sqlglot.dialects.postgres
1from __future__ import annotations 2 3from sqlglot import exp, generator, parser, tokens, transforms 4from sqlglot.dialects.dialect import ( 5 Dialect, 6 arrow_json_extract_scalar_sql, 7 arrow_json_extract_sql, 8 datestrtodate_sql, 9 format_time_lambda, 10 max_or_greatest, 11 min_or_least, 12 no_paren_current_date_sql, 13 no_tablesample_sql, 14 no_trycast_sql, 15 rename_func, 16 str_position_sql, 17 timestamptrunc_sql, 18 trim_sql, 19) 20from sqlglot.helper import seq_get 21from sqlglot.parser import binary_range_parser 22from sqlglot.tokens import TokenType 23 24DATE_DIFF_FACTOR = { 25 "MICROSECOND": " * 1000000", 26 "MILLISECOND": " * 1000", 27 "SECOND": "", 28 "MINUTE": " / 60", 29 "HOUR": " / 3600", 30 "DAY": " / 86400", 31} 32 33 34def _date_add_sql(kind): 35 def func(self, expression): 36 from sqlglot.optimizer.simplify import simplify 37 38 this = self.sql(expression, "this") 39 unit = expression.args.get("unit") 40 expression = simplify(expression.args["expression"]) 41 42 if not isinstance(expression, exp.Literal): 43 self.unsupported("Cannot add non literal") 44 45 expression = expression.copy() 46 expression.args["is_string"] = True 47 return f"{this} {kind} {self.sql(exp.Interval(this=expression, unit=unit))}" 48 49 return func 50 51 52def _date_diff_sql(self, expression): 53 unit = expression.text("unit").upper() 54 factor = DATE_DIFF_FACTOR.get(unit) 55 56 end = f"CAST({expression.this} AS TIMESTAMP)" 57 start = f"CAST({expression.expression} AS TIMESTAMP)" 58 59 if factor is not None: 60 return f"CAST(EXTRACT(epoch FROM {end} - {start}){factor} AS BIGINT)" 61 62 age = f"AGE({end}, {start})" 63 64 if unit == "WEEK": 65 unit = f"EXTRACT(year FROM {age}) * 48 + EXTRACT(month FROM {age}) * 4 + EXTRACT(day FROM {age}) / 7" 66 elif unit == "MONTH": 67 unit = f"EXTRACT(year FROM {age}) * 12 + EXTRACT(month FROM {age})" 68 elif unit == "QUARTER": 69 unit = f"EXTRACT(year FROM {age}) * 4 + EXTRACT(month FROM {age}) / 3" 70 elif unit == "YEAR": 71 unit = f"EXTRACT(year FROM {age})" 72 else: 73 unit = age 74 75 return f"CAST({unit} AS BIGINT)" 76 77 78def _substring_sql(self, expression): 79 this = self.sql(expression, "this") 80 start = self.sql(expression, "start") 81 length = self.sql(expression, "length") 82 83 from_part = f" FROM {start}" if start else "" 84 for_part = f" FOR {length}" if length else "" 85 86 return f"SUBSTRING({this}{from_part}{for_part})" 87 88 89def _string_agg_sql(self, expression): 90 expression = expression.copy() 91 separator = expression.args.get("separator") or exp.Literal.string(",") 92 93 order = "" 94 this = expression.this 95 if isinstance(this, exp.Order): 96 if this.this: 97 this = this.this.pop() 98 order = self.sql(expression.this) # Order has a leading space 99 100 return f"STRING_AGG({self.format_args(this, separator)}{order})" 101 102 103def _datatype_sql(self, expression): 104 if expression.this == exp.DataType.Type.ARRAY: 105 return f"{self.expressions(expression, flat=True)}[]" 106 return self.datatype_sql(expression) 107 108 109def _auto_increment_to_serial(expression): 110 auto = expression.find(exp.AutoIncrementColumnConstraint) 111 112 if auto: 113 expression = expression.copy() 114 expression.args["constraints"].remove(auto.parent) 115 kind = expression.args["kind"] 116 117 if kind.this == exp.DataType.Type.INT: 118 kind.replace(exp.DataType(this=exp.DataType.Type.SERIAL)) 119 elif kind.this == exp.DataType.Type.SMALLINT: 120 kind.replace(exp.DataType(this=exp.DataType.Type.SMALLSERIAL)) 121 elif kind.this == exp.DataType.Type.BIGINT: 122 kind.replace(exp.DataType(this=exp.DataType.Type.BIGSERIAL)) 123 124 return expression 125 126 127def _serial_to_generated(expression): 128 kind = expression.args["kind"] 129 130 if kind.this == exp.DataType.Type.SERIAL: 131 data_type = exp.DataType(this=exp.DataType.Type.INT) 132 elif kind.this == exp.DataType.Type.SMALLSERIAL: 133 data_type = exp.DataType(this=exp.DataType.Type.SMALLINT) 134 elif kind.this == exp.DataType.Type.BIGSERIAL: 135 data_type = exp.DataType(this=exp.DataType.Type.BIGINT) 136 else: 137 data_type = None 138 139 if data_type: 140 expression = expression.copy() 141 expression.args["kind"].replace(data_type) 142 constraints = expression.args["constraints"] 143 generated = exp.ColumnConstraint(kind=exp.GeneratedAsIdentityColumnConstraint(this=False)) 144 notnull = exp.ColumnConstraint(kind=exp.NotNullColumnConstraint()) 145 if notnull not in constraints: 146 constraints.insert(0, notnull) 147 if generated not in constraints: 148 constraints.insert(0, generated) 149 150 return expression 151 152 153def _generate_series(args): 154 # The goal is to convert step values like '1 day' or INTERVAL '1 day' into INTERVAL '1' day 155 step = seq_get(args, 2) 156 157 if step is None: 158 # Postgres allows calls with just two arguments -- the "step" argument defaults to 1 159 return exp.GenerateSeries.from_arg_list(args) 160 161 if step.is_string: 162 args[2] = exp.to_interval(step.this) 163 elif isinstance(step, exp.Interval) and not step.args.get("unit"): 164 args[2] = exp.to_interval(step.this.this) 165 166 return exp.GenerateSeries.from_arg_list(args) 167 168 169def _to_timestamp(args): 170 # TO_TIMESTAMP accepts either a single double argument or (text, text) 171 if len(args) == 1: 172 # https://www.postgresql.org/docs/current/functions-datetime.html#FUNCTIONS-DATETIME-TABLE 173 return exp.UnixToTime.from_arg_list(args) 174 # https://www.postgresql.org/docs/current/functions-formatting.html 175 return format_time_lambda(exp.StrToTime, "postgres")(args) 176 177 178class Postgres(Dialect): 179 null_ordering = "nulls_are_large" 180 time_format = "'YYYY-MM-DD HH24:MI:SS'" 181 time_mapping = { 182 "AM": "%p", 183 "PM": "%p", 184 "D": "%u", # 1-based day of week 185 "DD": "%d", # day of month 186 "DDD": "%j", # zero padded day of year 187 "FMDD": "%-d", # - is no leading zero for Python; same for FM in postgres 188 "FMDDD": "%-j", # day of year 189 "FMHH12": "%-I", # 9 190 "FMHH24": "%-H", # 9 191 "FMMI": "%-M", # Minute 192 "FMMM": "%-m", # 1 193 "FMSS": "%-S", # Second 194 "HH12": "%I", # 09 195 "HH24": "%H", # 09 196 "MI": "%M", # zero padded minute 197 "MM": "%m", # 01 198 "OF": "%z", # utc offset 199 "SS": "%S", # zero padded second 200 "TMDay": "%A", # TM is locale dependent 201 "TMDy": "%a", 202 "TMMon": "%b", # Sep 203 "TMMonth": "%B", # September 204 "TZ": "%Z", # uppercase timezone name 205 "US": "%f", # zero padded microsecond 206 "WW": "%U", # 1-based week of year 207 "YY": "%y", # 15 208 "YYYY": "%Y", # 2015 209 } 210 211 class Tokenizer(tokens.Tokenizer): 212 QUOTES = ["'", "$$"] 213 214 BIT_STRINGS = [("b'", "'"), ("B'", "'")] 215 HEX_STRINGS = [("x'", "'"), ("X'", "'")] 216 BYTE_STRINGS = [("e'", "'"), ("E'", "'")] 217 218 KEYWORDS = { 219 **tokens.Tokenizer.KEYWORDS, 220 "~~": TokenType.LIKE, 221 "~~*": TokenType.ILIKE, 222 "~*": TokenType.IRLIKE, 223 "~": TokenType.RLIKE, 224 "@>": TokenType.AT_GT, 225 "<@": TokenType.LT_AT, 226 "BEGIN": TokenType.COMMAND, 227 "BEGIN TRANSACTION": TokenType.BEGIN, 228 "BIGSERIAL": TokenType.BIGSERIAL, 229 "CHARACTER VARYING": TokenType.VARCHAR, 230 "DECLARE": TokenType.COMMAND, 231 "DO": TokenType.COMMAND, 232 "HSTORE": TokenType.HSTORE, 233 "JSONB": TokenType.JSONB, 234 "REFRESH": TokenType.COMMAND, 235 "REINDEX": TokenType.COMMAND, 236 "RESET": TokenType.COMMAND, 237 "RETURNING": TokenType.RETURNING, 238 "REVOKE": TokenType.COMMAND, 239 "SERIAL": TokenType.SERIAL, 240 "SMALLSERIAL": TokenType.SMALLSERIAL, 241 "TEMP": TokenType.TEMPORARY, 242 "CSTRING": TokenType.PSEUDO_TYPE, 243 } 244 245 SINGLE_TOKENS = { 246 **tokens.Tokenizer.SINGLE_TOKENS, 247 "$": TokenType.PARAMETER, 248 } 249 250 VAR_SINGLE_TOKENS = {"$"} 251 252 class Parser(parser.Parser): 253 STRICT_CAST = False 254 255 FUNCTIONS = { 256 **parser.Parser.FUNCTIONS, # type: ignore 257 "DATE_TRUNC": lambda args: exp.TimestampTrunc( 258 this=seq_get(args, 1), unit=seq_get(args, 0) 259 ), 260 "GENERATE_SERIES": _generate_series, 261 "NOW": exp.CurrentTimestamp.from_arg_list, 262 "TO_CHAR": format_time_lambda(exp.TimeToStr, "postgres"), 263 "TO_TIMESTAMP": _to_timestamp, 264 } 265 266 FUNCTION_PARSERS = { 267 **parser.Parser.FUNCTION_PARSERS, 268 "DATE_PART": lambda self: self._parse_date_part(), 269 } 270 271 BITWISE = { 272 **parser.Parser.BITWISE, # type: ignore 273 TokenType.HASH: exp.BitwiseXor, 274 } 275 276 FACTOR = { 277 **parser.Parser.FACTOR, 278 TokenType.CARET: exp.Pow, 279 } 280 281 RANGE_PARSERS = { 282 **parser.Parser.RANGE_PARSERS, # type: ignore 283 TokenType.DAMP: binary_range_parser(exp.ArrayOverlaps), 284 TokenType.AT_GT: binary_range_parser(exp.ArrayContains), 285 TokenType.LT_AT: binary_range_parser(exp.ArrayContained), 286 } 287 288 def _parse_date_part(self) -> exp.Expression: 289 part = self._parse_type() 290 self._match(TokenType.COMMA) 291 value = self._parse_bitwise() 292 293 if part and part.is_string: 294 part = exp.Var(this=part.name) 295 296 return self.expression(exp.Extract, this=part, expression=value) 297 298 class Generator(generator.Generator): 299 INTERVAL_ALLOWS_PLURAL_FORM = False 300 LOCKING_READS_SUPPORTED = True 301 JOIN_HINTS = False 302 TABLE_HINTS = False 303 PARAMETER_TOKEN = "$" 304 305 TYPE_MAPPING = { 306 **generator.Generator.TYPE_MAPPING, # type: ignore 307 exp.DataType.Type.TINYINT: "SMALLINT", 308 exp.DataType.Type.FLOAT: "REAL", 309 exp.DataType.Type.DOUBLE: "DOUBLE PRECISION", 310 exp.DataType.Type.BINARY: "BYTEA", 311 exp.DataType.Type.VARBINARY: "BYTEA", 312 exp.DataType.Type.DATETIME: "TIMESTAMP", 313 } 314 315 TRANSFORMS = { 316 **generator.Generator.TRANSFORMS, # type: ignore 317 exp.BitwiseXor: lambda self, e: self.binary(e, "#"), 318 exp.ColumnDef: transforms.preprocess( 319 [ 320 _auto_increment_to_serial, 321 _serial_to_generated, 322 ], 323 ), 324 exp.JSONExtract: arrow_json_extract_sql, 325 exp.JSONExtractScalar: arrow_json_extract_scalar_sql, 326 exp.JSONBExtract: lambda self, e: self.binary(e, "#>"), 327 exp.JSONBExtractScalar: lambda self, e: self.binary(e, "#>>"), 328 exp.JSONBContains: lambda self, e: self.binary(e, "?"), 329 exp.Pow: lambda self, e: self.binary(e, "^"), 330 exp.CurrentDate: no_paren_current_date_sql, 331 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 332 exp.DateAdd: _date_add_sql("+"), 333 exp.DateStrToDate: datestrtodate_sql, 334 exp.DateSub: _date_add_sql("-"), 335 exp.DateDiff: _date_diff_sql, 336 exp.LogicalOr: rename_func("BOOL_OR"), 337 exp.LogicalAnd: rename_func("BOOL_AND"), 338 exp.Max: max_or_greatest, 339 exp.Min: min_or_least, 340 exp.ArrayOverlaps: lambda self, e: self.binary(e, "&&"), 341 exp.ArrayContains: lambda self, e: self.binary(e, "@>"), 342 exp.ArrayContained: lambda self, e: self.binary(e, "<@"), 343 exp.Merge: transforms.preprocess([transforms.remove_target_from_merge]), 344 exp.RegexpLike: lambda self, e: self.binary(e, "~"), 345 exp.RegexpILike: lambda self, e: self.binary(e, "~*"), 346 exp.StrPosition: str_position_sql, 347 exp.StrToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')}, {self.format_time(e)})", 348 exp.Substring: _substring_sql, 349 exp.TimestampTrunc: timestamptrunc_sql, 350 exp.TimeStrToTime: lambda self, e: f"CAST({self.sql(e, 'this')} AS TIMESTAMP)", 351 exp.TimeToStr: lambda self, e: f"TO_CHAR({self.sql(e, 'this')}, {self.format_time(e)})", 352 exp.TableSample: no_tablesample_sql, 353 exp.ToChar: lambda self, e: self.function_fallback_sql(e), 354 exp.Trim: trim_sql, 355 exp.TryCast: no_trycast_sql, 356 exp.UnixToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')})", 357 exp.DataType: _datatype_sql, 358 exp.GroupConcat: _string_agg_sql, 359 exp.Array: lambda self, e: f"{self.normalize_func('ARRAY')}({self.sql(e.expressions[0])})" 360 if isinstance(seq_get(e.expressions, 0), exp.Select) 361 else f"{self.normalize_func('ARRAY')}[{self.expressions(e, flat=True)}]", 362 } 363 364 PROPERTIES_LOCATION = { 365 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 366 exp.TransientProperty: exp.Properties.Location.UNSUPPORTED, 367 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 368 }
179class Postgres(Dialect): 180 null_ordering = "nulls_are_large" 181 time_format = "'YYYY-MM-DD HH24:MI:SS'" 182 time_mapping = { 183 "AM": "%p", 184 "PM": "%p", 185 "D": "%u", # 1-based day of week 186 "DD": "%d", # day of month 187 "DDD": "%j", # zero padded day of year 188 "FMDD": "%-d", # - is no leading zero for Python; same for FM in postgres 189 "FMDDD": "%-j", # day of year 190 "FMHH12": "%-I", # 9 191 "FMHH24": "%-H", # 9 192 "FMMI": "%-M", # Minute 193 "FMMM": "%-m", # 1 194 "FMSS": "%-S", # Second 195 "HH12": "%I", # 09 196 "HH24": "%H", # 09 197 "MI": "%M", # zero padded minute 198 "MM": "%m", # 01 199 "OF": "%z", # utc offset 200 "SS": "%S", # zero padded second 201 "TMDay": "%A", # TM is locale dependent 202 "TMDy": "%a", 203 "TMMon": "%b", # Sep 204 "TMMonth": "%B", # September 205 "TZ": "%Z", # uppercase timezone name 206 "US": "%f", # zero padded microsecond 207 "WW": "%U", # 1-based week of year 208 "YY": "%y", # 15 209 "YYYY": "%Y", # 2015 210 } 211 212 class Tokenizer(tokens.Tokenizer): 213 QUOTES = ["'", "$$"] 214 215 BIT_STRINGS = [("b'", "'"), ("B'", "'")] 216 HEX_STRINGS = [("x'", "'"), ("X'", "'")] 217 BYTE_STRINGS = [("e'", "'"), ("E'", "'")] 218 219 KEYWORDS = { 220 **tokens.Tokenizer.KEYWORDS, 221 "~~": TokenType.LIKE, 222 "~~*": TokenType.ILIKE, 223 "~*": TokenType.IRLIKE, 224 "~": TokenType.RLIKE, 225 "@>": TokenType.AT_GT, 226 "<@": TokenType.LT_AT, 227 "BEGIN": TokenType.COMMAND, 228 "BEGIN TRANSACTION": TokenType.BEGIN, 229 "BIGSERIAL": TokenType.BIGSERIAL, 230 "CHARACTER VARYING": TokenType.VARCHAR, 231 "DECLARE": TokenType.COMMAND, 232 "DO": TokenType.COMMAND, 233 "HSTORE": TokenType.HSTORE, 234 "JSONB": TokenType.JSONB, 235 "REFRESH": TokenType.COMMAND, 236 "REINDEX": TokenType.COMMAND, 237 "RESET": TokenType.COMMAND, 238 "RETURNING": TokenType.RETURNING, 239 "REVOKE": TokenType.COMMAND, 240 "SERIAL": TokenType.SERIAL, 241 "SMALLSERIAL": TokenType.SMALLSERIAL, 242 "TEMP": TokenType.TEMPORARY, 243 "CSTRING": TokenType.PSEUDO_TYPE, 244 } 245 246 SINGLE_TOKENS = { 247 **tokens.Tokenizer.SINGLE_TOKENS, 248 "$": TokenType.PARAMETER, 249 } 250 251 VAR_SINGLE_TOKENS = {"$"} 252 253 class Parser(parser.Parser): 254 STRICT_CAST = False 255 256 FUNCTIONS = { 257 **parser.Parser.FUNCTIONS, # type: ignore 258 "DATE_TRUNC": lambda args: exp.TimestampTrunc( 259 this=seq_get(args, 1), unit=seq_get(args, 0) 260 ), 261 "GENERATE_SERIES": _generate_series, 262 "NOW": exp.CurrentTimestamp.from_arg_list, 263 "TO_CHAR": format_time_lambda(exp.TimeToStr, "postgres"), 264 "TO_TIMESTAMP": _to_timestamp, 265 } 266 267 FUNCTION_PARSERS = { 268 **parser.Parser.FUNCTION_PARSERS, 269 "DATE_PART": lambda self: self._parse_date_part(), 270 } 271 272 BITWISE = { 273 **parser.Parser.BITWISE, # type: ignore 274 TokenType.HASH: exp.BitwiseXor, 275 } 276 277 FACTOR = { 278 **parser.Parser.FACTOR, 279 TokenType.CARET: exp.Pow, 280 } 281 282 RANGE_PARSERS = { 283 **parser.Parser.RANGE_PARSERS, # type: ignore 284 TokenType.DAMP: binary_range_parser(exp.ArrayOverlaps), 285 TokenType.AT_GT: binary_range_parser(exp.ArrayContains), 286 TokenType.LT_AT: binary_range_parser(exp.ArrayContained), 287 } 288 289 def _parse_date_part(self) -> exp.Expression: 290 part = self._parse_type() 291 self._match(TokenType.COMMA) 292 value = self._parse_bitwise() 293 294 if part and part.is_string: 295 part = exp.Var(this=part.name) 296 297 return self.expression(exp.Extract, this=part, expression=value) 298 299 class Generator(generator.Generator): 300 INTERVAL_ALLOWS_PLURAL_FORM = False 301 LOCKING_READS_SUPPORTED = True 302 JOIN_HINTS = False 303 TABLE_HINTS = False 304 PARAMETER_TOKEN = "$" 305 306 TYPE_MAPPING = { 307 **generator.Generator.TYPE_MAPPING, # type: ignore 308 exp.DataType.Type.TINYINT: "SMALLINT", 309 exp.DataType.Type.FLOAT: "REAL", 310 exp.DataType.Type.DOUBLE: "DOUBLE PRECISION", 311 exp.DataType.Type.BINARY: "BYTEA", 312 exp.DataType.Type.VARBINARY: "BYTEA", 313 exp.DataType.Type.DATETIME: "TIMESTAMP", 314 } 315 316 TRANSFORMS = { 317 **generator.Generator.TRANSFORMS, # type: ignore 318 exp.BitwiseXor: lambda self, e: self.binary(e, "#"), 319 exp.ColumnDef: transforms.preprocess( 320 [ 321 _auto_increment_to_serial, 322 _serial_to_generated, 323 ], 324 ), 325 exp.JSONExtract: arrow_json_extract_sql, 326 exp.JSONExtractScalar: arrow_json_extract_scalar_sql, 327 exp.JSONBExtract: lambda self, e: self.binary(e, "#>"), 328 exp.JSONBExtractScalar: lambda self, e: self.binary(e, "#>>"), 329 exp.JSONBContains: lambda self, e: self.binary(e, "?"), 330 exp.Pow: lambda self, e: self.binary(e, "^"), 331 exp.CurrentDate: no_paren_current_date_sql, 332 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 333 exp.DateAdd: _date_add_sql("+"), 334 exp.DateStrToDate: datestrtodate_sql, 335 exp.DateSub: _date_add_sql("-"), 336 exp.DateDiff: _date_diff_sql, 337 exp.LogicalOr: rename_func("BOOL_OR"), 338 exp.LogicalAnd: rename_func("BOOL_AND"), 339 exp.Max: max_or_greatest, 340 exp.Min: min_or_least, 341 exp.ArrayOverlaps: lambda self, e: self.binary(e, "&&"), 342 exp.ArrayContains: lambda self, e: self.binary(e, "@>"), 343 exp.ArrayContained: lambda self, e: self.binary(e, "<@"), 344 exp.Merge: transforms.preprocess([transforms.remove_target_from_merge]), 345 exp.RegexpLike: lambda self, e: self.binary(e, "~"), 346 exp.RegexpILike: lambda self, e: self.binary(e, "~*"), 347 exp.StrPosition: str_position_sql, 348 exp.StrToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')}, {self.format_time(e)})", 349 exp.Substring: _substring_sql, 350 exp.TimestampTrunc: timestamptrunc_sql, 351 exp.TimeStrToTime: lambda self, e: f"CAST({self.sql(e, 'this')} AS TIMESTAMP)", 352 exp.TimeToStr: lambda self, e: f"TO_CHAR({self.sql(e, 'this')}, {self.format_time(e)})", 353 exp.TableSample: no_tablesample_sql, 354 exp.ToChar: lambda self, e: self.function_fallback_sql(e), 355 exp.Trim: trim_sql, 356 exp.TryCast: no_trycast_sql, 357 exp.UnixToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')})", 358 exp.DataType: _datatype_sql, 359 exp.GroupConcat: _string_agg_sql, 360 exp.Array: lambda self, e: f"{self.normalize_func('ARRAY')}({self.sql(e.expressions[0])})" 361 if isinstance(seq_get(e.expressions, 0), exp.Select) 362 else f"{self.normalize_func('ARRAY')}[{self.expressions(e, flat=True)}]", 363 } 364 365 PROPERTIES_LOCATION = { 366 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 367 exp.TransientProperty: exp.Properties.Location.UNSUPPORTED, 368 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 369 }
212 class Tokenizer(tokens.Tokenizer): 213 QUOTES = ["'", "$$"] 214 215 BIT_STRINGS = [("b'", "'"), ("B'", "'")] 216 HEX_STRINGS = [("x'", "'"), ("X'", "'")] 217 BYTE_STRINGS = [("e'", "'"), ("E'", "'")] 218 219 KEYWORDS = { 220 **tokens.Tokenizer.KEYWORDS, 221 "~~": TokenType.LIKE, 222 "~~*": TokenType.ILIKE, 223 "~*": TokenType.IRLIKE, 224 "~": TokenType.RLIKE, 225 "@>": TokenType.AT_GT, 226 "<@": TokenType.LT_AT, 227 "BEGIN": TokenType.COMMAND, 228 "BEGIN TRANSACTION": TokenType.BEGIN, 229 "BIGSERIAL": TokenType.BIGSERIAL, 230 "CHARACTER VARYING": TokenType.VARCHAR, 231 "DECLARE": TokenType.COMMAND, 232 "DO": TokenType.COMMAND, 233 "HSTORE": TokenType.HSTORE, 234 "JSONB": TokenType.JSONB, 235 "REFRESH": TokenType.COMMAND, 236 "REINDEX": TokenType.COMMAND, 237 "RESET": TokenType.COMMAND, 238 "RETURNING": TokenType.RETURNING, 239 "REVOKE": TokenType.COMMAND, 240 "SERIAL": TokenType.SERIAL, 241 "SMALLSERIAL": TokenType.SMALLSERIAL, 242 "TEMP": TokenType.TEMPORARY, 243 "CSTRING": TokenType.PSEUDO_TYPE, 244 } 245 246 SINGLE_TOKENS = { 247 **tokens.Tokenizer.SINGLE_TOKENS, 248 "$": TokenType.PARAMETER, 249 } 250 251 VAR_SINGLE_TOKENS = {"$"}
Inherited Members
253 class Parser(parser.Parser): 254 STRICT_CAST = False 255 256 FUNCTIONS = { 257 **parser.Parser.FUNCTIONS, # type: ignore 258 "DATE_TRUNC": lambda args: exp.TimestampTrunc( 259 this=seq_get(args, 1), unit=seq_get(args, 0) 260 ), 261 "GENERATE_SERIES": _generate_series, 262 "NOW": exp.CurrentTimestamp.from_arg_list, 263 "TO_CHAR": format_time_lambda(exp.TimeToStr, "postgres"), 264 "TO_TIMESTAMP": _to_timestamp, 265 } 266 267 FUNCTION_PARSERS = { 268 **parser.Parser.FUNCTION_PARSERS, 269 "DATE_PART": lambda self: self._parse_date_part(), 270 } 271 272 BITWISE = { 273 **parser.Parser.BITWISE, # type: ignore 274 TokenType.HASH: exp.BitwiseXor, 275 } 276 277 FACTOR = { 278 **parser.Parser.FACTOR, 279 TokenType.CARET: exp.Pow, 280 } 281 282 RANGE_PARSERS = { 283 **parser.Parser.RANGE_PARSERS, # type: ignore 284 TokenType.DAMP: binary_range_parser(exp.ArrayOverlaps), 285 TokenType.AT_GT: binary_range_parser(exp.ArrayContains), 286 TokenType.LT_AT: binary_range_parser(exp.ArrayContained), 287 } 288 289 def _parse_date_part(self) -> exp.Expression: 290 part = self._parse_type() 291 self._match(TokenType.COMMA) 292 value = self._parse_bitwise() 293 294 if part and part.is_string: 295 part = exp.Var(this=part.name) 296 297 return self.expression(exp.Extract, this=part, expression=value)
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
299 class Generator(generator.Generator): 300 INTERVAL_ALLOWS_PLURAL_FORM = False 301 LOCKING_READS_SUPPORTED = True 302 JOIN_HINTS = False 303 TABLE_HINTS = False 304 PARAMETER_TOKEN = "$" 305 306 TYPE_MAPPING = { 307 **generator.Generator.TYPE_MAPPING, # type: ignore 308 exp.DataType.Type.TINYINT: "SMALLINT", 309 exp.DataType.Type.FLOAT: "REAL", 310 exp.DataType.Type.DOUBLE: "DOUBLE PRECISION", 311 exp.DataType.Type.BINARY: "BYTEA", 312 exp.DataType.Type.VARBINARY: "BYTEA", 313 exp.DataType.Type.DATETIME: "TIMESTAMP", 314 } 315 316 TRANSFORMS = { 317 **generator.Generator.TRANSFORMS, # type: ignore 318 exp.BitwiseXor: lambda self, e: self.binary(e, "#"), 319 exp.ColumnDef: transforms.preprocess( 320 [ 321 _auto_increment_to_serial, 322 _serial_to_generated, 323 ], 324 ), 325 exp.JSONExtract: arrow_json_extract_sql, 326 exp.JSONExtractScalar: arrow_json_extract_scalar_sql, 327 exp.JSONBExtract: lambda self, e: self.binary(e, "#>"), 328 exp.JSONBExtractScalar: lambda self, e: self.binary(e, "#>>"), 329 exp.JSONBContains: lambda self, e: self.binary(e, "?"), 330 exp.Pow: lambda self, e: self.binary(e, "^"), 331 exp.CurrentDate: no_paren_current_date_sql, 332 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 333 exp.DateAdd: _date_add_sql("+"), 334 exp.DateStrToDate: datestrtodate_sql, 335 exp.DateSub: _date_add_sql("-"), 336 exp.DateDiff: _date_diff_sql, 337 exp.LogicalOr: rename_func("BOOL_OR"), 338 exp.LogicalAnd: rename_func("BOOL_AND"), 339 exp.Max: max_or_greatest, 340 exp.Min: min_or_least, 341 exp.ArrayOverlaps: lambda self, e: self.binary(e, "&&"), 342 exp.ArrayContains: lambda self, e: self.binary(e, "@>"), 343 exp.ArrayContained: lambda self, e: self.binary(e, "<@"), 344 exp.Merge: transforms.preprocess([transforms.remove_target_from_merge]), 345 exp.RegexpLike: lambda self, e: self.binary(e, "~"), 346 exp.RegexpILike: lambda self, e: self.binary(e, "~*"), 347 exp.StrPosition: str_position_sql, 348 exp.StrToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')}, {self.format_time(e)})", 349 exp.Substring: _substring_sql, 350 exp.TimestampTrunc: timestamptrunc_sql, 351 exp.TimeStrToTime: lambda self, e: f"CAST({self.sql(e, 'this')} AS TIMESTAMP)", 352 exp.TimeToStr: lambda self, e: f"TO_CHAR({self.sql(e, 'this')}, {self.format_time(e)})", 353 exp.TableSample: no_tablesample_sql, 354 exp.ToChar: lambda self, e: self.function_fallback_sql(e), 355 exp.Trim: trim_sql, 356 exp.TryCast: no_trycast_sql, 357 exp.UnixToTime: lambda self, e: f"TO_TIMESTAMP({self.sql(e, 'this')})", 358 exp.DataType: _datatype_sql, 359 exp.GroupConcat: _string_agg_sql, 360 exp.Array: lambda self, e: f"{self.normalize_func('ARRAY')}({self.sql(e.expressions[0])})" 361 if isinstance(seq_get(e.expressions, 0), exp.Select) 362 else f"{self.normalize_func('ARRAY')}[{self.expressions(e, flat=True)}]", 363 } 364 365 PROPERTIES_LOCATION = { 366 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 367 exp.TransientProperty: exp.Properties.Location.UNSUPPORTED, 368 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 369 }
Generator interprets the given syntax tree and produces a SQL string as an output.
Arguments:
- time_mapping (dict): the dictionary of custom time mappings in which the key represents a python time format and the output the target time format
- time_trie (trie): a trie of the time_mapping keys
- pretty (bool): if set to True the returned string will be formatted. Default: False.
- quote_start (str): specifies which starting character to use to delimit quotes. Default: '.
- quote_end (str): specifies which ending character to use to delimit quotes. Default: '.
- identifier_start (str): specifies which starting character to use to delimit identifiers. Default: ".
- identifier_end (str): specifies which ending character to use to delimit identifiers. Default: ".
- identify (bool | str): 'always': always quote, 'safe': quote identifiers if they don't contain an upcase, True defaults to always.
- normalize (bool): if set to True all identifiers will lower cased
- string_escape (str): specifies a string escape character. Default: '.
- identifier_escape (str): specifies an identifier escape character. Default: ".
- pad (int): determines padding in a formatted string. Default: 2.
- indent (int): determines the size of indentation in a formatted string. Default: 4.
- unnest_column_only (bool): if true unnest table aliases are considered only as column aliases
- normalize_functions (str): normalize function names, "upper", "lower", or None Default: "upper"
- alias_post_tablesample (bool): if the table alias comes after tablesample Default: False
- unsupported_level (ErrorLevel): determines the generator's behavior when it encounters unsupported expressions. Default ErrorLevel.WARN.
- null_ordering (str): Indicates the default null ordering method to use if not explicitly set. Options are "nulls_are_small", "nulls_are_large", "nulls_are_last". Default: "nulls_are_small"
- max_unsupported (int): Maximum number of unsupported messages to include in a raised UnsupportedError. This is only relevant if unsupported_level is ErrorLevel.RAISE. Default: 3
- leading_comma (bool): if the the comma is leading or trailing in select statements Default: False
- max_text_width: The max number of characters in a segment before creating new lines in pretty mode. The default is on the smaller end because the length only represents a segment and not the true line length. Default: 80
- comments: Whether or not to preserve comments in the output SQL code. Default: True
Inherited Members
- sqlglot.generator.Generator
- Generator
- generate
- unsupported
- sep
- seg
- pad_comment
- maybe_comment
- wrap
- no_identify
- normalize_func
- indent
- sql
- uncache_sql
- cache_sql
- characterset_sql
- column_sql
- columnposition_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
- bytestring_sql
- datatype_sql
- directory_sql
- delete_sql
- drop_sql
- except_sql
- except_op
- fetch_sql
- filter_sql
- hint_sql
- index_sql
- identifier_sql
- inputoutputformat_sql
- national_sql
- partition_sql
- properties_sql
- root_properties
- properties
- with_properties
- locate_properties
- property_sql
- likeproperty_sql
- fallbackproperty_sql
- journalproperty_sql
- freespaceproperty_sql
- afterjournalproperty_sql
- checksumproperty_sql
- mergeblockratioproperty_sql
- datablocksizeproperty_sql
- blockcompressionproperty_sql
- isolatedloadingproperty_sql
- lockingproperty_sql
- withdataproperty_sql
- insert_sql
- intersect_sql
- intersect_op
- introducer_sql
- pseudotype_sql
- onconflict_sql
- returning_sql
- rowformatdelimitedproperty_sql
- table_sql
- tablesample_sql
- pivot_sql
- tuple_sql
- update_sql
- values_sql
- var_sql
- into_sql
- from_sql
- group_sql
- having_sql
- join_sql
- lambda_sql
- lateral_sql
- limit_sql
- offset_sql
- setitem_sql
- set_sql
- pragma_sql
- lock_sql
- literal_sql
- loaddata_sql
- null_sql
- boolean_sql
- order_sql
- cluster_sql
- distribute_sql
- sort_sql
- ordered_sql
- matchrecognize_sql
- query_modifiers
- 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
- windowspec_sql
- withingroup_sql
- between_sql
- bracket_sql
- all_sql
- any_sql
- exists_sql
- case_sql
- constraint_sql
- nextvaluefor_sql
- extract_sql
- trim_sql
- concat_sql
- check_sql
- foreignkey_sql
- primarykey_sql
- unique_sql
- if_sql
- matchagainst_sql
- jsonkeyvalue_sql
- jsonobject_sql
- in_sql
- in_unnest_op
- interval_sql
- return_sql
- reference_sql
- anonymous_sql
- paren_sql
- neg_sql
- not_sql
- alias_sql
- aliases_sql
- attimezone_sql
- add_sql
- and_sql
- connector_sql
- bitwiseand_sql
- bitwiseleftshift_sql
- bitwisenot_sql
- bitwiseor_sql
- bitwiserightshift_sql
- bitwisexor_sql
- cast_sql
- currentdate_sql
- collate_sql
- command_sql
- comment_sql
- transaction_sql
- commit_sql
- rollback_sql
- altercolumn_sql
- renametable_sql
- altertable_sql
- droppartition_sql
- addconstraint_sql
- distinct_sql
- ignorenulls_sql
- respectnulls_sql
- intdiv_sql
- dpipe_sql
- div_sql
- overlaps_sql
- distance_sql
- dot_sql
- eq_sql
- escape_sql
- glob_sql
- gt_sql
- gte_sql
- ilike_sql
- ilikeany_sql
- is_sql
- like_sql
- likeany_sql
- similarto_sql
- lt_sql
- lte_sql
- mod_sql
- mul_sql
- neq_sql
- nullsafeeq_sql
- nullsafeneq_sql
- or_sql
- slice_sql
- sub_sql
- trycast_sql
- use_sql
- binary
- function_fallback_sql
- func
- format_args
- text_width
- format_time
- expressions
- op_expressions
- naked_property
- set_operation
- tag_sql
- token_sql
- userdefinedfunction_sql
- joinhint_sql
- kwarg_sql
- when_sql
- merge_sql
- tochar_sql