fix(pretty-print): normalize associative boolean expressions before comparison

crates/pgls_pretty_print/src/normalize.rs

@@ -18,6 +18,7 @@ use pgls_query::{NodeEnum, NodeMut};
 /// the original AST and the AST of reparsed formatted output.
 pub fn normalize_ast(node: &mut NodeEnum) {
     clear_location(node);
+    normalize_bool_expr_associativity(node);
     normalize_a_indirection(node);
     normalize_object_with_args(node);
     normalize_join_expr(node);
@@ -27,6 +28,61 @@ pub fn normalize_ast(node: &mut NodeEnum) {
     normalize_function_body(node);
 }
 
+/// Flatten nested boolean expressions which PostgreSQL represents differently
+/// depending on redundant parentheses.
+///
+/// Only same-operator AND and OR expressions are associative. NOT and mixed
+/// AND/OR expressions retain their original tree shape because parentheses
+/// affect their semantics.
+fn normalize_bool_expr_associativity(node: &mut NodeEnum) {
+    let bool_exprs = node
+        .iter_mut()
+        .filter_map(|node| match node {
+            NodeMut::BoolExpr(bool_expr) => Some(bool_expr),
+            _ => None,
+        })
+        .collect::<Vec<_>>();
+
+    // Process descendants first. Flattening a parent moves its child nodes, so
+    // child pointers must not be used after their parent has been normalized.
+    for bool_expr in bool_exprs.into_iter().rev() {
+        unsafe {
+            flatten_bool_expr_args(&mut *bool_expr);
+        }
+    }
+}
+
+fn flatten_bool_expr_args(bool_expr: &mut pgls_query::protobuf::BoolExpr) {
+    use pgls_query::protobuf::BoolExprType;
+
+    if !matches!(
+        BoolExprType::try_from(bool_expr.boolop),
+        Ok(BoolExprType::AndExpr | BoolExprType::OrExpr)
+    ) {
+        return;
+    }
+
+    let boolop = bool_expr.boolop;
+    let mut flattened = Vec::with_capacity(bool_expr.args.len());
+
+    for mut arg in std::mem::take(&mut bool_expr.args) {
+        let nested_args = match arg.node.as_mut() {
+            Some(NodeEnum::BoolExpr(nested)) if nested.boolop == boolop => {
+                Some(std::mem::take(&mut nested.args))
+            }
+            _ => None,
+        };
+
+        if let Some(mut nested_args) = nested_args {
+            flattened.append(&mut nested_args);
+        } else {
+            flattened.push(arg);
+        }
+    }
+
+    bool_expr.args = flattened;
+}
+
 /// Clear location fields in AST nodes.
 ///
 /// Location fields record the byte offset in the original source,

crates/pgls_pretty_print/tests/data/multi/bool_expr_associativity.sql

@@ -0,0 +1,19 @@
+CREATE TABLE demo_output AS
+SELECT
+	*
+FROM
+	demo
+WHERE
+	(
+		demo.accounting_class ~ '^[0-9]+$'
+		AND (
+			demo.accounting_class >= '6000'
+			AND demo.accounting_class <= '6629'
+		)
+		OR starts_with(demo.accounting_class, '71')
+	)
+	AND demo.changed_at > demo.created_at;
+
+SELECT * FROM demo WHERE (a OR (b OR c)) AND d;
+
+SELECT * FROM demo WHERE a AND (b OR c);

crates/pgls_pretty_print/tests/snapshots/multi/tests__bool_expr_associativity_100.snap

@@ -0,0 +1,20 @@
+---
+source: crates/pgls_pretty_print/tests/tests.rs
+input_file: crates/pgls_pretty_print/tests/data/multi/bool_expr_associativity.sql
+---
+create table demo_output
+as
+  select
+    *
+  from
+    demo
+  where
+    (demo.accounting_class ~ '^[0-9]+$' and
+    demo.accounting_class >= '6000' and
+    demo.accounting_class <= '6629' or
+    starts_with(demo.accounting_class, '71')) and
+    demo.changed_at > demo.created_at;
+
+select * from demo where (a or b or c) and d;
+
+select * from demo where a and (b or c);

crates/pgls_pretty_print/tests/snapshots/multi/tests__bool_expr_associativity_80.snap

@@ -0,0 +1,20 @@
+---
+source: crates/pgls_pretty_print/tests/tests.rs
+input_file: crates/pgls_pretty_print/tests/data/multi/bool_expr_associativity.sql
+---
+create table demo_output
+as
+  select
+    *
+  from
+    demo
+  where
+    (demo.accounting_class ~ '^[0-9]+$' and
+    demo.accounting_class >= '6000' and
+    demo.accounting_class <= '6629' or
+    starts_with(demo.accounting_class, '71')) and
+    demo.changed_at > demo.created_at;
+
+select * from demo where (a or b or c) and d;
+
+select * from demo where a and (b or c);