test

core/Cargo.toml

@@ -93,6 +93,8 @@ criterion = "0.5.1"
 jni = { version = "0.21", features = ["invocation"] }
 lazy_static = "1.4"
 assertables = "7"
+hex = "0.4.3"
+
 
 [features]
 default = []

core/src/parquet/read/values.rs

@@ -978,3 +978,37 @@ impl Decoder for DictDecoder {
         Encoding::RLE_DICTIONARY
     }
 }
+
+fn copy_i32_to_i16(src: &[u8], dst: &mut [u8], num: usize) {
+    let mut src_offset = 0;
+    let mut dst_offset = 0;
+    for _ in 0..num {
+        unsafe {
+            copy_nonoverlapping(
+                &src[src_offset..] as *const [u8] as *const u8 as *const i16,
+                &mut dst[dst_offset] as *mut u8 as *mut i16,
+                1,
+            );
+        }
+        src_offset += 4; // Parquet stores Int8/Int16 using 4 bytes
+        dst_offset += 2;
+    }
+}
+
+#[cfg(test)]
+mod test {
+    use super::copy_i32_to_i16;
+    use parquet::data_type::AsBytes;
+
+    #[test]
+    fn test_i32_to_i16() {
+        let source =
+            hex::decode("8a0e0000db93ffff1826000034f4ffff300200001d2b0000abe3ffff378dfffff1470000")
+                .unwrap();
+        let expected = hex::decode("8a0edb93182634f430021d2babe3378df147").unwrap();
+        let num = source.len() / 4;
+        let mut dest: Vec<u8> = vec![b' '; num * 2];
+        copy_i32_to_i16(&source.as_bytes(), dest.as_mut_slice(), num);
+        assert_eq!(expected.as_bytes(), dest.as_bytes());
+    }
+}