Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
7 changes: 7 additions & 0 deletions CHANGELOG.md
Original file line number Diff line number Diff line change
Expand Up @@ -30,6 +30,13 @@ The format is based on Keep a Changelog, and this project adheres to Semantic Ve
`(a JOIN b) AS j`, registers only `j`. `ScopeVisitor::reference` reports
the items inside the join at `depth + 1`, before the join itself, and
subqueries in an unaliased join's conditions now get subquery scopes.
- `build_scope` now registers an aliased table function FROM item, such as
`my_func(x) AS g(y, z)`, as a virtual source, so `g.y` qualifies and its
lineage reaches the function's arguments under a `_N` virtual source name.
Without a schema, an unqualified column is resolved among all FROM items,
the function included, so `SELECT a FROM t, my_func(t.x) AS g` no longer
assumes `t.a`. An aliased `MATCH_RECOGNIZE` is registered as its alias,
which hides its input relation.
- `build_scope` now gives a subquery scope to a query in an expression field
the scope walk skipped: an aggregate's `FILTER (WHERE ...)`, `ORDER BY` and
`LIMIT`, a window frame bound, a `CAST` `FORMAT` or `DEFAULT` expression,
Expand Down
23 changes: 23 additions & 0 deletions crates/polyglot-sql/src/lineage.rs
Original file line number Diff line number Diff line change
Expand Up @@ -5022,6 +5022,29 @@ FROM UNNEST(GENERATE_DATE_ARRAY('2024-01-01', '2024-12-31', INTERVAL 1 WEEK)) AS
);
}

#[test]
fn test_lineage_aliased_table_function_is_a_virtual_source() {
let expr = parse_one(
"SELECT g.y AS out FROM t, my_func(t.x) AS g(y)",
DialectType::PostgreSQL,
)
.expect("parse");

let node = lineage("out", &expr, Some(DialectType::PostgreSQL), false).expect("lineage");
let child = node.downstream.first().expect("out should have lineage");

assert_eq!(child.name, "_0.y");
assert_eq!(child.source_kind, SourceKind::Virtual);
assert_eq!(child.source_alias.as_deref(), Some("g"));
assert!(
child
.downstream
.iter()
.any(|argument| argument.name == "t.x"),
"the function's argument should be lineage of its output"
);
}

#[test]
fn test_lineage_real_table_named_like_unnest_alias_is_not_virtual() {
let expr =
Expand Down
25 changes: 25 additions & 0 deletions crates/polyglot-sql/src/optimizer/qualify_columns.rs
Original file line number Diff line number Diff line change
Expand Up @@ -3216,6 +3216,31 @@ mod tests {
);
}

#[test]
fn test_qualify_aliased_table_function_columns() {
let mut schema = MappingSchema::new();
schema
.add_table("t", &[("x".into(), DataType::Text)], None)
.unwrap();
let sql = "SELECT x, y FROM t, my_func(t.x) AS g(y)";
let qualified =
qualify_columns(parse(sql), &schema, &QualifyColumnsOptions::new()).unwrap();
assert_eq!(
gen(&qualified),
"SELECT t.x AS x, g.y AS y FROM t, MY_FUNC(t.x) AS g(y)"
);

// Without a column list the function's columns are unknown, but a
// reference qualified with its alias still resolves.
let sql = "SELECT x, g.z FROM t, my_func(t.x) AS g";
let qualified =
qualify_columns(parse(sql), &schema, &QualifyColumnsOptions::new()).unwrap();
assert_eq!(
gen(&qualified),
"SELECT t.x AS x, g.z AS z FROM t, MY_FUNC(t.x) AS g"
);
}

#[test]
fn test_get_scope_columns() {
let expr = parse("SELECT a, b FROM t WHERE c = 1");
Expand Down
26 changes: 18 additions & 8 deletions crates/polyglot-sql/src/resolver.rs
Original file line number Diff line number Diff line change
Expand Up @@ -271,6 +271,9 @@ impl<'a> Resolver<'a> {
Expression::Alias(alias) if matches!(&alias.this, Expression::Unnest(_)) => {
alias_output_columns(alias)
}
Expression::Alias(alias) if matches!(&alias.this, Expression::Function(_)) => {
table_function_output_columns(alias)
}
Expression::Alias(alias) => {
let columns = self.get_source_columns_for_expression(&alias.this)?;
apply_alias_columns(columns, &alias.column_aliases)
Expand Down Expand Up @@ -469,6 +472,9 @@ impl<'a> Resolver<'a> {
Expression::Alias(alias) if matches!(&alias.this, Expression::Unnest(_)) => {
alias_output_columns(alias)
}
Expression::Alias(alias) if matches!(&alias.this, Expression::Function(_)) => {
table_function_output_columns(alias)
}
Expression::Alias(alias) => {
let columns = self.get_source_output_columns(&alias.this);
apply_alias_columns(columns, &alias.column_aliases)
Expand Down Expand Up @@ -697,15 +703,16 @@ fn unnest_output_columns(unnest: &crate::expressions::UnnestFunc) -> Vec<String>
.collect()
}

/// A table function's output columns are unknown unless its alias lists them.
fn table_function_output_columns(alias: &crate::expressions::Alias) -> Vec<String> {
Comment thread
ravenac95 marked this conversation as resolved.
identifier_names(&alias.column_aliases)
}

fn alias_output_columns(alias: &crate::expressions::Alias) -> Vec<String> {
if alias.column_aliases.is_empty() {
vec![alias.alias.name.clone()]
} else {
alias
.column_aliases
.iter()
.map(|column| column.name.clone())
.collect()
identifier_names(&alias.column_aliases)
}
}

Expand All @@ -718,10 +725,13 @@ fn lateral_output_columns(lateral: &crate::expressions::Lateral) -> Vec<String>
}

fn lateral_view_output_columns(lateral_view: &crate::expressions::LateralView) -> Vec<String> {
lateral_view
.column_aliases
identifier_names(&lateral_view.column_aliases)
}

fn identifier_names(identifiers: &[Identifier]) -> Vec<String> {
identifiers
.iter()
.map(|column| column.name.clone())
.map(|identifier| identifier.name.clone())
.collect()
}

Expand Down
Loading
Loading