Edit on GitHub

sqlglot.dialects.presto

View Source

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

class Presto(sqlglot.dialects.dialect.Dialect): View Source

176class Presto(Dialect):
177    index_offset = 1
178    null_ordering = "nulls_are_last"
179    time_format = MySQL.time_format  # type: ignore
180    time_mapping = MySQL.time_mapping  # type: ignore
181
182    class Tokenizer(tokens.Tokenizer):
183        KEYWORDS = {
184            **tokens.Tokenizer.KEYWORDS,
185            "START": TokenType.BEGIN,
186            "ROW": TokenType.STRUCT,
187        }
188
189    class Parser(parser.Parser):
190        FUNCTIONS = {
191            **parser.Parser.FUNCTIONS,  # type: ignore
192            "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list,
193            "CARDINALITY": exp.ArraySize.from_arg_list,
194            "CONTAINS": exp.ArrayContains.from_arg_list,
195            "DATE_ADD": lambda args: exp.DateAdd(
196                this=seq_get(args, 2),
197                expression=seq_get(args, 1),
198                unit=seq_get(args, 0),
199            ),
200            "DATE_DIFF": lambda args: exp.DateDiff(
201                this=seq_get(args, 2),
202                expression=seq_get(args, 1),
203                unit=seq_get(args, 0),
204            ),
205            "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"),
206            "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"),
207            "DATE_TRUNC": date_trunc_to_time,
208            "FROM_UNIXTIME": _from_unixtime,
209            "NOW": exp.CurrentTimestamp.from_arg_list,
210            "STRPOS": lambda args: exp.StrPosition(
211                this=seq_get(args, 0),
212                substr=seq_get(args, 1),
213                instance=seq_get(args, 2),
214            ),
215            "TO_UNIXTIME": exp.TimeToUnix.from_arg_list,
216            "APPROX_PERCENTILE": _approx_percentile,
217            "FROM_HEX": exp.Unhex.from_arg_list,
218            "TO_HEX": exp.Hex.from_arg_list,
219            "TO_UTF8": lambda args: exp.Encode(
220                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
221            ),
222            "FROM_UTF8": lambda args: exp.Decode(
223                this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8")
224            ),
225        }
226        FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy()
227        FUNCTION_PARSERS.pop("TRIM")
228
229    class Generator(generator.Generator):
230        STRUCT_DELIMITER = ("(", ")")
231
232        PROPERTIES_LOCATION = {
233            **generator.Generator.PROPERTIES_LOCATION,  # type: ignore
234            exp.LocationProperty: exp.Properties.Location.UNSUPPORTED,
235        }
236
237        TYPE_MAPPING = {
238            **generator.Generator.TYPE_MAPPING,  # type: ignore
239            exp.DataType.Type.INT: "INTEGER",
240            exp.DataType.Type.FLOAT: "REAL",
241            exp.DataType.Type.BINARY: "VARBINARY",
242            exp.DataType.Type.TEXT: "VARCHAR",
243            exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP",
244            exp.DataType.Type.STRUCT: "ROW",
245        }
246
247        TRANSFORMS = {
248            **generator.Generator.TRANSFORMS,  # type: ignore
249            **transforms.UNALIAS_GROUP,  # type: ignore
250            **transforms.ELIMINATE_QUALIFY,  # type: ignore
251            exp.ApproxDistinct: _approx_distinct_sql,
252            exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]",
253            exp.ArrayConcat: rename_func("CONCAT"),
254            exp.ArrayContains: rename_func("CONTAINS"),
255            exp.ArraySize: rename_func("CARDINALITY"),
256            exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
257            exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
258            exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})",
259            exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
260            exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
261            exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
262            exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP",
263            exp.DataType: _datatype_sql,
264            exp.DateAdd: lambda self, e: self.func(
265                "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
266            ),
267            exp.DateDiff: lambda self, e: self.func(
268                "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
269            ),
270            exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)",
271            exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)",
272            exp.Decode: _decode_sql,
273            exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)",
274            exp.Encode: _encode_sql,
275            exp.GenerateSeries: _sequence_sql,
276            exp.Hex: rename_func("TO_HEX"),
277            exp.If: if_sql,
278            exp.ILike: no_ilike_sql,
279            exp.Initcap: _initcap_sql,
280            exp.Lateral: _explode_to_unnest_sql,
281            exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"),
282            exp.LogicalOr: rename_func("BOOL_OR"),
283            exp.LogicalAnd: rename_func("BOOL_AND"),
284            exp.Quantile: _quantile_sql,
285            exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"),
286            exp.SafeDivide: no_safe_divide_sql,
287            exp.Schema: _schema_sql,
288            exp.SortArray: _no_sort_array,
289            exp.StrPosition: rename_func("STRPOS"),
290            exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)",
291            exp.StrToTime: _str_to_time_sql,
292            exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))",
293            exp.StructExtract: struct_extract_sql,
294            exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT='{e.name.upper()}'",
295            exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'",
296            exp.TimestampTrunc: timestamptrunc_sql,
297            exp.TimeStrToDate: timestrtotime_sql,
298            exp.TimeStrToTime: timestrtotime_sql,
299            exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))",
300            exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})",
301            exp.TimeToUnix: rename_func("TO_UNIXTIME"),
302            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)",
303            exp.TsOrDsAdd: _ts_or_ds_add_sql,
304            exp.TsOrDsToDate: _ts_or_ds_to_date_sql,
305            exp.Unhex: rename_func("FROM_HEX"),
306            exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})",
307            exp.UnixToTime: rename_func("FROM_UNIXTIME"),
308            exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)",
309            exp.VariancePop: rename_func("VAR_POP"),
310        }
311
312        def transaction_sql(self, expression):
313            modes = expression.args.get("modes")
314            modes = f" {', '.join(modes)}" if modes else ""
315            return f"START TRANSACTION{modes}"

Inherited Members

sqlglot.dialects.dialect.Dialect: get_or_raise; format_time; parse; parse_into; generate; transpile; tokenize; parser; generator

class Presto.Tokenizer(sqlglot.tokens.Tokenizer): View Source

182    class Tokenizer(tokens.Tokenizer):
183        KEYWORDS = {
184            **tokens.Tokenizer.KEYWORDS,
185            "START": TokenType.BEGIN,
186            "ROW": TokenType.STRUCT,
187        }

Inherited Members

sqlglot.tokens.Tokenizer: reset; tokenize

class Presto.Parser(sqlglot.parser.Parser): View Source

189    class Parser(parser.Parser):
190        FUNCTIONS = {
191            **parser.Parser.FUNCTIONS,  # type: ignore
192            "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list,
193            "CARDINALITY": exp.ArraySize.from_arg_list,
194            "CONTAINS": exp.ArrayContains.from_arg_list,
195            "DATE_ADD": lambda args: exp.DateAdd(
196                this=seq_get(args, 2),
197                expression=seq_get(args, 1),
198                unit=seq_get(args, 0),
199            ),
200            "DATE_DIFF": lambda args: exp.DateDiff(
201                this=seq_get(args, 2),
202                expression=seq_get(args, 1),
203                unit=seq_get(args, 0),
204            ),
205            "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"),
206            "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"),
207            "DATE_TRUNC": date_trunc_to_time,
208            "FROM_UNIXTIME": _from_unixtime,
209            "NOW": exp.CurrentTimestamp.from_arg_list,
210            "STRPOS": lambda args: exp.StrPosition(
211                this=seq_get(args, 0),
212                substr=seq_get(args, 1),
213                instance=seq_get(args, 2),
214            ),
215            "TO_UNIXTIME": exp.TimeToUnix.from_arg_list,
216            "APPROX_PERCENTILE": _approx_percentile,
217            "FROM_HEX": exp.Unhex.from_arg_list,
218            "TO_HEX": exp.Hex.from_arg_list,
219            "TO_UTF8": lambda args: exp.Encode(
220                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
221            ),
222            "FROM_UTF8": lambda args: exp.Decode(
223                this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8")
224            ),
225        }
226        FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy()
227        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

sqlglot.parser.Parser: Parser; reset; parse; parse_into; check_errors; raise_error; expression; validate_expression

class Presto.Generator(sqlglot.generator.Generator): View Source

229    class Generator(generator.Generator):
230        STRUCT_DELIMITER = ("(", ")")
231
232        PROPERTIES_LOCATION = {
233            **generator.Generator.PROPERTIES_LOCATION,  # type: ignore
234            exp.LocationProperty: exp.Properties.Location.UNSUPPORTED,
235        }
236
237        TYPE_MAPPING = {
238            **generator.Generator.TYPE_MAPPING,  # type: ignore
239            exp.DataType.Type.INT: "INTEGER",
240            exp.DataType.Type.FLOAT: "REAL",
241            exp.DataType.Type.BINARY: "VARBINARY",
242            exp.DataType.Type.TEXT: "VARCHAR",
243            exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP",
244            exp.DataType.Type.STRUCT: "ROW",
245        }
246
247        TRANSFORMS = {
248            **generator.Generator.TRANSFORMS,  # type: ignore
249            **transforms.UNALIAS_GROUP,  # type: ignore
250            **transforms.ELIMINATE_QUALIFY,  # type: ignore
251            exp.ApproxDistinct: _approx_distinct_sql,
252            exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]",
253            exp.ArrayConcat: rename_func("CONCAT"),
254            exp.ArrayContains: rename_func("CONTAINS"),
255            exp.ArraySize: rename_func("CARDINALITY"),
256            exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
257            exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
258            exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})",
259            exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
260            exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
261            exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
262            exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP",
263            exp.DataType: _datatype_sql,
264            exp.DateAdd: lambda self, e: self.func(
265                "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
266            ),
267            exp.DateDiff: lambda self, e: self.func(
268                "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
269            ),
270            exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.date_format}) AS DATE)",
271            exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.dateint_format}) AS INT)",
272            exp.Decode: _decode_sql,
273            exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.dateint_format}) AS DATE)",
274            exp.Encode: _encode_sql,
275            exp.GenerateSeries: _sequence_sql,
276            exp.Hex: rename_func("TO_HEX"),
277            exp.If: if_sql,
278            exp.ILike: no_ilike_sql,
279            exp.Initcap: _initcap_sql,
280            exp.Lateral: _explode_to_unnest_sql,
281            exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"),
282            exp.LogicalOr: rename_func("BOOL_OR"),
283            exp.LogicalAnd: rename_func("BOOL_AND"),
284            exp.Quantile: _quantile_sql,
285            exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"),
286            exp.SafeDivide: no_safe_divide_sql,
287            exp.Schema: _schema_sql,
288            exp.SortArray: _no_sort_array,
289            exp.StrPosition: rename_func("STRPOS"),
290            exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)",
291            exp.StrToTime: _str_to_time_sql,
292            exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))",
293            exp.StructExtract: struct_extract_sql,
294            exp.TableFormatProperty: lambda self, e: f"TABLE_FORMAT='{e.name.upper()}'",
295            exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'",
296            exp.TimestampTrunc: timestamptrunc_sql,
297            exp.TimeStrToDate: timestrtotime_sql,
298            exp.TimeStrToTime: timestrtotime_sql,
299            exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.time_format}))",
300            exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})",
301            exp.TimeToUnix: rename_func("TO_UNIXTIME"),
302            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)",
303            exp.TsOrDsAdd: _ts_or_ds_add_sql,
304            exp.TsOrDsToDate: _ts_or_ds_to_date_sql,
305            exp.Unhex: rename_func("FROM_HEX"),
306            exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})",
307            exp.UnixToTime: rename_func("FROM_UNIXTIME"),
308            exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)",
309            exp.VariancePop: rename_func("VAR_POP"),
310        }
311
312        def transaction_sql(self, expression):
313            modes = expression.args.get("modes")
314            modes = f" {', '.join(modes)}" if modes else ""
315            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

def transaction_sql(self, expression): View Source

312        def transaction_sql(self, expression):
313            modes = expression.args.get("modes")
314            modes = f" {', '.join(modes)}" if modes else ""
315            return f"START TRANSACTION{modes}"

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; datatype_sql; directory_sql; delete_sql; drop_sql; except_sql; except_op; fetch_sql; filter_sql; hint_sql; index_sql; identifier_sql; national_sql; partition_sql; properties_sql; root_properties; properties; with_properties; locate_properties; property_sql; likeproperty_sql; fallbackproperty_sql; journalproperty_sql; freespaceproperty_sql; afterjournalproperty_sql; checksumproperty_sql; mergeblockratioproperty_sql; datablocksizeproperty_sql; blockcompressionproperty_sql; isolatedloadingproperty_sql; lockingproperty_sql; withdataproperty_sql; insert_sql; intersect_sql; intersect_op; introducer_sql; pseudotype_sql; returning_sql; rowformatdelimitedproperty_sql; table_sql; tablesample_sql; pivot_sql; tuple_sql; update_sql; values_sql; var_sql; into_sql; from_sql; group_sql; having_sql; join_sql; lambda_sql; lateral_sql; limit_sql; offset_sql; 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; window_spec_sql; withingroup_sql; between_sql; bracket_sql; all_sql; any_sql; exists_sql; case_sql; constraint_sql; extract_sql; trim_sql; concat_sql; check_sql; foreignkey_sql; primarykey_sql; unique_sql; if_sql; 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; 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