fix(rust): VariableByte not matching the cpp version (#60)

`VariableByte` currently matches none of the schemes in the Cpp impl.
Now it matches `MaskedVByteCodec` or `VByteCodec`.

I don't quite know why VB is implemented differently compared to "the
reference".
The codec appears in
6e232dd#diff-01152c24d242e2ecf9520ee8a8bbb63b7aa622e3d4a530a19f4acfe9f970f7ad

I am not entirely sure if it is 100% correct, but it is better than
before (as in: actually passes tests).

I am sure that there are 2-3 correctness issues in this code.. grumble..
I am also not sure about some of the chocies in Cpp, for example why 10?
.. 🤷🏻‍♂️

Author: CommanderStorm

fuzz/fuzz_targets/common.rs

@@ -11,9 +11,9 @@ pub struct FuzzInput<C> {
 
 impl<C: std::fmt::Debug> std::fmt::Debug for FuzzInput<C> {
     fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
-        f.debug_struct("FuzzInput<C>")
-            .field("data_length", &self.data.len())
+        f.debug_struct("FuzzInput")
             .field("codec", &self.codec)
+            .field("data", &HexSlice(&self.data))
             .finish()
     }
 }
@@ -117,3 +117,26 @@ impl From<CppCodec> for BoxedCppCodec {
         }
     }
 }
+
+pub struct HexSlice<'a>(pub &'a [u32]);
+
+impl<'a> std::fmt::Debug for HexSlice<'a> {
+    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
+        const MAX: usize = 20;
+
+        let total = self.0.len();
+        let shown = total.min(MAX);
+
+        let mut list = f.debug_list();
+
+        for v in &self.0[..shown] {
+            list.entry(&format_args!("{:#010x}", v));
+        }
+
+        if total > MAX {
+            list.entry(&format_args!(".. out of {} total", total));
+        }
+
+        list.finish()
+    }
+}

fuzz/fuzz_targets/rust_compress_oracle.rs

@@ -13,11 +13,6 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
         return;
     }
 
-    // TODO: Behaviour differs
-    if data.codec == RustCodec::VariableByte {
-        return;
-    }
-
     // TODO: To make the encoder not crash ->  Skip inputs smaller than block size
     let block_size = match data.codec {
         RustCodec::FastPFOR256 => 256,
@@ -44,7 +39,7 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
         .expect("Rust compression failed");
 
     // Compress with C++ implementation
-    let cpp_result = match data.codec {
+    let compressed_oracle_from_cpp = match data.codec {
         RustCodec::FastPFOR256 => {
             let mut cpp_codec = cpp::FastPFor256Codec::new();
             cpp_codec
@@ -58,7 +53,7 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
                 .expect("C++ compression failed")
         }
         RustCodec::VariableByte => {
-            let mut cpp_codec = cpp::VByteCodec::new();
+            let mut cpp_codec = cpp::MaskedVByteCodec::new();
             cpp_codec
                 .compress_to_slice(input, &mut cpp_compressed)
                 .expect("C++ compression failed")
@@ -74,13 +69,13 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
     // Compare compressed outputs
     assert_eq!(
         rust_result.len(),
-        cpp_result.len(),
+        compressed_oracle_from_cpp.len(),
         "Compressed length mismatch: Rust={}, C++={}",
         rust_result.len(),
-        cpp_result.len()
+        compressed_oracle_from_cpp.len()
     );
 
-    for (i, (&rust_val, &cpp_val)) in rust_result.iter().zip(cpp_result.iter()).enumerate() {
+    for (i, (&rust_val, &cpp_val)) in rust_result.iter().zip(compressed_oracle_from_cpp.iter()).enumerate() {
         assert_eq!(
             rust_val, cpp_val,
             "Compressed data mismatch at position {}: Rust={}, C++={}",

fuzz/fuzz_targets/rust_decompress_oracle.rs

@@ -1,6 +1,6 @@
 #![no_main]
 
-use fastpfor::{CodecToSlice, cpp, rust};
+use fastpfor::{cpp, rust, CodecToSlice};
 use libfuzzer_sys::fuzz_target;
 mod common;
 use common::*;
@@ -13,11 +13,6 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
         return;
     }
 
-    // TODO: Behaviour differs
-    if data.codec == RustCodec::VariableByte {
-        return;
-    }
-
     // TODO: To make the decoder not crash ->  Skip inputs smaller than block size
     let block_size = match data.codec {
         RustCodec::FastPFOR256 => 256,
@@ -35,7 +30,7 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
 
     // First, compress with C++ implementation to get valid compressed data
     let mut cpp_compressed = vec![0u32; input.len() * 2 + 1024];
-    let compressed_data = match data.codec {
+    let compressed_oracle_from_cpp = match data.codec {
         RustCodec::FastPFOR256 => {
             let mut cpp_codec = cpp::FastPFor256Codec::new();
             cpp_codec
@@ -49,7 +44,7 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
                 .expect("C++ compression failed")
         }
         RustCodec::VariableByte => {
-            let mut cpp_codec = cpp::VByteCodec::new();
+            let mut cpp_codec = cpp::MaskedVByteCodec::new();
             cpp_codec
                 .compress_to_slice(input, &mut cpp_compressed)
                 .expect("C++ compression failed")
@@ -66,7 +61,7 @@ fuzz_target!(|data: FuzzInput<RustCodec>| {
     let mut rust_decompressed = vec![0u32; input.len()];
     let mut rust_codec = rust::Codec::from(data.codec);
     let rust_result = rust_codec
-        .decompress_to_slice(compressed_data, &mut rust_decompressed)
+        .decompress_to_slice(compressed_oracle_from_cpp, &mut rust_decompressed)
         .expect("Rust decompression failed");
 
     // Compare decompressed outputs

src/rust/integer_compression/helpers.rs

@@ -35,15 +35,3 @@ fn bitlen(x: u64) -> i32 {
     }
     64 - clz(x) as i32
 }
-
-/// Extracts 7 bits from `val` at the position specified by `i` (0-indexed, each position is 7 bits).
-/// The result is masked to ensure only 7 bits are returned.
-pub fn extract7bits(i: i32, val: i64) -> u8 {
-    ((val >> (7 * i)) & ((1 << 7) - 1)) as u8
-}
-
-/// Extracts 7 bits from `val` at the position specified by `i` without masking.
-/// Caller must ensure proper masking if needed.
-pub fn extract_7bits_maskless(i: i32, val: i64) -> u8 {
-    (val >> (7 * i)) as u8
-}