Merging upstream version 10.0.8.

Signed-off-by: Daniel Baumann <daniel@debian.org>

sqlglot/expressions.py

@@ -641,9 +641,11 @@ class Set(Expression):
 
 class SetItem(Expression):
     arg_types = {
-        "this": True,
+        "this": False,
+        "expressions": False,
         "kind": False,
         "collate": False,  # MySQL SET NAMES statement
+        "global": False,
     }
 
 
@@ -787,6 +789,7 @@ class Drop(Expression):
         "exists": False,
         "temporary": False,
         "materialized": False,
+        "cascade": False,
     }
 
 
@@ -1073,6 +1076,18 @@ class FileFormatProperty(Property):
     pass
 
 
+class DistKeyProperty(Property):
+    pass
+
+
+class SortKeyProperty(Property):
+    pass
+
+
+class DistStyleProperty(Property):
+    pass
+
+
 class LocationProperty(Property):
     pass
 
@@ -1130,6 +1145,9 @@ class Properties(Expression):
         "LOCATION": LocationProperty,
         "PARTITIONED_BY": PartitionedByProperty,
         "TABLE_FORMAT": TableFormatProperty,
+        "DISTKEY": DistKeyProperty,
+        "DISTSTYLE": DistStyleProperty,
+        "SORTKEY": SortKeyProperty,
     }
 
     @classmethod
@@ -1356,7 +1374,7 @@ class Var(Expression):
 
 
 class Schema(Expression):
-    arg_types = {"this": False, "expressions": True}
+    arg_types = {"this": False, "expressions": False}
 
 
 class Select(Subqueryable):
@@ -1741,7 +1759,7 @@ class Select(Subqueryable):
             )
 
         if join_alias:
-            join.set("this", alias_(join.args["this"], join_alias, table=True))
+            join.set("this", alias_(join.this, join_alias, table=True))
         return _apply_list_builder(
             join,
             instance=self,
@@ -1884,6 +1902,7 @@ class Subquery(DerivedTable, Unionable):
     arg_types = {
         "this": True,
         "alias": False,
+        "with": False,
         **QUERY_MODIFIERS,
     }
 
@@ -2025,6 +2044,31 @@ class DataType(Expression):
         NULL = auto()
         UNKNOWN = auto()  # Sentinel value, useful for type annotation
 
+    TEXT_TYPES = {
+        Type.CHAR,
+        Type.NCHAR,
+        Type.VARCHAR,
+        Type.NVARCHAR,
+        Type.TEXT,
+    }
+
+    NUMERIC_TYPES = {
+        Type.INT,
+        Type.TINYINT,
+        Type.SMALLINT,
+        Type.BIGINT,
+        Type.FLOAT,
+        Type.DOUBLE,
+    }
+
+    TEMPORAL_TYPES = {
+        Type.TIMESTAMP,
+        Type.TIMESTAMPTZ,
+        Type.TIMESTAMPLTZ,
+        Type.DATE,
+        Type.DATETIME,
+    }
+
     @classmethod
     def build(cls, dtype, **kwargs) -> DataType:
         return DataType(
@@ -2054,16 +2098,25 @@ class Exists(SubqueryPredicate):
     pass
 
 
-# Commands to interact with the databases or engines
-# These expressions don't truly parse the expression and consume
-# whatever exists as a string until the end or a semicolon
+# Commands to interact with the databases or engines. For most of the command
+# expressions we parse whatever comes after the command's name as a string.
 class Command(Expression):
     arg_types = {"this": True, "expression": False}
 
 
-# Binary Expressions
-# (ADD a b)
-# (FROM table selects)
+class Transaction(Command):
+    arg_types = {"this": False, "modes": False}
+
+
+class Commit(Command):
+    arg_types = {}  # type: ignore
+
+
+class Rollback(Command):
+    arg_types = {"savepoint": False}
+
+
+# Binary expressions like (ADD a b)
 class Binary(Expression):
     arg_types = {"this": True, "expression": True}
 
@@ -2215,7 +2268,7 @@ class Not(Unary, Condition):
 
 
 class Paren(Unary, Condition):
-    pass
+    arg_types = {"this": True, "with": False}
 
 
 class Neg(Unary):
@@ -2428,6 +2481,10 @@ class Cast(Func):
         return self.args["to"]
 
 
+class Collate(Binary):
+    pass
+
+
 class TryCast(Cast):
     pass
 
@@ -2442,13 +2499,17 @@ class Coalesce(Func):
     is_var_len_args = True
 
 
-class ConcatWs(Func):
-    arg_types = {"expressions": False}
+class Concat(Func):
+    arg_types = {"expressions": True}
     is_var_len_args = True
 
 
+class ConcatWs(Concat):
+    _sql_names = ["CONCAT_WS"]
+
+
 class Count(AggFunc):
-    pass
+    arg_types = {"this": False}
 
 
 class CurrentDate(Func):
@@ -2556,10 +2617,18 @@ class Day(Func):
     pass
 
 
+class Decode(Func):
+    arg_types = {"this": True, "charset": True}
+
+
 class DiToDate(Func):
     pass
 
 
+class Encode(Func):
+    arg_types = {"this": True, "charset": True}
+
+
 class Exp(Func):
     pass
 
@@ -2581,6 +2650,10 @@ class GroupConcat(Func):
     arg_types = {"this": True, "separator": False}
 
 
+class Hex(Func):
+    pass
+
+
 class If(Func):
     arg_types = {"this": True, "true": True, "false": False}
 
@@ -2641,7 +2714,7 @@ class Log10(Func):
 
 
 class Lower(Func):
-    pass
+    _sql_names = ["LOWER", "LCASE"]
 
 
 class Map(Func):
@@ -2686,6 +2759,12 @@ class ApproxQuantile(Quantile):
     arg_types = {"this": True, "quantile": True, "accuracy": False}
 
 
+class ReadCSV(Func):
+    _sql_names = ["READ_CSV"]
+    is_var_len_args = True
+    arg_types = {"this": True, "expressions": False}
+
+
 class Reduce(Func):
     arg_types = {"this": True, "initial": True, "merge": True, "finish": True}
 
@@ -2804,8 +2883,8 @@ class TimeStrToUnix(Func):
 class Trim(Func):
     arg_types = {
         "this": True,
-        "position": False,
         "expression": False,
+        "position": False,
         "collation": False,
     }
 
@@ -2826,6 +2905,10 @@ class TsOrDiToDi(Func):
     pass
 
 
+class Unhex(Func):
+    pass
+
+
 class UnixToStr(Func):
     arg_types = {"this": True, "format": False}
 
@@ -2843,7 +2926,7 @@ class UnixToTimeStr(Func):
 
 
 class Upper(Func):
-    pass
+    _sql_names = ["UPPER", "UCASE"]
 
 
 class Variance(AggFunc):
@@ -3701,6 +3784,19 @@ def replace_placeholders(expression, *args, **kwargs):
     return expression.transform(_replace_placeholders, iter(args), **kwargs)
 
 
+def true():
+    return Boolean(this=True)
+
+
+def false():
+    return Boolean(this=False)
+
+
+def null():
+    return Null()
+
+
+# TODO: deprecate this
 TRUE = Boolean(this=True)
 FALSE = Boolean(this=False)
 NULL = Null()