reorgnize file

Author: xudong963

datafusion/optimizer/src/lib.rs

@@ -61,7 +61,6 @@ pub mod push_down_limit;
 pub mod replace_distinct_aggregate;
 pub mod scalar_subquery_to_join;
 pub mod simplify_expressions;
-mod simplify_predicates;
 pub mod single_distinct_to_groupby;
 pub mod utils;
 

datafusion/optimizer/src/push_down_filter.rs

@@ -40,7 +40,7 @@ use datafusion_expr::{
 };
 
 use crate::optimizer::ApplyOrder;
-use crate::simplify_predicates::simplify_predicates;
+use crate::simplify_expressions::simplify_predicates;
 use crate::utils::{has_all_column_refs, is_restrict_null_predicate};
 use crate::{OptimizerConfig, OptimizerRule};
 

datafusion/optimizer/src/simplify_expressions/mod.rs

@@ -23,6 +23,7 @@ mod guarantees;
 mod inlist_simplifier;
 mod regex;
 pub mod simplify_exprs;
+mod simplify_predicates;
 mod unwrap_cast;
 mod utils;
 
@@ -31,6 +32,7 @@ pub use datafusion_expr::simplify::{SimplifyContext, SimplifyInfo};
 
 pub use expr_simplifier::*;
 pub use simplify_exprs::*;
+pub use simplify_predicates::simplify_predicates;
 
 // Export for test in datafusion/core/tests/optimizer_integration.rs
 pub use guarantees::GuaranteeRewriter;

datafusion/optimizer/src/simplify_predicates.rs renamed to datafusion/optimizer/src/simplify_expressions/simplify_predicates.rs

@@ -41,7 +41,7 @@ use std::collections::BTreeMap;
 ///
 /// # Returns
 /// A `Result` containing a vector of simplified `Expr` predicates.
-pub(crate) fn simplify_predicates(predicates: Vec<Expr>) -> Result<Vec<Expr>> {
+pub fn simplify_predicates(predicates: Vec<Expr>) -> Result<Vec<Expr>> {
     // Early return for simple cases
     if predicates.len() <= 1 {
         return Ok(predicates);
@@ -132,34 +132,38 @@ fn simplify_column_predicates(predicates: Vec<Expr>) -> Result<Vec<Expr>> {
 
     let mut result = Vec::new();
 
-    // If we have equality predicates, they're the most restrictive
     if !eq_predicates.is_empty() {
-        if eq_predicates.len() > 1 {
-            result.push(Expr::Literal(ScalarValue::Boolean(Some(false)), None));
+        // If there are many equality predicates, we can only keep one if they are all the same
+        // If they are not the same, add false to the result
+        if eq_predicates.len() == 1
+            || eq_predicates.iter().all(|e| e == &eq_predicates[0])
+        {
+            result.push(eq_predicates.pop().unwrap());
         } else {
-            result.push(eq_predicates[0].clone());
+            // If they are not the same, add a false predicate
+            result.push(Expr::Literal(ScalarValue::Boolean(Some(false)), None));
         }
-    } else {
-        // Handle all greater-than-style predicates (keep the most restrictive - highest value)
-        if !greater_predicates.is_empty() {
-            if let Some(most_restrictive) =
-                find_most_restrictive_predicate(&greater_predicates, true)?
-            {
-                result.push(most_restrictive);
-            } else {
-                result.extend(greater_predicates);
-            }
+    }
+
+    // Handle all greater-than-style predicates (keep the most restrictive - highest value)
+    if !greater_predicates.is_empty() {
+        if let Some(most_restrictive) =
+            find_most_restrictive_predicate(&greater_predicates, true)?
+        {
+            result.push(most_restrictive);
+        } else {
+            result.extend(greater_predicates);
         }
+    }
 
-        // Handle all less-than-style predicates (keep the most restrictive - lowest value)
-        if !less_predicates.is_empty() {
-            if let Some(most_restrictive) =
-                find_most_restrictive_predicate(&less_predicates, false)?
-            {
-                result.push(most_restrictive);
-            } else {
-                result.extend(less_predicates);
-            }
+    // Handle all less-than-style predicates (keep the most restrictive - lowest value)
+    if !less_predicates.is_empty() {
+        if let Some(most_restrictive) =
+            find_most_restrictive_predicate(&less_predicates, false)?
+        {
+            result.push(most_restrictive);
+        } else {
+            result.extend(less_predicates);
         }
     }
 

datafusion/sqllogictest/test_files/simplify_predicates.slt

@@ -70,13 +70,35 @@ logical_plan
 01)Filter: test_data.float_col < Float32(8) AND test_data.int_col > Int32(6)
 02)--TableScan: test_data projection=[int_col, float_col, str_col, date_col, bool_col]
 
+# x = 7 AND x = 7 should simplify to x = 7
+query TT
+EXPLAIN SELECT * FROM test_data WHERE int_col = 7 AND int_col = 7;
+----
+logical_plan
+01)Filter: test_data.int_col = Int32(7)
+02)--TableScan: test_data projection=[int_col, float_col, str_col, date_col, bool_col]
+
+# x = 7 AND x = 6 should simplify to false
+query TT
+EXPLAIN SELECT * FROM test_data WHERE int_col = 7 AND int_col = 6;
+----
+logical_plan EmptyRelation
+
+# TODO: x = 7 AND x < 2 should simplify to false
+query TT
+EXPLAIN SELECT * FROM test_data WHERE int_col = 7 AND int_col < 2;
+----
+logical_plan
+01)Filter: test_data.int_col = Int32(7) AND test_data.int_col < Int32(2)
+02)--TableScan: test_data projection=[int_col, float_col, str_col, date_col, bool_col]
 
-# x = 7 AND x > 5 should simplify to x = 7
+
+# TODO: x = 7 AND x > 5 should simplify to x = 7
 query TT
 EXPLAIN SELECT * FROM test_data WHERE int_col = 7 AND int_col > 5;
 ----
 logical_plan
-01)Filter: test_data.int_col = Int32(7)
+01)Filter: test_data.int_col = Int32(7) AND test_data.int_col > Int32(5)
 02)--TableScan: test_data projection=[int_col, float_col, str_col, date_col, bool_col]
 
 # str_col > 'apple' AND str_col > 'banana' should simplify to str_col > 'banana'
@@ -148,7 +170,7 @@ logical_plan
 07)------Filter: test_data2.value < Int32(50) AND test_data2.id > Int32(10)
 08)--------TableScan: test_data2 projection=[id, value]
 
-# Case 13: Handling negated predicates
+# Handling negated predicates
 # NOT (x < 10) AND NOT (x < 5) should simplify to NOT (x < 10)
 query TT
 EXPLAIN SELECT * FROM test_data WHERE NOT (int_col < 10) AND NOT (int_col < 5);
@@ -198,7 +220,7 @@ logical_plan
 01)Filter: (test_data.int_col > Int32(5) OR test_data.float_col < Float32(10)) AND (test_data.int_col > Int32(6) OR test_data.float_col < Float32(8))
 02)--TableScan: test_data projection=[int_col, float_col, str_col, date_col, bool_col]
 
-# Case 20: Combination of AND and OR with simplifiable predicates
+# Combination of AND and OR with simplifiable predicates
 query TT
 EXPLAIN SELECT * FROM test_data
 WHERE (int_col > 5 AND int_col > 6)