| 1 | //! All IO functionality needed for TIFF decoding |
| 2 | |
| 3 | use std::convert::TryFrom; |
| 4 | use std::io::{self, BufRead, BufReader, Read, Seek, Take}; |
| 5 | |
| 6 | /// Byte order of the TIFF file. |
| 7 | #[derive (Clone, Copy, Debug)] |
| 8 | pub enum ByteOrder { |
| 9 | /// little endian byte order |
| 10 | LittleEndian, |
| 11 | /// big endian byte order |
| 12 | BigEndian, |
| 13 | } |
| 14 | |
| 15 | /// Reader that is aware of the byte order. |
| 16 | pub trait EndianReader: Read { |
| 17 | /// Byte order that should be adhered to |
| 18 | fn byte_order(&self) -> ByteOrder; |
| 19 | |
| 20 | /// Reads an u16 |
| 21 | #[inline (always)] |
| 22 | fn read_u16(&mut self) -> Result<u16, io::Error> { |
| 23 | let mut n = [0u8; 2]; |
| 24 | self.read_exact(&mut n)?; |
| 25 | Ok(match self.byte_order() { |
| 26 | ByteOrder::LittleEndian => u16::from_le_bytes(n), |
| 27 | ByteOrder::BigEndian => u16::from_be_bytes(n), |
| 28 | }) |
| 29 | } |
| 30 | |
| 31 | /// Reads an i8 |
| 32 | #[inline (always)] |
| 33 | fn read_i8(&mut self) -> Result<i8, io::Error> { |
| 34 | let mut n = [0u8; 1]; |
| 35 | self.read_exact(&mut n)?; |
| 36 | Ok(match self.byte_order() { |
| 37 | ByteOrder::LittleEndian => i8::from_le_bytes(n), |
| 38 | ByteOrder::BigEndian => i8::from_be_bytes(n), |
| 39 | }) |
| 40 | } |
| 41 | |
| 42 | /// Reads an i16 |
| 43 | #[inline (always)] |
| 44 | fn read_i16(&mut self) -> Result<i16, io::Error> { |
| 45 | let mut n = [0u8; 2]; |
| 46 | self.read_exact(&mut n)?; |
| 47 | Ok(match self.byte_order() { |
| 48 | ByteOrder::LittleEndian => i16::from_le_bytes(n), |
| 49 | ByteOrder::BigEndian => i16::from_be_bytes(n), |
| 50 | }) |
| 51 | } |
| 52 | |
| 53 | /// Reads an u32 |
| 54 | #[inline (always)] |
| 55 | fn read_u32(&mut self) -> Result<u32, io::Error> { |
| 56 | let mut n = [0u8; 4]; |
| 57 | self.read_exact(&mut n)?; |
| 58 | Ok(match self.byte_order() { |
| 59 | ByteOrder::LittleEndian => u32::from_le_bytes(n), |
| 60 | ByteOrder::BigEndian => u32::from_be_bytes(n), |
| 61 | }) |
| 62 | } |
| 63 | |
| 64 | /// Reads an i32 |
| 65 | #[inline (always)] |
| 66 | fn read_i32(&mut self) -> Result<i32, io::Error> { |
| 67 | let mut n = [0u8; 4]; |
| 68 | self.read_exact(&mut n)?; |
| 69 | Ok(match self.byte_order() { |
| 70 | ByteOrder::LittleEndian => i32::from_le_bytes(n), |
| 71 | ByteOrder::BigEndian => i32::from_be_bytes(n), |
| 72 | }) |
| 73 | } |
| 74 | |
| 75 | /// Reads an u64 |
| 76 | #[inline (always)] |
| 77 | fn read_u64(&mut self) -> Result<u64, io::Error> { |
| 78 | let mut n = [0u8; 8]; |
| 79 | self.read_exact(&mut n)?; |
| 80 | Ok(match self.byte_order() { |
| 81 | ByteOrder::LittleEndian => u64::from_le_bytes(n), |
| 82 | ByteOrder::BigEndian => u64::from_be_bytes(n), |
| 83 | }) |
| 84 | } |
| 85 | |
| 86 | /// Reads an i64 |
| 87 | #[inline (always)] |
| 88 | fn read_i64(&mut self) -> Result<i64, io::Error> { |
| 89 | let mut n = [0u8; 8]; |
| 90 | self.read_exact(&mut n)?; |
| 91 | Ok(match self.byte_order() { |
| 92 | ByteOrder::LittleEndian => i64::from_le_bytes(n), |
| 93 | ByteOrder::BigEndian => i64::from_be_bytes(n), |
| 94 | }) |
| 95 | } |
| 96 | |
| 97 | /// Reads an f32 |
| 98 | #[inline (always)] |
| 99 | fn read_f32(&mut self) -> Result<f32, io::Error> { |
| 100 | let mut n = [0u8; 4]; |
| 101 | self.read_exact(&mut n)?; |
| 102 | Ok(f32::from_bits(match self.byte_order() { |
| 103 | ByteOrder::LittleEndian => u32::from_le_bytes(n), |
| 104 | ByteOrder::BigEndian => u32::from_be_bytes(n), |
| 105 | })) |
| 106 | } |
| 107 | |
| 108 | /// Reads an f64 |
| 109 | #[inline (always)] |
| 110 | fn read_f64(&mut self) -> Result<f64, io::Error> { |
| 111 | let mut n = [0u8; 8]; |
| 112 | self.read_exact(&mut n)?; |
| 113 | Ok(f64::from_bits(match self.byte_order() { |
| 114 | ByteOrder::LittleEndian => u64::from_le_bytes(n), |
| 115 | ByteOrder::BigEndian => u64::from_be_bytes(n), |
| 116 | })) |
| 117 | } |
| 118 | } |
| 119 | |
| 120 | /// |
| 121 | /// # READERS |
| 122 | /// |
| 123 | |
| 124 | /// |
| 125 | /// ## Deflate Reader |
| 126 | /// |
| 127 | |
| 128 | pub type DeflateReader<R> = flate2::read::ZlibDecoder<R>; |
| 129 | |
| 130 | /// |
| 131 | /// ## LZW Reader |
| 132 | /// |
| 133 | |
| 134 | /// Reader that decompresses LZW streams |
| 135 | pub struct LZWReader<R: Read> { |
| 136 | reader: BufReader<Take<R>>, |
| 137 | decoder: weezl::decode::Decoder, |
| 138 | } |
| 139 | |
| 140 | impl<R: Read> LZWReader<R> { |
| 141 | /// Wraps a reader |
| 142 | pub fn new(reader: R, compressed_length: usize) -> LZWReader<R> { |
| 143 | Self { |
| 144 | reader: BufReader::with_capacity( |
| 145 | (32 * 1024).min(compressed_length), |
| 146 | inner:reader.take(limit:u64::try_from(compressed_length).unwrap()), |
| 147 | ), |
| 148 | decoder: weezl::decode::Decoder::with_tiff_size_switch(order:weezl::BitOrder::Msb, size:8), |
| 149 | } |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | impl<R: Read> Read for LZWReader<R> { |
| 154 | fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> { |
| 155 | loop { |
| 156 | let result = self.decoder.decode_bytes(self.reader.fill_buf()?, buf); |
| 157 | self.reader.consume(result.consumed_in); |
| 158 | |
| 159 | match result.status { |
| 160 | Ok(weezl::LzwStatus::Ok) => { |
| 161 | if result.consumed_out == 0 { |
| 162 | continue; |
| 163 | } else { |
| 164 | return Ok(result.consumed_out); |
| 165 | } |
| 166 | } |
| 167 | Ok(weezl::LzwStatus::NoProgress) => { |
| 168 | assert_eq!(result.consumed_in, 0); |
| 169 | assert_eq!(result.consumed_out, 0); |
| 170 | assert!(self.reader.buffer().is_empty()); |
| 171 | return Err(io::Error::new( |
| 172 | io::ErrorKind::UnexpectedEof, |
| 173 | "no lzw end code found" , |
| 174 | )); |
| 175 | } |
| 176 | Ok(weezl::LzwStatus::Done) => { |
| 177 | return Ok(result.consumed_out); |
| 178 | } |
| 179 | Err(err) => return Err(io::Error::new(io::ErrorKind::InvalidData, err)), |
| 180 | } |
| 181 | } |
| 182 | } |
| 183 | } |
| 184 | |
| 185 | /// |
| 186 | /// ## PackBits Reader |
| 187 | /// |
| 188 | |
| 189 | enum PackBitsReaderState { |
| 190 | Header, |
| 191 | Literal, |
| 192 | Repeat { value: u8 }, |
| 193 | } |
| 194 | |
| 195 | /// Reader that unpacks Apple's `PackBits` format |
| 196 | pub struct PackBitsReader<R: Read> { |
| 197 | reader: Take<R>, |
| 198 | state: PackBitsReaderState, |
| 199 | count: usize, |
| 200 | } |
| 201 | |
| 202 | impl<R: Read> PackBitsReader<R> { |
| 203 | /// Wraps a reader |
| 204 | pub fn new(reader: R, length: u64) -> Self { |
| 205 | Self { |
| 206 | reader: reader.take(limit:length), |
| 207 | state: PackBitsReaderState::Header, |
| 208 | count: 0, |
| 209 | } |
| 210 | } |
| 211 | } |
| 212 | |
| 213 | impl<R: Read> Read for PackBitsReader<R> { |
| 214 | fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> { |
| 215 | while let PackBitsReaderState::Header = self.state { |
| 216 | if self.reader.limit() == 0 { |
| 217 | return Ok(0); |
| 218 | } |
| 219 | let mut header: [u8; 1] = [0]; |
| 220 | self.reader.read_exact(&mut header)?; |
| 221 | let h = header[0] as i8; |
| 222 | if (-127..=-1).contains(&h) { |
| 223 | let mut data: [u8; 1] = [0]; |
| 224 | self.reader.read_exact(&mut data)?; |
| 225 | self.state = PackBitsReaderState::Repeat { value: data[0] }; |
| 226 | self.count = (1 - h as isize) as usize; |
| 227 | } else if h >= 0 { |
| 228 | self.state = PackBitsReaderState::Literal; |
| 229 | self.count = h as usize + 1; |
| 230 | } else { |
| 231 | // h = -128 is a no-op. |
| 232 | } |
| 233 | } |
| 234 | |
| 235 | let length = buf.len().min(self.count); |
| 236 | let actual = match self.state { |
| 237 | PackBitsReaderState::Literal => self.reader.read(&mut buf[..length])?, |
| 238 | PackBitsReaderState::Repeat { value } => { |
| 239 | for b in &mut buf[..length] { |
| 240 | *b = value; |
| 241 | } |
| 242 | |
| 243 | length |
| 244 | } |
| 245 | PackBitsReaderState::Header => unreachable!(), |
| 246 | }; |
| 247 | |
| 248 | self.count -= actual; |
| 249 | if self.count == 0 { |
| 250 | self.state = PackBitsReaderState::Header; |
| 251 | } |
| 252 | Ok(actual) |
| 253 | } |
| 254 | } |
| 255 | |
| 256 | /// |
| 257 | /// ## SmartReader Reader |
| 258 | /// |
| 259 | |
| 260 | /// Reader that is aware of the byte order. |
| 261 | #[derive (Debug)] |
| 262 | pub struct SmartReader<R> |
| 263 | where |
| 264 | R: Read, |
| 265 | { |
| 266 | reader: R, |
| 267 | pub byte_order: ByteOrder, |
| 268 | } |
| 269 | |
| 270 | impl<R> SmartReader<R> |
| 271 | where |
| 272 | R: Read, |
| 273 | { |
| 274 | /// Wraps a reader |
| 275 | pub fn wrap(reader: R, byte_order: ByteOrder) -> SmartReader<R> { |
| 276 | SmartReader { reader, byte_order } |
| 277 | } |
| 278 | pub fn into_inner(self) -> R { |
| 279 | self.reader |
| 280 | } |
| 281 | } |
| 282 | impl<R: Read + Seek> SmartReader<R> { |
| 283 | pub fn goto_offset(&mut self, offset: u64) -> io::Result<()> { |
| 284 | self.seek(io::SeekFrom::Start(offset)).map(|_| ()) |
| 285 | } |
| 286 | } |
| 287 | |
| 288 | impl<R> EndianReader for SmartReader<R> |
| 289 | where |
| 290 | R: Read, |
| 291 | { |
| 292 | #[inline (always)] |
| 293 | fn byte_order(&self) -> ByteOrder { |
| 294 | self.byte_order |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | impl<R: Read> Read for SmartReader<R> { |
| 299 | #[inline ] |
| 300 | fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> { |
| 301 | self.reader.read(buf) |
| 302 | } |
| 303 | } |
| 304 | |
| 305 | impl<R: Read + Seek> Seek for SmartReader<R> { |
| 306 | #[inline ] |
| 307 | fn seek(&mut self, pos: io::SeekFrom) -> io::Result<u64> { |
| 308 | self.reader.seek(pos) |
| 309 | } |
| 310 | } |
| 311 | |
| 312 | #[cfg (test)] |
| 313 | mod test { |
| 314 | use super::*; |
| 315 | |
| 316 | #[test ] |
| 317 | fn test_packbits() { |
| 318 | let encoded = vec![ |
| 319 | 0xFE, 0xAA, 0x02, 0x80, 0x00, 0x2A, 0xFD, 0xAA, 0x03, 0x80, 0x00, 0x2A, 0x22, 0xF7, |
| 320 | 0xAA, |
| 321 | ]; |
| 322 | let encoded_len = encoded.len(); |
| 323 | |
| 324 | let buff = io::Cursor::new(encoded); |
| 325 | let mut decoder = PackBitsReader::new(buff, encoded_len as u64); |
| 326 | |
| 327 | let mut decoded = Vec::new(); |
| 328 | decoder.read_to_end(&mut decoded).unwrap(); |
| 329 | |
| 330 | let expected = vec![ |
| 331 | 0xAA, 0xAA, 0xAA, 0x80, 0x00, 0x2A, 0xAA, 0xAA, 0xAA, 0xAA, 0x80, 0x00, 0x2A, 0x22, |
| 332 | 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, 0xAA, |
| 333 | ]; |
| 334 | assert_eq!(decoded, expected); |
| 335 | } |
| 336 | } |
| 337 | |