fixed regression in compression accuracy and added tests to avoid it …

…from happening again

src/hash_chain_holder.rs

@@ -377,6 +377,10 @@ impl<H: HashImplementation> HashChainHolderImpl<H> {
 
 #[inline(always)]
 fn prefix_compare(s1: &[u8], s2: &[u8], best_len: u32, max_len: u32) -> u32 {
+    prefix_cmp_odd_size(max_len, s1, s2, best_len)
+    /*
+    not working yet
+
     if max_len == 258 {
         assert!(s1.len() >= 258 && s2.len() >= 258);
 
@@ -408,7 +412,7 @@ fn prefix_compare(s1: &[u8], s2: &[u8], best_len: u32, max_len: u32) -> u32 {
         return 258;
     } else {
         prefix_cmp_odd_size(max_len, s1, s2, best_len)
-    }
+    }*/
 }
 
 #[cold]

src/lib.rs

@@ -44,7 +44,7 @@ pub use preflate_container::{
 use preflate_error::ExitCode;
 pub use preflate_error::{PreflateError, Result};
 
-use std::io::{Cursor, Write};
+use std::io::Write;
 
 pub struct PreflateCompressionContext {
     content: Vec<u8>,

src/process.rs

@@ -366,6 +366,31 @@ fn do_analyze(crc: Option<u32>, compressed_data: &[u8], verify: bool) {
     }
 }
 
+/// verify that levels 1-6 of zlib are compressed without any correction data
+///
+/// Future work: figure out why level 7 and above are not perfect
+#[test]
+fn verify_zlib_perfect_compression() {
+    for i in 1..6 {
+        println!("iteration {}", i);
+        let compressed_data: &[u8] =
+            &read_file(format!("compressed_zlib_level{i}.deflate").as_str());
+
+        let compressed_data = compressed_data;
+        use crate::preflate_parameter_estimator::estimate_preflate_parameters;
+
+        let contents = parse_deflate(compressed_data, 1).unwrap();
+
+        let params = estimate_preflate_parameters(&contents.plain_text, &contents.blocks).unwrap();
+
+        println!("params: {:?}", params);
+
+        // this "encoder" just asserts if anything gets passed to it
+        let mut verify_encoder = crate::statistical_codec::AssertDefaultOnlyEncoder {};
+        encode_mispredictions(&contents, &params, &mut verify_encoder).unwrap();
+    }
+}
+
 #[test]
 fn verify_longmatch() {
     do_analyze(
@@ -379,3 +404,9 @@ fn verify_longmatch() {
 fn verify_zlibng() {
     do_analyze(None, &read_file("compressed_zlibng_level1.deflate"), false);
 }
+
+// this is the deflate stream extracted out of the
+#[test]
+fn verify_png_deflate() {
+    do_analyze(None, &read_file("treegdi.extract.deflate"), false);
+}

src/token_predictor.rs

@@ -11,7 +11,6 @@ use crate::{
     hash_algorithm::HashAlgorithm,
     hash_chain_holder::{new_hash_chain_holder, HashChainHolder, MatchResult},
     huffman_calc::HufftreeBitCalc,
-    huffman_encoding,
     preflate_constants::MIN_MATCH,
     preflate_error::{err_exit_code, AddContext, ExitCode, Result},
     preflate_input::PreflateInput,

tests/end_to_end.rs

@@ -39,6 +39,12 @@ fn end_to_end_compressed() {
     }
 }
 
+#[test]
+fn treepng_deflate() {
+    let v = read_file("treepng.deflate");
+    verifyresult(&v);
+}
+
 #[test]
 fn test_matchnotfound() {
     test_file("sample3.bin");