sqlglot 28.4.0__py3-none-any.whl → 28.8.0__py3-none-any.whl

sqlglot/expressions.py

@@ -280,6 +280,8 @@ class Expression(metaclass=_Expression):
 
     @property
     def type(self) -> t.Optional[DataType]:
+        if isinstance(self, Cast):
+            return self._type or self.to
         return self._type
 
     @type.setter
@@ -1616,6 +1618,7 @@ class Describe(Expression):
         "expressions": False,
         "partition": False,
         "format": False,
+        "as_json": False,
     }
 
 
@@ -2134,6 +2137,11 @@ class ComputedColumnConstraint(ColumnConstraintKind):
     arg_types = {"this": True, "persisted": False, "not_null": False, "data_type": False}
 
 
+# https://docs.oracle.com/en/database/other-databases/timesten/22.1/plsql-developer/examples-using-input-and-output-parameters-and-bind-variables.html#GUID-4B20426E-F93F-4835-88CB-6A79829A8D7F
+class InOutColumnConstraint(ColumnConstraintKind):
+    arg_types = {"input_": False, "output": False, "variadic": False}
+
+
 class Constraint(Expression):
     arg_types = {"this": True, "expressions": True}
 
@@ -2481,6 +2489,7 @@ class OnConflict(Expression):
         "expressions": False,
         "action": False,
         "conflict_keys": False,
+        "index_predicate": False,
         "constraint": False,
         "where": False,
     }
@@ -2600,8 +2609,19 @@ class Literal(Condition):
     arg_types = {"this": True, "is_string": True}
 
     @classmethod
-    def number(cls, number) -> Literal:
-        return cls(this=str(number), is_string=False)
+    def number(cls, number) -> Literal | Neg:
+        expr: Literal | Neg = cls(this=str(number), is_string=False)
+
+        try:
+            to_py = expr.to_py()
+
+            if not isinstance(to_py, str) and to_py < 0:
+                expr.set("this", str(abs(to_py)))
+                expr = Neg(this=expr)
+        except Exception:
+            pass
+
+        return expr
 
     @classmethod
     def string(cls, string) -> Literal:
@@ -2631,6 +2651,7 @@ class Join(Expression):
         "global_": False,
         "hint": False,
         "match_condition": False,  # Snowflake
+        "directed": False,  # Snowflake
         "expressions": False,
         "pivots": False,
     }
@@ -3115,6 +3136,16 @@ class PartitionByRangePropertyDynamic(Expression):
     arg_types = {"this": False, "start": True, "end": True, "every": True}
 
 
+# https://docs.starrocks.io/docs/sql-reference/sql-statements/table_bucket_part_index/CREATE_TABLE/#rollup-index
+class RollupProperty(Property):
+    arg_types = {"expressions": True}
+
+
+# https://docs.starrocks.io/docs/sql-reference/sql-statements/table_bucket_part_index/CREATE_TABLE/#rollup-index
+class RollupIndex(Expression):
+    arg_types = {"this": True, "expressions": True, "from_index": False, "properties": False}
+
+
 # https://doris.apache.org/docs/table-design/data-partitioning/manual-partitioning
 class PartitionByListProperty(Property):
     arg_types = {"partition_expressions": True, "create_expressions": True}
@@ -3128,7 +3159,7 @@ class PartitionList(Expression):
 # https://doris.apache.org/docs/sql-manual/sql-statements/table-and-view/async-materialized-view/CREATE-ASYNC-MATERIALIZED-VIEW
 class RefreshTriggerProperty(Property):
     arg_types = {
-        "method": True,
+        "method": False,
         "kind": False,
         "every": False,
         "unit": False,
@@ -5143,7 +5174,7 @@ class BitwiseAnd(Binary):
 
 
 class BitwiseLeftShift(Binary):
-    pass
+    arg_types = {"this": True, "expression": True, "requires_int128": False}
 
 
 class BitwiseOr(Binary):
@@ -5151,7 +5182,7 @@ class BitwiseOr(Binary):
 
 
 class BitwiseRightShift(Binary):
-    pass
+    arg_types = {"this": True, "expression": True, "requires_int128": False}
 
 
 class BitwiseXor(Binary):
@@ -5313,6 +5344,12 @@ class Sub(Binary):
     pass
 
 
+# https://www.postgresql.org/docs/current/functions-range.html
+# Represents range adjacency operator: -|-
+class Adjacent(Binary):
+    pass
+
+
 # Unary Expressions
 # (NOT a)
 class Unary(Condition):
@@ -5555,6 +5592,12 @@ class Func(Condition):
         return {name: cls.from_arg_list for name in cls.sql_names()}
 
 
+# Function returns NULL instead of error
+# https://docs.cloud.google.com/bigquery/docs/reference/standard-sql/functions-reference#safe_prefix
+class SafeFunc(Func):
+    pass
+
+
 class Typeof(Func):
     pass
 
@@ -5708,15 +5751,15 @@ class ByteLength(Func):
 
 
 class Boolnot(Func):
-    pass
+    arg_types = {"this": True, "round_input": False}
 
 
 class Booland(Func):
-    arg_types = {"this": True, "expression": True}
+    arg_types = {"this": True, "expression": True, "round_input": False}
 
 
 class Boolor(Func):
-    arg_types = {"this": True, "expression": True}
+    arg_types = {"this": True, "expression": True, "round_input": False}
 
 
 # https://cloud.google.com/bigquery/docs/reference/standard-sql/json_functions#bool_for_json
@@ -5725,7 +5768,7 @@ class JSONBool(Func):
 
 
 class ArrayRemove(Func):
-    arg_types = {"this": True, "expression": True}
+    arg_types = {"this": True, "expression": True, "null_propagation": False}
 
 
 class ParameterizedAgg(AggFunc):
@@ -5823,7 +5866,7 @@ class Grouping(AggFunc):
 
 
 class GroupingId(AggFunc):
-    arg_types = {"expressions": True}
+    arg_types = {"expressions": False}
     is_var_len_args = True
 
 
@@ -6010,6 +6053,11 @@ class ExplodingGenerateSeries(GenerateSeries):
     pass
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/generator
+class Generator(Func, UDTF):
+    arg_types = {"rowcount": False, "timelimit": False}
+
+
 class ArrayAgg(AggFunc):
     arg_types = {"this": True, "nulls_excluded": False}
 
@@ -6041,9 +6089,17 @@ class ArrayAny(Func):
     arg_types = {"this": True, "expression": True}
 
 
+class ArrayAppend(Func):
+    arg_types = {"this": True, "expression": True, "null_propagation": False}
+
+
+class ArrayPrepend(Func):
+    arg_types = {"this": True, "expression": True, "null_propagation": False}
+
+
 class ArrayConcat(Func):
     _sql_names = ["ARRAY_CONCAT", "ARRAY_CAT"]
-    arg_types = {"this": True, "expressions": False}
+    arg_types = {"this": True, "expressions": False, "null_propagation": False}
     is_var_len_args = True
 
 
@@ -6051,6 +6107,14 @@ class ArrayConcatAgg(AggFunc):
     pass
 
 
+class ArrayCompact(Func):
+    pass
+
+
+class ArrayInsert(Func):
+    arg_types = {"this": True, "position": True, "expression": True, "offset": False}
+
+
 class ArrayConstructCompact(Func):
     arg_types = {"expressions": False}
     is_var_len_args = True
@@ -6137,6 +6201,11 @@ class ArrayUnionAgg(AggFunc):
     pass
 
 
+class ArraysZip(Func):
+    arg_types = {"expressions": False}
+    is_var_len_args = True
+
+
 class Avg(AggFunc):
     pass
 
@@ -6410,6 +6479,10 @@ class Localtimestamp(Func):
     arg_types = {"this": False}
 
 
+class Systimestamp(Func):
+    arg_types = {"this": False}
+
+
 class CurrentTimestamp(Func):
     arg_types = {"this": False, "sysdate": False}
 
@@ -6422,10 +6495,6 @@ class CurrentTimezone(Func):
     arg_types = {}
 
 
-class Sysdate(Func):
-    arg_types = {}
-
-
 class CurrentOrganizationName(Func):
     arg_types = {}
 
@@ -6499,7 +6568,7 @@ class DateDiff(Func, TimeUnit):
 
 
 class DateTrunc(Func):
-    arg_types = {"unit": True, "this": True, "zone": False}
+    arg_types = {"unit": True, "this": True, "zone": False, "input_type_preserved": False}
 
     def __init__(self, **args):
         # Across most dialects it's safe to unabbreviate the unit (e.g. 'Q' -> 'QUARTER') except Oracle
@@ -6567,6 +6636,10 @@ class DayOfYear(Func):
     _sql_names = ["DAY_OF_YEAR", "DAYOFYEAR"]
 
 
+class Dayname(Func):
+    arg_types = {"this": True, "abbreviated": False}
+
+
 class ToDays(Func):
     pass
 
@@ -6591,6 +6664,7 @@ class MakeInterval(Func):
     arg_types = {
         "year": False,
         "month": False,
+        "week": False,
         "day": False,
         "hour": False,
         "minute": False,
@@ -6637,7 +6711,7 @@ class Elt(Func):
 
 
 class Timestamp(Func):
-    arg_types = {"this": False, "zone": False, "with_tz": False, "safe": False}
+    arg_types = {"this": False, "zone": False, "with_tz": False}
 
 
 class TimestampAdd(Func, TimeUnit):
@@ -6654,7 +6728,7 @@ class TimestampDiff(Func, TimeUnit):
 
 
 class TimestampTrunc(Func, TimeUnit):
-    arg_types = {"this": True, "unit": True, "zone": False}
+    arg_types = {"this": True, "unit": True, "zone": False, "input_type_preserved": False}
 
 
 class TimeSlice(Func, TimeUnit):
@@ -6679,7 +6753,7 @@ class TimeTrunc(Func, TimeUnit):
 
 class DateFromParts(Func):
     _sql_names = ["DATE_FROM_PARTS", "DATEFROMPARTS"]
-    arg_types = {"year": True, "month": False, "day": False}
+    arg_types = {"year": True, "month": False, "day": False, "allow_overflow": False}
 
 
 class TimeFromParts(Func):
@@ -6691,6 +6765,7 @@ class TimeFromParts(Func):
         "nano": False,
         "fractions": False,
         "precision": False,
+        "overflow": False,
     }
 
 
@@ -6725,6 +6800,30 @@ class DecodeCase(Func):
     is_var_len_args = True
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/decrypt
+class Decrypt(Func):
+    arg_types = {
+        "this": True,
+        "passphrase": True,
+        "aad": False,
+        "encryption_method": False,
+        "safe": False,
+    }
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/decrypt_raw
+class DecryptRaw(Func):
+    arg_types = {
+        "this": True,
+        "key": True,
+        "iv": True,
+        "aad": False,
+        "encryption_method": False,
+        "aead": False,
+        "safe": False,
+    }
+
+
 class DenseRank(AggFunc):
     arg_types = {"expressions": False}
     is_var_len_args = True
@@ -6738,6 +6837,16 @@ class Encode(Func):
     arg_types = {"this": True, "charset": True}
 
 
+# https://docs.snowflake.com/en/sql-reference/functions/encrypt
+class Encrypt(Func):
+    arg_types = {"this": True, "passphrase": True, "aad": False, "encryption_method": False}
+
+
+# https://docs.snowflake.com/en/sql-reference/functions/encrypt_raw
+class EncryptRaw(Func):
+    arg_types = {"this": True, "key": True, "iv": True, "aad": False, "encryption_method": False}
+
+
 class EqualNull(Func):
     arg_types = {"this": True, "expression": True}
 
@@ -6886,21 +6995,13 @@ class GetExtract(Func):
 
 
 class Getbit(Func):
-    arg_types = {"this": True, "expression": True}
+    _sql_names = ["GETBIT", "GET_BIT"]
+    # zero_is_msb means the most significant bit is indexed 0
+    arg_types = {"this": True, "expression": True, "zero_is_msb": False}
 
 
 class Greatest(Func):
-    arg_types = {"this": True, "expressions": False, "null_if_any_null": False}
-    is_var_len_args = True
-
-
-class GreatestIgnoreNulls(Func):
-    arg_types = {"expressions": True}
-    is_var_len_args = True
-
-
-class LeastIgnoreNulls(Func):
-    arg_types = {"expressions": True}
+    arg_types = {"this": True, "expressions": False, "ignore_nulls": True}
     is_var_len_args = True
 
 
@@ -6969,7 +7070,8 @@ class Or(Connector, Func):
 
 
 class Xor(Connector, Func):
-    arg_types = {"this": False, "expression": False, "expressions": False}
+    arg_types = {"this": False, "expression": False, "expressions": False, "round_input": False}
+    is_var_len_args = True
 
 
 class If(Func):
@@ -7006,6 +7108,10 @@ class IsNullValue(Func):
     pass
 
 
+class IsArray(Func):
+    pass
+
+
 # https://www.postgresql.org/docs/current/functions-json.html
 class JSON(Expression):
     arg_types = {"this": False, "with_": False, "unique": False}
@@ -7073,6 +7179,12 @@ class Format(Func):
     is_var_len_args = True
 
 
+class JSONKeys(Func):
+    arg_types = {"this": True, "expression": False, "expressions": False}
+    is_var_len_args = True
+    _sql_names = ["JSON_KEYS"]
+
+
 class JSONKeyValue(Expression):
     arg_types = {"this": True, "expression": True}
 
@@ -7118,7 +7230,7 @@ class JSONArray(Func):
 
 
 # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/JSON_ARRAYAGG.html
-class JSONArrayAgg(Func):
+class JSONArrayAgg(AggFunc):
     arg_types = {
         "this": True,
         "order": False,
@@ -7368,7 +7480,7 @@ class ParseDatetime(Func):
 
 
 class Least(Func):
-    arg_types = {"this": True, "expressions": False, "null_if_any_null": False}
+    arg_types = {"this": True, "expressions": False, "ignore_nulls": True}
     is_var_len_args = True
 
 
@@ -7451,6 +7563,36 @@ class MapFromEntries(Func):
     pass
 
 
+class MapCat(Func):
+    arg_types = {"this": True, "expression": True}
+
+
+class MapContainsKey(Func):
+    arg_types = {"this": True, "key": True}
+
+
+class MapDelete(Func):
+    arg_types = {"this": True, "expressions": True}
+    is_var_len_args = True
+
+
+class MapInsert(Func):
+    arg_types = {"this": True, "key": False, "value": True, "update_flag": False}
+
+
+class MapKeys(Func):
+    pass
+
+
+class MapPick(Func):
+    arg_types = {"this": True, "expressions": True}
+    is_var_len_args = True
+
+
+class MapSize(Func):
+    pass
+
+
 # https://learn.microsoft.com/en-us/sql/t-sql/language-elements/scope-resolution-operator-transact-sql?view=sql-server-ver16
 class ScopeResolution(Expression):
     arg_types = {"this": False, "expression": True}
@@ -7496,7 +7638,11 @@ class MD5(Func):
 
 
 # Represents the variant of the MD5 function that returns a binary value
+# Var len args due to Exasol:
+# https://docs.exasol.com/db/latest/sql_references/functions/alphabeticallistfunctions/hashtype_md5.htm
 class MD5Digest(Func):
+    arg_types = {"this": True, "expressions": False}
+    is_var_len_args = True
     _sql_names = ["MD5_DIGEST"]
 
 
@@ -7528,11 +7674,11 @@ class Month(Func):
 
 
 class Monthname(Func):
-    pass
+    arg_types = {"this": True, "abbreviated": False}
 
 
 class AddMonths(Func):
-    arg_types = {"this": True, "expression": True}
+    arg_types = {"this": True, "expression": True, "preserve_end_of_month": False}
 
 
 class Nvl2(Func):
@@ -7551,9 +7697,19 @@ class Normal(Func):
     arg_types = {"this": True, "stddev": True, "gen": True}
 
 
+# https://docs.cloud.google.com/bigquery/docs/reference/standard-sql/net_functions
+class NetFunc(Func):
+    pass
+
+
 # https://cloud.google.com/bigquery/docs/reference/standard-sql/net_functions#nethost
-class NetHost(Func):
-    _sql_names = ["NET.HOST"]
+class Host(Func):
+    pass
+
+
+# https://docs.cloud.google.com/bigquery/docs/reference/standard-sql/net_functions#netreg_domain
+class RegDomain(Func):
+    pass
 
 
 class Overlay(Func):
@@ -7841,6 +7997,22 @@ class RowNumber(Func):
     arg_types = {"this": False}
 
 
+class Seq1(Func):
+    arg_types = {"this": False}
+
+
+class Seq2(Func):
+    arg_types = {"this": False}
+
+
+class Seq4(Func):
+    arg_types = {"this": False}
+
+
+class Seq8(Func):
+    arg_types = {"this": False}
+
+
 class SafeAdd(Func):
     arg_types = {"this": True, "expression": True}
 
@@ -7976,11 +8148,11 @@ class StrToDate(Func):
 
 
 class StrToTime(Func):
-    arg_types = {"this": True, "format": True, "zone": False, "safe": False}
+    arg_types = {"this": True, "format": True, "zone": False, "safe": False, "target_type": False}
 
 
 # Spark allows unix_timestamp()
-# https://spark.apache.org/docs/3.1.3/api/python/reference/api/pyspark.sql.functions.unix_timestamp.html
+# https://spark.apache.org/docs/latest/api/python/reference/pyspark.sql/api/pyspark.sql.functions.unix_timestamp.html
 class StrToUnix(Func):
     arg_types = {"this": False, "format": False}
 
@@ -8154,6 +8326,7 @@ class UnixToTime(Func):
         "hours": False,
         "minutes": False,
         "format": False,
+        "target_type": False,
     }
 
     SECONDS = Literal.number(0)
@@ -8230,7 +8403,10 @@ class Upper(Func):
 
 
 class Corr(Binary, AggFunc):
-    pass
+    # Correlation divides by variance(column). If a column has 0 variance, the denominator
+    # is 0 - some dialects return NaN (DuckDB) while others return NULL (Snowflake).
+    # `null_on_zero_variance` is set to True at parse time for dialects that return NULL.
+    arg_types = {"this": True, "expression": True, "null_on_zero_variance": False}
 
 
 # https://docs.oracle.com/en/database/oracle/oracle-database/19/sqlrf/CUME_DIST.html
@@ -8247,20 +8423,30 @@ class VariancePop(AggFunc):
     _sql_names = ["VARIANCE_POP", "VAR_POP"]
 
 
+class Kurtosis(AggFunc):
+    pass
+
+
 class Skewness(AggFunc):
     pass
 
 
 class WidthBucket(Func):
-    arg_types = {"this": True, "min_value": True, "max_value": True, "num_buckets": True}
+    arg_types = {
+        "this": True,
+        "min_value": False,
+        "max_value": False,
+        "num_buckets": False,
+        "threshold": False,
+    }
 
 
-class CovarSamp(Binary, AggFunc):
-    pass
+class CovarSamp(AggFunc):
+    arg_types = {"this": True, "expression": True}
 
 
-class CovarPop(Binary, AggFunc):
-    pass
+class CovarPop(AggFunc):
+    arg_types = {"this": True, "expression": True}
 
 
 class Week(Func):
@@ -8277,7 +8463,7 @@ class NextDay(Func):
 
 class XMLElement(Func):
     _sql_names = ["XMLELEMENT"]
-    arg_types = {"this": True, "expressions": False}
+    arg_types = {"this": True, "expressions": False, "evalname": False}
 
 
 class XMLGet(Func):
@@ -8356,6 +8542,12 @@ class TableColumn(Expression):
     pass
 
 
+# https://www.postgresql.org/docs/current/typeconv-func.html
+# https://www.postgresql.org/docs/current/xfunc-sql.html
+class Variadic(Expression):
+    pass
+
+
 ALL_FUNCTIONS = subclasses(__name__, Func, {AggFunc, Anonymous, Func})
 FUNCTION_BY_NAME = {name: func for func in ALL_FUNCTIONS for name in func.sql_names()}