add benchmark

delta-io · Aug 29, 2022 · dab6556 · dab6556
1 parent f91462c
commit dab6556
Show file tree

Hide file tree

Showing 4 changed files with 37 additions and 23 deletions.
diff --git a/rust/Cargo.toml b/rust/Cargo.toml
@@ -147,3 +147,8 @@ tempfile = "3"
 maplit = { version = "1" }
 anyhow = "1"
 rand = "0.8"
+criterion = "0"
+
+[[bench]]
+name = "read_checkpoint"
+harness = false
diff --git a/rust/src/action/parquet2_read/boolean.rs b/rust/src/action/parquet2_read/boolean.rs
@@ -1,7 +1,6 @@
 use parquet2::encoding::hybrid_rle::BitmapIter;
 use parquet2::metadata::ColumnDescriptor;
 use parquet2::page::DataPage;
-use parquet2::schema::types::PhysicalType;
 
 use super::validity::ValidityRowIndexIter;
 use super::{split_page, ActionVariant, ParseError};
@@ -56,11 +55,14 @@ where
     SetFn: Fn(&mut ActType, bool),
 {
     #[cfg(debug_assertions)]
-    if page.descriptor.primitive_type.physical_type != PhysicalType::Boolean {
-        return Err(ParseError::InvalidAction(format!(
-            "expect physical parquet type boolean, got {:?}",
-            page.descriptor.primitive_type,
-        )));
+    {
+        use parquet2::schema::types::PhysicalType;
+        if page.descriptor.primitive_type.physical_type != PhysicalType::Boolean {
+            return Err(ParseError::InvalidAction(format!(
+                "expect physical parquet type boolean, got {:?}",
+                page.descriptor.primitive_type,
+            )));
+        }
     }
 
     let some_value_iter = SomeBooleanValueIter::try_new(page, descriptor)?;

diff --git a/rust/src/action/parquet2_read/map.rs b/rust/src/action/parquet2_read/map.rs
@@ -2,7 +2,6 @@
 
 use parquet2::metadata::ColumnDescriptor;
 use parquet2::page::{DataPage, DictPage};
-use parquet2::schema::types::PhysicalType;
 
 use super::string::for_each_repeated_string_field_value_with_idx;
 use super::{ActionVariant, ParseError};
@@ -29,11 +28,14 @@ where
 {
     debug_assert!(field[0] == "key_value");
     #[cfg(debug_assertions)]
-    if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
-        return Err(ParseError::InvalidAction(format!(
-            "expect parquet utf8 type for map key/value, got primitive type: {:?}",
-            page.descriptor.primitive_type,
-        )));
+    {
+        use parquet2::schema::types::PhysicalType;
+        if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
+            return Err(ParseError::InvalidAction(format!(
+                "expect parquet utf8 type for map key/value, got primitive type: {:?}",
+                page.descriptor.primitive_type,
+            )));
+        }
     }
 
     match field[1].as_str() {

diff --git a/rust/src/action/parquet2_read/string.rs b/rust/src/action/parquet2_read/string.rs
@@ -4,7 +4,6 @@ use parquet2::encoding::hybrid_rle::HybridRleDecoder;
 use parquet2::encoding::Encoding;
 use parquet2::metadata::ColumnDescriptor;
 use parquet2::page::{DataPage, DictPage};
-use parquet2::schema::types::PhysicalType;
 
 use super::dictionary;
 use super::dictionary::binary::BinaryPageDict;
@@ -206,11 +205,14 @@ where
     MapFn: FnMut(Result<(usize, Vec<String>), ParseError>) -> Result<(), ParseError>,
 {
     #[cfg(debug_assertions)]
-    if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
-        return Err(ParseError::InvalidAction(format!(
-            "expect parquet utf8 type, got primitive type: {:?}",
-            page.descriptor.primitive_type,
-        )));
+    {
+        use parquet2::schema::types::PhysicalType;
+        if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
+            return Err(ParseError::InvalidAction(format!(
+                "expect parquet utf8 type, got primitive type: {:?}",
+                page.descriptor.primitive_type,
+            )));
+        }
     }
 
     match page.encoding() {
@@ -271,11 +273,14 @@ where
     SetFn: Fn(&mut ActType, String),
 {
     #[cfg(debug_assertions)]
-    if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
-        return Err(ParseError::InvalidAction(format!(
-            "expect parquet utf8 type, got primitive type: {:?}",
-            page.descriptor.primitive_type,
-        )));
+    {
+        use parquet2::schema::types::PhysicalType;
+        if page.descriptor.primitive_type.physical_type != PhysicalType::ByteArray {
+            return Err(ParseError::InvalidAction(format!(
+                "expect parquet utf8 type, got primitive type: {:?}",
+                page.descriptor.primitive_type,
+            )));
+        }
     }
 
     let map_fn = |entry: Result<(usize, String), ParseError>| -> Result<(), ParseError> {