feat: Filter Pushdown Rule

optd-datafusion-bridge/src/into_optd.rs

@@ -15,28 +15,6 @@ use optd_datafusion_repr::properties::schema::Schema as OPTDSchema;
 
 use crate::OptdPlanContext;
 
-// flatten_nested_logical is a helper function to flatten nested logical operators with same op type
-// eg. (a AND (b AND c)) => ExprList([a, b, c])
-//    (a OR (b OR c)) => ExprList([a, b, c])
-// It assume the children of the input expr_list are already flattened
-//  and can only be used in bottom up manner
-fn flatten_nested_logical(op: LogOpType, expr_list: ExprList) -> ExprList {
-    // conv_into_optd_expr is building the children bottom up so there is no need to
-    // call flatten_nested_logical recursively
-    let mut new_expr_list = Vec::new();
-    for child in expr_list.to_vec() {
-        if let OptRelNodeTyp::LogOp(child_op) = child.typ() {
-            if child_op == op {
-                let child_log_op_expr = LogOpExpr::from_rel_node(child.into_rel_node()).unwrap();
-                new_expr_list.extend(child_log_op_expr.children().to_vec());
-                continue;
-            }
-        }
-        new_expr_list.push(child.clone());
-    }
-    ExprList::new(new_expr_list)
-}
-
 impl OptdPlanContext<'_> {
     fn conv_into_optd_table_scan(&mut self, node: &logical_plan::TableScan) -> Result<PlanNode> {
         let table_name = node.table_name.to_string();
@@ -73,14 +51,16 @@ impl OptdPlanContext<'_> {
                     Operator::And => {
                         let op = LogOpType::And;
                         let expr_list = ExprList::new(vec![left, right]);
-                        let expr_list = flatten_nested_logical(op, expr_list);
-                        return Ok(LogOpExpr::new(op, expr_list).into_expr());
+                        return Ok(
+                            LogOpExpr::new_flattened_nested_logical(op, expr_list).into_expr()
+                        );
                     }
                     Operator::Or => {
                         let op = LogOpType::Or;
                         let expr_list = ExprList::new(vec![left, right]);
-                        let expr_list = flatten_nested_logical(op, expr_list);
-                        return Ok(LogOpExpr::new(op, expr_list).into_expr());
+                        return Ok(
+                            LogOpExpr::new_flattened_nested_logical(op, expr_list).into_expr()
+                        );
                     }
                     _ => {}
                 }
@@ -331,13 +311,8 @@ impl OptdPlanContext<'_> {
         } else {
             let expr_list = ExprList::new(log_ops);
             // the expr from filter is already flattened in conv_into_optd_expr
-            let expr_list = flatten_nested_logical(LogOpType::And, expr_list);
-            Ok(LogicalJoin::new(
-                left,
-                right,
-                LogOpExpr::new(LogOpType::And, expr_list).into_expr(),
-                join_type,
-            ))
+            let log_op = LogOpExpr::new_flattened_nested_logical(LogOpType::And, expr_list);
+            Ok(LogicalJoin::new(left, right, log_op.into_expr(), join_type))
         }
     }
 

optd-datafusion-repr/src/lib.rs

@@ -29,11 +29,18 @@ use rules::{
 
 pub use optd_core::rel_node::Value;
 
+use crate::rules::{
+    FilterAggTransposeRule, FilterCrossJoinTransposeRule, FilterInnerJoinTransposeRule,
+    FilterMergeRule, FilterProjectTransposeRule, FilterSortTransposeRule,
+};
+
 pub mod cost;
 mod explain;
 pub mod plan_nodes;
 pub mod properties;
 pub mod rules;
+#[cfg(test)]
+mod testing;
 
 pub struct DatafusionOptimizer {
     hueristic_optimizer: HeuristicsOptimizer<OptRelNodeTyp>,
@@ -92,6 +99,23 @@ impl DatafusionOptimizer {
         for rule in rules {
             rule_wrappers.push(RuleWrapper::new_cascades(rule));
         }
+        // add all filter pushdown rules as heuristic rules
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(
+            FilterProjectTransposeRule::new(),
+        )));
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(FilterMergeRule::new())));
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(
+            FilterCrossJoinTransposeRule::new(),
+        )));
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(
+            FilterInnerJoinTransposeRule::new(),
+        )));
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(
+            FilterSortTransposeRule::new(),
+        )));
+        rule_wrappers.push(RuleWrapper::new_heuristic(Arc::new(
+            FilterAggTransposeRule::new(),
+        )));
         rule_wrappers.push(RuleWrapper::new_cascades(Arc::new(HashJoinRule::new()))); // 17
         rule_wrappers.push(RuleWrapper::new_cascades(Arc::new(JoinCommuteRule::new()))); // 18
         rule_wrappers.push(RuleWrapper::new_cascades(Arc::new(JoinAssocRule::new())));

optd-datafusion-repr/src/plan_nodes.rs

@@ -16,6 +16,7 @@ use std::fmt::Debug;
 use std::sync::Arc;
 
 use arrow_schema::DataType;
+use itertools::Itertools;
 use optd_core::{
     cascades::{CascadesOptimizer, GroupId},
     rel_node::{RelNode, RelNodeMeta, RelNodeMetaMap, RelNodeRef, RelNodeTyp},
@@ -282,6 +283,56 @@ impl Expr {
     pub fn child(&self, idx: usize) -> OptRelNodeRef {
         self.0.child(idx)
     }
+
+    /// Recursively rewrite all column references in the expression.using a provided
+    /// function that replaces a column index.
+    /// The provided function will, given a ColumnRefExpr's index,
+    /// return either Some(usize) or None.
+    /// - If it is Some, the column index can be rewritten with the value.
+    /// - If any of the columns is None, we will return None all the way up
+    /// the call stack, and no expression will be returned.
+    pub fn rewrite_column_refs(
+        &self,
+        rewrite_fn: &impl Fn(usize) -> Option<usize>,
+    ) -> Option<Self> {
+        assert!(self.typ().is_expression());
+        if let OptRelNodeTyp::ColumnRef = self.typ() {
+            let col_ref = ColumnRefExpr::from_rel_node(self.0.clone()).unwrap();
+            let rewritten = rewrite_fn(col_ref.index());
+            return if let Some(rewritten_idx) = rewritten {
+                let new_col_ref = ColumnRefExpr::new(rewritten_idx);
+                Some(Self(new_col_ref.into_rel_node()))
+            } else {
+                None
+            };
+        }
+
+        let children = self.0.children.clone();
+        let children = children
+            .into_iter()
+            .map(|child| {
+                if child.typ == OptRelNodeTyp::List {
+                    // TODO: What should we do with List?
+                    return Some(child);
+                }
+                Expr::from_rel_node(child.clone())
+                    .unwrap()
+                    .rewrite_column_refs(rewrite_fn)
+                    .map(|x| x.into_rel_node())
+            })
+            .collect::<Option<Vec<_>>>()?;
+        Some(
+            Expr::from_rel_node(
+                RelNode {
+                    typ: self.0.typ.clone(),
+                    children: children.into_iter().collect_vec(),
+                    data: self.0.data.clone(),
+                }
+                .into(),
+            )
+            .unwrap(),
+        )
+    }
 }
 
 impl OptRelNode for Expr {

optd-datafusion-repr/src/plan_nodes/expr.rs

@@ -608,6 +608,29 @@ impl LogOpExpr {
         ))
     }
 
+    /// flatten_nested_logical is a helper function to flatten nested logical operators with same op type
+    /// eg. (a AND (b AND c)) => ExprList([a, b, c])
+    ///    (a OR (b OR c)) => ExprList([a, b, c])
+    /// It assume the children of the input expr_list are already flattened
+    ///  and can only be used in bottom up manner
+    pub fn new_flattened_nested_logical(op: LogOpType, expr_list: ExprList) -> Self {
+        // Since we assume that we are building the children bottom up,
+        // there is no need to call flatten_nested_logical recursively
+        let mut new_expr_list = Vec::new();
+        for child in expr_list.to_vec() {
+            if let OptRelNodeTyp::LogOp(child_op) = child.typ() {
+                if child_op == op {
+                    let child_log_op_expr =
+                        LogOpExpr::from_rel_node(child.into_rel_node()).unwrap();
+                    new_expr_list.extend(child_log_op_expr.children().to_vec());
+                    continue;
+                }
+            }
+            new_expr_list.push(child.clone());
+        }
+        LogOpExpr::new(op, ExprList::new(new_expr_list))
+    }
+
     pub fn children(&self) -> Vec<Expr> {
         self.0
              .0

optd-datafusion-repr/src/plan_nodes/join.rs

@@ -62,3 +62,20 @@ define_plan_node!(
         { 3, right_keys: ExprList }
     ], { join_type: JoinType }
 );
+
+impl LogicalJoin {
+    /// Takes in left/right schema sizes, and maps an index to be as if it were
+    /// pushed down to the left or right side of a join accordingly.
+    pub fn map_through_join(
+        index: usize,
+        left_schema_size: usize,
+        right_schema_size: usize,
+    ) -> usize {
+        assert!(index < left_schema_size + right_schema_size);
+        if index < left_schema_size {
+            index
+        } else {
+            index - left_schema_size
+        }
+    }
+}

optd-datafusion-repr/src/plan_nodes/projection.rs

@@ -1,7 +1,7 @@
 use super::expr::ExprList;
 use super::macros::define_plan_node;
 
-use super::{OptRelNode, OptRelNodeRef, OptRelNodeTyp, PlanNode};
+use super::{ColumnRefExpr, Expr, OptRelNode, OptRelNodeRef, OptRelNodeTyp, PlanNode};
 
 #[derive(Clone, Debug)]
 pub struct LogicalProjection(pub PlanNode);
@@ -26,3 +26,57 @@ define_plan_node!(
         { 1, exprs: ExprList }
     ]
 );
+
+pub struct ProjectionMapping {
+    forward: Vec<usize>,
+    _backward: Vec<Option<usize>>,
+}
+
+impl ProjectionMapping {
+    pub fn build(mapping: Vec<usize>) -> Option<Self> {
+        let mut backward = vec![];
+        for (i, &x) in mapping.iter().enumerate() {
+            if x >= backward.len() {
+                backward.resize(x + 1, None);
+            }
+            backward[x] = Some(i);
+        }
+        Some(Self {
+            forward: mapping,
+            _backward: backward,
+        })
+    }
+
+    pub fn projection_col_refers_to(&self, col: usize) -> usize {
+        self.forward[col]
+    }
+
+    pub fn _original_col_maps_to(&self, col: usize) -> Option<usize> {
+        self._backward[col]
+    }
+
+    /// Recursively rewrites all ColumnRefs in an Expr to what the projection
+    /// node is rewriting. E.g. if Projection is A -> B, B will be rewritten as A
+    pub fn rewrite_condition(&self, cond: Expr, child_schema_len: usize) -> Expr {
+        let proj_schema_size = self.forward.len();
+        cond.rewrite_column_refs(&|idx| {
+            Some(if idx < proj_schema_size {
+                self.projection_col_refers_to(idx)
+            } else {
+                idx - proj_schema_size + child_schema_len
+            })
+        })
+        .unwrap()
+    }
+}
+
+impl LogicalProjection {
+    pub fn compute_column_mapping(exprs: &ExprList) -> Option<ProjectionMapping> {
+        let mut mapping = vec![];
+        for expr in exprs.to_vec() {
+            let col_expr = ColumnRefExpr::from_rel_node(expr.into_rel_node())?;
+            mapping.push(col_expr.index());
+        }
+        ProjectionMapping::build(mapping)
+    }
+}

optd-datafusion-repr/src/rules.rs

@@ -2,6 +2,7 @@
 mod eliminate_duplicated_expr;
 mod eliminate_limit;
 mod filter;
+mod filter_pushdown;
 mod joins;
 mod macros;
 mod physical;
@@ -12,6 +13,10 @@ pub use eliminate_duplicated_expr::{
 };
 pub use eliminate_limit::EliminateLimitRule;
 pub use filter::{EliminateFilterRule, SimplifyFilterRule, SimplifyJoinCondRule};
+pub use filter_pushdown::{
+    FilterAggTransposeRule, FilterCrossJoinTransposeRule, FilterInnerJoinTransposeRule,
+    FilterMergeRule, FilterProjectTransposeRule, FilterSortTransposeRule,
+};
 pub use joins::{
     EliminateJoinRule, HashJoinRule, JoinAssocRule, JoinCommuteRule, ProjectionPullUpJoin,
 };