Skip to main content

ppmd_rust/
encoder_7a.rs

1use std::io::Write;
2
3use crate::{
4    internal::ppmd7::{PPMd7, RangeEncoder7a},
5    Error, PPMD7_MAX_MEM_SIZE, PPMD7_MAX_ORDER, PPMD7_MIN_MEM_SIZE, PPMD7_MIN_ORDER, SYM_END,
6};
7
8/// An encoder to compress data using PPMd7 (PPMdH) with the carryless range coder (PPMd7a).
9pub struct Ppmd7aEncoder<W: Write> {
10    ppmd: PPMd7<RangeEncoder7a<W>>,
11}
12
13impl<W: Write> Ppmd7aEncoder<W> {
14    /// Creates a new [`Ppmd7aEncoder`] which provides a writer over the compressed data.
15    ///
16    /// The given `order` must be between [`PPMD7_MIN_ORDER`] and [`PPMD7_MAX_ORDER`] (inclusive).
17    /// The given `mem_size` must be between [`PPMD7_MIN_MEM_SIZE`] and [`PPMD7_MAX_MEM_SIZE`] (inclusive).
18    pub fn new(writer: W, order: u32, mem_size: u32) -> crate::Result<Self> {
19        if !(PPMD7_MIN_ORDER..=PPMD7_MAX_ORDER).contains(&order)
20            || !(PPMD7_MIN_MEM_SIZE..=PPMD7_MAX_MEM_SIZE).contains(&mem_size)
21        {
22            return Err(Error::InvalidParameter);
23        }
24
25        let ppmd = PPMd7::new_encoder_7a(writer, order, mem_size)?;
26
27        Ok(Self { ppmd })
28    }
29
30    /// Gets a reference to the underlying writer.
31    pub fn get_ref(&self) -> &W {
32        self.ppmd.get_ref()
33    }
34
35    /// Gets a mutable reference to the underlying writer.
36    ///
37    /// Note that mutating the output/input state of the stream may corrupt
38    /// this object, so care must be taken when using this method.
39    pub fn get_mut(&mut self) -> &mut W {
40        self.ppmd.get_mut()
41    }
42
43    /// Returns the inner writer.
44    pub fn into_inner(self) -> W {
45        self.ppmd.into_inner()
46    }
47
48    /// Finishes the encoding process.
49    ///
50    /// Adds an end marker to the data if `with_end_marker` is set to `true`.
51    pub fn finish(mut self, with_end_marker: bool) -> std::io::Result<W> {
52        if with_end_marker {
53            self.ppmd.encode_symbol(SYM_END)?;
54        }
55        self.ppmd.flush_range_encoder()?;
56        Ok(self.into_inner())
57    }
58}
59
60impl<W: Write> Write for Ppmd7aEncoder<W> {
61    fn write(&mut self, buf: &[u8]) -> std::io::Result<usize> {
62        if buf.is_empty() {
63            return Ok(0);
64        }
65
66        for &byte in buf.iter() {
67            self.ppmd.encode_symbol(byte as i32)?;
68        }
69
70        Ok(buf.len())
71    }
72
73    /// Flushes the inner writer. The range coder writes its final bytes in
74    /// [`finish`](Self::finish), so the stream continues after a flush.
75    fn flush(&mut self) -> std::io::Result<()> {
76        self.ppmd.get_mut().flush()
77    }
78}
79
80#[cfg(test)]
81mod test {
82    use std::io::{Read, Write};
83
84    use super::{super::decoder_7a::Ppmd7aDecoder, Ppmd7aEncoder};
85
86    const ORDER: u32 = 8;
87    const MEM_SIZE: u32 = 262144;
88
89    #[test]
90    fn ppmd7aencoder_without_end_marker() {
91        let test_data = include_str!("../tests/fixtures/text/apache2.txt");
92
93        let mut data = Vec::new();
94        {
95            let mut encoder = Ppmd7aEncoder::new(&mut data, ORDER, MEM_SIZE).unwrap();
96            encoder.write_all(test_data.as_bytes()).unwrap();
97            encoder.finish(false).unwrap();
98        }
99
100        let mut decoder = Ppmd7aDecoder::new(data.as_slice(), ORDER, MEM_SIZE).unwrap();
101
102        let mut decoded = vec![0; test_data.len()];
103        decoder.read_exact(&mut decoded).unwrap();
104
105        let decoded_data = String::from_utf8(decoded).unwrap();
106        assert_eq!(decoded_data, test_data);
107    }
108
109    #[test]
110    fn ppmd7aencoder_with_end_marker() {
111        let test_data = include_str!("../tests/fixtures/text/apache2.txt");
112
113        let mut data = Vec::new();
114        {
115            let mut encoder = Ppmd7aEncoder::new(&mut data, ORDER, MEM_SIZE).unwrap();
116            encoder.write_all(test_data.as_bytes()).unwrap();
117            encoder.finish(true).unwrap();
118        }
119
120        let mut decoder = Ppmd7aDecoder::new(data.as_slice(), ORDER, MEM_SIZE).unwrap();
121
122        let mut decoded = Vec::new();
123        decoder.read_to_end(&mut decoded).unwrap();
124
125        let decoded_data = String::from_utf8(decoded).unwrap();
126        assert_eq!(decoded_data, test_data);
127    }
128
129    #[test]
130    fn ppmd7aencoder_flush_between_writes() {
131        let test_data = include_str!("../tests/fixtures/text/apache2.txt");
132        let (first, second) = test_data.as_bytes().split_at(test_data.len() / 2);
133
134        let encode = |flush: bool| {
135            let mut data = Vec::new();
136            {
137                let mut encoder = Ppmd7aEncoder::new(&mut data, ORDER, MEM_SIZE).unwrap();
138                encoder.write_all(first).unwrap();
139                if flush {
140                    encoder.flush().unwrap();
141                }
142                encoder.write_all(second).unwrap();
143                encoder.finish(true).unwrap();
144            }
145            data
146        };
147
148        let data = encode(true);
149        assert_eq!(data, encode(false));
150
151        let mut decoder = Ppmd7aDecoder::new(data.as_slice(), ORDER, MEM_SIZE).unwrap();
152
153        let mut decoded = Vec::new();
154        decoder.read_to_end(&mut decoded).unwrap();
155
156        let decoded_data = String::from_utf8(decoded).unwrap();
157        assert_eq!(decoded_data, test_data);
158    }
159}