sqlglot.dialects.redshift
1from __future__ import annotations 2 3import typing as t 4 5from sqlglot import exp, transforms 6from sqlglot.dialects.dialect import ( 7 NormalizationStrategy, 8 concat_to_dpipe_sql, 9 concat_ws_to_dpipe_sql, 10 date_delta_sql, 11 generatedasidentitycolumnconstraint_sql, 12 json_extract_segments, 13 no_tablesample_sql, 14 rename_func, 15 map_date_part, 16) 17from sqlglot.dialects.postgres import Postgres 18from sqlglot.helper import seq_get 19from sqlglot.tokens import TokenType 20from sqlglot.parser import build_convert_timezone 21 22if t.TYPE_CHECKING: 23 from sqlglot._typing import E 24 25 26def _build_date_delta(expr_type: t.Type[E]) -> t.Callable[[t.List], E]: 27 def _builder(args: t.List) -> E: 28 expr = expr_type( 29 this=seq_get(args, 2), 30 expression=seq_get(args, 1), 31 unit=map_date_part(seq_get(args, 0)), 32 ) 33 if expr_type is exp.TsOrDsAdd: 34 expr.set("return_type", exp.DataType.build("TIMESTAMP")) 35 36 return expr 37 38 return _builder 39 40 41class Redshift(Postgres): 42 # https://docs.aws.amazon.com/redshift/latest/dg/r_names.html 43 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 44 45 SUPPORTS_USER_DEFINED_TYPES = False 46 INDEX_OFFSET = 0 47 COPY_PARAMS_ARE_CSV = False 48 HEX_LOWERCASE = True 49 HAS_DISTINCT_ARRAY_CONSTRUCTORS = True 50 51 # ref: https://docs.aws.amazon.com/redshift/latest/dg/r_FORMAT_strings.html 52 TIME_FORMAT = "'YYYY-MM-DD HH24:MI:SS'" 53 TIME_MAPPING = {**Postgres.TIME_MAPPING, "MON": "%b", "HH24": "%H", "HH": "%I"} 54 55 class Parser(Postgres.Parser): 56 FUNCTIONS = { 57 **Postgres.Parser.FUNCTIONS, 58 "ADD_MONTHS": lambda args: exp.TsOrDsAdd( 59 this=seq_get(args, 0), 60 expression=seq_get(args, 1), 61 unit=exp.var("month"), 62 return_type=exp.DataType.build("TIMESTAMP"), 63 ), 64 "CONVERT_TIMEZONE": lambda args: build_convert_timezone(args, "UTC"), 65 "DATEADD": _build_date_delta(exp.TsOrDsAdd), 66 "DATE_ADD": _build_date_delta(exp.TsOrDsAdd), 67 "DATEDIFF": _build_date_delta(exp.TsOrDsDiff), 68 "DATE_DIFF": _build_date_delta(exp.TsOrDsDiff), 69 "GETDATE": exp.CurrentTimestamp.from_arg_list, 70 "LISTAGG": exp.GroupConcat.from_arg_list, 71 "SPLIT_TO_ARRAY": lambda args: exp.StringToArray( 72 this=seq_get(args, 0), expression=seq_get(args, 1) or exp.Literal.string(",") 73 ), 74 "STRTOL": exp.FromBase.from_arg_list, 75 } 76 77 NO_PAREN_FUNCTION_PARSERS = { 78 **Postgres.Parser.NO_PAREN_FUNCTION_PARSERS, 79 "APPROXIMATE": lambda self: self._parse_approximate_count(), 80 "SYSDATE": lambda self: self.expression(exp.CurrentTimestamp, sysdate=True), 81 } 82 83 SUPPORTS_IMPLICIT_UNNEST = True 84 85 def _parse_table( 86 self, 87 schema: bool = False, 88 joins: bool = False, 89 alias_tokens: t.Optional[t.Collection[TokenType]] = None, 90 parse_bracket: bool = False, 91 is_db_reference: bool = False, 92 parse_partition: bool = False, 93 ) -> t.Optional[exp.Expression]: 94 # Redshift supports UNPIVOTing SUPER objects, e.g. `UNPIVOT foo.obj[0] AS val AT attr` 95 unpivot = self._match(TokenType.UNPIVOT) 96 table = super()._parse_table( 97 schema=schema, 98 joins=joins, 99 alias_tokens=alias_tokens, 100 parse_bracket=parse_bracket, 101 is_db_reference=is_db_reference, 102 ) 103 104 return self.expression(exp.Pivot, this=table, unpivot=True) if unpivot else table 105 106 def _parse_convert( 107 self, strict: bool, safe: t.Optional[bool] = None 108 ) -> t.Optional[exp.Expression]: 109 to = self._parse_types() 110 self._match(TokenType.COMMA) 111 this = self._parse_bitwise() 112 return self.expression(exp.TryCast, this=this, to=to, safe=safe) 113 114 def _parse_approximate_count(self) -> t.Optional[exp.ApproxDistinct]: 115 index = self._index - 1 116 func = self._parse_function() 117 118 if isinstance(func, exp.Count) and isinstance(func.this, exp.Distinct): 119 return self.expression(exp.ApproxDistinct, this=seq_get(func.this.expressions, 0)) 120 self._retreat(index) 121 return None 122 123 class Tokenizer(Postgres.Tokenizer): 124 BIT_STRINGS = [] 125 HEX_STRINGS = [] 126 STRING_ESCAPES = ["\\", "'"] 127 128 KEYWORDS = { 129 **Postgres.Tokenizer.KEYWORDS, 130 "(+)": TokenType.JOIN_MARKER, 131 "HLLSKETCH": TokenType.HLLSKETCH, 132 "MINUS": TokenType.EXCEPT, 133 "SUPER": TokenType.SUPER, 134 "TOP": TokenType.TOP, 135 "UNLOAD": TokenType.COMMAND, 136 "VARBYTE": TokenType.VARBINARY, 137 "BINARY VARYING": TokenType.VARBINARY, 138 } 139 KEYWORDS.pop("VALUES") 140 141 # Redshift allows # to appear as a table identifier prefix 142 SINGLE_TOKENS = Postgres.Tokenizer.SINGLE_TOKENS.copy() 143 SINGLE_TOKENS.pop("#") 144 145 class Generator(Postgres.Generator): 146 LOCKING_READS_SUPPORTED = False 147 QUERY_HINTS = False 148 VALUES_AS_TABLE = False 149 TZ_TO_WITH_TIME_ZONE = True 150 NVL2_SUPPORTED = True 151 LAST_DAY_SUPPORTS_DATE_PART = False 152 CAN_IMPLEMENT_ARRAY_ANY = False 153 MULTI_ARG_DISTINCT = True 154 COPY_PARAMS_ARE_WRAPPED = False 155 HEX_FUNC = "TO_HEX" 156 PARSE_JSON_NAME = "JSON_PARSE" 157 ARRAY_CONCAT_IS_VAR_LEN = False 158 SUPPORTS_CONVERT_TIMEZONE = True 159 EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False 160 SUPPORTS_MEDIAN = True 161 ALTER_SET_TYPE = "TYPE" 162 163 # Redshift doesn't have `WITH` as part of their with_properties so we remove it 164 WITH_PROPERTIES_PREFIX = " " 165 166 TYPE_MAPPING = { 167 **Postgres.Generator.TYPE_MAPPING, 168 exp.DataType.Type.BINARY: "VARBYTE", 169 exp.DataType.Type.BLOB: "VARBYTE", 170 exp.DataType.Type.INT: "INTEGER", 171 exp.DataType.Type.TIMETZ: "TIME", 172 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 173 exp.DataType.Type.VARBINARY: "VARBYTE", 174 exp.DataType.Type.ROWVERSION: "VARBYTE", 175 } 176 177 TRANSFORMS = { 178 **Postgres.Generator.TRANSFORMS, 179 exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CONCAT"), 180 exp.Concat: concat_to_dpipe_sql, 181 exp.ConcatWs: concat_ws_to_dpipe_sql, 182 exp.ApproxDistinct: lambda self, 183 e: f"APPROXIMATE COUNT(DISTINCT {self.sql(e, 'this')})", 184 exp.CurrentTimestamp: lambda self, e: ( 185 "SYSDATE" if e.args.get("sysdate") else "GETDATE()" 186 ), 187 exp.DateAdd: date_delta_sql("DATEADD"), 188 exp.DateDiff: date_delta_sql("DATEDIFF"), 189 exp.DistKeyProperty: lambda self, e: self.func("DISTKEY", e.this), 190 exp.DistStyleProperty: lambda self, e: self.naked_property(e), 191 exp.Explode: lambda self, e: self.explode_sql(e), 192 exp.FromBase: rename_func("STRTOL"), 193 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 194 exp.JSONExtract: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 195 exp.JSONExtractScalar: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 196 exp.GroupConcat: rename_func("LISTAGG"), 197 exp.Hex: lambda self, e: self.func("UPPER", self.func("TO_HEX", self.sql(e, "this"))), 198 exp.Select: transforms.preprocess( 199 [ 200 transforms.eliminate_distinct_on, 201 transforms.eliminate_semi_and_anti_joins, 202 transforms.unqualify_unnest, 203 transforms.unnest_generate_date_array_using_recursive_cte, 204 ] 205 ), 206 exp.SortKeyProperty: lambda self, 207 e: f"{'COMPOUND ' if e.args['compound'] else ''}SORTKEY({self.format_args(*e.this)})", 208 exp.StartsWith: lambda self, 209 e: f"{self.sql(e.this)} LIKE {self.sql(e.expression)} || '%'", 210 exp.StringToArray: rename_func("SPLIT_TO_ARRAY"), 211 exp.TableSample: no_tablesample_sql, 212 exp.TsOrDsAdd: date_delta_sql("DATEADD"), 213 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 214 exp.UnixToTime: lambda self, 215 e: f"(TIMESTAMP 'epoch' + {self.sql(e.this)} * INTERVAL '1 SECOND')", 216 } 217 218 # Postgres maps exp.Pivot to no_pivot_sql, but Redshift support pivots 219 TRANSFORMS.pop(exp.Pivot) 220 221 # Postgres doesn't support JSON_PARSE, but Redshift does 222 TRANSFORMS.pop(exp.ParseJSON) 223 224 # Redshift uses the POW | POWER (expr1, expr2) syntax instead of expr1 ^ expr2 (postgres) 225 TRANSFORMS.pop(exp.Pow) 226 227 # Redshift supports these functions 228 TRANSFORMS.pop(exp.AnyValue) 229 TRANSFORMS.pop(exp.LastDay) 230 TRANSFORMS.pop(exp.SHA2) 231 232 RESERVED_KEYWORDS = { 233 "aes128", 234 "aes256", 235 "all", 236 "allowoverwrite", 237 "analyse", 238 "analyze", 239 "and", 240 "any", 241 "array", 242 "as", 243 "asc", 244 "authorization", 245 "az64", 246 "backup", 247 "between", 248 "binary", 249 "blanksasnull", 250 "both", 251 "bytedict", 252 "bzip2", 253 "case", 254 "cast", 255 "check", 256 "collate", 257 "column", 258 "constraint", 259 "create", 260 "credentials", 261 "cross", 262 "current_date", 263 "current_time", 264 "current_timestamp", 265 "current_user", 266 "current_user_id", 267 "default", 268 "deferrable", 269 "deflate", 270 "defrag", 271 "delta", 272 "delta32k", 273 "desc", 274 "disable", 275 "distinct", 276 "do", 277 "else", 278 "emptyasnull", 279 "enable", 280 "encode", 281 "encrypt ", 282 "encryption", 283 "end", 284 "except", 285 "explicit", 286 "false", 287 "for", 288 "foreign", 289 "freeze", 290 "from", 291 "full", 292 "globaldict256", 293 "globaldict64k", 294 "grant", 295 "group", 296 "gzip", 297 "having", 298 "identity", 299 "ignore", 300 "ilike", 301 "in", 302 "initially", 303 "inner", 304 "intersect", 305 "interval", 306 "into", 307 "is", 308 "isnull", 309 "join", 310 "leading", 311 "left", 312 "like", 313 "limit", 314 "localtime", 315 "localtimestamp", 316 "lun", 317 "luns", 318 "lzo", 319 "lzop", 320 "minus", 321 "mostly16", 322 "mostly32", 323 "mostly8", 324 "natural", 325 "new", 326 "not", 327 "notnull", 328 "null", 329 "nulls", 330 "off", 331 "offline", 332 "offset", 333 "oid", 334 "old", 335 "on", 336 "only", 337 "open", 338 "or", 339 "order", 340 "outer", 341 "overlaps", 342 "parallel", 343 "partition", 344 "percent", 345 "permissions", 346 "pivot", 347 "placing", 348 "primary", 349 "raw", 350 "readratio", 351 "recover", 352 "references", 353 "rejectlog", 354 "resort", 355 "respect", 356 "restore", 357 "right", 358 "select", 359 "session_user", 360 "similar", 361 "snapshot", 362 "some", 363 "sysdate", 364 "system", 365 "table", 366 "tag", 367 "tdes", 368 "text255", 369 "text32k", 370 "then", 371 "timestamp", 372 "to", 373 "top", 374 "trailing", 375 "true", 376 "truncatecolumns", 377 "type", 378 "union", 379 "unique", 380 "unnest", 381 "unpivot", 382 "user", 383 "using", 384 "verbose", 385 "wallet", 386 "when", 387 "where", 388 "with", 389 "without", 390 } 391 392 def unnest_sql(self, expression: exp.Unnest) -> str: 393 args = expression.expressions 394 num_args = len(args) 395 396 if num_args != 1: 397 self.unsupported(f"Unsupported number of arguments in UNNEST: {num_args}") 398 return "" 399 400 if isinstance(expression.find_ancestor(exp.From, exp.Join, exp.Select), exp.Select): 401 self.unsupported("Unsupported UNNEST when not used in FROM/JOIN clauses") 402 return "" 403 404 arg = self.sql(seq_get(args, 0)) 405 406 alias = self.expressions(expression.args.get("alias"), key="columns", flat=True) 407 return f"{arg} AS {alias}" if alias else arg 408 409 def cast_sql(self, expression: exp.Cast, safe_prefix: t.Optional[str] = None) -> str: 410 if expression.is_type(exp.DataType.Type.JSON): 411 # Redshift doesn't support a JSON type, so casting to it is treated as a noop 412 return self.sql(expression, "this") 413 414 return super().cast_sql(expression, safe_prefix=safe_prefix) 415 416 def datatype_sql(self, expression: exp.DataType) -> str: 417 """ 418 Redshift converts the `TEXT` data type to `VARCHAR(255)` by default when people more generally mean 419 VARCHAR of max length which is `VARCHAR(max)` in Redshift. Therefore if we get a `TEXT` data type 420 without precision we convert it to `VARCHAR(max)` and if it does have precision then we just convert 421 `TEXT` to `VARCHAR`. 422 """ 423 if expression.is_type("text"): 424 expression.set("this", exp.DataType.Type.VARCHAR) 425 precision = expression.args.get("expressions") 426 427 if not precision: 428 expression.append("expressions", exp.var("MAX")) 429 430 return super().datatype_sql(expression) 431 432 def alterset_sql(self, expression: exp.AlterSet) -> str: 433 exprs = self.expressions(expression, flat=True) 434 exprs = f" TABLE PROPERTIES ({exprs})" if exprs else "" 435 location = self.sql(expression, "location") 436 location = f" LOCATION {location}" if location else "" 437 file_format = self.expressions(expression, key="file_format", flat=True, sep=" ") 438 file_format = f" FILE FORMAT {file_format}" if file_format else "" 439 440 return f"SET{exprs}{location}{file_format}" 441 442 def array_sql(self, expression: exp.Array) -> str: 443 if expression.args.get("bracket_notation"): 444 return super().array_sql(expression) 445 446 return rename_func("ARRAY")(self, expression) 447 448 def explode_sql(self, expression: exp.Explode) -> str: 449 self.unsupported("Unsupported EXPLODE() function") 450 return ""
42class Redshift(Postgres): 43 # https://docs.aws.amazon.com/redshift/latest/dg/r_names.html 44 NORMALIZATION_STRATEGY = NormalizationStrategy.CASE_INSENSITIVE 45 46 SUPPORTS_USER_DEFINED_TYPES = False 47 INDEX_OFFSET = 0 48 COPY_PARAMS_ARE_CSV = False 49 HEX_LOWERCASE = True 50 HAS_DISTINCT_ARRAY_CONSTRUCTORS = True 51 52 # ref: https://docs.aws.amazon.com/redshift/latest/dg/r_FORMAT_strings.html 53 TIME_FORMAT = "'YYYY-MM-DD HH24:MI:SS'" 54 TIME_MAPPING = {**Postgres.TIME_MAPPING, "MON": "%b", "HH24": "%H", "HH": "%I"} 55 56 class Parser(Postgres.Parser): 57 FUNCTIONS = { 58 **Postgres.Parser.FUNCTIONS, 59 "ADD_MONTHS": lambda args: exp.TsOrDsAdd( 60 this=seq_get(args, 0), 61 expression=seq_get(args, 1), 62 unit=exp.var("month"), 63 return_type=exp.DataType.build("TIMESTAMP"), 64 ), 65 "CONVERT_TIMEZONE": lambda args: build_convert_timezone(args, "UTC"), 66 "DATEADD": _build_date_delta(exp.TsOrDsAdd), 67 "DATE_ADD": _build_date_delta(exp.TsOrDsAdd), 68 "DATEDIFF": _build_date_delta(exp.TsOrDsDiff), 69 "DATE_DIFF": _build_date_delta(exp.TsOrDsDiff), 70 "GETDATE": exp.CurrentTimestamp.from_arg_list, 71 "LISTAGG": exp.GroupConcat.from_arg_list, 72 "SPLIT_TO_ARRAY": lambda args: exp.StringToArray( 73 this=seq_get(args, 0), expression=seq_get(args, 1) or exp.Literal.string(",") 74 ), 75 "STRTOL": exp.FromBase.from_arg_list, 76 } 77 78 NO_PAREN_FUNCTION_PARSERS = { 79 **Postgres.Parser.NO_PAREN_FUNCTION_PARSERS, 80 "APPROXIMATE": lambda self: self._parse_approximate_count(), 81 "SYSDATE": lambda self: self.expression(exp.CurrentTimestamp, sysdate=True), 82 } 83 84 SUPPORTS_IMPLICIT_UNNEST = True 85 86 def _parse_table( 87 self, 88 schema: bool = False, 89 joins: bool = False, 90 alias_tokens: t.Optional[t.Collection[TokenType]] = None, 91 parse_bracket: bool = False, 92 is_db_reference: bool = False, 93 parse_partition: bool = False, 94 ) -> t.Optional[exp.Expression]: 95 # Redshift supports UNPIVOTing SUPER objects, e.g. `UNPIVOT foo.obj[0] AS val AT attr` 96 unpivot = self._match(TokenType.UNPIVOT) 97 table = super()._parse_table( 98 schema=schema, 99 joins=joins, 100 alias_tokens=alias_tokens, 101 parse_bracket=parse_bracket, 102 is_db_reference=is_db_reference, 103 ) 104 105 return self.expression(exp.Pivot, this=table, unpivot=True) if unpivot else table 106 107 def _parse_convert( 108 self, strict: bool, safe: t.Optional[bool] = None 109 ) -> t.Optional[exp.Expression]: 110 to = self._parse_types() 111 self._match(TokenType.COMMA) 112 this = self._parse_bitwise() 113 return self.expression(exp.TryCast, this=this, to=to, safe=safe) 114 115 def _parse_approximate_count(self) -> t.Optional[exp.ApproxDistinct]: 116 index = self._index - 1 117 func = self._parse_function() 118 119 if isinstance(func, exp.Count) and isinstance(func.this, exp.Distinct): 120 return self.expression(exp.ApproxDistinct, this=seq_get(func.this.expressions, 0)) 121 self._retreat(index) 122 return None 123 124 class Tokenizer(Postgres.Tokenizer): 125 BIT_STRINGS = [] 126 HEX_STRINGS = [] 127 STRING_ESCAPES = ["\\", "'"] 128 129 KEYWORDS = { 130 **Postgres.Tokenizer.KEYWORDS, 131 "(+)": TokenType.JOIN_MARKER, 132 "HLLSKETCH": TokenType.HLLSKETCH, 133 "MINUS": TokenType.EXCEPT, 134 "SUPER": TokenType.SUPER, 135 "TOP": TokenType.TOP, 136 "UNLOAD": TokenType.COMMAND, 137 "VARBYTE": TokenType.VARBINARY, 138 "BINARY VARYING": TokenType.VARBINARY, 139 } 140 KEYWORDS.pop("VALUES") 141 142 # Redshift allows # to appear as a table identifier prefix 143 SINGLE_TOKENS = Postgres.Tokenizer.SINGLE_TOKENS.copy() 144 SINGLE_TOKENS.pop("#") 145 146 class Generator(Postgres.Generator): 147 LOCKING_READS_SUPPORTED = False 148 QUERY_HINTS = False 149 VALUES_AS_TABLE = False 150 TZ_TO_WITH_TIME_ZONE = True 151 NVL2_SUPPORTED = True 152 LAST_DAY_SUPPORTS_DATE_PART = False 153 CAN_IMPLEMENT_ARRAY_ANY = False 154 MULTI_ARG_DISTINCT = True 155 COPY_PARAMS_ARE_WRAPPED = False 156 HEX_FUNC = "TO_HEX" 157 PARSE_JSON_NAME = "JSON_PARSE" 158 ARRAY_CONCAT_IS_VAR_LEN = False 159 SUPPORTS_CONVERT_TIMEZONE = True 160 EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False 161 SUPPORTS_MEDIAN = True 162 ALTER_SET_TYPE = "TYPE" 163 164 # Redshift doesn't have `WITH` as part of their with_properties so we remove it 165 WITH_PROPERTIES_PREFIX = " " 166 167 TYPE_MAPPING = { 168 **Postgres.Generator.TYPE_MAPPING, 169 exp.DataType.Type.BINARY: "VARBYTE", 170 exp.DataType.Type.BLOB: "VARBYTE", 171 exp.DataType.Type.INT: "INTEGER", 172 exp.DataType.Type.TIMETZ: "TIME", 173 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 174 exp.DataType.Type.VARBINARY: "VARBYTE", 175 exp.DataType.Type.ROWVERSION: "VARBYTE", 176 } 177 178 TRANSFORMS = { 179 **Postgres.Generator.TRANSFORMS, 180 exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CONCAT"), 181 exp.Concat: concat_to_dpipe_sql, 182 exp.ConcatWs: concat_ws_to_dpipe_sql, 183 exp.ApproxDistinct: lambda self, 184 e: f"APPROXIMATE COUNT(DISTINCT {self.sql(e, 'this')})", 185 exp.CurrentTimestamp: lambda self, e: ( 186 "SYSDATE" if e.args.get("sysdate") else "GETDATE()" 187 ), 188 exp.DateAdd: date_delta_sql("DATEADD"), 189 exp.DateDiff: date_delta_sql("DATEDIFF"), 190 exp.DistKeyProperty: lambda self, e: self.func("DISTKEY", e.this), 191 exp.DistStyleProperty: lambda self, e: self.naked_property(e), 192 exp.Explode: lambda self, e: self.explode_sql(e), 193 exp.FromBase: rename_func("STRTOL"), 194 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 195 exp.JSONExtract: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 196 exp.JSONExtractScalar: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 197 exp.GroupConcat: rename_func("LISTAGG"), 198 exp.Hex: lambda self, e: self.func("UPPER", self.func("TO_HEX", self.sql(e, "this"))), 199 exp.Select: transforms.preprocess( 200 [ 201 transforms.eliminate_distinct_on, 202 transforms.eliminate_semi_and_anti_joins, 203 transforms.unqualify_unnest, 204 transforms.unnest_generate_date_array_using_recursive_cte, 205 ] 206 ), 207 exp.SortKeyProperty: lambda self, 208 e: f"{'COMPOUND ' if e.args['compound'] else ''}SORTKEY({self.format_args(*e.this)})", 209 exp.StartsWith: lambda self, 210 e: f"{self.sql(e.this)} LIKE {self.sql(e.expression)} || '%'", 211 exp.StringToArray: rename_func("SPLIT_TO_ARRAY"), 212 exp.TableSample: no_tablesample_sql, 213 exp.TsOrDsAdd: date_delta_sql("DATEADD"), 214 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 215 exp.UnixToTime: lambda self, 216 e: f"(TIMESTAMP 'epoch' + {self.sql(e.this)} * INTERVAL '1 SECOND')", 217 } 218 219 # Postgres maps exp.Pivot to no_pivot_sql, but Redshift support pivots 220 TRANSFORMS.pop(exp.Pivot) 221 222 # Postgres doesn't support JSON_PARSE, but Redshift does 223 TRANSFORMS.pop(exp.ParseJSON) 224 225 # Redshift uses the POW | POWER (expr1, expr2) syntax instead of expr1 ^ expr2 (postgres) 226 TRANSFORMS.pop(exp.Pow) 227 228 # Redshift supports these functions 229 TRANSFORMS.pop(exp.AnyValue) 230 TRANSFORMS.pop(exp.LastDay) 231 TRANSFORMS.pop(exp.SHA2) 232 233 RESERVED_KEYWORDS = { 234 "aes128", 235 "aes256", 236 "all", 237 "allowoverwrite", 238 "analyse", 239 "analyze", 240 "and", 241 "any", 242 "array", 243 "as", 244 "asc", 245 "authorization", 246 "az64", 247 "backup", 248 "between", 249 "binary", 250 "blanksasnull", 251 "both", 252 "bytedict", 253 "bzip2", 254 "case", 255 "cast", 256 "check", 257 "collate", 258 "column", 259 "constraint", 260 "create", 261 "credentials", 262 "cross", 263 "current_date", 264 "current_time", 265 "current_timestamp", 266 "current_user", 267 "current_user_id", 268 "default", 269 "deferrable", 270 "deflate", 271 "defrag", 272 "delta", 273 "delta32k", 274 "desc", 275 "disable", 276 "distinct", 277 "do", 278 "else", 279 "emptyasnull", 280 "enable", 281 "encode", 282 "encrypt ", 283 "encryption", 284 "end", 285 "except", 286 "explicit", 287 "false", 288 "for", 289 "foreign", 290 "freeze", 291 "from", 292 "full", 293 "globaldict256", 294 "globaldict64k", 295 "grant", 296 "group", 297 "gzip", 298 "having", 299 "identity", 300 "ignore", 301 "ilike", 302 "in", 303 "initially", 304 "inner", 305 "intersect", 306 "interval", 307 "into", 308 "is", 309 "isnull", 310 "join", 311 "leading", 312 "left", 313 "like", 314 "limit", 315 "localtime", 316 "localtimestamp", 317 "lun", 318 "luns", 319 "lzo", 320 "lzop", 321 "minus", 322 "mostly16", 323 "mostly32", 324 "mostly8", 325 "natural", 326 "new", 327 "not", 328 "notnull", 329 "null", 330 "nulls", 331 "off", 332 "offline", 333 "offset", 334 "oid", 335 "old", 336 "on", 337 "only", 338 "open", 339 "or", 340 "order", 341 "outer", 342 "overlaps", 343 "parallel", 344 "partition", 345 "percent", 346 "permissions", 347 "pivot", 348 "placing", 349 "primary", 350 "raw", 351 "readratio", 352 "recover", 353 "references", 354 "rejectlog", 355 "resort", 356 "respect", 357 "restore", 358 "right", 359 "select", 360 "session_user", 361 "similar", 362 "snapshot", 363 "some", 364 "sysdate", 365 "system", 366 "table", 367 "tag", 368 "tdes", 369 "text255", 370 "text32k", 371 "then", 372 "timestamp", 373 "to", 374 "top", 375 "trailing", 376 "true", 377 "truncatecolumns", 378 "type", 379 "union", 380 "unique", 381 "unnest", 382 "unpivot", 383 "user", 384 "using", 385 "verbose", 386 "wallet", 387 "when", 388 "where", 389 "with", 390 "without", 391 } 392 393 def unnest_sql(self, expression: exp.Unnest) -> str: 394 args = expression.expressions 395 num_args = len(args) 396 397 if num_args != 1: 398 self.unsupported(f"Unsupported number of arguments in UNNEST: {num_args}") 399 return "" 400 401 if isinstance(expression.find_ancestor(exp.From, exp.Join, exp.Select), exp.Select): 402 self.unsupported("Unsupported UNNEST when not used in FROM/JOIN clauses") 403 return "" 404 405 arg = self.sql(seq_get(args, 0)) 406 407 alias = self.expressions(expression.args.get("alias"), key="columns", flat=True) 408 return f"{arg} AS {alias}" if alias else arg 409 410 def cast_sql(self, expression: exp.Cast, safe_prefix: t.Optional[str] = None) -> str: 411 if expression.is_type(exp.DataType.Type.JSON): 412 # Redshift doesn't support a JSON type, so casting to it is treated as a noop 413 return self.sql(expression, "this") 414 415 return super().cast_sql(expression, safe_prefix=safe_prefix) 416 417 def datatype_sql(self, expression: exp.DataType) -> str: 418 """ 419 Redshift converts the `TEXT` data type to `VARCHAR(255)` by default when people more generally mean 420 VARCHAR of max length which is `VARCHAR(max)` in Redshift. Therefore if we get a `TEXT` data type 421 without precision we convert it to `VARCHAR(max)` and if it does have precision then we just convert 422 `TEXT` to `VARCHAR`. 423 """ 424 if expression.is_type("text"): 425 expression.set("this", exp.DataType.Type.VARCHAR) 426 precision = expression.args.get("expressions") 427 428 if not precision: 429 expression.append("expressions", exp.var("MAX")) 430 431 return super().datatype_sql(expression) 432 433 def alterset_sql(self, expression: exp.AlterSet) -> str: 434 exprs = self.expressions(expression, flat=True) 435 exprs = f" TABLE PROPERTIES ({exprs})" if exprs else "" 436 location = self.sql(expression, "location") 437 location = f" LOCATION {location}" if location else "" 438 file_format = self.expressions(expression, key="file_format", flat=True, sep=" ") 439 file_format = f" FILE FORMAT {file_format}" if file_format else "" 440 441 return f"SET{exprs}{location}{file_format}" 442 443 def array_sql(self, expression: exp.Array) -> str: 444 if expression.args.get("bracket_notation"): 445 return super().array_sql(expression) 446 447 return rename_func("ARRAY")(self, expression) 448 449 def explode_sql(self, expression: exp.Explode) -> str: 450 self.unsupported("Unsupported EXPLODE() function") 451 return ""
Specifies the strategy according to which identifiers should be normalized.
Whether the ARRAY constructor is context-sensitive, i.e in Redshift ARRAY[1, 2, 3] != ARRAY(1, 2, 3) as the former is of type INT[] vs the latter which is SUPER
Associates this dialect's time formats with their equivalent Python strftime
formats.
Mapping of an escaped sequence (\n
) to its unescaped version (
).
56 class Parser(Postgres.Parser): 57 FUNCTIONS = { 58 **Postgres.Parser.FUNCTIONS, 59 "ADD_MONTHS": lambda args: exp.TsOrDsAdd( 60 this=seq_get(args, 0), 61 expression=seq_get(args, 1), 62 unit=exp.var("month"), 63 return_type=exp.DataType.build("TIMESTAMP"), 64 ), 65 "CONVERT_TIMEZONE": lambda args: build_convert_timezone(args, "UTC"), 66 "DATEADD": _build_date_delta(exp.TsOrDsAdd), 67 "DATE_ADD": _build_date_delta(exp.TsOrDsAdd), 68 "DATEDIFF": _build_date_delta(exp.TsOrDsDiff), 69 "DATE_DIFF": _build_date_delta(exp.TsOrDsDiff), 70 "GETDATE": exp.CurrentTimestamp.from_arg_list, 71 "LISTAGG": exp.GroupConcat.from_arg_list, 72 "SPLIT_TO_ARRAY": lambda args: exp.StringToArray( 73 this=seq_get(args, 0), expression=seq_get(args, 1) or exp.Literal.string(",") 74 ), 75 "STRTOL": exp.FromBase.from_arg_list, 76 } 77 78 NO_PAREN_FUNCTION_PARSERS = { 79 **Postgres.Parser.NO_PAREN_FUNCTION_PARSERS, 80 "APPROXIMATE": lambda self: self._parse_approximate_count(), 81 "SYSDATE": lambda self: self.expression(exp.CurrentTimestamp, sysdate=True), 82 } 83 84 SUPPORTS_IMPLICIT_UNNEST = True 85 86 def _parse_table( 87 self, 88 schema: bool = False, 89 joins: bool = False, 90 alias_tokens: t.Optional[t.Collection[TokenType]] = None, 91 parse_bracket: bool = False, 92 is_db_reference: bool = False, 93 parse_partition: bool = False, 94 ) -> t.Optional[exp.Expression]: 95 # Redshift supports UNPIVOTing SUPER objects, e.g. `UNPIVOT foo.obj[0] AS val AT attr` 96 unpivot = self._match(TokenType.UNPIVOT) 97 table = super()._parse_table( 98 schema=schema, 99 joins=joins, 100 alias_tokens=alias_tokens, 101 parse_bracket=parse_bracket, 102 is_db_reference=is_db_reference, 103 ) 104 105 return self.expression(exp.Pivot, this=table, unpivot=True) if unpivot else table 106 107 def _parse_convert( 108 self, strict: bool, safe: t.Optional[bool] = None 109 ) -> t.Optional[exp.Expression]: 110 to = self._parse_types() 111 self._match(TokenType.COMMA) 112 this = self._parse_bitwise() 113 return self.expression(exp.TryCast, this=this, to=to, safe=safe) 114 115 def _parse_approximate_count(self) -> t.Optional[exp.ApproxDistinct]: 116 index = self._index - 1 117 func = self._parse_function() 118 119 if isinstance(func, exp.Count) and isinstance(func.this, exp.Distinct): 120 return self.expression(exp.ApproxDistinct, this=seq_get(func.this.expressions, 0)) 121 self._retreat(index) 122 return None
Parser consumes a list of tokens produced by the Tokenizer and produces a parsed syntax tree.
Arguments:
- error_level: The desired error level. Default: ErrorLevel.IMMEDIATE
- error_message_context: The amount of context to capture from a query string when displaying the error message (in number of characters). Default: 100
- 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
Inherited Members
- sqlglot.parser.Parser
- Parser
- STRUCT_TYPE_TOKENS
- NESTED_TYPE_TOKENS
- ENUM_TYPE_TOKENS
- AGGREGATE_TYPE_TOKENS
- TYPE_TOKENS
- SIGNED_TO_UNSIGNED_TYPE_TOKEN
- SUBQUERY_PREDICATES
- RESERVED_TOKENS
- DB_CREATABLES
- CREATABLES
- ALTERABLES
- ALIAS_TOKENS
- ARRAY_CONSTRUCTORS
- COMMENT_TABLE_ALIAS_TOKENS
- UPDATE_ALIAS_TOKENS
- TRIM_TYPES
- FUNC_TOKENS
- CONJUNCTION
- ASSIGNMENT
- DISJUNCTION
- EQUALITY
- COMPARISON
- TERM
- FACTOR
- TIMES
- TIMESTAMPS
- SET_OPERATIONS
- JOIN_METHODS
- JOIN_SIDES
- JOIN_KINDS
- JOIN_HINTS
- LAMBDAS
- EXPRESSION_PARSERS
- UNARY_PARSERS
- STRING_PARSERS
- NUMERIC_PARSERS
- PRIMARY_PARSERS
- PLACEHOLDER_PARSERS
- CONSTRAINT_PARSERS
- ALTER_PARSERS
- ALTER_ALTER_PARSERS
- SCHEMA_UNNAMED_CONSTRAINTS
- INVALID_FUNC_NAME_TOKENS
- FUNCTIONS_WITH_ALIASED_ARGS
- KEY_VALUE_DEFINITIONS
- QUERY_MODIFIER_PARSERS
- SET_PARSERS
- SHOW_PARSERS
- TYPE_LITERAL_PARSERS
- TYPE_CONVERTERS
- DDL_SELECT_TOKENS
- PRE_VOLATILE_TOKENS
- TRANSACTION_KIND
- TRANSACTION_CHARACTERISTICS
- CONFLICT_ACTIONS
- CREATE_SEQUENCE
- ISOLATED_LOADING_OPTIONS
- USABLES
- CAST_ACTIONS
- SCHEMA_BINDING_OPTIONS
- PROCEDURE_OPTIONS
- EXECUTE_AS_OPTIONS
- KEY_CONSTRAINT_OPTIONS
- INSERT_ALTERNATIVES
- CLONE_KEYWORDS
- HISTORICAL_DATA_PREFIX
- HISTORICAL_DATA_KIND
- OPCLASS_FOLLOW_KEYWORDS
- OPTYPE_FOLLOW_TOKENS
- TABLE_INDEX_HINT_TOKENS
- VIEW_ATTRIBUTES
- WINDOW_ALIAS_TOKENS
- WINDOW_BEFORE_PAREN_TOKENS
- WINDOW_SIDES
- JSON_KEY_VALUE_SEPARATOR_TOKENS
- FETCH_TOKENS
- ADD_CONSTRAINT_TOKENS
- DISTINCT_TOKENS
- NULL_TOKENS
- UNNEST_OFFSET_ALIAS_TOKENS
- SELECT_START_TOKENS
- COPY_INTO_VARLEN_OPTIONS
- IS_JSON_PREDICATE_KIND
- ODBC_DATETIME_LITERALS
- ON_CONDITION_TOKENS
- PRIVILEGE_FOLLOW_TOKENS
- DESCRIBE_STYLES
- ANALYZE_STYLES
- ANALYZE_EXPRESSION_PARSERS
- PARTITION_KEYWORDS
- AMBIGUOUS_ALIAS_TOKENS
- OPERATION_MODIFIERS
- RECURSIVE_CTE_SEARCH_KIND
- MODIFIABLES
- STRICT_CAST
- PREFIXED_PIVOT_COLUMNS
- IDENTIFY_PIVOT_STRINGS
- LOG_DEFAULTS_TO_LN
- ALTER_TABLE_ADD_REQUIRED_FOR_EACH_COLUMN
- TABLESAMPLE_CSV
- DEFAULT_SAMPLING_METHOD
- SET_REQUIRES_ASSIGNMENT_DELIMITER
- TRIM_PATTERN_FIRST
- STRING_ALIASES
- MODIFIERS_ATTACHED_TO_SET_OP
- SET_OP_MODIFIERS
- NO_PAREN_IF_COMMANDS
- COLON_IS_VARIANT_EXTRACT
- VALUES_FOLLOWED_BY_PAREN
- INTERVAL_SPANS
- SUPPORTS_PARTITION_SELECTION
- WRAPPED_TRANSFORM_COLUMN_CONSTRAINT
- OPTIONAL_ALIAS_TOKEN_CTE
- error_level
- error_message_context
- max_errors
- dialect
- reset
- parse
- parse_into
- check_errors
- raise_error
- expression
- validate_expression
- errors
- sql
124 class Tokenizer(Postgres.Tokenizer): 125 BIT_STRINGS = [] 126 HEX_STRINGS = [] 127 STRING_ESCAPES = ["\\", "'"] 128 129 KEYWORDS = { 130 **Postgres.Tokenizer.KEYWORDS, 131 "(+)": TokenType.JOIN_MARKER, 132 "HLLSKETCH": TokenType.HLLSKETCH, 133 "MINUS": TokenType.EXCEPT, 134 "SUPER": TokenType.SUPER, 135 "TOP": TokenType.TOP, 136 "UNLOAD": TokenType.COMMAND, 137 "VARBYTE": TokenType.VARBINARY, 138 "BINARY VARYING": TokenType.VARBINARY, 139 } 140 KEYWORDS.pop("VALUES") 141 142 # Redshift allows # to appear as a table identifier prefix 143 SINGLE_TOKENS = Postgres.Tokenizer.SINGLE_TOKENS.copy() 144 SINGLE_TOKENS.pop("#")
Inherited Members
- sqlglot.tokens.Tokenizer
- Tokenizer
- RAW_STRINGS
- UNICODE_STRINGS
- IDENTIFIERS
- QUOTES
- IDENTIFIER_ESCAPES
- STRING_ESCAPES_ALLOWED_IN_RAW_STRINGS
- NESTED_COMMENTS
- HINT_START
- TOKENS_PRECEDING_HINT
- WHITE_SPACE
- COMMANDS
- COMMAND_PREFIX_TOKENS
- NUMERIC_LITERALS
- COMMENTS
- dialect
- use_rs_tokenizer
- reset
- tokenize
- tokenize_rs
- size
- sql
- tokens
146 class Generator(Postgres.Generator): 147 LOCKING_READS_SUPPORTED = False 148 QUERY_HINTS = False 149 VALUES_AS_TABLE = False 150 TZ_TO_WITH_TIME_ZONE = True 151 NVL2_SUPPORTED = True 152 LAST_DAY_SUPPORTS_DATE_PART = False 153 CAN_IMPLEMENT_ARRAY_ANY = False 154 MULTI_ARG_DISTINCT = True 155 COPY_PARAMS_ARE_WRAPPED = False 156 HEX_FUNC = "TO_HEX" 157 PARSE_JSON_NAME = "JSON_PARSE" 158 ARRAY_CONCAT_IS_VAR_LEN = False 159 SUPPORTS_CONVERT_TIMEZONE = True 160 EXCEPT_INTERSECT_SUPPORT_ALL_CLAUSE = False 161 SUPPORTS_MEDIAN = True 162 ALTER_SET_TYPE = "TYPE" 163 164 # Redshift doesn't have `WITH` as part of their with_properties so we remove it 165 WITH_PROPERTIES_PREFIX = " " 166 167 TYPE_MAPPING = { 168 **Postgres.Generator.TYPE_MAPPING, 169 exp.DataType.Type.BINARY: "VARBYTE", 170 exp.DataType.Type.BLOB: "VARBYTE", 171 exp.DataType.Type.INT: "INTEGER", 172 exp.DataType.Type.TIMETZ: "TIME", 173 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 174 exp.DataType.Type.VARBINARY: "VARBYTE", 175 exp.DataType.Type.ROWVERSION: "VARBYTE", 176 } 177 178 TRANSFORMS = { 179 **Postgres.Generator.TRANSFORMS, 180 exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CONCAT"), 181 exp.Concat: concat_to_dpipe_sql, 182 exp.ConcatWs: concat_ws_to_dpipe_sql, 183 exp.ApproxDistinct: lambda self, 184 e: f"APPROXIMATE COUNT(DISTINCT {self.sql(e, 'this')})", 185 exp.CurrentTimestamp: lambda self, e: ( 186 "SYSDATE" if e.args.get("sysdate") else "GETDATE()" 187 ), 188 exp.DateAdd: date_delta_sql("DATEADD"), 189 exp.DateDiff: date_delta_sql("DATEDIFF"), 190 exp.DistKeyProperty: lambda self, e: self.func("DISTKEY", e.this), 191 exp.DistStyleProperty: lambda self, e: self.naked_property(e), 192 exp.Explode: lambda self, e: self.explode_sql(e), 193 exp.FromBase: rename_func("STRTOL"), 194 exp.GeneratedAsIdentityColumnConstraint: generatedasidentitycolumnconstraint_sql, 195 exp.JSONExtract: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 196 exp.JSONExtractScalar: json_extract_segments("JSON_EXTRACT_PATH_TEXT"), 197 exp.GroupConcat: rename_func("LISTAGG"), 198 exp.Hex: lambda self, e: self.func("UPPER", self.func("TO_HEX", self.sql(e, "this"))), 199 exp.Select: transforms.preprocess( 200 [ 201 transforms.eliminate_distinct_on, 202 transforms.eliminate_semi_and_anti_joins, 203 transforms.unqualify_unnest, 204 transforms.unnest_generate_date_array_using_recursive_cte, 205 ] 206 ), 207 exp.SortKeyProperty: lambda self, 208 e: f"{'COMPOUND ' if e.args['compound'] else ''}SORTKEY({self.format_args(*e.this)})", 209 exp.StartsWith: lambda self, 210 e: f"{self.sql(e.this)} LIKE {self.sql(e.expression)} || '%'", 211 exp.StringToArray: rename_func("SPLIT_TO_ARRAY"), 212 exp.TableSample: no_tablesample_sql, 213 exp.TsOrDsAdd: date_delta_sql("DATEADD"), 214 exp.TsOrDsDiff: date_delta_sql("DATEDIFF"), 215 exp.UnixToTime: lambda self, 216 e: f"(TIMESTAMP 'epoch' + {self.sql(e.this)} * INTERVAL '1 SECOND')", 217 } 218 219 # Postgres maps exp.Pivot to no_pivot_sql, but Redshift support pivots 220 TRANSFORMS.pop(exp.Pivot) 221 222 # Postgres doesn't support JSON_PARSE, but Redshift does 223 TRANSFORMS.pop(exp.ParseJSON) 224 225 # Redshift uses the POW | POWER (expr1, expr2) syntax instead of expr1 ^ expr2 (postgres) 226 TRANSFORMS.pop(exp.Pow) 227 228 # Redshift supports these functions 229 TRANSFORMS.pop(exp.AnyValue) 230 TRANSFORMS.pop(exp.LastDay) 231 TRANSFORMS.pop(exp.SHA2) 232 233 RESERVED_KEYWORDS = { 234 "aes128", 235 "aes256", 236 "all", 237 "allowoverwrite", 238 "analyse", 239 "analyze", 240 "and", 241 "any", 242 "array", 243 "as", 244 "asc", 245 "authorization", 246 "az64", 247 "backup", 248 "between", 249 "binary", 250 "blanksasnull", 251 "both", 252 "bytedict", 253 "bzip2", 254 "case", 255 "cast", 256 "check", 257 "collate", 258 "column", 259 "constraint", 260 "create", 261 "credentials", 262 "cross", 263 "current_date", 264 "current_time", 265 "current_timestamp", 266 "current_user", 267 "current_user_id", 268 "default", 269 "deferrable", 270 "deflate", 271 "defrag", 272 "delta", 273 "delta32k", 274 "desc", 275 "disable", 276 "distinct", 277 "do", 278 "else", 279 "emptyasnull", 280 "enable", 281 "encode", 282 "encrypt ", 283 "encryption", 284 "end", 285 "except", 286 "explicit", 287 "false", 288 "for", 289 "foreign", 290 "freeze", 291 "from", 292 "full", 293 "globaldict256", 294 "globaldict64k", 295 "grant", 296 "group", 297 "gzip", 298 "having", 299 "identity", 300 "ignore", 301 "ilike", 302 "in", 303 "initially", 304 "inner", 305 "intersect", 306 "interval", 307 "into", 308 "is", 309 "isnull", 310 "join", 311 "leading", 312 "left", 313 "like", 314 "limit", 315 "localtime", 316 "localtimestamp", 317 "lun", 318 "luns", 319 "lzo", 320 "lzop", 321 "minus", 322 "mostly16", 323 "mostly32", 324 "mostly8", 325 "natural", 326 "new", 327 "not", 328 "notnull", 329 "null", 330 "nulls", 331 "off", 332 "offline", 333 "offset", 334 "oid", 335 "old", 336 "on", 337 "only", 338 "open", 339 "or", 340 "order", 341 "outer", 342 "overlaps", 343 "parallel", 344 "partition", 345 "percent", 346 "permissions", 347 "pivot", 348 "placing", 349 "primary", 350 "raw", 351 "readratio", 352 "recover", 353 "references", 354 "rejectlog", 355 "resort", 356 "respect", 357 "restore", 358 "right", 359 "select", 360 "session_user", 361 "similar", 362 "snapshot", 363 "some", 364 "sysdate", 365 "system", 366 "table", 367 "tag", 368 "tdes", 369 "text255", 370 "text32k", 371 "then", 372 "timestamp", 373 "to", 374 "top", 375 "trailing", 376 "true", 377 "truncatecolumns", 378 "type", 379 "union", 380 "unique", 381 "unnest", 382 "unpivot", 383 "user", 384 "using", 385 "verbose", 386 "wallet", 387 "when", 388 "where", 389 "with", 390 "without", 391 } 392 393 def unnest_sql(self, expression: exp.Unnest) -> str: 394 args = expression.expressions 395 num_args = len(args) 396 397 if num_args != 1: 398 self.unsupported(f"Unsupported number of arguments in UNNEST: {num_args}") 399 return "" 400 401 if isinstance(expression.find_ancestor(exp.From, exp.Join, exp.Select), exp.Select): 402 self.unsupported("Unsupported UNNEST when not used in FROM/JOIN clauses") 403 return "" 404 405 arg = self.sql(seq_get(args, 0)) 406 407 alias = self.expressions(expression.args.get("alias"), key="columns", flat=True) 408 return f"{arg} AS {alias}" if alias else arg 409 410 def cast_sql(self, expression: exp.Cast, safe_prefix: t.Optional[str] = None) -> str: 411 if expression.is_type(exp.DataType.Type.JSON): 412 # Redshift doesn't support a JSON type, so casting to it is treated as a noop 413 return self.sql(expression, "this") 414 415 return super().cast_sql(expression, safe_prefix=safe_prefix) 416 417 def datatype_sql(self, expression: exp.DataType) -> str: 418 """ 419 Redshift converts the `TEXT` data type to `VARCHAR(255)` by default when people more generally mean 420 VARCHAR of max length which is `VARCHAR(max)` in Redshift. Therefore if we get a `TEXT` data type 421 without precision we convert it to `VARCHAR(max)` and if it does have precision then we just convert 422 `TEXT` to `VARCHAR`. 423 """ 424 if expression.is_type("text"): 425 expression.set("this", exp.DataType.Type.VARCHAR) 426 precision = expression.args.get("expressions") 427 428 if not precision: 429 expression.append("expressions", exp.var("MAX")) 430 431 return super().datatype_sql(expression) 432 433 def alterset_sql(self, expression: exp.AlterSet) -> str: 434 exprs = self.expressions(expression, flat=True) 435 exprs = f" TABLE PROPERTIES ({exprs})" if exprs else "" 436 location = self.sql(expression, "location") 437 location = f" LOCATION {location}" if location else "" 438 file_format = self.expressions(expression, key="file_format", flat=True, sep=" ") 439 file_format = f" FILE FORMAT {file_format}" if file_format else "" 440 441 return f"SET{exprs}{location}{file_format}" 442 443 def array_sql(self, expression: exp.Array) -> str: 444 if expression.args.get("bracket_notation"): 445 return super().array_sql(expression) 446 447 return rename_func("ARRAY")(self, expression) 448 449 def explode_sql(self, expression: exp.Explode) -> str: 450 self.unsupported("Unsupported EXPLODE() function") 451 return ""
Generator converts a given syntax tree to the corresponding SQL string.
Arguments:
- pretty: Whether to format the produced SQL string. Default: False.
- identify: Determines when an identifier should be quoted. Possible values are: False (default): Never quote, except in cases where it's mandatory by the dialect. True or 'always': Always quote. 'safe': Only quote identifiers that are case insensitive.
- normalize: Whether to normalize identifiers to lowercase. Default: False.
- pad: The pad size in a formatted string. For example, this affects the indentation of a projection in a query, relative to its nesting level. Default: 2.
- indent: The indentation size in a formatted string. For example, this affects the
indentation of subqueries and filters under a
WHERE
clause. Default: 2. - normalize_functions: How to normalize function names. Possible values are: "upper" or True (default): Convert names to uppercase. "lower": Convert names to lowercase. False: Disables function name normalization.
- unsupported_level: Determines the generator's behavior when it encounters unsupported expressions. Default ErrorLevel.WARN.
- max_unsupported: 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: Whether the comma is leading or trailing in select expressions. This is only relevant when generating in pretty mode. 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 to preserve comments in the output SQL code. Default: True
393 def unnest_sql(self, expression: exp.Unnest) -> str: 394 args = expression.expressions 395 num_args = len(args) 396 397 if num_args != 1: 398 self.unsupported(f"Unsupported number of arguments in UNNEST: {num_args}") 399 return "" 400 401 if isinstance(expression.find_ancestor(exp.From, exp.Join, exp.Select), exp.Select): 402 self.unsupported("Unsupported UNNEST when not used in FROM/JOIN clauses") 403 return "" 404 405 arg = self.sql(seq_get(args, 0)) 406 407 alias = self.expressions(expression.args.get("alias"), key="columns", flat=True) 408 return f"{arg} AS {alias}" if alias else arg
410 def cast_sql(self, expression: exp.Cast, safe_prefix: t.Optional[str] = None) -> str: 411 if expression.is_type(exp.DataType.Type.JSON): 412 # Redshift doesn't support a JSON type, so casting to it is treated as a noop 413 return self.sql(expression, "this") 414 415 return super().cast_sql(expression, safe_prefix=safe_prefix)
417 def datatype_sql(self, expression: exp.DataType) -> str: 418 """ 419 Redshift converts the `TEXT` data type to `VARCHAR(255)` by default when people more generally mean 420 VARCHAR of max length which is `VARCHAR(max)` in Redshift. Therefore if we get a `TEXT` data type 421 without precision we convert it to `VARCHAR(max)` and if it does have precision then we just convert 422 `TEXT` to `VARCHAR`. 423 """ 424 if expression.is_type("text"): 425 expression.set("this", exp.DataType.Type.VARCHAR) 426 precision = expression.args.get("expressions") 427 428 if not precision: 429 expression.append("expressions", exp.var("MAX")) 430 431 return super().datatype_sql(expression)
Redshift converts the TEXT
data type to VARCHAR(255)
by default when people more generally mean
VARCHAR of max length which is VARCHAR(max)
in Redshift. Therefore if we get a TEXT
data type
without precision we convert it to VARCHAR(max)
and if it does have precision then we just convert
TEXT
to VARCHAR
.
433 def alterset_sql(self, expression: exp.AlterSet) -> str: 434 exprs = self.expressions(expression, flat=True) 435 exprs = f" TABLE PROPERTIES ({exprs})" if exprs else "" 436 location = self.sql(expression, "location") 437 location = f" LOCATION {location}" if location else "" 438 file_format = self.expressions(expression, key="file_format", flat=True, sep=" ") 439 file_format = f" FILE FORMAT {file_format}" if file_format else "" 440 441 return f"SET{exprs}{location}{file_format}"
Inherited Members
- sqlglot.generator.Generator
- Generator
- NULL_ORDERING_SUPPORTED
- IGNORE_NULLS_IN_FUNC
- WRAP_DERIVED_VALUES
- CREATE_FUNCTION_RETURN_AS
- MATCHED_BY_SOURCE
- INTERVAL_ALLOWS_PLURAL_FORM
- LIMIT_FETCH
- LIMIT_ONLY_LITERALS
- GROUPINGS_SEP
- INDEX_ON
- QUERY_HINT_SEP
- IS_BOOL_ALLOWED
- DUPLICATE_KEY_UPDATE_WITH_SET
- LIMIT_IS_TOP
- RETURNING_END
- EXTRACT_ALLOWS_QUOTES
- ALTER_TABLE_INCLUDE_COLUMN_KEYWORD
- UNNEST_WITH_ORDINALITY
- AGGREGATE_FILTER_SUPPORTED
- SEMI_ANTI_JOIN_WITH_SIDE
- COMPUTED_COLUMN_WITH_TYPE
- SUPPORTS_TABLE_COPY
- TABLESAMPLE_REQUIRES_PARENS
- TABLESAMPLE_KEYWORDS
- TABLESAMPLE_WITH_METHOD
- COLLATE_IS_FUNC
- DATA_TYPE_SPECIFIERS_ALLOWED
- ENSURE_BOOLS
- CTE_RECURSIVE_KEYWORD_REQUIRED
- SUPPORTS_SINGLE_ARG_CONCAT
- SUPPORTS_TABLE_ALIAS_COLUMNS
- UNPIVOT_ALIASES_ARE_IDENTIFIERS
- JSON_KEY_VALUE_PAIR_SEP
- INSERT_OVERWRITE
- SUPPORTS_CREATE_TABLE_LIKE
- JSON_PATH_BRACKETED_KEY_SUPPORTED
- JSON_PATH_SINGLE_QUOTE_ESCAPE
- SUPPORTS_TO_NUMBER
- SET_OP_MODIFIERS
- COPY_PARAMS_EQ_REQUIRED
- STAR_EXCEPT
- QUOTE_JSON_PATH
- PAD_FILL_PATTERN_IS_REQUIRED
- SUPPORTS_EXPLODING_PROJECTIONS
- SUPPORTS_UNIX_SECONDS
- ARRAY_SIZE_NAME
- TIME_PART_SINGULARS
- TOKEN_MAPPING
- STRUCT_DELIMITER
- NAMED_PLACEHOLDER_TOKEN
- EXPRESSION_PRECEDES_PROPERTIES_CREATABLES
- WITH_SEPARATED_COMMENTS
- EXCLUDE_COMMENTS
- UNWRAPPED_INTERVAL_VALUES
- PARAMETERIZABLE_TEXT_TYPES
- EXPRESSIONS_WITHOUT_NESTED_CTES
- SENTINEL_LINE_BREAK
- pretty
- identify
- normalize
- pad
- unsupported_level
- max_unsupported
- leading_comma
- max_text_width
- comments
- dialect
- normalize_functions
- unsupported_messages
- generate
- preprocess
- unsupported
- sep
- seg
- pad_comment
- maybe_comment
- wrap
- no_identify
- normalize_func
- indent
- sql
- uncache_sql
- cache_sql
- characterset_sql
- column_parts
- column_sql
- columnposition_sql
- columndef_sql
- columnconstraint_sql
- autoincrementcolumnconstraint_sql
- compresscolumnconstraint_sql
- generatedasidentitycolumnconstraint_sql
- generatedasrowcolumnconstraint_sql
- periodforsystemtimeconstraint_sql
- notnullcolumnconstraint_sql
- transformcolumnconstraint_sql
- primarykeycolumnconstraint_sql
- uniquecolumnconstraint_sql
- createable_sql
- create_sql
- sequenceproperties_sql
- clone_sql
- describe_sql
- heredoc_sql
- prepend_ctes
- with_sql
- cte_sql
- tablealias_sql
- bitstring_sql
- hexstring_sql
- bytestring_sql
- unicodestring_sql
- rawstring_sql
- datatypeparam_sql
- directory_sql
- delete_sql
- drop_sql
- set_operation
- set_operations
- fetch_sql
- limitoptions_sql
- filter_sql
- hint_sql
- indexparameters_sql
- index_sql
- identifier_sql
- hex_sql
- lowerhex_sql
- inputoutputformat_sql
- national_sql
- partition_sql
- properties_sql
- root_properties
- properties
- with_properties
- locate_properties
- property_name
- property_sql
- likeproperty_sql
- fallbackproperty_sql
- journalproperty_sql
- freespaceproperty_sql
- checksumproperty_sql
- mergeblockratioproperty_sql
- datablocksizeproperty_sql
- blockcompressionproperty_sql
- isolatedloadingproperty_sql
- partitionboundspec_sql
- partitionedofproperty_sql
- lockingproperty_sql
- withdataproperty_sql
- withsystemversioningproperty_sql
- insert_sql
- introducer_sql
- kill_sql
- pseudotype_sql
- objectidentifier_sql
- onconflict_sql
- returning_sql
- rowformatdelimitedproperty_sql
- withtablehint_sql
- indextablehint_sql
- historicaldata_sql
- table_parts
- table_sql
- tablefromrows_sql
- tablesample_sql
- pivot_sql
- version_sql
- tuple_sql
- update_sql
- values_sql
- var_sql
- into_sql
- from_sql
- groupingsets_sql
- rollup_sql
- cube_sql
- group_sql
- having_sql
- connect_sql
- prior_sql
- join_sql
- lambda_sql
- lateral_op
- lateral_sql
- limit_sql
- offset_sql
- setitem_sql
- set_sql
- pragma_sql
- lock_sql
- literal_sql
- escape_str
- loaddata_sql
- null_sql
- boolean_sql
- order_sql
- withfill_sql
- cluster_sql
- distribute_sql
- sort_sql
- ordered_sql
- matchrecognizemeasure_sql
- matchrecognize_sql
- query_modifiers
- options_modifier
- queryoption_sql
- offset_limit_modifiers
- after_limit_modifiers
- select_sql
- schema_sql
- schema_columns_sql
- star_sql
- parameter_sql
- sessionparameter_sql
- placeholder_sql
- subquery_sql
- qualify_sql
- prewhere_sql
- where_sql
- window_sql
- partition_by_sql
- windowspec_sql
- withingroup_sql
- between_sql
- bracket_offset_expressions
- all_sql
- any_sql
- exists_sql
- case_sql
- constraint_sql
- nextvaluefor_sql
- extract_sql
- trim_sql
- convert_concat_args
- concat_sql
- concatws_sql
- check_sql
- foreignkey_sql
- primarykey_sql
- if_sql
- jsonkeyvalue_sql
- jsonpath_sql
- json_path_part
- formatjson_sql
- jsonobject_sql
- jsonobjectagg_sql
- jsonarray_sql
- jsonarrayagg_sql
- jsoncolumndef_sql
- jsonschema_sql
- jsontable_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
- pivotalias_sql
- aliases_sql
- atindex_sql
- attimezone_sql
- fromtimezone_sql
- add_sql
- and_sql
- or_sql
- xor_sql
- connector_sql
- bitwiseand_sql
- bitwiseleftshift_sql
- bitwisenot_sql
- bitwiseor_sql
- bitwiserightshift_sql
- bitwisexor_sql
- currentdate_sql
- collate_sql
- command_sql
- comment_sql
- mergetreettlaction_sql
- mergetreettl_sql
- transaction_sql
- commit_sql
- rollback_sql
- altercolumn_sql
- alterindex_sql
- alterdiststyle_sql
- altersortkey_sql
- alterrename_sql
- renamecolumn_sql
- alter_sql
- add_column_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- havingmax_sql
- intdiv_sql
- dpipe_sql
- div_sql
- safedivide_sql
- overlaps_sql
- distance_sql
- dot_sql
- eq_sql
- propertyeq_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
- slice_sql
- sub_sql
- trycast_sql
- jsoncast_sql
- try_sql
- log_sql
- use_sql
- binary
- ceil_floor
- function_fallback_sql
- func
- format_args
- too_wide
- format_time
- expressions
- op_expressions
- naked_property
- tag_sql
- token_sql
- userdefinedfunction_sql
- joinhint_sql
- kwarg_sql
- when_sql
- whens_sql
- merge_sql
- tochar_sql
- tonumber_sql
- dictproperty_sql
- dictrange_sql
- dictsubproperty_sql
- duplicatekeyproperty_sql
- uniquekeyproperty_sql
- distributedbyproperty_sql
- oncluster_sql
- clusteredbyproperty_sql
- anyvalue_sql
- querytransform_sql
- indexconstraintoption_sql
- checkcolumnconstraint_sql
- indexcolumnconstraint_sql
- nvl2_sql
- comprehension_sql
- columnprefix_sql
- opclass_sql
- predict_sql
- forin_sql
- refresh_sql
- toarray_sql
- tsordstotime_sql
- tsordstotimestamp_sql
- tsordstodatetime_sql
- tsordstodate_sql
- unixdate_sql
- lastday_sql
- dateadd_sql
- arrayany_sql
- struct_sql
- partitionrange_sql
- truncatetable_sql
- convert_sql
- copyparameter_sql
- credentials_sql
- copy_sql
- semicolon_sql
- datadeletionproperty_sql
- maskingpolicycolumnconstraint_sql
- gapfill_sql
- scope_resolution
- scoperesolution_sql
- parsejson_sql
- rand_sql
- changes_sql
- pad_sql
- summarize_sql
- explodinggenerateseries_sql
- arrayconcat_sql
- converttimezone_sql
- json_sql
- jsonvalue_sql
- conditionalinsert_sql
- multitableinserts_sql
- oncondition_sql
- jsonextractquote_sql
- jsonexists_sql
- arrayagg_sql
- apply_sql
- grant_sql
- grantprivilege_sql
- grantprincipal_sql
- columns_sql
- overlay_sql
- todouble_sql
- string_sql
- median_sql
- overflowtruncatebehavior_sql
- unixseconds_sql
- arraysize_sql
- attach_sql
- detach_sql
- attachoption_sql
- featuresattime_sql
- watermarkcolumnconstraint_sql
- encodeproperty_sql
- includeproperty_sql
- xmlelement_sql
- partitionbyrangeproperty_sql
- partitionbyrangepropertydynamic_sql
- unpivotcolumns_sql
- analyzesample_sql
- analyzestatistics_sql
- analyzehistogram_sql
- analyzedelete_sql
- analyzelistchainedrows_sql
- analyzevalidate_sql
- analyze_sql
- xmltable_sql
- xmlnamespace_sql
- export_sql
- declare_sql
- declareitem_sql
- recursivewithsearch_sql
- parameterizedagg_sql
- anonymousaggfunc_sql
- combinedaggfunc_sql
- combinedparameterizedagg_sql
- show_sql
- put_sql
- sqlglot.dialects.postgres.Postgres.Generator
- SINGLE_STRING_INTERVAL
- RENAME_TABLE_WITH_DB
- JOIN_HINTS
- TABLE_HINTS
- PARAMETER_TOKEN
- TABLESAMPLE_SIZE_IS_ROWS
- TABLESAMPLE_SEED_KEYWORD
- SUPPORTS_SELECT_INTO
- JSON_TYPE_REQUIRED_FOR_EXTRACTION
- SUPPORTS_UNLOGGED_TABLES
- LIKE_PROPERTY_INSIDE_SCHEMA
- COPY_HAS_INTO_KEYWORD
- ARRAY_SIZE_DIM_REQUIRED
- SUPPORTED_JSON_PATH_PARTS
- PROPERTIES_LOCATION
- schemacommentproperty_sql
- commentcolumnconstraint_sql
- bracket_sql
- matchagainst_sql
- computedcolumnconstraint_sql
- isascii_sql
- currentschema_sql