited Members
132 class Tokenizer(tokens.Tokenizer): 133 QUOTES = [ 134 (prefix + quote, quote) if prefix else quote 135 for quote in ["'", '"', '"""', "'''"] 136 for prefix in ["", "r", "R"] 137 ] 138 COMMENTS = ["--", "#", ("/*", "*/")] 139 IDENTIFIERS = ["`"] 140 STRING_ESCAPES = ["\\"] 141 HEX_STRINGS = [("0x", ""), ("0X", "")] 142 143 KEYWORDS = { 144 **tokens.Tokenizer.KEYWORDS, 145 "BEGIN": TokenType.COMMAND, 146 "BEGIN TRANSACTION": TokenType.BEGIN, 147 "CURRENT_DATETIME": TokenType.CURRENT_DATETIME, 148 "CURRENT_TIME": TokenType.CURRENT_TIME, 149 "DECLARE": TokenType.COMMAND, 150 "GEOGRAPHY": TokenType.GEOGRAPHY, 151 "FLOAT64": TokenType.DOUBLE, 152 "INT64": TokenType.BIGINT, 153 "NOT DETERMINISTIC": TokenType.VOLATILE, 154 "UNKNOWN": TokenType.NULL, 155 } 156 KEYWORDS.pop("DIV")
Inherited Members
158 class Parser(parser.Parser): 159 FUNCTIONS = { 160 **parser.Parser.FUNCTIONS, # type: ignore 161 "DATE_TRUNC": lambda args: exp.DateTrunc( 162 unit=exp.Literal.string(seq_get(args, 1).name), # type: ignore 163 this=seq_get(args, 0), 164 ), 165 "DATE_ADD": _date_add(exp.DateAdd), 166 "DATETIME_ADD": _date_add(exp.DatetimeAdd), 167 "DIV": lambda args: exp.IntDiv(this=seq_get(args, 0), expression=seq_get(args, 1)), 168 "REGEXP_CONTAINS": exp.RegexpLike.from_arg_list, 169 "REGEXP_EXTRACT": lambda args: exp.RegexpExtract( 170 this=seq_get(args, 0), 171 expression=seq_get(args, 1), 172 position=seq_get(args, 2), 173 occurrence=seq_get(args, 3), 174 group=exp.Literal.number(1) 175 if re.compile(str(seq_get(args, 1))).groups == 1 176 else None, 177 ), 178 "TIME_ADD": _date_add(exp.TimeAdd), 179 "TIMESTAMP_ADD": _date_add(exp.TimestampAdd), 180 "DATE_SUB": _date_add(exp.DateSub), 181 "DATETIME_SUB": _date_add(exp.DatetimeSub), 182 "TIME_SUB": _date_add(exp.TimeSub), 183 "TIMESTAMP_SUB": _date_add(exp.TimestampSub), 184 "PARSE_TIMESTAMP": lambda args: exp.StrToTime( 185 this=seq_get(args, 1), format=seq_get(args, 0) 186 ), 187 } 188 189 FUNCTION_PARSERS = { 190 **parser.Parser.FUNCTION_PARSERS, # type: ignore 191 "ARRAY": lambda self: self.expression(exp.Array, expressions=[self._parse_statement()]), 192 } 193 FUNCTION_PARSERS.pop("TRIM") 194 195 NO_PAREN_FUNCTIONS = { 196 **parser.Parser.NO_PAREN_FUNCTIONS, # type: ignore 197 TokenType.CURRENT_DATETIME: exp.CurrentDatetime, 198 TokenType.CURRENT_TIME: exp.CurrentTime, 199 } 200 201 NESTED_TYPE_TOKENS = { 202 *parser.Parser.NESTED_TYPE_TOKENS, # type: ignore 203 TokenType.TABLE, 204 } 205 206 ID_VAR_TOKENS = { 207 *parser.Parser.ID_VAR_TOKENS, # type: ignore 208 TokenType.VALUES, 209 } 210 211 PROPERTY_PARSERS = { 212 **parser.Parser.PROPERTY_PARSERS, # type: ignore 213 "NOT DETERMINISTIC": lambda self: self.expression( 214 exp.VolatilityProperty, this=exp.Literal.string("VOLATILE") 215 ), 216 } 217 218 INTEGER_DIVISION = False
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
220 class Generator(generator.Generator): 221 INTEGER_DIVISION = False 222 223 TRANSFORMS = { 224 **generator.Generator.TRANSFORMS, # type: ignore 225 **transforms.REMOVE_PRECISION_PARAMETERIZED_TYPES, # type: ignore 226 exp.ArraySize: rename_func("ARRAY_LENGTH"), 227 exp.DateAdd: _date_add_sql("DATE", "ADD"), 228 exp.DateSub: _date_add_sql("DATE", "SUB"), 229 exp.DatetimeAdd: _date_add_sql("DATETIME", "ADD"), 230 exp.DatetimeSub: _date_add_sql("DATETIME", "SUB"), 231 exp.DateDiff: lambda self, e: f"DATE_DIFF({self.sql(e, 'this')}, {self.sql(e, 'expression')}, {self.sql(e.args.get('unit', 'DAY'))})", 232 exp.DateStrToDate: datestrtodate_sql, 233 exp.DateTrunc: lambda self, e: self.func("DATE_TRUNC", e.this, e.text("unit")), 234 exp.GroupConcat: rename_func("STRING_AGG"), 235 exp.ILike: no_ilike_sql, 236 exp.IntDiv: rename_func("DIV"), 237 exp.Min: min_or_least, 238 exp.Select: transforms.preprocess( 239 [_unqualify_unnest], transforms.delegate("select_sql") 240 ), 241 exp.StrToTime: lambda self, e: f"PARSE_TIMESTAMP({self.format_time(e)}, {self.sql(e, 'this')})", 242 exp.TimeAdd: _date_add_sql("TIME", "ADD"), 243 exp.TimeSub: _date_add_sql("TIME", "SUB"), 244 exp.TimestampAdd: _date_add_sql("TIMESTAMP", "ADD"), 245 exp.TimestampSub: _date_add_sql("TIMESTAMP", "SUB"), 246 exp.TimeStrToTime: timestrtotime_sql, 247 exp.TsOrDsToDate: ts_or_ds_to_date_sql("bigquery"), 248 exp.TsOrDsAdd: _date_add_sql("DATE", "ADD"), 249 exp.PartitionedByProperty: lambda self, e: f"PARTITION BY {self.sql(e, 'this')}", 250 exp.VariancePop: rename_func("VAR_POP"), 251 exp.Values: _derived_table_values_to_unnest, 252 exp.ReturnsProperty: _returnsproperty_sql, 253 exp.Create: _create_sql, 254 exp.Trim: lambda self, e: self.func(f"TRIM", e.this, e.expression), 255 exp.VolatilityProperty: lambda self, e: f"DETERMINISTIC" 256 if e.name == "IMMUTABLE" 257 else "NOT DETERMINISTIC", 258 exp.RegexpLike: rename_func("REGEXP_CONTAINS"), 259 } 260 261 TYPE_MAPPING = { 262 **generator.Generator.TYPE_MAPPING, # type: ignore 263 exp.DataType.Type.TINYINT: "INT64", 264 exp.DataType.Type.SMALLINT: "INT64", 265 exp.DataType.Type.INT: "INT64", 266 exp.DataType.Type.BIGINT: "INT64", 267 exp.DataType.Type.DECIMAL: "NUMERIC", 268 exp.DataType.Type.FLOAT: "FLOAT64", 269 exp.DataType.Type.DOUBLE: "FLOAT64", 270 exp.DataType.Type.BOOLEAN: "BOOL", 271 exp.DataType.Type.TEXT: "STRING", 272 exp.DataType.Type.VARCHAR: "STRING", 273 exp.DataType.Type.NVARCHAR: "STRING", 274 } 275 PROPERTIES_LOCATION = { 276 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 277 exp.PartitionedByProperty: exp.Properties.Location.POST_SCHEMA, 278 } 279 280 EXPLICIT_UNION = True 281 282 def array_sql(self, expression: exp.Array) -> str: 283 first_arg = seq_get(expression.expressions, 0) 284 if isinstance(first_arg, exp.Subqueryable): 285 return f"ARRAY{self.wrap(self.sql(first_arg))}" 286 287 return inline_array_sql(self, expression) 288 289 def transaction_sql(self, *_) -> str: 290 return "BEGIN TRANSACTION" 291 292 def commit_sql(self, *_) -> str: 293 return "COMMIT TRANSACTION" 294 295 def rollback_sql(self, *_) -> str: 296 return "ROLLBACK TRANSACTION" 297 298 def in_unnest_op(self, expression: exp.Unnest) -> str: 299 return self.sql(expression) 300 301 def except_op(self, expression: exp.Except) -> str: 302 if not expression.args.get("distinct", False): 303 self.unsupported("EXCEPT without DISTINCT is not supported in BigQuery") 304 return f"EXCEPT{' DISTINCT' if expression.args.get('distinct') else ' ALL'}" 305 306 def intersect_op(self, expression: exp.Intersect) -> str: 307 if not expression.args.get("distinct", False): 308 self.unsupported("INTERSECT without DISTINCT is not supported in BigQuery") 309 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
- compresscolumnconstraint_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
- withdataproperty_sql
- insert_sql
- intersect_sql
- 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
- 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
- comment_sql
- altercolumn_sql
- renametable_sql
- altertable_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- intdiv_sql
- dpipe_sql
- div_sql
- floatdiv_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