Skip to content

feat(Snowflake)!: set default window frame for certain Snowflake ranking functions during transpilation #7195

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
geooo109 merged 3 commits into from
Mar 4, 2026
+78 −34
Merged

feat(Snowflake)!: set default window frame for certain Snowflake ranking functions during transpilation #7195

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
7245661
set default window frame for certain Snowflake ranking functions duri…
fivetran-felixhuang Mar 3, 2026
aa3d33c
remove logic around null ordering
fivetran-felixhuang Mar 4, 2026
3bb8dff
code cleanup
fivetran-felixhuang Mar 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
42 changes: 42 additions & 0 deletions sqlglot/dialects/snowflake.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,14 @@
exp.DType.TIMESTAMPTZ: "TO_TIMESTAMP_TZ",
}

RANKING_WINDOW_FUNCTIONS_WITH_FRAME = (
exp.FirstValue,
exp.LastValue,
exp.NthValue,
# Technically DenseRank and Rank also support window frame, but when it comes to transpilation, they behave differently in BigQuery and DuckDB.
# We should handle them on a more granular level when we need to support their transpilation
)


def _build_strtok(args: t.List) -> exp.SplitPart:
# Add default delimiter (space) if missing - per Snowflake docs
Expand Down Expand Up @@ -1563,6 +1571,20 @@ def _parse_window(

result = super()._parse_window(this, alias)

# Set default window frame for ranking functions if not present
if (
isinstance(result, exp.Window)
and isinstance(this, RANKING_WINDOW_FUNCTIONS_WITH_FRAME)
and not result.args.get("spec")
):
frame = exp.WindowSpec(
kind="ROWS",
start="UNBOUNDED",
start_side="PRECEDING",
end="UNBOUNDED",
end_side="FOLLOWING",
)
result.set("spec", frame)
return result

class Tokenizer(tokens.Tokenizer):
Expand Down Expand Up @@ -2340,3 +2362,23 @@ def uniform_sql(self, expression: exp.Uniform) -> str:
gen = exp.Rand()

return self.func("UNIFORM", expression.this, expression.expression, gen)

def window_sql(self, expression: exp.Window) -> str:
spec = expression.args.get("spec")
this = expression.this

if isinstance(this, RANKING_WINDOW_FUNCTIONS_WITH_FRAME) or (
isinstance(this, (exp.RespectNulls, exp.IgnoreNulls))
and isinstance(this.this, RANKING_WINDOW_FUNCTIONS_WITH_FRAME)
):
if spec:
# omit the default window from window ranknig functions
if (
spec.text("kind").upper() == "ROWS"
and spec.text("start").upper() == "UNBOUNDED"
and spec.text("start_side").upper() == "PRECEDING"
and spec.text("end").upper() == "UNBOUNDED"
and spec.text("end_side").upper() == "FOLLOWING"
):
expression.set("spec", None)
return super().window_sql(expression)
70 changes: 36 additions & 34 deletions tests/dialects/test_snowflake.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1354,61 +1354,63 @@ def test_snowflake(self):
},
)

# NTH_VALUE FROM FIRST not supported in DuckDB
self.validate_all(
"SELECT NTH_VALUE(is_deleted, 2) FROM FIRST IGNORE NULLS OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
write={
"snowflake": "SELECT NTH_VALUE(is_deleted, 2) FROM FIRST IGNORE NULLS OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2 IGNORE NULLS) OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2 IGNORE NULLS) OVER (PARTITION BY id ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS nth_is_deleted FROM my_table",
},
)

# NTH_VALUE FROM LAST not supported in DuckDB
self.validate_all(
"SELECT NTH_VALUE(is_deleted, 2) FROM LAST RESPECT NULLS OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
write={
"snowflake": "SELECT NTH_VALUE(is_deleted, 2) FROM LAST RESPECT NULLS OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2 RESPECT NULLS) OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2 RESPECT NULLS) OVER (PARTITION BY id ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS nth_is_deleted FROM my_table",
},
)

self.validate_all(
"SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
write={
"snowflake": "SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS nth_is_deleted FROM my_table",
},
)

self.validate_all(
"SELECT FIRST_VALUE(is_deleted) RESPECT NULLS OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
"SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
write={
"snowflake": "SELECT FIRST_VALUE(is_deleted) RESPECT NULLS OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
"duckdb": "SELECT FIRST_VALUE(is_deleted RESPECT NULLS) OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
"snowflake": "SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
"duckdb": "SELECT NTH_VALUE(is_deleted, 2) OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
},
)

self.validate_all(
"SELECT FIRST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
write={
"snowflake": "SELECT FIRST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
"duckdb": "SELECT FIRST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
},
)

self.validate_all(
"SELECT LAST_VALUE(is_deleted) RESPECT NULLS OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
write={
"snowflake": "SELECT LAST_VALUE(is_deleted) RESPECT NULLS OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
"duckdb": "SELECT LAST_VALUE(is_deleted RESPECT NULLS) OVER (PARTITION BY id ORDER BY c ASC) AS f FROM my_table",
},
)

self.validate_all(
"SELECT LAST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
write={
"snowflake": "SELECT LAST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
"duckdb": "SELECT LAST_VALUE(is_deleted) OVER (PARTITION BY id) AS f FROM my_table",
},
)
for func in (
"FIRST_VALUE",
"LAST_VALUE",
):
for options in (
" IGNORE NULLS",
" RESPECT NULLS",
"",
):
self.validate_all(
f"SELECT {func}(is_deleted){options} OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
write={
"snowflake": f"SELECT {func}(is_deleted){options} OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
"duckdb": f"SELECT {func}(is_deleted{options}) OVER (PARTITION BY id ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS nth_is_deleted FROM my_table",
},
)
self.validate_all(
f"SELECT {func}(is_deleted){options} OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
write={
"snowflake": f"SELECT {func}(is_deleted){options} OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
"duckdb": f"SELECT {func}(is_deleted{options}) OVER (PARTITION BY id ROWS BETWEEN 1 PRECEDING AND 1 FOLLOWING) AS nth_is_deleted FROM my_table",
},
)

self.validate_all(
"SELECT LEAD(is_deleted, 2, -10) RESPECT NULLS OVER (PARTITION BY id) AS nth_is_deleted FROM my_table",
Expand Down Expand Up @@ -3128,31 +3130,31 @@ def test_null_treatment(self):
self.validate_all(
r"SELECT FIRST_VALUE(TABLE1.COLUMN1) OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1",
write={
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1"
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2) AS MY_ALIAS FROM TABLE1"
},
)
self.validate_all(
r"SELECT FIRST_VALUE(TABLE1.COLUMN1 RESPECT NULLS) OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1",
write={
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) RESPECT NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1"
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) RESPECT NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2) AS MY_ALIAS FROM TABLE1"
},
)
self.validate_all(
r"SELECT FIRST_VALUE(TABLE1.COLUMN1) RESPECT NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1",
write={
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) RESPECT NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1"
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) RESPECT NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2) AS MY_ALIAS FROM TABLE1"
},
)
self.validate_all(
r"SELECT FIRST_VALUE(TABLE1.COLUMN1 IGNORE NULLS) OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1",
write={
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) IGNORE NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1"
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) IGNORE NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2) AS MY_ALIAS FROM TABLE1"
},
)
self.validate_all(
r"SELECT FIRST_VALUE(TABLE1.COLUMN1) IGNORE NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1",
write={
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) IGNORE NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2 ROWS BETWEEN UNBOUNDED PRECEDING AND UNBOUNDED FOLLOWING) AS MY_ALIAS FROM TABLE1"
"snowflake": r"SELECT FIRST_VALUE(TABLE1.COLUMN1) IGNORE NULLS OVER (PARTITION BY RANDOM_COLUMN1, RANDOM_COLUMN2) AS MY_ALIAS FROM TABLE1"
},
)
self.validate_all(
Expand Down