Fix LZSS length base (#3291)

Author: raviqqe

cmd/decode/src/main.scm

@@ -62,7 +62,7 @@
             (decompressor-push! decompressor y)
             y)
           (begin
-            (decompressor-set-length! decompressor y)
+            (decompressor-set-length! decompressor (+ y 1))
             (decompressor-set-offset! decompressor (read-u8))
             (decompressor-read decompressor)))))
     (else

cmd/minimal/Cargo.lock

@@ -1695,7 +1695,7 @@
 
     (define window-size 256)
     (define minimum-match 2) ; exclusive
-    (define maximum-match 127) ; inclusive
+    (define maximum-match 128) ; inclusive
 
     ;; Compressor
 
@@ -1758,7 +1758,7 @@
              (n (cdr match)))
        (if (> n minimum-match)
         (begin
-         (write-u8 (+ 1 (* 2 n)))
+         (write-u8 (+ 1 (* 2 (- n 1))))
          (write-u8 (- back (car match) 1))
          (compressor-pop! compressor n))
         (write-u8 (* 2 (compressor-pop! compressor 1)))))))

compile.scm

@@ -1,8 +1,6 @@
-use crate::ring_buffer::RingBuffer;
+use crate::{MAX_LENGTH, ring_buffer::RingBuffer};
 
 const MIN_LENGTH: usize = 2;
-/// The maximum match length.
-pub const MAX_LENGTH: usize = (u8::MAX / 2) as _;
 
 /// LZSS compression iterator.
 pub struct LzssCompressionIterator<const B: usize, I: Iterator<Item = u8>> {
@@ -83,7 +81,7 @@ impl<const B: usize, I: Iterator<Item = u8>> Iterator for LzssCompressionIterato
                 self.ahead -= m;
                 self.next = Some(n as _);
 
-                (m as u8) << 1 | 1
+                ((m - 1) << 1) as u8 | 1
             } else {
                 self.next()? << 1
             }
@@ -180,7 +178,7 @@ mod tests {
         assert_eq!(
             LzssCompressionIterator::<BUFFER_SIZE, _>::new([42, 42, 42, 42].into_iter())
                 .collect::<Vec<_>>(),
-            [84, 7, 0]
+            [84, 5, 0]
         );
     }
 
@@ -191,7 +189,7 @@ mod tests {
                 [42, 42, 42, 42, 7, 7, 7, 127, 127, 127, 127, 127].into_iter()
             )
             .collect::<Vec<_>>(),
-            [84, 7, 0, 14, 14, 14, 254, 9, 0]
+            [84, 5, 0, 14, 14, 14, 254, 7, 0]
         );
     }
 
@@ -200,7 +198,7 @@ mod tests {
         assert_eq!(
             LzssCompressionIterator::<BUFFER_SIZE, _>::new([0, 0, 0].into_iter())
                 .collect::<Vec<_>>(),
-            [7, 0]
+            [5, 0]
         );
     }
 
@@ -233,8 +231,7 @@ mod tests {
                 .chain(&chunk)
                 .copied()
                 .map(|x| x << 1)
-                // TODO Set the length base to 0.
-                .chain([u8::MAX, u8::MAX, 0])
+                .chain([u8::MAX, u8::MAX])
                 .collect::<Vec<_>>()
         );
     }

lzss/src/compress.rs

@@ -38,7 +38,7 @@ impl<const W: usize, I: Iterator<Item = u8>> Iterator for LzssDecompressionItera
                 self.buffer.push(y);
                 Some(y)
             } else {
-                self.length = y;
+                self.length = y + 1;
                 self.offset = self.iterator.next()?;
 
                 self.next()
@@ -58,7 +58,7 @@ mod tests {
     #[test]
     fn repetition() {
         assert_eq!(
-            LzssDecompressionIterator::<8, _>::new([2, 4, 6, 8, 11, 3].into_iter())
+            LzssDecompressionIterator::<8, _>::new([2, 4, 6, 8, 9, 3].into_iter())
                 .collect::<Vec<_>>(),
             [1, 2, 3, 4, 1, 2, 3, 4, 1]
         );
@@ -67,7 +67,7 @@ mod tests {
     #[test]
     fn repetitions() {
         assert_eq!(
-            LzssDecompressionIterator::<8, _>::new([2, 4, 6, 8, 11, 3, 10, 12, 7, 1].into_iter())
+            LzssDecompressionIterator::<8, _>::new([2, 4, 6, 8, 9, 3, 10, 12, 5, 1].into_iter())
                 .collect::<Vec<_>>(),
             [1, 2, 3, 4, 1, 2, 3, 4, 1, 5, 6, 5, 6, 5]
         );
@@ -77,7 +77,7 @@ mod tests {
     fn max_length() {
         assert_eq!(
             LzssDecompressionIterator::<1, _>::new(
-                [84, (MAX_LENGTH as u8) << 1 | 1, 0].into_iter()
+                [84, (MAX_LENGTH - 1 << 1) as u8 | 1, 0].into_iter()
             )
             .collect::<Vec<_>>(),
             repeat(42).take(MAX_LENGTH + 1).collect::<Vec<_>>()

lzss/src/decompress.rs

@@ -11,12 +11,14 @@ mod compress;
 mod decompress;
 mod ring_buffer;
 
-pub use self::compress::MAX_LENGTH;
 use self::{compress::LzssCompressionIterator, decompress::LzssDecompressionIterator};
 
 /// The maximum window size.
 pub const MAX_WINDOW_SIZE: usize = 1 << 8;
 
+/// The maximum match length.
+pub const MAX_LENGTH: usize = 1 << 7;
+
 /// LZSS compression for 7-bit bytes.
 pub trait Lzss {
     /// Compresses bytes.
@@ -27,8 +29,8 @@ pub trait Lzss {
 }
 
 impl<I: IntoIterator<Item = u8>> Lzss for I {
-    fn compress<const W: usize>(self) -> impl Iterator<Item = u8> {
-        LzssCompressionIterator::<W, _>::new(self.into_iter())
+    fn compress<const B: usize>(self) -> impl Iterator<Item = u8> {
+        LzssCompressionIterator::<B, _>::new(self.into_iter())
     }
 
     fn decompress<const W: usize>(self) -> impl Iterator<Item = u8> {
@@ -134,7 +136,7 @@ mod tests {
     #[test]
     fn max_length() {
         const WINDOW_SIZE: usize = 1;
-        let data = repeat(42).take(256).collect::<Vec<_>>();
+        let data = repeat(42).take(MAX_LENGTH + 1).collect::<Vec<_>>();
 
         assert_eq!(
             data.iter()
@@ -149,7 +151,7 @@ mod tests {
     #[test]
     fn max_offset() {
         const WINDOW_SIZE: usize = 128;
-        let data = (0..128).chain(0..128).collect::<Vec<_>>();
+        let data = repeat(0..128).take(2).flatten().collect::<Vec<_>>();
 
         assert_eq!(
             data.iter()
@@ -172,4 +174,16 @@ mod tests {
             .collect::<Vec<_>>()
             == data
     }
+
+    #[quickcheck]
+    fn random_max(data: Vec<u8>) -> bool {
+        let data = data.into_iter().map(|x| x >> 1).collect::<Vec<_>>();
+
+        data.iter()
+            .copied()
+            .compress::<{ MAX_WINDOW_SIZE + MAX_LENGTH }>()
+            .decompress::<MAX_WINDOW_SIZE>()
+            .collect::<Vec<_>>()
+            == data
+    }
 }

lzss/src/lib.rs

@@ -2293,6 +2293,8 @@
       - get 0
     - get 2
     - get 1
+    - constant 1
+    - call 2 #f ||
     - call 2 #f ||
     - set 0
     - get 2

snapshots/cmd/decode/src/main.md

@@ -19183,7 +19183,7 @@
   - list
     - define
     - maximum-match
-    - 127
+    - 128
   - list
     - define-record-type
     - compressor
@@ -19474,7 +19474,10 @@
                 - list
                   - -
                   - 2
-                  - n
+                  - list
+                    - -
+                    - n
+                    - 1
             - list
               - write-u8
               - list