parquet filter pushdown correctness tests

Author: alamb

datafusion/core/Cargo.toml

@@ -108,6 +108,7 @@ csv = "1.1.6"
 ctor = "0.1.22"
 doc-comment = "0.3"
 env_logger = "0.9"
+parquet-test-utils = { path = "../../parquet-test-utils" }
 rstest = "0.15.0"
 test-utils = { path = "../../test-utils" }
 

datafusion/core/src/physical_plan/file_format/parquet.rs

@@ -1198,7 +1198,6 @@ mod tests {
     use crate::datasource::listing::{FileRange, PartitionedFile};
     use crate::datasource::object_store::ObjectStoreUrl;
     use crate::execution::options::CsvReadOptions;
-    use crate::physical_plan::metrics::MetricValue;
     use crate::prelude::{ParquetReadOptions, SessionConfig, SessionContext};
     use crate::test::object_store::local_unpartitioned_file;
     use crate::{
@@ -2029,15 +2028,8 @@ mod tests {
     ///
     /// Panics if no such metric.
     fn get_value(metrics: &MetricsSet, metric_name: &str) -> usize {
-        let sum = metrics.sum(|m| match m.value() {
-            MetricValue::Count { name, .. } if name == metric_name => true,
-            MetricValue::Time { name, .. } if name == metric_name => true,
-            _ => false,
-        });
-
-        match sum {
-            Some(MetricValue::Count { count, .. }) => count.value(),
-            Some(MetricValue::Time { time, .. }) => time.value(),
+        match metrics.sum_by_name(metric_name) {
+            Some(v) => v.as_usize(),
             _ => {
                 panic!(
                     "Expected metric not found. Looking for '{}' in\n\n{:#?}",

datafusion/core/src/physical_plan/metrics/mod.rs

@@ -241,6 +241,23 @@ impl MetricsSet {
         Some(accum)
     }
 
+    /// returns the sum of all the metrics with the specified name
+    /// the returned set.
+    pub fn sum_by_name(&self, metric_name: &str) -> Option<MetricValue> {
+        self.sum(|m| match m.value() {
+            MetricValue::Count { name, .. } => name == metric_name,
+            MetricValue::Time { name, .. } => name == metric_name,
+            MetricValue::OutputRows(_) => false,
+            MetricValue::ElapsedCompute(_) => false,
+            MetricValue::SpillCount(_) => false,
+            MetricValue::SpilledBytes(_) => false,
+            MetricValue::CurrentMemoryUsage(_) => false,
+            MetricValue::Gauge { name, .. } => name == metric_name,
+            MetricValue::StartTimestamp(_) => false,
+            MetricValue::EndTimestamp(_) => false,
+        })
+    }
+
     /// Returns returns a new derived `MetricsSet` where all metrics
     /// that had the same name and partition=`Some(..)` have been
     /// aggregated together. The resulting `MetricsSet` has all