dbcli/sqlglot
1
0
Fork 0
Code Issues Activity

Merging 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 21:52:32 +01:00
parent 7688e2bdf8
commit bad79d1f7c
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

69
sqlglot/dialects/dialect.py
View file

@ -133,6 +133,10 @@ class _Dialect(type):
klass.INVERSE_FORMAT_MAPPING = {v: k for k, v in klass.FORMAT_MAPPING.items()}
klass.INVERSE_FORMAT_TRIE = new_trie(klass.INVERSE_FORMAT_MAPPING)
klass.INVERSE_CREATABLE_KIND_MAPPING = {
v: k for k, v in klass.CREATABLE_KIND_MAPPING.items()
}
base = seq_get(bases, 0)
base_tokenizer = (getattr(base, "tokenizer_class", Tokenizer),)
base_jsonpath_tokenizer = (getattr(base, "jsonpath_tokenizer_class", JSONPathTokenizer),)
@ -183,6 +187,9 @@ class _Dialect(type):
if enum not in ("", "bigquery"):
klass.generator_class.SELECT_KINDS = ()
if enum not in ("", "clickhouse"):
klass.generator_class.SUPPORTS_NULLABLE_TYPES = False
if enum not in ("", "athena", "presto", "trino"):
klass.generator_class.TRY_SUPPORTED = False
klass.generator_class.SUPPORTS_UESCAPE = False
@ -369,6 +376,24 @@ class Dialect(metaclass=_Dialect):
Whether ORDER BY ALL is supported (expands to all the selected columns) as in DuckDB, Spark3/Databricks
"""
HAS_DISTINCT_ARRAY_CONSTRUCTORS = False
"""
Whether the ARRAY constructor is context-sensitive, i.e in Redshift ARRAY[1, 2, 3] != ARRAY(1, 2, 3)
as the former is of type INT[] vs the latter which is SUPER
"""
SUPPORTS_FIXED_SIZE_ARRAYS = False
"""
Whether expressions such as x::INT[5] should be parsed as fixed-size array defs/casts e.g. in DuckDB. In
dialects which don't support fixed size arrays such as Snowflake, this should be interpreted as a subscript/index operator
"""
CREATABLE_KIND_MAPPING: dict[str, str] = {}
"""
Helper for dialects that use a different name for the same creatable kind. For example, the Clickhouse
equivalent of CREATE SCHEMA is CREATE DATABASE.
"""
# --- Autofilled ---
tokenizer_class = Tokenizer
@ -385,6 +410,8 @@ class Dialect(metaclass=_Dialect):
INVERSE_FORMAT_MAPPING: t.Dict[str, str] = {}
INVERSE_FORMAT_TRIE: t.Dict = {}
INVERSE_CREATABLE_KIND_MAPPING: dict[str, str] = {}
ESCAPED_SEQUENCES: t.Dict[str, str] = {}
# Delimiters for string literals and identifiers
@ -635,6 +662,9 @@ class Dialect(metaclass=_Dialect):
exp.GenerateDateArray: lambda self, e: self._annotate_with_type(
e, exp.DataType.build("ARRAY<DATE>")
),
exp.GenerateTimestampArray: lambda self, e: self._annotate_with_type(
e, exp.DataType.build("ARRAY<TIMESTAMP>")
),
exp.If: lambda self, e: self._annotate_by_args(e, "true", "false"),
exp.Interval: lambda self, e: self._annotate_with_type(e, exp.DataType.Type.INTERVAL),
exp.Least: lambda self, e: self._annotate_by_args(e, "expressions"),
@ -1214,7 +1244,13 @@ def right_to_substring_sql(self: Generator, expression: exp.Left) -> str:
def timestrtotime_sql(self: Generator, expression: exp.TimeStrToTime) -> str:
return self.sql(exp.cast(expression.this, exp.DataType.Type.TIMESTAMP))
datatype = (
exp.DataType.Type.TIMESTAMPTZ
if expression.args.get("zone")
else exp.DataType.Type.TIMESTAMP
)
return self.sql(exp.cast(expression.this, datatype, dialect=self.dialect))
def datestrtodate_sql(self: Generator, expression: exp.DateStrToDate) -> str:
@ -1464,14 +1500,19 @@ def merge_without_target_sql(self: Generator, expression: exp.Merge) -> str:
targets.add(normalize(alias.this))
for when in expression.expressions:
when.transform(
lambda node: (
exp.column(node.this)
if isinstance(node, exp.Column) and normalize(node.args.get("table")) in targets
else node
),
copy=False,
)
# only remove the target names from the THEN clause
# theyre still valid in the <condition> part of WHEN MATCHED / WHEN NOT MATCHED
# ref: https://github.com/TobikoData/sqlmesh/issues/2934
then = when.args.get("then")
if then:
then.transform(
lambda node: (
exp.column(node.this)
if isinstance(node, exp.Column) and normalize(node.args.get("table")) in targets
else node
),
copy=False,
)
return self.merge_sql(expression)
@ -1590,9 +1631,9 @@ def sha256_sql(self: Generator, expression: exp.SHA2) -> str:
return self.func(f"SHA{expression.text('length') or '256'}", expression.this)
def sequence_sql(self: Generator, expression: exp.GenerateSeries):
start = expression.args["start"]
end = expression.args["end"]
def sequence_sql(self: Generator, expression: exp.GenerateSeries | exp.GenerateDateArray) -> str:
start = expression.args.get("start")
end = expression.args.get("end")
step = expression.args.get("step")
if isinstance(start, exp.Cast):
@ -1602,8 +1643,8 @@ def sequence_sql(self: Generator, expression: exp.GenerateSeries):
else:
target_type = None
if target_type and target_type.is_type("timestamp"):
if target_type is start.to:
if start and end and target_type and target_type.is_type("date", "timestamp"):
if isinstance(start, exp.Cast) and target_type is start.to:
end = exp.cast(end, target_type)
else:
start = exp.cast(start, target_type)