dbcli/sqlglot
1
0
Fork 0
Code Issues Activity

Adding upstream version 25.16.1.

Browse source 
Signed-off-by: Daniel Baumann <daniel@debian.org>
This commit is contained in:
Daniel Baumann 2025-02-13 21:52:21 +01:00
parent a500eebbbb
commit 1341bc6477
Signed by: daniel
GPG key ID: FBB4F0E80A80222F
110 changed files with 75353 additions and 68092 deletions
Download patch file Download diff file Expand all files Collapse all files

77
sqlglot/dialects/snowflake.py
View file

@ -122,12 +122,6 @@ def _regexpilike_sql(self: Snowflake.Generator, expression: exp.RegexpILike) ->
)
def _build_convert_timezone(args: t.List) -> t.Union[exp.Anonymous, exp.AtTimeZone]:
if len(args) == 3:
return exp.Anonymous(this="CONVERT_TIMEZONE", expressions=args)
return exp.AtTimeZone(this=seq_get(args, 1), zone=seq_get(args, 0))
def _build_regexp_replace(args: t.List) -> exp.RegexpReplace:
regexp_replace = exp.RegexpReplace.from_arg_list(args)
@ -186,6 +180,47 @@ def _flatten_structured_types_unless_iceberg(expression: exp.Expression) -> exp.
return expression
def _unnest_generate_date_array(expression: exp.Expression) -> exp.Expression:
if isinstance(expression, exp.Select):
for unnest in expression.find_all(exp.Unnest):
if (
isinstance(unnest.parent, (exp.From, exp.Join))
and len(unnest.expressions) == 1
and isinstance(unnest.expressions[0], exp.GenerateDateArray)
):
generate_date_array = unnest.expressions[0]
start = generate_date_array.args.get("start")
end = generate_date_array.args.get("end")
step = generate_date_array.args.get("step")
if not start or not end or not isinstance(step, exp.Interval) or step.name != "1":
continue
unit = step.args.get("unit")
unnest_alias = unnest.args.get("alias")
if unnest_alias:
unnest_alias = unnest_alias.copy()
sequence_value_name = seq_get(unnest_alias.columns, 0) or "value"
else:
sequence_value_name = "value"
# We'll add the next sequence value to the starting date and project the result
date_add = _build_date_time_add(exp.DateAdd)(
[unit, exp.cast(sequence_value_name, "int"), exp.cast(start, "date")]
).as_(sequence_value_name)
# We use DATEDIFF to compute the number of sequence values needed
number_sequence = Snowflake.Parser.FUNCTIONS["ARRAY_GENERATE_RANGE"](
[exp.Literal.number(0), _build_datediff([unit, start, end]) + 1]
)
unnest.set("expressions", [number_sequence])
unnest.replace(exp.select(date_add).from_(unnest.copy()).subquery(unnest_alias))
return expression
class Snowflake(Dialect):
# https://docs.snowflake.com/en/sql-reference/identifiers-syntax
NORMALIZATION_STRATEGY = NormalizationStrategy.UPPERCASE
@ -255,7 +290,7 @@ class Snowflake(Dialect):
**parser.Parser.FUNCTIONS,
"APPROX_PERCENTILE": exp.ApproxQuantile.from_arg_list,
"ARRAYAGG": exp.ArrayAgg.from_arg_list,
"ARRAY_CONSTRUCT": exp.Array.from_arg_list,
"ARRAY_CONSTRUCT": lambda args: exp.Array(expressions=args),
"ARRAY_CONTAINS": lambda args: exp.ArrayContains(
this=seq_get(args, 1), expression=seq_get(args, 0)
),
@ -268,7 +303,6 @@ class Snowflake(Dialect):
"BITXOR": binary_from_function(exp.BitwiseXor),
"BIT_XOR": binary_from_function(exp.BitwiseXor),
"BOOLXOR": binary_from_function(exp.Xor),
"CONVERT_TIMEZONE": _build_convert_timezone,
"DATE": _build_datetime("DATE", exp.DataType.Type.DATE),
"DATE_TRUNC": _date_trunc_to_time,
"DATEADD": _build_date_time_add(exp.DateAdd),
@ -413,6 +447,26 @@ class Snowflake(Dialect):
),
}
def _negate_range(
self, this: t.Optional[exp.Expression] = None
) -> t.Optional[exp.Expression]:
if not this:
return this
query = this.args.get("query")
if isinstance(this, exp.In) and isinstance(query, exp.Query):
# Snowflake treats `value NOT IN (subquery)` as `VALUE <> ALL (subquery)`, so
# we do this conversion here to avoid parsing it into `NOT value IN (subquery)`
# which can produce different results (most likely a SnowFlake bug).
#
# https://docs.snowflake.com/en/sql-reference/functions/in
# Context: https://github.com/tobymao/sqlglot/issues/3890
return self.expression(
exp.NEQ, this=this.this, expression=exp.All(this=query.unnest())
)
return self.expression(exp.Not, this=this)
def _parse_with_constraint(self) -> t.Optional[exp.Expression]:
if self._prev.token_type != TokenType.WITH:
self._retreat(self._index - 1)
@ -638,6 +692,7 @@ class Snowflake(Dialect):
HEX_STRINGS = [("x'", "'"), ("X'", "'")]
RAW_STRINGS = ["$$"]
COMMENTS = ["--", "//", ("/*", "*/")]
NESTED_COMMENTS = False
KEYWORDS = {
**tokens.Tokenizer.KEYWORDS,
@ -692,6 +747,9 @@ class Snowflake(Dialect):
COPY_PARAMS_ARE_WRAPPED = False
COPY_PARAMS_EQ_REQUIRED = True
STAR_EXCEPT = "EXCLUDE"
SUPPORTS_EXPLODING_PROJECTIONS = False
ARRAY_CONCAT_IS_VAR_LEN = False
SUPPORTS_CONVERT_TIMEZONE = True
TRANSFORMS = {
**generator.Generator.TRANSFORMS,
@ -699,7 +757,7 @@ class Snowflake(Dialect):
exp.ArgMax: rename_func("MAX_BY"),
exp.ArgMin: rename_func("MIN_BY"),
exp.Array: inline_array_sql,
exp.ArrayConcat: rename_func("ARRAY_CAT"),
exp.ArrayConcat: lambda self, e: self.arrayconcat_sql(e, name="ARRAY_CAT"),
exp.ArrayContains: lambda self, e: self.func("ARRAY_CONTAINS", e.expression, e.this),
exp.AtTimeZone: lambda self, e: self.func(
"CONVERT_TIMEZONE", e.args.get("zone"), e.this
@ -751,6 +809,7 @@ class Snowflake(Dialect):
transforms.eliminate_distinct_on,
transforms.explode_to_unnest(),
transforms.eliminate_semi_and_anti_joins,
_unnest_generate_date_array,
]
),
exp.SHA: rename_func("SHA1"),