Add Parquet arrow_reader benchmarks for {u}int{8,16} columns (#7484)

* allow for reading improperly encode UINT_8 and UINT_16 parquet data

* add some benchmarks

* remove print

* checkpoint some experimental code

* checkpoint

* add a few more types

* modify comment

* another edit

* Revert non benchmark code

---------

Co-authored-by: Ed Seidl <[email protected]>
Co-authored-by: seidl <[email protected]>

Author: alamb

parquet/benches/arrow_reader.rs

@@ -76,6 +76,18 @@ fn build_test_schema() -> SchemaDescPtr {
             OPTIONAL FIXED_LEN_BYTE_ARRAY (8) optional_flba8_leaf;
             REQUIRED FIXED_LEN_BYTE_ARRAY (16) mandatory_flba16_leaf;
             OPTIONAL FIXED_LEN_BYTE_ARRAY (16) optional_flba16_leaf;
+            REQUIRED INT32 mandatory_uint8_leaf (INTEGER(8, false));
+            OPTIONAL INT32 optional_uint8_leaf (INTEGER(8, false));
+            REQUIRED INT32 mandatory_uint16_leaf (INTEGER(16, false));
+            OPTIONAL INT32 optional_uint16_leaf (INTEGER(16, false));
+            REQUIRED INT32 mandatory_uint32_leaf (INTEGER(32, false));
+            OPTIONAL INT32 optional_uint32_leaf (INTEGER(32, false));
+            REQUIRED INT32 mandatory_int8_leaf (INTEGER(8, true));
+            OPTIONAL INT32 optional_int8_leaf (INTEGER(8, true));
+            REQUIRED INT32 mandatory_int16_leaf (INTEGER(16, true));
+            OPTIONAL INT32 optional_int16_leaf (INTEGER(16, true));
+            REQUIRED INT64 mandatory_uint64_leaf (INTEGER(64, false));
+            OPTIONAL INT64 optional_uint64_leaf (INTEGER(64, false));
         }
         ";
     parse_message_type(message_type)
@@ -1280,6 +1292,18 @@ fn add_benches(c: &mut Criterion) {
     let string_list_desc = schema.column(14);
     let mandatory_binary_column_desc = schema.column(15);
     let optional_binary_column_desc = schema.column(16);
+    let mandatory_uint8_column_desc = schema.column(27);
+    let optional_uint8_column_desc = schema.column(28);
+    let mandatory_uint16_column_desc = schema.column(29);
+    let optional_uint16_column_desc = schema.column(30);
+    let mandatory_uint32_column_desc = schema.column(31);
+    let optional_uint32_column_desc = schema.column(32);
+    let mandatory_int8_column_desc = schema.column(33);
+    let optional_int8_column_desc = schema.column(34);
+    let mandatory_int16_column_desc = schema.column(35);
+    let optional_int16_column_desc = schema.column(36);
+    let mandatory_uint64_column_desc = schema.column(37);
+    let optional_uint64_column_desc = schema.column(38);
 
     // primitive / int32 benchmarks
     // =============================
@@ -1294,6 +1318,61 @@ fn add_benches(c: &mut Criterion) {
     );
     group.finish();
 
+    // primitive int32 / logical uint8 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/UInt8Array");
+    bench_primitive::<Int32Type>(
+        &mut group,
+        &mandatory_uint8_column_desc,
+        &optional_uint8_column_desc,
+        0,
+        256,
+    );
+    group.finish();
+
+    // primitive int32 / logical int8 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/Int8Array");
+    bench_primitive::<Int32Type>(
+        &mut group,
+        &mandatory_int8_column_desc,
+        &optional_int8_column_desc,
+        0,
+        128,
+    );
+    group.finish();
+
+    // primitive int32 / logical uint16 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/UInt16Array");
+    bench_primitive::<Int32Type>(
+        &mut group,
+        &mandatory_uint16_column_desc,
+        &optional_uint16_column_desc,
+        0,
+        65536,
+    );
+    group.finish();
+
+    // primitive int32 / logical int16 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/Int16Array");
+    bench_primitive::<Int32Type>(
+        &mut group,
+        &mandatory_int16_column_desc,
+        &optional_int16_column_desc,
+        0,
+        32768,
+    );
+    group.finish();
+
+    // primitive int32 / logical uint32 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/UInt32Array");
+    bench_primitive::<Int32Type>(
+        &mut group,
+        &mandatory_uint32_column_desc,
+        &optional_uint32_column_desc,
+        0,
+        1000,
+    );
+    group.finish();
+
     // primitive / int64 benchmarks
     // =============================
 
@@ -1307,6 +1386,17 @@ fn add_benches(c: &mut Criterion) {
     );
     group.finish();
 
+    // primitive int64 / logical uint64 benchmarks
+    let mut group = c.benchmark_group("arrow_array_reader/UInt64Array");
+    bench_primitive::<Int64Type>(
+        &mut group,
+        &mandatory_uint64_column_desc,
+        &optional_uint64_column_desc,
+        0,
+        1000,
+    );
+    group.finish();
+
     // string benchmarks
     //==============================