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    binary_from_function,
  9    bool_xor_sql,
 10    date_trunc_to_time,
 11    encode_decode_sql,
 12    format_time_lambda,
 13    if_sql,
 14    left_to_substring_sql,
 15    no_ilike_sql,
 16    no_pivot_sql,
 17    no_safe_divide_sql,
 18    regexp_extract_sql,
 19    rename_func,
 20    right_to_substring_sql,
 21    struct_extract_sql,
 22    timestamptrunc_sql,
 23    timestrtotime_sql,
 24)
 25from sqlglot.dialects.mysql import MySQL
 26from sqlglot.helper import apply_index_offset, seq_get
 27from sqlglot.tokens import TokenType
 28
 29
 30def _approx_distinct_sql(self: Presto.Generator, expression: exp.ApproxDistinct) -> str:
 31    accuracy = expression.args.get("accuracy")
 32    accuracy = ", " + self.sql(accuracy) if accuracy else ""
 33    return f"APPROX_DISTINCT({self.sql(expression, 'this')}{accuracy})"
 34
 35
 36def _explode_to_unnest_sql(self: Presto.Generator, expression: exp.Lateral) -> str:
 37    if isinstance(expression.this, (exp.Explode, exp.Posexplode)):
 38        expression = expression.copy()
 39        return self.sql(
 40            exp.Join(
 41                this=exp.Unnest(
 42                    expressions=[expression.this.this],
 43                    alias=expression.args.get("alias"),
 44                    offset=isinstance(expression.this, exp.Posexplode),
 45                ),
 46                kind="cross",
 47            )
 48        )
 49    return self.lateral_sql(expression)
 50
 51
 52def _initcap_sql(self: Presto.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 _no_sort_array(self: Presto.Generator, expression: exp.SortArray) -> str:
 58    if expression.args.get("asc") == exp.false():
 59        comparator = "(a, b) -> CASE WHEN a < b THEN 1 WHEN a > b THEN -1 ELSE 0 END"
 60    else:
 61        comparator = None
 62    return self.func("ARRAY_SORT", expression.this, comparator)
 63
 64
 65def _schema_sql(self: Presto.Generator, expression: exp.Schema) -> str:
 66    if isinstance(expression.parent, exp.Property):
 67        columns = ", ".join(f"'{c.name}'" for c in expression.expressions)
 68        return f"ARRAY[{columns}]"
 69
 70    if expression.parent:
 71        for schema in expression.parent.find_all(exp.Schema):
 72            if isinstance(schema.parent, exp.Property):
 73                expression = expression.copy()
 74                expression.expressions.extend(schema.expressions)
 75
 76    return self.schema_sql(expression)
 77
 78
 79def _quantile_sql(self: Presto.Generator, expression: exp.Quantile) -> str:
 80    self.unsupported("Presto does not support exact quantiles")
 81    return f"APPROX_PERCENTILE({self.sql(expression, 'this')}, {self.sql(expression, 'quantile')})"
 82
 83
 84def _str_to_time_sql(
 85    self: Presto.Generator, expression: exp.StrToDate | exp.StrToTime | exp.TsOrDsToDate
 86) -> str:
 87    return f"DATE_PARSE({self.sql(expression, 'this')}, {self.format_time(expression)})"
 88
 89
 90def _ts_or_ds_to_date_sql(self: Presto.Generator, expression: exp.TsOrDsToDate) -> str:
 91    time_format = self.format_time(expression)
 92    if time_format and time_format not in (Presto.TIME_FORMAT, Presto.DATE_FORMAT):
 93        return exp.cast(_str_to_time_sql(self, expression), "DATE").sql(dialect="presto")
 94    return exp.cast(exp.cast(expression.this, "TIMESTAMP", copy=True), "DATE").sql(dialect="presto")
 95
 96
 97def _ts_or_ds_add_sql(self: Presto.Generator, expression: exp.TsOrDsAdd) -> str:
 98    this = expression.this
 99
100    if not isinstance(this, exp.CurrentDate):
101        this = exp.cast(exp.cast(expression.this, "TIMESTAMP", copy=True), "DATE")
102
103    return self.func(
104        "DATE_ADD",
105        exp.Literal.string(expression.text("unit") or "day"),
106        expression.expression,
107        this,
108    )
109
110
111def _approx_percentile(args: t.List) -> exp.Expression:
112    if len(args) == 4:
113        return exp.ApproxQuantile(
114            this=seq_get(args, 0),
115            weight=seq_get(args, 1),
116            quantile=seq_get(args, 2),
117            accuracy=seq_get(args, 3),
118        )
119    if len(args) == 3:
120        return exp.ApproxQuantile(
121            this=seq_get(args, 0), quantile=seq_get(args, 1), accuracy=seq_get(args, 2)
122        )
123    return exp.ApproxQuantile.from_arg_list(args)
124
125
126def _from_unixtime(args: t.List) -> exp.Expression:
127    if len(args) == 3:
128        return exp.UnixToTime(
129            this=seq_get(args, 0),
130            hours=seq_get(args, 1),
131            minutes=seq_get(args, 2),
132        )
133    if len(args) == 2:
134        return exp.UnixToTime(this=seq_get(args, 0), zone=seq_get(args, 1))
135
136    return exp.UnixToTime.from_arg_list(args)
137
138
139def _parse_element_at(args: t.List) -> exp.SafeBracket:
140    this = seq_get(args, 0)
141    index = seq_get(args, 1)
142    assert isinstance(this, exp.Expression) and isinstance(index, exp.Expression)
143    return exp.SafeBracket(this=this, expressions=apply_index_offset(this, [index], -1))
144
145
146def _unnest_sequence(expression: exp.Expression) -> exp.Expression:
147    if isinstance(expression, exp.Table):
148        if isinstance(expression.this, exp.GenerateSeries):
149            unnest = exp.Unnest(expressions=[expression.this])
150
151            if expression.alias:
152                return exp.alias_(unnest, alias="_u", table=[expression.alias], copy=False)
153            return unnest
154    return expression
155
156
157def _first_last_sql(self: Presto.Generator, expression: exp.First | exp.Last) -> str:
158    """
159    Trino doesn't support FIRST / LAST as functions, but they're valid in the context
160    of MATCH_RECOGNIZE, so we need to preserve them in that case. In all other cases
161    they're converted into an ARBITRARY call.
162
163    Reference: https://trino.io/docs/current/sql/match-recognize.html#logical-navigation-functions
164    """
165    if isinstance(expression.find_ancestor(exp.MatchRecognize, exp.Select), exp.MatchRecognize):
166        return self.function_fallback_sql(expression)
167
168    return rename_func("ARBITRARY")(self, expression)
169
170
171class Presto(Dialect):
172    INDEX_OFFSET = 1
173    NULL_ORDERING = "nulls_are_last"
174    TIME_FORMAT = MySQL.TIME_FORMAT
175    TIME_MAPPING = MySQL.TIME_MAPPING
176    STRICT_STRING_CONCAT = True
177    SUPPORTS_SEMI_ANTI_JOIN = False
178
179    # https://github.com/trinodb/trino/issues/17
180    # https://github.com/trinodb/trino/issues/12289
181    # https://github.com/prestodb/presto/issues/2863
182    RESOLVES_IDENTIFIERS_AS_UPPERCASE = None
183
184    class Tokenizer(tokens.Tokenizer):
185        KEYWORDS = {
186            **tokens.Tokenizer.KEYWORDS,
187            "START": TokenType.BEGIN,
188            "MATCH_RECOGNIZE": TokenType.MATCH_RECOGNIZE,
189            "ROW": TokenType.STRUCT,
190            "IPADDRESS": TokenType.IPADDRESS,
191            "IPPREFIX": TokenType.IPPREFIX,
192        }
193
194    class Parser(parser.Parser):
195        FUNCTIONS = {
196            **parser.Parser.FUNCTIONS,
197            "ARBITRARY": exp.AnyValue.from_arg_list,
198            "APPROX_DISTINCT": exp.ApproxDistinct.from_arg_list,
199            "APPROX_PERCENTILE": _approx_percentile,
200            "BITWISE_AND": binary_from_function(exp.BitwiseAnd),
201            "BITWISE_NOT": lambda args: exp.BitwiseNot(this=seq_get(args, 0)),
202            "BITWISE_OR": binary_from_function(exp.BitwiseOr),
203            "BITWISE_XOR": binary_from_function(exp.BitwiseXor),
204            "CARDINALITY": exp.ArraySize.from_arg_list,
205            "CONTAINS": exp.ArrayContains.from_arg_list,
206            "DATE_ADD": lambda args: exp.DateAdd(
207                this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0)
208            ),
209            "DATE_DIFF": lambda args: exp.DateDiff(
210                this=seq_get(args, 2), expression=seq_get(args, 1), unit=seq_get(args, 0)
211            ),
212            "DATE_FORMAT": format_time_lambda(exp.TimeToStr, "presto"),
213            "DATE_PARSE": format_time_lambda(exp.StrToTime, "presto"),
214            "DATE_TRUNC": date_trunc_to_time,
215            "ELEMENT_AT": _parse_element_at,
216            "FROM_HEX": exp.Unhex.from_arg_list,
217            "FROM_UNIXTIME": _from_unixtime,
218            "FROM_UTF8": lambda args: exp.Decode(
219                this=seq_get(args, 0), replace=seq_get(args, 1), charset=exp.Literal.string("utf-8")
220            ),
221            "NOW": exp.CurrentTimestamp.from_arg_list,
222            "REGEXP_EXTRACT": lambda args: exp.RegexpExtract(
223                this=seq_get(args, 0), expression=seq_get(args, 1), group=seq_get(args, 2)
224            ),
225            "REGEXP_REPLACE": lambda args: exp.RegexpReplace(
226                this=seq_get(args, 0),
227                expression=seq_get(args, 1),
228                replacement=seq_get(args, 2) or exp.Literal.string(""),
229            ),
230            "ROW": exp.Struct.from_arg_list,
231            "SEQUENCE": exp.GenerateSeries.from_arg_list,
232            "SPLIT_TO_MAP": exp.StrToMap.from_arg_list,
233            "STRPOS": lambda args: exp.StrPosition(
234                this=seq_get(args, 0), substr=seq_get(args, 1), instance=seq_get(args, 2)
235            ),
236            "TO_UNIXTIME": exp.TimeToUnix.from_arg_list,
237            "TO_HEX": exp.Hex.from_arg_list,
238            "TO_UTF8": lambda args: exp.Encode(
239                this=seq_get(args, 0), charset=exp.Literal.string("utf-8")
240            ),
241        }
242
243        FUNCTION_PARSERS = parser.Parser.FUNCTION_PARSERS.copy()
244        FUNCTION_PARSERS.pop("TRIM")
245
246    class Generator(generator.Generator):
247        INTERVAL_ALLOWS_PLURAL_FORM = False
248        JOIN_HINTS = False
249        TABLE_HINTS = False
250        QUERY_HINTS = False
251        IS_BOOL_ALLOWED = False
252        TZ_TO_WITH_TIME_ZONE = True
253        NVL2_SUPPORTED = False
254        STRUCT_DELIMITER = ("(", ")")
255
256        PROPERTIES_LOCATION = {
257            **generator.Generator.PROPERTIES_LOCATION,
258            exp.LocationProperty: exp.Properties.Location.UNSUPPORTED,
259            exp.VolatileProperty: exp.Properties.Location.UNSUPPORTED,
260        }
261
262        TYPE_MAPPING = {
263            **generator.Generator.TYPE_MAPPING,
264            exp.DataType.Type.INT: "INTEGER",
265            exp.DataType.Type.FLOAT: "REAL",
266            exp.DataType.Type.BINARY: "VARBINARY",
267            exp.DataType.Type.TEXT: "VARCHAR",
268            exp.DataType.Type.TIMETZ: "TIME",
269            exp.DataType.Type.TIMESTAMPTZ: "TIMESTAMP",
270            exp.DataType.Type.STRUCT: "ROW",
271            exp.DataType.Type.DATETIME: "TIMESTAMP",
272            exp.DataType.Type.DATETIME64: "TIMESTAMP",
273        }
274
275        TRANSFORMS = {
276            **generator.Generator.TRANSFORMS,
277            exp.AnyValue: rename_func("ARBITRARY"),
278            exp.ApproxDistinct: _approx_distinct_sql,
279            exp.ApproxQuantile: rename_func("APPROX_PERCENTILE"),
280            exp.Array: lambda self, e: f"ARRAY[{self.expressions(e, flat=True)}]",
281            exp.ArrayConcat: rename_func("CONCAT"),
282            exp.ArrayContains: rename_func("CONTAINS"),
283            exp.ArraySize: rename_func("CARDINALITY"),
284            exp.BitwiseAnd: lambda self, e: f"BITWISE_AND({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
285            exp.BitwiseLeftShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_LEFT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
286            exp.BitwiseNot: lambda self, e: f"BITWISE_NOT({self.sql(e, 'this')})",
287            exp.BitwiseOr: lambda self, e: f"BITWISE_OR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
288            exp.BitwiseRightShift: lambda self, e: f"BITWISE_ARITHMETIC_SHIFT_RIGHT({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
289            exp.BitwiseXor: lambda self, e: f"BITWISE_XOR({self.sql(e, 'this')}, {self.sql(e, 'expression')})",
290            exp.Cast: transforms.preprocess([transforms.epoch_cast_to_ts]),
291            exp.CurrentTimestamp: lambda *_: "CURRENT_TIMESTAMP",
292            exp.DateAdd: lambda self, e: self.func(
293                "DATE_ADD", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
294            ),
295            exp.DateDiff: lambda self, e: self.func(
296                "DATE_DIFF", exp.Literal.string(e.text("unit") or "day"), e.expression, e.this
297            ),
298            exp.DateStrToDate: lambda self, e: f"CAST(DATE_PARSE({self.sql(e, 'this')}, {Presto.DATE_FORMAT}) AS DATE)",
299            exp.DateToDi: lambda self, e: f"CAST(DATE_FORMAT({self.sql(e, 'this')}, {Presto.DATEINT_FORMAT}) AS INT)",
300            exp.DateSub: lambda self, e: self.func(
301                "DATE_ADD",
302                exp.Literal.string(e.text("unit") or "day"),
303                e.expression * -1,
304                e.this,
305            ),
306            exp.Decode: lambda self, e: encode_decode_sql(self, e, "FROM_UTF8"),
307            exp.DiToDate: lambda self, e: f"CAST(DATE_PARSE(CAST({self.sql(e, 'this')} AS VARCHAR), {Presto.DATEINT_FORMAT}) AS DATE)",
308            exp.Encode: lambda self, e: encode_decode_sql(self, e, "TO_UTF8"),
309            exp.FileFormatProperty: lambda self, e: f"FORMAT='{e.name.upper()}'",
310            exp.First: _first_last_sql,
311            exp.Group: transforms.preprocess([transforms.unalias_group]),
312            exp.GroupConcat: lambda self, e: self.func(
313                "ARRAY_JOIN", self.func("ARRAY_AGG", e.this), e.args.get("separator")
314            ),
315            exp.Hex: rename_func("TO_HEX"),
316            exp.If: if_sql(),
317            exp.ILike: no_ilike_sql,
318            exp.Initcap: _initcap_sql,
319            exp.ParseJSON: rename_func("JSON_PARSE"),
320            exp.Last: _first_last_sql,
321            exp.Lateral: _explode_to_unnest_sql,
322            exp.Left: left_to_substring_sql,
323            exp.Levenshtein: rename_func("LEVENSHTEIN_DISTANCE"),
324            exp.LogicalAnd: rename_func("BOOL_AND"),
325            exp.LogicalOr: rename_func("BOOL_OR"),
326            exp.Pivot: no_pivot_sql,
327            exp.Quantile: _quantile_sql,
328            exp.RegexpExtract: regexp_extract_sql,
329            exp.Right: right_to_substring_sql,
330            exp.SafeBracket: lambda self, e: self.func(
331                "ELEMENT_AT", e.this, seq_get(apply_index_offset(e.this, e.expressions, 1), 0)
332            ),
333            exp.SafeDivide: no_safe_divide_sql,
334            exp.Schema: _schema_sql,
335            exp.Select: transforms.preprocess(
336                [
337                    transforms.eliminate_qualify,
338                    transforms.eliminate_distinct_on,
339                    transforms.explode_to_unnest(1),
340                    transforms.eliminate_semi_and_anti_joins,
341                ]
342            ),
343            exp.SortArray: _no_sort_array,
344            exp.StrPosition: rename_func("STRPOS"),
345            exp.StrToDate: lambda self, e: f"CAST({_str_to_time_sql(self, e)} AS DATE)",
346            exp.StrToMap: rename_func("SPLIT_TO_MAP"),
347            exp.StrToTime: _str_to_time_sql,
348            exp.StrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {self.format_time(e)}))",
349            exp.StructExtract: struct_extract_sql,
350            exp.Table: transforms.preprocess([_unnest_sequence]),
351            exp.TimestampTrunc: timestamptrunc_sql,
352            exp.TimeStrToDate: timestrtotime_sql,
353            exp.TimeStrToTime: timestrtotime_sql,
354            exp.TimeStrToUnix: lambda self, e: f"TO_UNIXTIME(DATE_PARSE({self.sql(e, 'this')}, {Presto.TIME_FORMAT}))",
355            exp.TimeToStr: lambda self, e: f"DATE_FORMAT({self.sql(e, 'this')}, {self.format_time(e)})",
356            exp.TimeToUnix: rename_func("TO_UNIXTIME"),
357            exp.TryCast: transforms.preprocess([transforms.epoch_cast_to_ts]),
358            exp.TsOrDiToDi: lambda self, e: f"CAST(SUBSTR(REPLACE(CAST({self.sql(e, 'this')} AS VARCHAR), '-', ''), 1, 8) AS INT)",
359            exp.TsOrDsAdd: _ts_or_ds_add_sql,
360            exp.TsOrDsToDate: _ts_or_ds_to_date_sql,
361            exp.Unhex: rename_func("FROM_HEX"),
362            exp.UnixToStr: lambda self, e: f"DATE_FORMAT(FROM_UNIXTIME({self.sql(e, 'this')}), {self.format_time(e)})",
363            exp.UnixToTime: rename_func("FROM_UNIXTIME"),
364            exp.UnixToTimeStr: lambda self, e: f"CAST(FROM_UNIXTIME({self.sql(e, 'this')}) AS VARCHAR)",
365            exp.VariancePop: rename_func("VAR_POP"),
366            exp.With: transforms.preprocess([transforms.add_recursive_cte_column_names]),
367            exp.WithinGroup: transforms.preprocess(
368                [transforms.remove_within_group_for_percentiles]
369            ),
370            exp.Timestamp: transforms.preprocess([transforms.timestamp_to_cast]),
371            exp.Xor: bool_xor_sql,
372        }
373
374        def struct_sql(self, expression: exp.Struct) -> str:
375            if any(isinstance(arg, (exp.EQ, exp.Slice)) for arg in expression.expressions):
376                self.unsupported("Struct with key-value definitions is unsupported.")
377                return self.function_fallback_sql(expression)
378
379            return rename_func("ROW")(self, expression)
380
381        def interval_sql(self, expression: exp.Interval) -> str:
382            unit = self.sql(expression, "unit")
383            if expression.this and unit.lower().startswith("week"):
384                return f"({expression.this.name} * INTERVAL '7' day)"
385            return super().interval_sql(expression)
386
387        def transaction_sql(self, expression: exp.Transaction) -> str:
388            modes = expression.args.get("modes")
389            modes = f" {', '.join(modes)}" if modes else ""
390            return f"START TRANSACTION{modes}"
391
392        def generateseries_sql(self, expression: exp.GenerateSeries) -> str:
393            start = expression.args["start"]
394            end = expression.args["end"]
395            step = expression.args.get("step")
396
397            if isinstance(start, exp.Cast):
398                target_type = start.to
399            elif isinstance(end, exp.Cast):
400                target_type = end.to
401            else:
402                target_type = None
403
404            if target_type and target_type.is_type("timestamp"):
405                to = target_type.copy()
406
407                if target_type is start.to:
408                    end = exp.cast(end, to)
409                else:
410                    start = exp.cast(start, to)
411
412            return self.func("SEQUENCE", start, end, step)
413
414        def offset_limit_modifiers(
415            self, expression: exp.Expression, fetch: bool, limit: t.Optional[exp.Fetch | exp.Limit]
416        ) -> t.List[str]:
417            return [
418                self.sql(expression, "offset"),
419                self.sql(limit),
420            ]