X-Git-Url: https://git.nihav.org/?a=blobdiff_plain;f=nihav-core%2Fsrc%2Fcompr%2Fdeflate.rs;h=8fc6a13e873f534f93bd35ef575f95012b50ea77;hb=353373a32fcbf842efb6fb11462b5c757b0d382e;hp=13d9ad53c2270082d8295666d6beed53992702a5;hpb=6130c973810971098fc6e4deb3ac71816a5ee093;p=nihav.git diff --git a/nihav-core/src/compr/deflate.rs b/nihav-core/src/compr/deflate.rs index 13d9ad5..8fc6a13 100644 --- a/nihav-core/src/compr/deflate.rs +++ b/nihav-core/src/compr/deflate.rs @@ -71,6 +71,7 @@ //! # } //! ``` +use crate::options::NAOptionDefinitionType; use crate::io::byteio::*; use crate::io::bitreader::*; use crate::io::codebook::*; @@ -171,6 +172,23 @@ impl<'a> CurrentSource<'a> { } Ok(()) } + fn skip_bytes(&mut self, nbytes: usize) -> BitReaderResult<()> { + self.align(); + let cached = usize::from(self.br.bits / 8); + if nbytes <= cached { + self.skip((nbytes as u32) * 8)?; + } else { + self.skip((cached as u32) * 8)?; + self.br.bits = 0; + self.br.bitbuf = 0; + self.br.pos += nbytes - cached; + if self.br.pos > self.src.len() { + return Err(BitReaderError::BitstreamEnd); + } + self.refill(); + } + Ok(()) + } fn align(&mut self) { let b = self.br.bits & 7; if b != 0 { @@ -180,6 +198,9 @@ impl<'a> CurrentSource<'a> { fn left(&self) -> isize { ((self.src.len() as isize) - (self.br.pos as isize)) * 8 + (self.br.bits as isize) } + fn tell(&self) -> usize { + self.br.pos - usize::from(self.br.bits / 8) + } } impl<'a, S: Copy> CodebookReader for CurrentSource<'a> { @@ -355,12 +376,17 @@ impl Inflate { self.full_pos += len; Ok(()) } + ///! Sets custom history for decoding an update for already decoded data. + pub fn set_dict(&mut self, dict: &[u8]) { + let len = dict.len().min(self.buf.len()); + let start = dict.len() - len; + self.buf[..len].copy_from_slice(&dict[start..]); + self.bpos = len; + self.full_pos = len; + } ///! Reports whether decoder has finished decoding the input. pub fn is_finished(&self) -> bool { - match self.state { - InflateState::End => true, - _ => false, - } + matches!(self.state, InflateState::End) } ///! Reports the current amount of bytes output into the destination buffer after the last run. pub fn get_current_output_size(&self) -> usize { self.output_idx } @@ -375,8 +401,15 @@ impl Inflate { ///! ///! [`DecompressError::ShortData`]: ../enum.DecompressError.html#variant.ShortData ///! [`DecompressError::OutputFull`]: ../enum.DecompressError.html#variant.OutputFull - #[allow(clippy::comparison_chain)] pub fn decompress_data(&mut self, src: &[u8], dst: &mut [u8], continue_block: bool) -> DecompressResult { + self.decompress_data_internal(src, dst, continue_block, false) + } + ///! Tries to decompress whole input chunk to the output buffer. + pub fn decompress_block(&mut self, src: &[u8], dst: &mut [u8]) -> DecompressResult { + self.decompress_data_internal(src, dst, false, true) + } + #[allow(clippy::comparison_chain)] + fn decompress_data_internal(&mut self, src: &[u8], dst: &mut [u8], continue_block: bool, do_one_block: bool) -> DecompressResult { if src.is_empty() || dst.is_empty() { return Err(DecompressError::InvalidArgument); } @@ -386,6 +419,9 @@ impl Inflate { self.output_idx = 0; CurrentSource::reinit(src, self.br) }; + if do_one_block { + self.output_idx = 0; + } // check for zlib stream header if let (&InflateState::Start, true) = (&self.state, src.len() > 2) { let cm = src[0] & 0xF; @@ -399,6 +435,9 @@ impl Inflate { match self.state { InflateState::Start | InflateState::BlockStart => { if csrc.left() == 0 { + if do_one_block { + return Ok(self.output_idx); + } self.br = csrc.br; return Err(DecompressError::ShortData); } @@ -602,7 +641,7 @@ impl Inflate { let (lit_lengths, dist_lengths) = self.all_lengths.split_at(self.hlit); let mut lit_codes = [ShortCodebookDesc { code: 0, bits: 0 }; NUM_LITERALS]; - lengths_to_codes(&lit_lengths, &mut lit_codes)?; + lengths_to_codes(lit_lengths, &mut lit_codes)?; let mut cr = ShortCodebookDescReader::new(lit_codes.to_vec()); let ret = Codebook::new(&mut cr, CodebookMode::LSB); if ret.is_err() { return Err(DecompressError::InvalidHeader); } @@ -743,10 +782,220 @@ impl Inflate { } } } + ///! Resets decoder state. + pub fn reset(&mut self) { + self.bpos = 0; + self.output_idx = 0; + self.full_pos = 0; + self.state = InflateState::Start; + } + ///! Decompresses input data into output returning the uncompressed data length. pub fn uncompress(src: &[u8], dst: &mut [u8]) -> DecompressResult { - let mut inflate = Self::new(); - inflate.decompress_data(src, dst, false) + let mut csrc = CurrentSource::new(src, BitReaderState::default()); + if src.len() > 2 { + let cm = src[0] & 0xF; + let cinfo = src[0] >> 4; + let hdr = (u16::from(src[0]) << 8) | u16::from(src[1]); + if cm == 8 && cinfo <= 7 && (hdr % 31) == 0 { + csrc.skip(16).unwrap(); + } + } + + let mut fix_len_cb = None; + + let mut dst_idx = 0; + let mut final_block = false; + while !final_block { + final_block = csrc.read_bool()?; + + let bmode = csrc.read(2)?; + match bmode { + 0 => { + csrc.align(); + let len = csrc.read(16)? as usize; + let inv_len = csrc.read(16)? as usize; + if (len ^ inv_len) != 0xFFFF { + return Err(DecompressError::InvalidHeader); + } + let src_pos = csrc.tell(); + if src_pos + len > src.len() { + return Err(DecompressError::ShortData); + } + if dst_idx + len > dst.len() { + return Err(DecompressError::OutputFull); + } + dst[dst_idx..][..len].copy_from_slice(&src[src_pos..][..len]); + dst_idx += len; + csrc.skip_bytes(len)?; + }, + 1 => { + if fix_len_cb.is_none() { + let mut cr = FixedLenCodeReader {}; + fix_len_cb = Some(Codebook::new(&mut cr, CodebookMode::LSB).unwrap()); + } + if let Some(ref len_cb) = &fix_len_cb { + loop { + let val = csrc.read_cb(len_cb)?; + if val < 256 { + if dst_idx >= dst.len() { + return Err(DecompressError::OutputFull); + } + dst[dst_idx] = val as u8; + dst_idx += 1; + } else if val == 256 { + break; + } else { + let len_idx = (val - 257) as usize; + if len_idx >= LENGTH_BASE.len() { + return Err(DecompressError::InvalidData); + } + let len_bits = LENGTH_ADD_BITS[len_idx]; + let mut length = LENGTH_BASE[len_idx] as usize; + if len_bits > 0 { + length += csrc.read(len_bits)? as usize; + } + let dist_idx = reverse_bits(csrc.read(5)?, 5) as usize; + if dist_idx >= DIST_BASE.len() { + return Err(DecompressError::InvalidData); + } + let dist_bits = DIST_ADD_BITS[dist_idx]; + let mut dist = DIST_BASE[dist_idx] as usize; + if dist_bits > 0 { + dist += csrc.read(dist_bits)? as usize; + } + + if dst_idx + length > dst.len() { + return Err(DecompressError::OutputFull); + } + if dist > dst_idx { + return Err(DecompressError::InvalidData); + } + lz_copy(dst, dst_idx, dist, length); + dst_idx += length; + } + } + } else { + unreachable!(); + } + }, + 2 => { + let hlit = csrc.read(5)? as usize + 257; + if hlit >= 287 { + return Err(DecompressError::InvalidHeader); + } + let hdist = csrc.read(5)? as usize + 1; + let hclen = csrc.read(4)? as usize + 4; + let mut cur_len_idx = 0; + let mut len_lengths = [0; 19]; + let mut all_lengths = [0; NUM_LITERALS + NUM_DISTS]; + + for _ in 0..hclen { + len_lengths[LEN_RECODE[cur_len_idx]] = csrc.read(3)? as u8; + cur_len_idx += 1; + } + let mut len_codes = [ShortCodebookDesc { code: 0, bits: 0 }; 19]; + lengths_to_codes(&len_lengths, &mut len_codes)?; + let mut cr = ShortCodebookDescReader::new(len_codes.to_vec()); + let ret = Codebook::new(&mut cr, CodebookMode::LSB); + if ret.is_err() { + return Err(DecompressError::InvalidHeader); + } + let dyn_len_cb = ret.unwrap(); + + let mut cur_len_idx = 0; + while cur_len_idx < hlit + hdist { + let val = csrc.read_cb(&dyn_len_cb)?; + if val < 16 { + all_lengths[cur_len_idx] = val as u8; + cur_len_idx += 1; + } else { + let mode = (val as usize) - 16; + if mode > 2 { + return Err(DecompressError::InvalidHeader); + } + let base = REPEAT_BASE[mode] as usize; + let bits = REPEAT_BITS[mode]; + let len = base + (csrc.read(bits)? as usize); + if cur_len_idx + len > hlit + hdist { + return Err(DecompressError::InvalidHeader); + } + let rpt = if mode == 0 { + if cur_len_idx == 0 { + return Err(DecompressError::InvalidHeader); + } + all_lengths[cur_len_idx - 1] + } else { + 0 + }; + for _ in 0..len { + all_lengths[cur_len_idx] = rpt; + cur_len_idx += 1; + } + } + } + let (lit_lengths, dist_lengths) = all_lengths.split_at(hlit); + + let mut lit_codes = [ShortCodebookDesc { code: 0, bits: 0 }; NUM_LITERALS]; + lengths_to_codes(lit_lengths, &mut lit_codes)?; + let mut cr = ShortCodebookDescReader::new(lit_codes.to_vec()); + let ret = Codebook::new(&mut cr, CodebookMode::LSB); + if ret.is_err() { return Err(DecompressError::InvalidHeader); } + let dyn_lit_cb = ret.unwrap(); + + let mut dist_codes = [ShortCodebookDesc { code: 0, bits: 0 }; NUM_DISTS]; + lengths_to_codes(&dist_lengths[..hdist], &mut dist_codes)?; + let mut cr = ShortCodebookDescReader::new(dist_codes.to_vec()); + let ret = Codebook::new(&mut cr, CodebookMode::LSB); + if ret.is_err() { return Err(DecompressError::InvalidHeader); } + let dyn_dist_cb = ret.unwrap(); + + loop { + let val = csrc.read_cb(&dyn_lit_cb)?; + if val < 256 { + if dst_idx >= dst.len() { + return Err(DecompressError::OutputFull); + } + dst[dst_idx] = val as u8; + dst_idx += 1; + } else if val == 256 { + break; + } else { + let len_idx = (val - 257) as usize; + if len_idx >= LENGTH_BASE.len() { + return Err(DecompressError::InvalidData); + } + let len_bits = LENGTH_ADD_BITS[len_idx]; + let mut length = LENGTH_BASE[len_idx] as usize; + if len_bits > 0 { + length += csrc.read(len_bits)? as usize; + } + + let dist_idx = csrc.read_cb(&dyn_dist_cb)? as usize; + if dist_idx >= DIST_BASE.len() { + return Err(DecompressError::InvalidData); + } + let dist_bits = DIST_ADD_BITS[dist_idx]; + let mut dist = DIST_BASE[dist_idx] as usize; + if dist_bits > 0 { + dist += csrc.read(dist_bits)? as usize; + } + + if dst_idx + length > dst.len() { + return Err(DecompressError::OutputFull); + } + if dist > dst_idx { + return Err(DecompressError::InvalidData); + } + lz_copy(dst, dst_idx, dist, length); + dst_idx += length; + } + } + }, + _ => return Err(DecompressError::InvalidHeader), + }; + } + Ok(dst_idx) } } @@ -1632,7 +1881,7 @@ impl LZParse for OptimalParser { } dst.reserve(src.len()); - self.trellis.truncate(0); + self.trellis.clear(); self.trellis.reserve(src.len() + 1); for _ in 0..=src.len() { self.trellis.push(TNode::default()); @@ -1700,6 +1949,48 @@ pub enum DeflateMode { Best, } +impl Default for DeflateMode { + fn default() -> Self { DeflateMode::Better } +} + +pub const DEFLATE_MODE_DESCRIPTION: &str = "Deflate compression level."; +///! Deflate option for no compression. +pub const DEFLATE_MODE_NONE: &str = "none"; +///! Deflate option for fast compression. +pub const DEFLATE_MODE_FAST: &str = "fast"; +///! Deflate option for better compression. +pub const DEFLATE_MODE_BETTER: &str = "better"; +///! Deflate option for best compression. +pub const DEFLATE_MODE_BEST: &str = "best"; + +///! All possible option values for deflate compression. +pub const DEFLATE_OPTION_VALUES: NAOptionDefinitionType = NAOptionDefinitionType::String(Some(&[DEFLATE_MODE_NONE, DEFLATE_MODE_FAST, DEFLATE_MODE_BETTER, DEFLATE_MODE_BEST])); + +impl std::str::FromStr for DeflateMode { + type Err = (); + + fn from_str(s: &str) -> Result { + match s { + DEFLATE_MODE_NONE => Ok(DeflateMode::NoCompr), + DEFLATE_MODE_FAST => Ok(DeflateMode::Fast), + DEFLATE_MODE_BETTER => Ok(DeflateMode::Better), + DEFLATE_MODE_BEST => Ok(DeflateMode::Best), + _ => Err(()), + } + } +} + +impl ToString for DeflateMode { + fn to_string(&self) -> String { + match *self { + DeflateMode::NoCompr => DEFLATE_MODE_NONE.to_string(), + DeflateMode::Fast => DEFLATE_MODE_FAST.to_string(), + DeflateMode::Better => DEFLATE_MODE_BETTER.to_string(), + DeflateMode::Best => DEFLATE_MODE_BEST.to_string(), + } + } +} + #[derive(Clone,Copy,Debug,PartialEq)] enum Mode { Copy, @@ -1718,17 +2009,17 @@ pub struct Deflate { sum1: u32, sum2: u32, zlib_mode: bool, - parser: Box, + parser: Box, } impl Deflate { ///! Creates a new instance of `Deflate`. pub fn new(mode: DeflateMode) -> Self { let (mode, parser) = match mode { - DeflateMode::NoCompr => (Mode::Copy, Box::new(NoParser{}) as Box), - DeflateMode::Fast => (Mode::Fixed, Box::new(GreedyParser{}) as Box), - DeflateMode::Better => (Mode::Dynamic, Box::new(LazyParser{}) as Box), - DeflateMode::Best => (Mode::Dynamic, Box::new(OptimalParser::new()) as Box), + DeflateMode::NoCompr => (Mode::Copy, Box::new(NoParser{}) as Box), + DeflateMode::Fast => (Mode::Fixed, Box::new(GreedyParser{}) as Box), + DeflateMode::Better => (Mode::Dynamic, Box::new(LazyParser{}) as Box), + DeflateMode::Best => (Mode::Dynamic, Box::new(OptimalParser::new()) as Box), }; Self { mode, parser, @@ -1793,6 +2084,20 @@ impl Deflate { self.write_zlib_footer(wr); } } + ///! Tells the encoder to compress the data it received and flush it. + pub fn compress_flush(&mut self, wr: &mut DeflateWriter) { + if self.ssize > 0 { + self.do_block(wr, false); + } + if (wr.bits & 7) != 0 { + // write zero-length copy block for byte-alignment + wr.write(0, 1); + wr.write(0, 2); + wr.align(); + wr.write(0, 16); + wr.write(0xFFFF, 16); + } + } fn do_block(&mut self, wr: &mut DeflateWriter, final_block: bool) { const CRC_BASE: u32 = 65521; for &b in self.srcbuf[..self.ssize].iter() { @@ -1819,7 +2124,7 @@ impl Deflate { Mode::Fixed => { wr.write(final_block as u16, 1); wr.write(1, 2); - self.tokens.truncate(0); + self.tokens.clear(); self.parser.parse(&self.srcbuf[..self.ssize], &mut self.tokens); let mut codes = CodeHuff::new(true); codes.make_codes(&self.tokens); @@ -1830,7 +2135,7 @@ impl Deflate { Mode::Dynamic => { wr.write(final_block as u16, 1); wr.write(2, 2); - self.tokens.truncate(0); + self.tokens.clear(); self.parser.parse(&self.srcbuf[..self.ssize], &mut self.tokens); let mut codes = CodeHuff::new(false); codes.make_codes(&self.tokens);