test: ensure rust and cpp compression matches

Author: Weixing-Zhang

tests/cpp_compat_tests.rs

@@ -5,42 +5,80 @@ use fastpfor::rust::Integer as _;
 use fastpfor::{cpp, rust};
 use std::io::Cursor;
 
+mod test_data;
+use test_data::get_test_cases;
+
 #[test]
 fn test_rust_decompresses_cpp_encoded_data() {
-    let input: Vec<u32> = (0..256).map(|i| (i * 32) ^ (i >> 1)).collect();
+    let codec_cpp = cpp::FastPFor128Codec::new();
+    let mut codec_rs = rust::FastPFOR::new(rust::DEFAULT_PAGE_SIZE, rust::BLOCK_SIZE_128);
 
-    // Buffer for the C++ encoded
-    let mut compressed_buffer = vec![0; input.len()];
+    for input in get_test_cases(512 + rust::BLOCK_SIZE_128 as usize) {
+        // Buffer for the C++ encoded
+        let mut compressed_buffer = vec![0; input.len()];
 
-    // C++ encoding
-    let codec_cpp = cpp::FastPFor128Codec::new();
-    let encoded_cpp = codec_cpp.encode32(&input, &mut compressed_buffer).unwrap();
-    let compressed_len = encoded_cpp.len();
+        // C++ encoding
+        let encoded_cpp = codec_cpp.encode32(&input, &mut compressed_buffer).unwrap();
+        let compressed_len = encoded_cpp.len();
 
-    // C++ decoding
-    let mut decoded_by_cpp = vec![0; input.len()];
-    let decoded_cpp = codec_cpp
-        .decode32(encoded_cpp, &mut decoded_by_cpp)
-        .unwrap();
+        // C++ decoding
+        let mut decoded_by_cpp = vec![0; input.len()];
+        let decoded_cpp = codec_cpp
+            .decode32(encoded_cpp, &mut decoded_by_cpp)
+            .unwrap();
+
+        // Rust decoding
+        let mut input_offset = Cursor::new(0u32);
+        let mut decoded_by_rust = vec![0; input.len()];
+        codec_rs
+            .uncompress(
+                &compressed_buffer,
+                compressed_len as u32,
+                &mut input_offset,
+                &mut decoded_by_rust,
+                &mut Cursor::new(0u32),
+            )
+            .unwrap();
+
+        assert_eq!(
+            decoded_cpp.len(),
+            decoded_by_rust.len(),
+            "Mismatched output lengths"
+        );
+        assert_eq!(decoded_cpp, decoded_by_rust);
+    }
+}
 
-    // Rust decoding
-    let mut input_offset = Cursor::new(0u32);
+#[test]
+fn test_rust_and_cpp_fastpfor32_compression_matches() {
+    let codec_cpp = cpp::FastPFor128Codec::new();
     let mut codec_rs = rust::FastPFOR::new(rust::DEFAULT_PAGE_SIZE, rust::BLOCK_SIZE_128);
-    let mut decoded_by_rust = vec![0; input.len()];
-    codec_rs
-        .uncompress(
-            &compressed_buffer,
-            compressed_len as u32,
-            &mut input_offset,
-            &mut decoded_by_rust,
-            &mut Cursor::new(0u32),
-        )
-        .unwrap();
-
-    assert_eq!(
-        decoded_cpp.len(),
-        decoded_by_rust.len(),
-        "Mismatched output lengths"
-    );
-    assert_eq!(decoded_cpp, decoded_by_rust);
+
+    for input in get_test_cases(512 + rust::BLOCK_SIZE_128 as usize) {
+        // Buffer for the C++ encoded
+        let mut compressed_buffer = vec![0; input.len()];
+
+        // C++ encoding
+        let encoded_cpp = codec_cpp.encode32(&input, &mut compressed_buffer).unwrap();
+        let compressed_len = encoded_cpp.len();
+
+        // Rust encoding
+        let mut input_offset_rs = Cursor::new(0u32);
+        let mut encoded_rs: Vec<u32> = vec![0; input.len()];
+        let mut output_offset_rs = Cursor::new(0u32);
+        codec_rs
+            .compress(
+                &input,
+                input.len() as u32,
+                &mut input_offset_rs,
+                &mut encoded_rs,
+                &mut output_offset_rs,
+            )
+            .unwrap();
+
+        assert_eq!(
+            &encoded_cpp[..compressed_len],
+            &encoded_rs[..compressed_len]
+        );
+    }
 }

tests/test_data.rs

@@ -0,0 +1,45 @@
+#![cfg(all(feature = "rust", feature = "cpp"))]
+
+use rand::rngs::StdRng;
+use rand::Rng as _;
+use rand::SeedableRng as _;
+
+pub fn get_test_cases(n: usize) -> Vec<Vec<u32>> {
+    let mut rng = StdRng::seed_from_u64(14);
+
+    vec![
+        // Zeroes
+        vec![0u32; n],
+        // Same non-zero
+        vec![14u32; n],
+        // Ascending values
+        (0..n).map(|i| i as u32).collect::<Vec<u32>>(),
+        // Descending values
+        (0..n).rev().map(|i| i as u32).collect::<Vec<u32>>(),
+        // Bit-flipping pattern
+        (0..n)
+            .map(|i| ((i as u32) * 32) ^ ((i as u32) >> 1))
+            .collect::<Vec<u32>>(),
+        // Alternating large and small values
+        (0..n)
+            .map(|i| {
+                let ui = i as u32;
+                if ui % 2 == 0 {
+                    1 << 30
+                } else {
+                    3
+                }
+            })
+            .collect::<Vec<u32>>(),
+        // Random u32 values
+        (0..n)
+            .map(|_| rng.random_range(0..(1 << 31)))
+            .collect::<Vec<u32>>(),
+        // Spike in the middle
+        (0..n)
+            .map(|i| if i == n / 2 { u32::MAX } else { 1 })
+            .collect::<Vec<u32>>(),
+        // An empty vector
+        Vec::new(),
+    ]
+}