sqlglot.dialects.presto
1from __future__ import annotations 2 3import typing as t 4 5from sqlglot import exp, generator, parser, tokens, transforms 6from sqlglot.dialects.dialect import ( 7 Dialect, 8 date_trunc_to_time, 9 format_time_lambda, 10 if_sql, 11 no_ilike_sql, 12 no_safe_divide_sql, 13 rename_func, 14 struct_extract_sql, 15 timestamptrunc_sql, 16 timestrtotime_sql, 17) 18from sqlglot.dialects.mysql import MySQL 19from sqlglot.errors import UnsupportedError 20from sqlglot.helper import seq_get 21from sqlglot.tokens import TokenType 22 23 24def _approx_distinct_sql(self: generator.Generator, expression: exp.ApproxDistinct) -> str: 25 accuracy = expression.args.get("accuracy") 26 accuracy = ", " + self.sql(accuracy) if accuracy else "" 27 return f"APPROX_DISTINCT({self.sql(expression, 'this')}{accuracy})" 28 29 30def _datatype_sql(self: generator.Generator, expression: exp.DataType) -> str: 31 sql = self.datatype_sql(expression) 32 if expression.this == exp.DataType.Type.TIMESTAMPTZ: 33 sql = f"{sql} WITH TIME ZONE" 34 return sql 35 36 37def _explode_to_unnest_sql(self: generator.Generator, expression: exp.Lateral) -> str: 38 if isinstance(expression.this, (exp.Explode, exp.Posexplode)): 39 return self.sql( 40 exp.Join( 41 this=exp.Unnest( 42 expressions=[expression.this.this], 43 alias=expression.args.get("alias"), 44 ordinality=isinstance(expression.this, exp.Posexplode), 45 ), 46 kind="cross", 47 ) 48 ) 49 return self.lateral_sql(expression) 50 51 52def _initcap_sql(self: generator.Generator, expression: exp.Initcap) -> str: 53 regex = r"(\w)(\w*)" 54 return f"REGEXP_REPLACE({self.sql(expression, 'this')}, '{regex}', x -> UPPER(x[1]) || LOWER(x[2]))" 55 56 57def _decode_sql(self: generator.Generator, expression: exp.Decode) -> str: 58 _ensure_utf8(expression.args["charset"]) 59 return self.func("FROM_UTF8", expression.this, expression.args.get("replace")) 60 61 62def _encode_sql(self: generator.Generator, expression: exp.Encode) -> str: 63 _ensure_utf8(expression.args["charset"]) 64 return f"TO_UTF8({self.sql(expression, 'this')})" 65 66 67def _no_sort_array(self: generator.Generator, expression: exp.SortArray) -> str: 68 if expression.args.get("asc") == exp.false(): 69 comparator = "(a, b) -> CASE WHEN a < b THEN 1 WHEN a > b THEN -1 ELSE 0 END" 70 else: 71 comparator = None 72 return self.func("ARRAY_SORT", expression.this, comparator) 73 74 75def _schema_sql(self: generator.Generator, expression: exp.Schema) -> str: 76 if isinstance(expression.parent, exp.Property): 77 columns = ", ".join(f"'{c.name}'" for c in expression.expressions) 78 return f"ARRAY[{columns}]" 79 80 if expression.parent: 81 for schema in expression.parent.find_all(exp.Schema): 82 if isinstance(schema.parent, exp.Property): 83 expression = expression.copy() 84 expression.expressions.extend(schema.expressions) 85 86 return self.schema_sql(expression) 87 88 89def _quantile_sql(self: generator.Generator, expression: exp.Quantile) -> str: 90 self.unsupported("Presto does not support exact quantiles") 91 return f"APPROX_PERCENTILE({self.sql(expression, 'this')}, {self.sql(expression, 'quantile')})" 92 93 94def _str_to_time_sql( 95 self: generator.Generator, expression: exp.StrToDate | exp.StrToTime | exp.TsOrDsToDate 96) -> str: 97 return f"DATE_PARSE({self.sql(expression, 'this')}, {self.format_time(expression)})" 98 99 100def _ts_or_ds_to_date_sql(self: generator.Generator, expression: exp.TsOrDsToDate) -> str: 101 time_format = self.format_time(expression) 102 if time_format and time_format not in (Presto.time_format, Presto.date_format): 103 return f"CAST({_str_to_time_sql(self, expression)} AS DATE)" 104 return f"CAST(SUBSTR(CAST({self.sql(expression, 'this')} AS VARCHAR), 1, 10) AS DATE)" 105 106 107def _ts_or_ds_add_sql(self: generator.Generator, expression: exp.TsOrDsAdd) -> str: 108 this = expression.this 109 110 if not isinstance(this, exp.CurrentDate): 111 this = self.func( 112 "DATE_PARSE", 113 self.func( 114 "SUBSTR", 115 this if this.is_string else exp.cast(this, "VARCHAR"), 116 exp.Literal.number(1), 117 exp.Literal.number(10), 118 ), 119 Presto.date_format, 120 ) 121 122 return self.func( 123 "DATE_ADD", 124 exp.Literal.string(expression.text("unit") or "day"), 125 expression.expression, 126 this, 127 ) 128 129 130def _sequence_sql(self: generator.Generator, expression: exp.GenerateSeries) -> str: 131 start = expression.args["start"] 132 end = expression.args["end"] 133 step = expression.args.get("step") 134 135 target_type = None 136 137 if isinstance(start, exp.Cast): 138 target_type = start.to 139 elif isinstance(end, exp.Cast): 140 target_type = end.to 141 142 if target_type and target_type.this == exp.DataType.Type.TIMESTAMP: 143 to = target_type.copy() 144 145 if target_type is start.to: 146 end = exp.Cast(this=end, to=to) 147 else: 148 start = exp.Cast(this=start, to=to) 149 150 sql = self.func("SEQUENCE", start, end, step) 151 if isinstance(expression.parent, exp.Table): 152 sql = f"UNNEST({sql})" 153 154 return sql 155 156 157def _ensure_utf8(charset: exp.Literal) -> None: 158 if charset.name.lower() != "utf-8": 159 raise UnsupportedError(f"Unsupported charset {charset}") 160 161 162def _approx_percentile(args: t.Sequence) -> exp.Expression: 163 if len(args) == 4: 164 return exp.ApproxQuantile( 165 this=seq_get(args, 0), 166 weight=seq_get(args, 1), 167 quantile=seq_get(args, 2), 168 accuracy=seq_get(args, 3), 169 ) 170 if len(args) == 3: 171 return exp.ApproxQuantile( 172 this=seq_get(args, 0), 173 quantile=seq_get(args, 1), 174 accuracy=seq_get(args, 2), 175 ) 176 return exp.ApproxQuantile.from_arg_list(args) 177 178 179def _from_unixtime(args: t.Sequence) -> exp.Expression: 180 if len(args) == 3: 181 return exp.UnixToTime( 182 this=seq_get(args, 0), 183 hours=seq_get(args, 1), 184 minutes=seq_get(args, 2), 185 ) 186 if len(args) == 2: 187 return exp.UnixToTime( 188 this=seq_get(args, 0), 189 zone=seq_get(args, 1), 190 ) 191 return exp.UnixToTime.from_arg_list(args) 192 193 194class Presto(Dialect): 195 index_offset = 1 196 null_ordering = "nulls_are_last" 197 time_format = MySQL.time_format # type: ignore 198 time_mapping = MySQL.time_mapping # type: ignore 199 200 class Tokenizer(tokens.Tokenizer): 201 KEYWORDS = { 202 **tokens.Tokenizer.KEYWORDS, 203 "START": TokenType.BEGIN, 204 "ROW": TokenType.STRUCT, 205 } 206 207 class Parser(parser.Parser): 208 FUNCTIONS = { 209 **parser.Parser.FUNCTIONS, # type: ignore 210 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 211 "APPROX_PERCENTILE": _approx_percentile, 212 "CARDINALITY": exp.ArraySize.from_arg_list, 213 "CONTAINS": exp.ArrayContains.from_arg_list, 214 "DATE_ADD": lambda args: exp.DateAdd( 215 this=seq_get(args, 2), 216 expression=seq_get(args, 1), 217 unit=seq_get(args, 0), 218 ), 219 "DATE_DIFF": lambda args: exp.DateDiff( 220 this=seq_get(args, 2), 221 expression=seq_get(args, 1), 222 unit=seq_get(args, 0), 223 ), 224 "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"), 225 "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"), 226 "DATE_TRUNC": date_trunc_to_time, 227 "FROM_HEX": exp.Unhex.from_arg_list, 228 "FROM_UNIXTIME": _from_unixtime, 229 "FROM_UTF8": lambda args: exp.Decode( 230 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 231 ), 232 "NOW": exp.CurrentTimestamp.from_arg_list, 233 "SEQUENCE": exp.GenerateSeries.from_arg_list, 234 "STRPOS": lambda args: exp.StrPosition( 235 this=seq_get(args, 0), 236 substr=seq_get(args, 1), 237 instance=seq_get(args, 2), 238 ), 239 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 240 "TO_HEX": exp.Hex.from_arg_list, 241 "TO_UTF8": lambda args: exp.Encode( 242 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 243 ), 244 } 245 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 246 FUNCTION_PARSERS.pop("TRIM") 247 248 class Generator(generator.Generator): 249 INTERVAL_ALLOWS_PLURAL_FORM = False 250 JOIN_HINTS = False 251 TABLE_HINTS = False 252 STRUCT_DELIMITER = ("(", ")") 253 254 PROPERTIES_LOCATION = { 255 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 256 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 257 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 258 } 259 260 TYPE_MAPPING = { 261 **generator.Generator.TYPE_MAPPING, # type: ignore 262 exp.DataType.Type.INT: "INTEGER", 263 exp.DataType.Type.FLOAT: "REAL", 264 exp.DataType.Type.BINARY: "VARBINARY", 265 exp.DataType.Type.TEXT: "VARCHAR", 266 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 267 exp.DataType.Type.STRUCT: "ROW", 268 } 269 270 TRANSFORMS = { 271 **generator.Generator.TRANSFORMS, # type: ignore 272 exp.ApproxDistinct: _approx_distinct_sql, 273 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 274 exp.ArrayConcat: rename_func("CONCAT"), 275 exp.ArrayContains: rename_func("CONTAINS"), 276 exp.ArraySize: rename_func("CARDINALITY"), 277 exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 278 exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 279 exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})", 280 exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 281 exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 282 exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 283 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 284 exp.DataType: _datatype_sql, 285 exp.DateAdd: lambda self, e: self.func( 286 "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 287 ), 288 exp.DateDiff: lambda self, e: self.func( 289 "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 290 ), 291 exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)", 292 exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)", 293 exp.Decode: _decode_sql, 294 exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)", 295 exp.Encode: _encode_sql, 296 exp.GenerateSeries: _sequence_sql, 297 exp.Group: transforms.preprocess([transforms.unalias_group]), 298 exp.Hex: rename_func("TO_HEX"), 299 exp.If: if_sql, 300 exp.ILike: no_ilike_sql, 301 exp.Initcap: _initcap_sql, 302 exp.Lateral: _explode_to_unnest_sql, 303 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 304 exp.LogicalOr: rename_func("BOOL_OR"), 305 exp.LogicalAnd: rename_func("BOOL_AND"), 306 exp.Quantile: _quantile_sql, 307 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 308 exp.SafeDivide: no_safe_divide_sql, 309 exp.Schema: _schema_sql, 310 exp.Select: transforms.preprocess( 311 [ 312 transforms.eliminate_qualify, 313 transforms.eliminate_distinct_on, 314 transforms.explode_to_unnest, 315 ] 316 ), 317 exp.SortArray: _no_sort_array, 318 exp.StrPosition: rename_func("STRPOS"), 319 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 320 exp.StrToTime: _str_to_time_sql, 321 exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))", 322 exp.StructExtract: struct_extract_sql, 323 exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT='{e.name.upper()}'", 324 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 325 exp.TimestampTrunc: timestamptrunc_sql, 326 exp.TimeStrToDate: timestrtotime_sql, 327 exp.TimeStrToTime: timestrtotime_sql, 328 exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))", 329 exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})", 330 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 331 exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 332 exp.TsOrDsAdd: _ts_or_ds_add_sql, 333 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 334 exp.Unhex: rename_func("FROM_HEX"), 335 exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 336 exp.UnixToTime: rename_func("FROM_UNIXTIME"), 337 exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 338 exp.VariancePop: rename_func("VAR_POP"), 339 } 340 341 def interval_sql(self, expression: exp.Interval) -> str: 342 unit = self.sql(expression, "unit") 343 if expression.this and unit.lower().startswith("week"): 344 return f"({expression.this.name} * INTERVAL '7' day)" 345 return super().interval_sql(expression) 346 347 def transaction_sql(self, expression: exp.Transaction) -> str: 348 modes = expression.args.get("modes") 349 modes = f" {', '.join(modes)}" if modes else "" 350 return f"START TRANSACTION{modes}"
195class Presto(Dialect): 196 index_offset = 1 197 null_ordering = "nulls_are_last" 198 time_format = MySQL.time_format # type: ignore 199 time_mapping = MySQL.time_mapping # type: ignore 200 201 class Tokenizer(tokens.Tokenizer): 202 KEYWORDS = { 203 **tokens.Tokenizer.KEYWORDS, 204 "START": TokenType.BEGIN, 205 "ROW": TokenType.STRUCT, 206 } 207 208 class Parser(parser.Parser): 209 FUNCTIONS = { 210 **parser.Parser.FUNCTIONS, # type: ignore 211 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 212 "APPROX_PERCENTILE": _approx_percentile, 213 "CARDINALITY": exp.ArraySize.from_arg_list, 214 "CONTAINS": exp.ArrayContains.from_arg_list, 215 "DATE_ADD": lambda args: exp.DateAdd( 216 this=seq_get(args, 2), 217 expression=seq_get(args, 1), 218 unit=seq_get(args, 0), 219 ), 220 "DATE_DIFF": lambda args: exp.DateDiff( 221 this=seq_get(args, 2), 222 expression=seq_get(args, 1), 223 unit=seq_get(args, 0), 224 ), 225 "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"), 226 "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"), 227 "DATE_TRUNC": date_trunc_to_time, 228 "FROM_HEX": exp.Unhex.from_arg_list, 229 "FROM_UNIXTIME": _from_unixtime, 230 "FROM_UTF8": lambda args: exp.Decode( 231 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 232 ), 233 "NOW": exp.CurrentTimestamp.from_arg_list, 234 "SEQUENCE": exp.GenerateSeries.from_arg_list, 235 "STRPOS": lambda args: exp.StrPosition( 236 this=seq_get(args, 0), 237 substr=seq_get(args, 1), 238 instance=seq_get(args, 2), 239 ), 240 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 241 "TO_HEX": exp.Hex.from_arg_list, 242 "TO_UTF8": lambda args: exp.Encode( 243 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 244 ), 245 } 246 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 247 FUNCTION_PARSERS.pop("TRIM") 248 249 class Generator(generator.Generator): 250 INTERVAL_ALLOWS_PLURAL_FORM = False 251 JOIN_HINTS = False 252 TABLE_HINTS = False 253 STRUCT_DELIMITER = ("(", ")") 254 255 PROPERTIES_LOCATION = { 256 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 257 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 258 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 259 } 260 261 TYPE_MAPPING = { 262 **generator.Generator.TYPE_MAPPING, # type: ignore 263 exp.DataType.Type.INT: "INTEGER", 264 exp.DataType.Type.FLOAT: "REAL", 265 exp.DataType.Type.BINARY: "VARBINARY", 266 exp.DataType.Type.TEXT: "VARCHAR", 267 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 268 exp.DataType.Type.STRUCT: "ROW", 269 } 270 271 TRANSFORMS = { 272 **generator.Generator.TRANSFORMS, # type: ignore 273 exp.ApproxDistinct: _approx_distinct_sql, 274 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 275 exp.ArrayConcat: rename_func("CONCAT"), 276 exp.ArrayContains: rename_func("CONTAINS"), 277 exp.ArraySize: rename_func("CARDINALITY"), 278 exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 279 exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 280 exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})", 281 exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 282 exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 283 exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 284 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 285 exp.DataType: _datatype_sql, 286 exp.DateAdd: lambda self, e: self.func( 287 "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 288 ), 289 exp.DateDiff: lambda self, e: self.func( 290 "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 291 ), 292 exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)", 293 exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)", 294 exp.Decode: _decode_sql, 295 exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)", 296 exp.Encode: _encode_sql, 297 exp.GenerateSeries: _sequence_sql, 298 exp.Group: transforms.preprocess([transforms.unalias_group]), 299 exp.Hex: rename_func("TO_HEX"), 300 exp.If: if_sql, 301 exp.ILike: no_ilike_sql, 302 exp.Initcap: _initcap_sql, 303 exp.Lateral: _explode_to_unnest_sql, 304 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 305 exp.LogicalOr: rename_func("BOOL_OR"), 306 exp.LogicalAnd: rename_func("BOOL_AND"), 307 exp.Quantile: _quantile_sql, 308 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 309 exp.SafeDivide: no_safe_divide_sql, 310 exp.Schema: _schema_sql, 311 exp.Select: transforms.preprocess( 312 [ 313 transforms.eliminate_qualify, 314 transforms.eliminate_distinct_on, 315 transforms.explode_to_unnest, 316 ] 317 ), 318 exp.SortArray: _no_sort_array, 319 exp.StrPosition: rename_func("STRPOS"), 320 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 321 exp.StrToTime: _str_to_time_sql, 322 exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))", 323 exp.StructExtract: struct_extract_sql, 324 exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT='{e.name.upper()}'", 325 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 326 exp.TimestampTrunc: timestamptrunc_sql, 327 exp.TimeStrToDate: timestrtotime_sql, 328 exp.TimeStrToTime: timestrtotime_sql, 329 exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))", 330 exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})", 331 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 332 exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 333 exp.TsOrDsAdd: _ts_or_ds_add_sql, 334 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 335 exp.Unhex: rename_func("FROM_HEX"), 336 exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 337 exp.UnixToTime: rename_func("FROM_UNIXTIME"), 338 exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 339 exp.VariancePop: rename_func("VAR_POP"), 340 } 341 342 def interval_sql(self, expression: exp.Interval) -> str: 343 unit = self.sql(expression, "unit") 344 if expression.this and unit.lower().startswith("week"): 345 return f"({expression.this.name} * INTERVAL '7' day)" 346 return super().interval_sql(expression) 347 348 def transaction_sql(self, expression: exp.Transaction) -> str: 349 modes = expression.args.get("modes") 350 modes = f" {', '.join(modes)}" if modes else "" 351 return f"START TRANSACTION{modes}"
201 class Tokenizer(tokens.Tokenizer): 202 KEYWORDS = { 203 **tokens.Tokenizer.KEYWORDS, 204 "START": TokenType.BEGIN, 205 "ROW": TokenType.STRUCT, 206 }
Inherited Members
208 class Parser(parser.Parser): 209 FUNCTIONS = { 210 **parser.Parser.FUNCTIONS, # type: ignore 211 "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list, 212 "APPROX_PERCENTILE": _approx_percentile, 213 "CARDINALITY": exp.ArraySize.from_arg_list, 214 "CONTAINS": exp.ArrayContains.from_arg_list, 215 "DATE_ADD": lambda args: exp.DateAdd( 216 this=seq_get(args, 2), 217 expression=seq_get(args, 1), 218 unit=seq_get(args, 0), 219 ), 220 "DATE_DIFF": lambda args: exp.DateDiff( 221 this=seq_get(args, 2), 222 expression=seq_get(args, 1), 223 unit=seq_get(args, 0), 224 ), 225 "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"), 226 "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"), 227 "DATE_TRUNC": date_trunc_to_time, 228 "FROM_HEX": exp.Unhex.from_arg_list, 229 "FROM_UNIXTIME": _from_unixtime, 230 "FROM_UTF8": lambda args: exp.Decode( 231 this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8") 232 ), 233 "NOW": exp.CurrentTimestamp.from_arg_list, 234 "SEQUENCE": exp.GenerateSeries.from_arg_list, 235 "STRPOS": lambda args: exp.StrPosition( 236 this=seq_get(args, 0), 237 substr=seq_get(args, 1), 238 instance=seq_get(args, 2), 239 ), 240 "TO_UNIXTIME": exp.TimeToUnix.from_arg_list, 241 "TO_HEX": exp.Hex.from_arg_list, 242 "TO_UTF8": lambda args: exp.Encode( 243 this=seq_get(args, 0), charset=exp.Literal.string("utf-8") 244 ), 245 } 246 FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy() 247 FUNCTION_PARSERS.pop("TRIM")
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
249 class Generator(generator.Generator): 250 INTERVAL_ALLOWS_PLURAL_FORM = False 251 JOIN_HINTS = False 252 TABLE_HINTS = False 253 STRUCT_DELIMITER = ("(", ")") 254 255 PROPERTIES_LOCATION = { 256 **generator.Generator.PROPERTIES_LOCATION, # type: ignore 257 exp.LocationProperty: exp.Properties.Location.UNSUPPORTED, 258 exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED, 259 } 260 261 TYPE_MAPPING = { 262 **generator.Generator.TYPE_MAPPING, # type: ignore 263 exp.DataType.Type.INT: "INTEGER", 264 exp.DataType.Type.FLOAT: "REAL", 265 exp.DataType.Type.BINARY: "VARBINARY", 266 exp.DataType.Type.TEXT: "VARCHAR", 267 exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP", 268 exp.DataType.Type.STRUCT: "ROW", 269 } 270 271 TRANSFORMS = { 272 **generator.Generator.TRANSFORMS, # type: ignore 273 exp.ApproxDistinct: _approx_distinct_sql, 274 exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]", 275 exp.ArrayConcat: rename_func("CONCAT"), 276 exp.ArrayContains: rename_func("CONTAINS"), 277 exp.ArraySize: rename_func("CARDINALITY"), 278 exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 279 exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 280 exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})", 281 exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 282 exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 283 exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})", 284 exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP", 285 exp.DataType: _datatype_sql, 286 exp.DateAdd: lambda self, e: self.func( 287 "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 288 ), 289 exp.DateDiff: lambda self, e: self.func( 290 "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this 291 ), 292 exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)", 293 exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)", 294 exp.Decode: _decode_sql, 295 exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)", 296 exp.Encode: _encode_sql, 297 exp.GenerateSeries: _sequence_sql, 298 exp.Group: transforms.preprocess([transforms.unalias_group]), 299 exp.Hex: rename_func("TO_HEX"), 300 exp.If: if_sql, 301 exp.ILike: no_ilike_sql, 302 exp.Initcap: _initcap_sql, 303 exp.Lateral: _explode_to_unnest_sql, 304 exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"), 305 exp.LogicalOr: rename_func("BOOL_OR"), 306 exp.LogicalAnd: rename_func("BOOL_AND"), 307 exp.Quantile: _quantile_sql, 308 exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"), 309 exp.SafeDivide: no_safe_divide_sql, 310 exp.Schema: _schema_sql, 311 exp.Select: transforms.preprocess( 312 [ 313 transforms.eliminate_qualify, 314 transforms.eliminate_distinct_on, 315 transforms.explode_to_unnest, 316 ] 317 ), 318 exp.SortArray: _no_sort_array, 319 exp.StrPosition: rename_func("STRPOS"), 320 exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)", 321 exp.StrToTime: _str_to_time_sql, 322 exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))", 323 exp.StructExtract: struct_extract_sql, 324 exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT='{e.name.upper()}'", 325 exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'", 326 exp.TimestampTrunc: timestamptrunc_sql, 327 exp.TimeStrToDate: timestrtotime_sql, 328 exp.TimeStrToTime: timestrtotime_sql, 329 exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))", 330 exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})", 331 exp.TimeToUnix: rename_func("TO_UNIXTIME"), 332 exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)", 333 exp.TsOrDsAdd: _ts_or_ds_add_sql, 334 exp.TsOrDsToDate: _ts_or_ds_to_date_sql, 335 exp.Unhex: rename_func("FROM_HEX"), 336 exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})", 337 exp.UnixToTime: rename_func("FROM_UNIXTIME"), 338 exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)", 339 exp.VariancePop: rename_func("VAR_POP"), 340 } 341 342 def interval_sql(self, expression: exp.Interval) -> str: 343 unit = self.sql(expression, "unit") 344 if expression.this and unit.lower().startswith("week"): 345 return f"({expression.this.name} * INTERVAL '7' day)" 346 return super().interval_sql(expression) 347 348 def transaction_sql(self, expression: exp.Transaction) -> str: 349 modes = expression.args.get("modes") 350 modes = f" {', '.join(modes)}" if modes else "" 351 return f"START TRANSACTION{modes}"
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
- 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
- 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