cz/src/binio.rs

@@ -1,70 +1,80 @@
-use std::io::{self, Read, Write};
+pub struct BitIo {
-
+    data: Vec<u8>,
-use byteorder::{ReadBytesExt, WriteBytesExt};
-
-/// A simple way to write individual bits to an input implementing [Write].
-pub struct BitWriter<'a, O: Write + WriteBytesExt> {
-    output: &'a mut O,
-
-    current_byte: u8,
-
     byte_offset: usize,
     bit_offset: usize,
 
     byte_size: usize,
 }
 
-impl<'a, O: Write + WriteBytesExt> BitWriter<'a, O> {
+impl BitIo {
-    /// Create a new BitWriter wrapper around something which
+    /// Create a new BitIO reader and writer over some data
-    /// implements [Write].
+    pub fn new(data: Vec<u8>) -> Self {
-    pub fn new(output: &'a mut O) -> Self {
         Self {
-            output,
+            data,
-
-            current_byte: 0,
-
             byte_offset: 0,
             bit_offset: 0,
-
             byte_size: 0,
         }
     }
 
-    /// Get the number of whole bytes written to the stream.
+    /// Get the byte offset of the reader
+    pub fn byte_offset(&self) -> usize {
+        self.byte_offset
+    }
+
+    /// Get the byte size of the reader
     pub fn byte_size(&self) -> usize {
         self.byte_size
     }
 
-    /// Get the bit offset within the current byte.
+    /// Get the current bytes up to `byte_size` in the reader
-    pub fn bit_offset(&self) -> u8 {
+    pub fn bytes(&self) -> Vec<u8> {
-        self.bit_offset as u8
+        self.data[..self.byte_size].to_vec()
     }
 
-    /// Check if the stream is aligned to a byte.
+    /// Read some bits from the buffer
-    pub fn aligned(&self) -> bool {
+    pub fn read_bit(&mut self, bit_len: usize) -> u64 {
-        self.bit_offset() == 0
+        if bit_len > 8 * 8 {
+            panic!("Cannot read more than 64 bits")
         }
 
-    /// Align the writer to the nearest byte by padding with zero bits.
+        if bit_len % 8 == 0 && self.bit_offset == 0 {
-    ///
+            return self.read(bit_len / 8);
-    /// Returns the number of zero bits
+        }
-    pub fn flush(&mut self) -> Result<usize, io::Error> {
+
+        let mut result = 0;
+        for i in 0..bit_len {
+            let bit_value = ((self.data[self.byte_offset] as usize >> self.bit_offset) & 1) as u64;
+            self.bit_offset += 1;
+
+            if self.bit_offset == 8 {
                 self.byte_offset += 1;
-
-        // Write out the current byte unfinished
-        self.output.write_u8(self.current_byte).unwrap();
-        self.current_byte = 0;
                 self.bit_offset = 0;
-
-        Ok(8 - self.bit_offset)
             }
 
-    /// Write some bits to the output.
+            result |= bit_value << i;
+        }
+
+        result
+    }
+
+    /// Read some bytes from the buffer
+    pub fn read(&mut self, byte_len: usize) -> u64 {
+        if byte_len > 8 {
+            panic!("Cannot read more than 8 bytes")
+        }
+
+        let mut padded_slice = [0u8; 8];
+        padded_slice.copy_from_slice(&self.data[self.byte_offset..self.byte_offset + byte_len]);
+        self.byte_offset += byte_len;
+
+        u64::from_le_bytes(padded_slice)
+    }
+
+    /// Write some bits to the buffer
     pub fn write_bit(&mut self, data: u64, bit_len: usize) {
-        if bit_len > 64 {
+        if bit_len > 8 * 8 {
-            panic!("Cannot write more than 64 bits at once.");
+            panic!("Cannot write more than 64 bits");
-        } else if bit_len == 0 {
-            panic!("Must write 1 or more bits.")
         }
 
         if bit_len % 8 == 0 && self.bit_offset == 0 {
@@ -75,115 +85,32 @@ impl<'a, O: Write + WriteBytesExt> BitWriter<'a, O> {
         for i in 0..bit_len {
             let bit_value = (data >> i) & 1;
 
-            self.current_byte &= !(1 << self.bit_offset);
+            self.data[self.byte_offset] &= !(1 << self.bit_offset);
 
-            self.current_byte |= (bit_value << self.bit_offset) as u8;
+            self.data[self.byte_offset] |= (bit_value << self.bit_offset) as u8;
 
             self.bit_offset += 1;
-            if self.bit_offset >= 8 {
-                self.byte_offset += 1;
-                self.bit_offset = 0;
-
-                self.output.write_u8(self.current_byte).unwrap();
-                self.current_byte = 0;
-            }
-        }
-
-        self.byte_size = self.byte_offset + (self.bit_offset + 7) / 8;
-    }
-
-    /// Write some bytes to the output.
-    pub fn write(&mut self, data: u64, byte_len: usize) {
-        if byte_len > 8 {
-            panic!("Cannot write more than 8 bytes at once.")
-        } else if byte_len == 0 {
-            panic!("Must write 1 or more bytes.")
-        }
-
-        self.output
-        .write_all(&data.to_le_bytes()[..byte_len])
-        .unwrap();
-        self.byte_offset += byte_len;
-
-        self.byte_size = self.byte_offset + (self.bit_offset + 7) / 8;
-    }
-}
-
-/// A simple way to read individual bits from an input implementing [Read].
-pub struct BitReader<'a, I: Read + ReadBytesExt> {
-    input: &'a mut I,
-
-    current_byte: Option<u8>,
-
-    byte_offset: usize,
-    bit_offset: usize,
-}
-
-impl<'a, I: Read + ReadBytesExt> BitReader<'a, I> {
-    /// Create a new BitReader wrapper around something which
-    /// implements [Write].
-    pub fn new(input: &'a mut I) -> Self {
-        let first = input.read_u8().unwrap();
-        Self {
-            input,
-
-            current_byte: Some(first),
-
-            byte_offset: 0,
-            bit_offset: 0,
-        }
-    }
-
-    /// Get the number of whole bytes read from the stream.
-    pub fn byte_offset(&self) -> usize {
-        self.byte_offset
-    }
-
-    /// Read some bits from the input.
-    pub fn read_bit(&mut self, bit_len: usize) -> u64 {
-        if bit_len > 64 {
-            panic!("Cannot read more than 64 bits at once.")
-        } else if bit_len == 0 {
-            panic!("Must read 1 or more bits.")
-        }
-
-        if bit_len % 8 == 0 && self.bit_offset == 0 {
-            return self.read(bit_len / 8);
-        }
-
-        let mut result = 0;
-        for i in 0..bit_len {
-            let bit_value = ((self.current_byte.unwrap() as usize >> self.bit_offset) & 1) as u64;
-            self.bit_offset += 1;
-
             if self.bit_offset == 8 {
                 self.byte_offset += 1;
                 self.bit_offset = 0;
-
+            }
-                self.current_byte = Some(self.input.read_u8().unwrap());
         }
 
-            result |= bit_value << i;
+        self.byte_size = self.byte_offset + (self.bit_offset + 7) / 8;
     }
 
-        result
+    pub fn write(&mut self, data: u64, byte_len: usize) {
-    }
-
-    /// Read some bytes from the input.
-    pub fn read(&mut self, byte_len: usize) -> u64 {
         if byte_len > 8 {
-            panic!("Cannot read more than 8 bytes at once.")
+            panic!("Cannot write more than 8 bytes")
-        } else if byte_len == 0 {
-            panic!("Must read 1 or more bytes")
         }
 
-        let mut padded_slice = vec![0u8; byte_len];
+        let mut padded_slice = [0u8; 8];
-        self.input.read_exact(&mut padded_slice).unwrap();
+        padded_slice.copy_from_slice(&data.to_le_bytes());
+
+        self.data[self.byte_offset..self.byte_offset + byte_len]
+            .copy_from_slice(&padded_slice[..byte_len]);
         self.byte_offset += byte_len;
 
-        let extra_length = padded_slice.len() - byte_len;
+        self.byte_size = self.byte_offset + (self.bit_offset + 7) / 8;
-        padded_slice.extend_from_slice(&vec![0u8; extra_length]);
-
-        u64::from_le_bytes(padded_slice.try_into().unwrap())
     }
 }

cz/src/compression.rs

@@ -1,10 +1,10 @@
 use byteorder::{ReadBytesExt, WriteBytesExt, LE};
 use std::{
     collections::HashMap,
-    io::{Cursor, Read, Seek, Write},
+    io::{Read, Seek, Write},
 };
 
-use crate::binio::{BitReader, BitWriter};
+use crate::binio::BitIo;
 use crate::common::CzError;
 
 /// The size of compressed data in each chunk
@@ -163,7 +163,7 @@ pub fn decompress2<T: Seek + ReadBytesExt + Read>(
 }
 
 fn decompress_lzw2(input_data: &[u8], size: usize) -> Vec<u8> {
-    let mut data = Cursor::new(input_data);
+    let mut data = input_data.to_vec();
     let mut dictionary = HashMap::new();
     for i in 0..256 {
         dictionary.insert(i as u64, vec![i as u8]);
@@ -172,15 +172,12 @@ fn decompress_lzw2(input_data: &[u8], size: usize) -> Vec<u8> {
     let mut result = Vec::with_capacity(size);
 
     let data_size = input_data.len();
-    let mut bit_io = BitReader::new(&mut data);
+    data.extend_from_slice(&[0, 0]);
+    let mut bit_io = BitIo::new(data);
     let mut w = dictionary.get(&0).unwrap().clone();
 
     let mut element;
     loop {
-        if bit_io.byte_offset() >= data_size - 1 {
-            break;
-        }
-
         let flag = bit_io.read_bit(1);
         if flag == 0 {
             element = bit_io.read_bit(15);
@@ -188,6 +185,10 @@ fn decompress_lzw2(input_data: &[u8], size: usize) -> Vec<u8> {
             element = bit_io.read_bit(18);
         }
 
+        if bit_io.byte_offset() > data_size {
+            break;
+        }
+
         let mut entry;
         if let Some(x) = dictionary.get(&element) {
             // If the element was already in the dict, get it
@@ -196,7 +197,7 @@ fn decompress_lzw2(input_data: &[u8], size: usize) -> Vec<u8> {
             entry = w.clone();
             entry.push(w[0])
         } else {
-            panic!("Bad compressed element {} at offset {}", element, bit_io.byte_offset())
+            panic!("Bad compressed element: {}", element)
         }
 
         //println!("{}", element);
@@ -362,9 +363,8 @@ fn compress_lzw2(data: &[u8], last: Vec<u8>) -> (usize, Vec<u8>, Vec<u8>) {
         element = last
     }
 
-    let mut output_buf = Vec::new();
+    let mut bit_io = BitIo::new(vec![0u8; 0xF0000]);
-    let mut bit_io = BitWriter::new(&mut output_buf);
+    let write_bit = |bit_io: &mut BitIo, code: u64| {
-    let write_bit = |bit_io: &mut BitWriter<Vec<u8>>, code: u64| {
         if code > 0x7FFF {
             bit_io.write_bit(1, 1);
             bit_io.write_bit(code, 18);
@@ -402,18 +402,13 @@ fn compress_lzw2(data: &[u8], last: Vec<u8>) -> (usize, Vec<u8>, Vec<u8>) {
                 write_bit(&mut bit_io, *dictionary.get(&vec![c]).unwrap());
             }
         }
-
+        return (count, bit_io.bytes(), Vec::new());
-        bit_io.flush().unwrap();
-        return (count, output_buf, Vec::new());
     } else if bit_io.byte_size() < 0x87BDF {
         if !last_element.is_empty() {
             write_bit(&mut bit_io, *dictionary.get(&last_element).unwrap());
         }
-
+        return (count, bit_io.bytes(), Vec::new());
-        bit_io.flush().unwrap();
-        return (count, output_buf, Vec::new());
     }
 
-    bit_io.flush().unwrap();
+    (count, bit_io.bytes(), last_element)
-    (count, output_buf, last_element)
 }

cz/tests/round_trip.rs

@@ -1,114 +0,0 @@
-use std::io::Cursor;
-
-use cz::{common::CzVersion, DynamicCz};
-
-const KODIM03: (u16, u16, &[u8]) = (128, 128, include_bytes!("test_images/kodim03.rgba"));
-const KODIM23: (u16, u16, &[u8]) = (225, 225, include_bytes!("test_images/kodim23.rgba"));
-const SQPTEXT: (u16, u16, &[u8]) = (2048, 810, include_bytes!("test_images/sqp_text.rgba"));
-const DPFLOGO: (u16, u16, &[u8]) = (1123, 639, include_bytes!("test_images/dpf_logo.rgba"));
-
-type TestImage = (u16, u16, &'static [u8]);
-const TEST_IMAGES: &[TestImage] = &[KODIM03, KODIM23, SQPTEXT, DPFLOGO];
-
-#[test]
-fn cz0_round_trip() {
-    for image in TEST_IMAGES {
-        let original_cz = DynamicCz::from_raw(
-            CzVersion::CZ0,
-            image.0,
-            image.1,
-            image.2.to_vec()
-        );
-
-        let mut cz_bytes = Vec::new();
-        original_cz.encode(&mut cz_bytes).unwrap();
-
-        let mut cz_bytes = Cursor::new(cz_bytes);
-        let decoded_cz = DynamicCz::decode(&mut cz_bytes).unwrap();
-
-        assert_eq!(original_cz.as_raw(), decoded_cz.as_raw());
-    }
-}
-
-#[test]
-fn cz1_round_trip() {
-    for image in TEST_IMAGES {
-        let original_cz = DynamicCz::from_raw(
-            CzVersion::CZ1,
-            image.0,
-            image.1,
-            image.2.to_vec()
-        );
-
-        let mut cz_bytes = Vec::new();
-        original_cz.encode(&mut cz_bytes).unwrap();
-
-        let mut cz_bytes = Cursor::new(cz_bytes);
-        let decoded_cz = DynamicCz::decode(&mut cz_bytes).unwrap();
-
-        assert_eq!(original_cz.as_raw(), decoded_cz.as_raw());
-    }
-}
-
-#[test]
-fn cz2_round_trip() {
-    let mut i = 0;
-    for image in TEST_IMAGES {
-        let original_cz = DynamicCz::from_raw(
-            CzVersion::CZ2,
-            image.0,
-            image.1,
-            image.2.to_vec()
-        );
-
-        let mut cz_bytes = Vec::new();
-        original_cz.encode(&mut cz_bytes).unwrap();
-
-        let mut cz_bytes = Cursor::new(cz_bytes);
-        let decoded_cz = DynamicCz::decode(&mut cz_bytes).unwrap();
-
-        assert_eq!(original_cz.as_raw(), decoded_cz.as_raw());
-
-        i += 1;
-    }
-}
-
-#[test]
-fn cz3_round_trip() {
-    for image in TEST_IMAGES {
-        let original_cz = DynamicCz::from_raw(
-            CzVersion::CZ3,
-            image.0,
-            image.1,
-            image.2.to_vec()
-        );
-
-        let mut cz_bytes = Vec::new();
-        original_cz.encode(&mut cz_bytes).unwrap();
-
-        let mut cz_bytes = Cursor::new(cz_bytes);
-        let decoded_cz = DynamicCz::decode(&mut cz_bytes).unwrap();
-
-        assert_eq!(original_cz.as_raw(), decoded_cz.as_raw());
-    }
-}
-
-#[test]
-fn cz4_round_trip() {
-    for image in TEST_IMAGES {
-        let original_cz = DynamicCz::from_raw(
-            CzVersion::CZ4,
-            image.0,
-            image.1,
-            image.2.to_vec()
-        );
-
-        let mut cz_bytes = Vec::new();
-        original_cz.encode(&mut cz_bytes).unwrap();
-
-        let mut cz_bytes = Cursor::new(cz_bytes);
-        let decoded_cz = DynamicCz::decode(&mut cz_bytes).unwrap();
-
-        assert_eq!(original_cz.as_raw(), decoded_cz.as_raw());
-    }
-}