python/sqlglot
1
0
Fork 0
Code Issues Activity

Adding upstream version 25.5.1.

Browse source 
Signed-off-by: Daniel Baumann <daniel@debian.org>
This commit is contained in:
Daniel Baumann 2025-02-13 21:41:00 +01:00
parent 147b6e06e8
commit 4e506fbac7
Signed by: daniel
GPG key ID: FBB4F0E80A80222F
136 changed files with 80990 additions and 72541 deletions
Download patch file Download diff file Expand all files Collapse all files

56
sqlglot/dialects/duckdb.py
View file

@ -15,11 +15,13 @@ from sqlglot.dialects.dialect import (
build_default_decimal_type,
date_trunc_to_time,
datestrtodate_sql,
no_datetime_sql,
encode_decode_sql,
build_formatted_time,
inline_array_unless_query,
no_comment_column_constraint_sql,
no_safe_divide_sql,
no_time_sql,
no_timestamp_sql,
pivot_column_names,
regexp_extract_sql,
@ -218,6 +220,7 @@ class DuckDB(Dialect):
"TIMESTAMP_US": TokenType.TIMESTAMP,
"VARCHAR": TokenType.TEXT,
}
KEYWORDS.pop("/*+")
SINGLE_TOKENS = {
**tokens.Tokenizer.SINGLE_TOKENS,
@ -407,6 +410,7 @@ class DuckDB(Dialect):
"DATE_DIFF", f"'{e.args.get('unit') or 'DAY'}'", e.expression, e.this
),
exp.DateStrToDate: datestrtodate_sql,
exp.Datetime: no_datetime_sql,
exp.DateToDi: lambda self,
e: f"CAST(STRFTIME({self.sql(e, 'this')}, {DuckDB.DATEINT_FORMAT}) AS INT)",
exp.Decode: lambda self, e: encode_decode_sql(self, e, "DECODE", replace=False),
@ -429,7 +433,6 @@ class DuckDB(Dialect):
exp.cast(e.expression, exp.DataType.Type.TIMESTAMP, copy=True),
exp.cast(e.this, exp.DataType.Type.TIMESTAMP, copy=True),
),
exp.ParseJSON: rename_func("JSON"),
exp.PercentileCont: rename_func("QUANTILE_CONT"),
exp.PercentileDisc: rename_func("QUANTILE_DISC"),
# DuckDB doesn't allow qualified columns inside of PIVOT expressions.
@ -450,13 +453,12 @@ class DuckDB(Dialect):
exp.Split: rename_func("STR_SPLIT"),
exp.SortArray: _sort_array_sql,
exp.StrPosition: str_position_sql,
exp.StrToDate: lambda self, e: f"CAST({str_to_time_sql(self, e)} AS DATE)",
exp.StrToTime: str_to_time_sql,
exp.StrToUnix: lambda self, e: self.func(
"EPOCH", self.func("STRPTIME", e.this, self.format_time(e))
),
exp.Struct: _struct_sql,
exp.TimeAdd: _date_delta_sql,
exp.Time: no_time_sql,
exp.Timestamp: no_timestamp_sql,
exp.TimestampDiff: lambda self, e: self.func(
"DATE_DIFF", exp.Literal.string(e.unit), e.expression, e.this
@ -608,6 +610,24 @@ class DuckDB(Dialect):
PROPERTIES_LOCATION[exp.LikeProperty] = exp.Properties.Location.POST_SCHEMA
PROPERTIES_LOCATION[exp.TemporaryProperty] = exp.Properties.Location.POST_CREATE
def strtotime_sql(self, expression: exp.StrToTime) -> str:
if expression.args.get("safe"):
formatted_time = self.format_time(expression)
return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS TIMESTAMP)"
return str_to_time_sql(self, expression)
def strtodate_sql(self, expression: exp.StrToDate) -> str:
if expression.args.get("safe"):
formatted_time = self.format_time(expression)
return f"CAST({self.func('TRY_STRPTIME', expression.this, formatted_time)} AS DATE)"
return f"CAST({str_to_time_sql(self, expression)} AS DATE)"
def parsejson_sql(self, expression: exp.ParseJSON) -> str:
arg = expression.this
if expression.args.get("safe"):
return self.sql(exp.case().when(exp.func("json_valid", arg), arg).else_(exp.null()))
return self.func("JSON", arg)
def timefromparts_sql(self, expression: exp.TimeFromParts) -> str:
nano = expression.args.get("nano")
if nano is not None:
@ -728,3 +748,33 @@ class DuckDB(Dialect):
this = self.sql(expression, "this").rstrip(")")
return f"{this}{expression_sql})"
def length_sql(self, expression: exp.Length) -> str:
arg = expression.this
# Dialects like BQ and Snowflake also accept binary values as args, so
# DDB will attempt to infer the type or resort to case/when resolution
if not expression.args.get("binary") or arg.is_string:
return self.func("LENGTH", arg)
if not arg.type:
from sqlglot.optimizer.annotate_types import annotate_types
arg = annotate_types(arg)
if arg.is_type(*exp.DataType.TEXT_TYPES):
return self.func("LENGTH", arg)
# We need these casts to make duckdb's static type checker happy
blob = exp.cast(arg, exp.DataType.Type.VARBINARY)
varchar = exp.cast(arg, exp.DataType.Type.VARCHAR)
case = (
exp.case(self.func("TYPEOF", arg))
.when(
"'VARCHAR'", exp.Anonymous(this="LENGTH", expressions=[varchar])
) # anonymous to break length_sql recursion
.when("'BLOB'", self.func("OCTET_LENGTH", blob))
)
return self.sql(case)