| 1 | //! Bit level parsers |
| 2 | //! |
| 3 | |
| 4 | #[cfg (test)] |
| 5 | mod tests; |
| 6 | |
| 7 | use crate::combinator::trace; |
| 8 | use crate::error::{ErrorConvert, Needed, ParserError}; |
| 9 | use crate::lib::std::ops::{AddAssign, Div, Shl, Shr}; |
| 10 | use crate::stream::{Stream, StreamIsPartial, ToUsize}; |
| 11 | use crate::{Parser, Result}; |
| 12 | |
| 13 | /// Number of bits in a byte |
| 14 | const BYTE: usize = u8::BITS as usize; |
| 15 | |
| 16 | /// Converts a byte-level input to a bit-level input |
| 17 | /// |
| 18 | /// See [`bytes`] to convert it back. |
| 19 | /// |
| 20 | /// # Example |
| 21 | /// ```rust |
| 22 | /// # use winnow::prelude::*; |
| 23 | /// # use winnow::Bytes; |
| 24 | /// # use winnow::binary::bits::{bits, take}; |
| 25 | /// # use winnow::error::ContextError; |
| 26 | /// # use winnow::error::ErrMode; |
| 27 | /// type Stream<'i> = &'i Bytes; |
| 28 | /// |
| 29 | /// fn stream(b: &[u8]) -> Stream<'_> { |
| 30 | /// Bytes::new(b) |
| 31 | /// } |
| 32 | /// |
| 33 | /// fn parse(input: &mut Stream<'_>) -> ModalResult<(u8, u8)> { |
| 34 | /// bits::<_, _, ErrMode<ContextError>, _, _>((take(4usize), take(8usize))).parse_next(input) |
| 35 | /// } |
| 36 | /// |
| 37 | /// let input = stream(&[0x12, 0x34, 0xff, 0xff]); |
| 38 | /// |
| 39 | /// let output = parse.parse_peek(input).expect("We take 1.5 bytes and the input is longer than 2 bytes" ); |
| 40 | /// |
| 41 | /// // The first byte is consumed, the second byte is partially consumed and dropped. |
| 42 | /// let remaining = output.0; |
| 43 | /// assert_eq!(remaining, stream(&[0xff, 0xff])); |
| 44 | /// |
| 45 | /// let parsed = output.1; |
| 46 | /// assert_eq!(parsed.0, 0x01); |
| 47 | /// assert_eq!(parsed.1, 0x23); |
| 48 | /// ``` |
| 49 | pub fn bits<Input, Output, BitError, ByteError, ParseNext>( |
| 50 | mut parser: ParseNext, |
| 51 | ) -> impl Parser<Input, Output, ByteError> |
| 52 | where |
| 53 | BitError: ParserError<(Input, usize)> + ErrorConvert<ByteError>, |
| 54 | ByteError: ParserError<Input>, |
| 55 | (Input, usize): Stream, |
| 56 | Input: Stream + Clone, |
| 57 | ParseNext: Parser<(Input, usize), Output, BitError>, |
| 58 | { |
| 59 | trace(name:"bits" , parser:move |input: &mut Input| { |
| 60 | let mut bit_input: (&mut Input, usize) = (input.clone(), 0); |
| 61 | match parser.parse_next(&mut bit_input) { |
| 62 | Ok(result: Output) => { |
| 63 | let (mut rest: &mut Input, offset: usize) = bit_input; |
| 64 | // If the next byte has been partially read, it will be sliced away as well. |
| 65 | // The parser functions might already slice away all fully read bytes. |
| 66 | // That's why `offset / BYTE` isn't necessarily needed at all times. |
| 67 | let remaining_bytes_index: usize = offset / BYTE + if offset % BYTE == 0 { 0 } else { 1 }; |
| 68 | let _ = rest.next_slice(offset:remaining_bytes_index); |
| 69 | *input = rest; |
| 70 | Ok(result) |
| 71 | } |
| 72 | Err(e: BitError) => match e.needed() { |
| 73 | Some(n: Needed) => Err(ParserError::incomplete( |
| 74 | input, |
| 75 | _needed:n.map(|u: NonZero| u.get() / BYTE + 1), |
| 76 | )), |
| 77 | None => Err(ErrorConvert::convert(self:e)), |
| 78 | }, |
| 79 | } |
| 80 | }) |
| 81 | } |
| 82 | |
| 83 | /// Convert a [`bits`] stream back into a byte stream |
| 84 | /// |
| 85 | /// <div class="warning"> |
| 86 | /// |
| 87 | /// **Warning:** A partial byte remaining in the input will be ignored and the given parser will |
| 88 | /// start parsing at the next full byte. |
| 89 | /// |
| 90 | /// </div> |
| 91 | /// |
| 92 | /// # Examples |
| 93 | /// |
| 94 | /// ``` |
| 95 | /// # use winnow::prelude::*; |
| 96 | /// # use winnow::Bytes; |
| 97 | /// # use winnow::token::rest; |
| 98 | /// # use winnow::error::ContextError; |
| 99 | /// # use winnow::error::ErrMode; |
| 100 | /// use winnow::binary::bits::{bits, bytes, take}; |
| 101 | /// |
| 102 | /// type Stream<'i> = &'i Bytes; |
| 103 | /// |
| 104 | /// fn stream(b: &[u8]) -> Stream<'_> { |
| 105 | /// Bytes::new(b) |
| 106 | /// } |
| 107 | /// |
| 108 | /// fn parse<'i>(input: &mut Stream<'i>) -> ModalResult<(u8, u8, &'i [u8])> { |
| 109 | /// bits::<_, _, ErrMode<ContextError>, _, _>(( |
| 110 | /// take(4usize), |
| 111 | /// take(8usize), |
| 112 | /// bytes::<_, _, ErrMode<ContextError>, _, _>(rest) |
| 113 | /// )).parse_next(input) |
| 114 | /// } |
| 115 | /// |
| 116 | /// let input = stream(&[0x12, 0x34, 0xff, 0xff]); |
| 117 | /// |
| 118 | /// assert_eq!(parse.parse_peek(input), Ok(( stream(&[]), (0x01, 0x23, &[0xff, 0xff][..]) ))); |
| 119 | /// ``` |
| 120 | pub fn bytes<Input, Output, ByteError, BitError, ParseNext>( |
| 121 | mut parser: ParseNext, |
| 122 | ) -> impl Parser<(Input, usize), Output, BitError> |
| 123 | where |
| 124 | ByteError: ParserError<Input> + ErrorConvert<BitError>, |
| 125 | BitError: ParserError<(Input, usize)>, |
| 126 | Input: Stream<Token = u8> + Clone, |
| 127 | ParseNext: Parser<Input, Output, ByteError>, |
| 128 | { |
| 129 | trace("bytes" , move |bit_input: &mut (Input, usize)| { |
| 130 | let (mut input, offset) = bit_input.clone(); |
| 131 | let _ = if offset % BYTE != 0 { |
| 132 | input.next_slice(1 + offset / BYTE) |
| 133 | } else { |
| 134 | input.next_slice(offset / BYTE) |
| 135 | }; |
| 136 | match parser.parse_next(&mut input) { |
| 137 | Ok(res) => { |
| 138 | *bit_input = (input, 0); |
| 139 | Ok(res) |
| 140 | } |
| 141 | Err(e) => match e.needed() { |
| 142 | Some(Needed::Unknown) => Err(ParserError::incomplete(bit_input, Needed::Unknown)), |
| 143 | Some(Needed::Size(sz)) => Err(match sz.get().checked_mul(BYTE) { |
| 144 | Some(v) => ParserError::incomplete(bit_input, Needed::new(v)), |
| 145 | None => ParserError::assert( |
| 146 | bit_input, |
| 147 | "overflow in turning needed bytes into needed bits" , |
| 148 | ), |
| 149 | }), |
| 150 | None => Err(ErrorConvert::convert(e)), |
| 151 | }, |
| 152 | } |
| 153 | }) |
| 154 | } |
| 155 | |
| 156 | /// Parse taking `count` bits |
| 157 | /// |
| 158 | /// # Effective Signature |
| 159 | /// |
| 160 | /// Assuming you are parsing a `(&[u8], usize)` bit [Stream]: |
| 161 | /// ```rust |
| 162 | /// # use winnow::prelude::*;; |
| 163 | /// # use winnow::error::ContextError; |
| 164 | /// pub fn take<'i>(count: usize) -> impl Parser<(&'i [u8], usize), u8, ContextError> |
| 165 | /// # { |
| 166 | /// # winnow::binary::bits::take(count) |
| 167 | /// # } |
| 168 | /// ``` |
| 169 | /// |
| 170 | /// # Example |
| 171 | /// ```rust |
| 172 | /// # use winnow::prelude::*; |
| 173 | /// # use winnow::Bytes; |
| 174 | /// # use winnow::error::ContextError; |
| 175 | /// use winnow::binary::bits::take; |
| 176 | /// |
| 177 | /// type Stream<'i> = &'i Bytes; |
| 178 | /// |
| 179 | /// fn stream(b: &[u8]) -> Stream<'_> { |
| 180 | /// Bytes::new(b) |
| 181 | /// } |
| 182 | /// |
| 183 | /// // Consumes 0 bits, returns 0 |
| 184 | /// assert_eq!(take::<_, usize, _, ContextError>(0usize).parse_peek((stream(&[0b00010010]), 0)), Ok(((stream(&[0b00010010]), 0), 0))); |
| 185 | /// |
| 186 | /// // Consumes 4 bits, returns their values and increase offset to 4 |
| 187 | /// assert_eq!(take::<_, usize, _, ContextError>(4usize).parse_peek((stream(&[0b00010010]), 0)), Ok(((stream(&[0b00010010]), 4), 0b00000001))); |
| 188 | /// |
| 189 | /// // Consumes 4 bits, offset is 4, returns their values and increase offset to 0 of next byte |
| 190 | /// assert_eq!(take::<_, usize, _, ContextError>(4usize).parse_peek((stream(&[0b00010010]), 4)), Ok(((stream(&[]), 0), 0b00000010))); |
| 191 | /// |
| 192 | /// // Tries to consume 12 bits but only 8 are available |
| 193 | /// assert!(take::<_, usize, _, ContextError>(12usize).parse_peek((stream(&[0b00010010]), 0)).is_err()); |
| 194 | /// ``` |
| 195 | #[inline (always)] |
| 196 | pub fn take<Input, Output, Count, Error>(count: Count) -> impl Parser<(Input, usize), Output, Error> |
| 197 | where |
| 198 | Input: Stream<Token = u8> + StreamIsPartial + Clone, |
| 199 | Output: From<u8> + AddAssign + Shl<usize, Output = Output> + Shr<usize, Output = Output>, |
| 200 | Count: ToUsize, |
| 201 | Error: ParserError<(Input, usize)>, |
| 202 | { |
| 203 | let count: usize = count.to_usize(); |
| 204 | trace(name:"take" , parser:move |input: &mut (Input, usize)| { |
| 205 | if <Input as StreamIsPartial>::is_partial_supported() { |
| 206 | take_::<_, _, _, true>(input, count) |
| 207 | } else { |
| 208 | take_::<_, _, _, false>(input, count) |
| 209 | } |
| 210 | }) |
| 211 | } |
| 212 | |
| 213 | fn take_<I, O, E: ParserError<(I, usize)>, const PARTIAL: bool>( |
| 214 | bit_input: &mut (I, usize), |
| 215 | count: usize, |
| 216 | ) -> Result<O, E> |
| 217 | where |
| 218 | I: StreamIsPartial, |
| 219 | I: Stream<Token = u8> + Clone, |
| 220 | O: From<u8> + AddAssign + Shl<usize, Output = O> + Shr<usize, Output = O>, |
| 221 | { |
| 222 | if count == 0 { |
| 223 | Ok(0u8.into()) |
| 224 | } else { |
| 225 | let (mut input, bit_offset) = bit_input.clone(); |
| 226 | if input.eof_offset() * BYTE < count + bit_offset { |
| 227 | if PARTIAL && input.is_partial() { |
| 228 | Err(ParserError::incomplete(bit_input, Needed::new(count))) |
| 229 | } else { |
| 230 | Err(ParserError::from_input(&(input, bit_offset))) |
| 231 | } |
| 232 | } else { |
| 233 | let cnt = (count + bit_offset).div(BYTE); |
| 234 | let mut acc: O = 0_u8.into(); |
| 235 | let mut offset: usize = bit_offset; |
| 236 | let mut remaining: usize = count; |
| 237 | let mut end_offset: usize = 0; |
| 238 | |
| 239 | for (_, byte) in input.iter_offsets().take(cnt + 1) { |
| 240 | if remaining == 0 { |
| 241 | break; |
| 242 | } |
| 243 | let val: O = if offset == 0 { |
| 244 | byte.into() |
| 245 | } else { |
| 246 | (byte << offset >> offset).into() |
| 247 | }; |
| 248 | |
| 249 | if remaining < BYTE - offset { |
| 250 | acc += val >> (BYTE - offset - remaining); |
| 251 | end_offset = remaining + offset; |
| 252 | break; |
| 253 | } else { |
| 254 | acc += val << (remaining - (BYTE - offset)); |
| 255 | remaining -= BYTE - offset; |
| 256 | offset = 0; |
| 257 | } |
| 258 | } |
| 259 | let _ = input.next_slice(cnt); |
| 260 | *bit_input = (input, end_offset); |
| 261 | Ok(acc) |
| 262 | } |
| 263 | } |
| 264 | } |
| 265 | |
| 266 | /// Parse taking `count` bits and comparing them to `pattern` |
| 267 | /// |
| 268 | /// # Effective Signature |
| 269 | /// |
| 270 | /// Assuming you are parsing a `(&[u8], usize)` bit [Stream]: |
| 271 | /// ```rust |
| 272 | /// # use winnow::prelude::*;; |
| 273 | /// # use winnow::error::ContextError; |
| 274 | /// pub fn pattern<'i>(pattern: u8, count: usize) -> impl Parser<(&'i [u8], usize), u8, ContextError> |
| 275 | /// # { |
| 276 | /// # winnow::binary::bits::pattern(pattern, count) |
| 277 | /// # } |
| 278 | /// ``` |
| 279 | /// |
| 280 | /// # Example |
| 281 | /// |
| 282 | /// ```rust |
| 283 | /// # use winnow::prelude::*; |
| 284 | /// # use winnow::Bytes; |
| 285 | /// # use winnow::error::ContextError; |
| 286 | /// use winnow::binary::bits::pattern; |
| 287 | /// |
| 288 | /// type Stream<'i> = &'i Bytes; |
| 289 | /// |
| 290 | /// fn stream(b: &[u8]) -> Stream<'_> { |
| 291 | /// Bytes::new(b) |
| 292 | /// } |
| 293 | /// |
| 294 | /// /// Compare the lowest `count` bits of `input` against the lowest `count` bits of `pattern`. |
| 295 | /// /// Return Ok and the matching section of `input` if there's a match. |
| 296 | /// /// Return Err if there's no match. |
| 297 | /// fn parser(bits: u8, count: u8, input: &mut (Stream<'_>, usize)) -> ModalResult<u8> { |
| 298 | /// pattern(bits, count).parse_next(input) |
| 299 | /// } |
| 300 | /// |
| 301 | /// // The lowest 4 bits of 0b00001111 match the lowest 4 bits of 0b11111111. |
| 302 | /// assert_eq!( |
| 303 | /// pattern::<_, usize, _, ContextError>(0b0000_1111, 4usize).parse_peek((stream(&[0b1111_1111]), 0)), |
| 304 | /// Ok(((stream(&[0b1111_1111]), 4), 0b0000_1111)) |
| 305 | /// ); |
| 306 | /// |
| 307 | /// // The lowest bit of 0b00001111 matches the lowest bit of 0b11111111 (both are 1). |
| 308 | /// assert_eq!( |
| 309 | /// pattern::<_, usize, _, ContextError>(0b00000001, 1usize).parse_peek((stream(&[0b11111111]), 0)), |
| 310 | /// Ok(((stream(&[0b11111111]), 1), 0b00000001)) |
| 311 | /// ); |
| 312 | /// |
| 313 | /// // The lowest 2 bits of 0b11111111 and 0b00000001 are different. |
| 314 | /// assert!(pattern::<_, usize, _, ContextError>(0b000000_01, 2usize).parse_peek((stream(&[0b111111_11]), 0)).is_err()); |
| 315 | /// |
| 316 | /// // The lowest 8 bits of 0b11111111 and 0b11111110 are different. |
| 317 | /// assert!(pattern::<_, usize, _, ContextError>(0b11111110, 8usize).parse_peek((stream(&[0b11111111]), 0)).is_err()); |
| 318 | /// ``` |
| 319 | #[inline (always)] |
| 320 | #[doc (alias = "literal" )] |
| 321 | #[doc (alias = "just" )] |
| 322 | #[doc (alias = "tag" )] |
| 323 | pub fn pattern<Input, Output, Count, Error: ParserError<(Input, usize)>>( |
| 324 | pattern: Output, |
| 325 | count: Count, |
| 326 | ) -> impl Parser<(Input, usize), Output, Error> |
| 327 | where |
| 328 | Input: Stream<Token = u8> + StreamIsPartial + Clone, |
| 329 | Count: ToUsize, |
| 330 | Output: From<u8> |
| 331 | + AddAssign |
| 332 | + Shl<usize, Output = Output> |
| 333 | + Shr<usize, Output = Output> |
| 334 | + PartialEq, |
| 335 | { |
| 336 | let count: usize = count.to_usize(); |
| 337 | trace(name:"pattern" , parser:move |input: &mut (Input, usize)| { |
| 338 | let start: Checkpoint<(::Checkpoint, …), …> = input.checkpoint(); |
| 339 | |
| 340 | take(count).parse_next(input).and_then(|o: Output| { |
| 341 | if pattern == o { |
| 342 | Ok(o) |
| 343 | } else { |
| 344 | input.reset(&start); |
| 345 | Err(ParserError::from_input(input)) |
| 346 | } |
| 347 | }) |
| 348 | }) |
| 349 | } |
| 350 | |
| 351 | /// Parses one specific bit as a bool. |
| 352 | /// |
| 353 | /// # Effective Signature |
| 354 | /// |
| 355 | /// Assuming you are parsing a `(&[u8], usize)` bit [Stream]: |
| 356 | /// ```rust |
| 357 | /// # use winnow::prelude::*;; |
| 358 | /// # use winnow::error::ContextError; |
| 359 | /// pub fn bool(input: &mut (&[u8], usize)) -> ModalResult<bool> |
| 360 | /// # { |
| 361 | /// # winnow::binary::bits::bool.parse_next(input) |
| 362 | /// # } |
| 363 | /// ``` |
| 364 | /// |
| 365 | /// # Example |
| 366 | /// |
| 367 | /// ```rust |
| 368 | /// # use winnow::prelude::*; |
| 369 | /// # use winnow::Bytes; |
| 370 | /// # use winnow::error::InputError; |
| 371 | /// use winnow::binary::bits::bool; |
| 372 | /// |
| 373 | /// type Stream<'i> = &'i Bytes; |
| 374 | /// |
| 375 | /// fn stream(b: &[u8]) -> Stream<'_> { |
| 376 | /// Bytes::new(b) |
| 377 | /// } |
| 378 | /// |
| 379 | /// fn parse(input: &mut (Stream<'_>, usize)) -> ModalResult<bool> { |
| 380 | /// bool.parse_next(input) |
| 381 | /// } |
| 382 | /// |
| 383 | /// assert_eq!(parse.parse_peek((stream(&[0b10000000]), 0)), Ok(((stream(&[0b10000000]), 1), true))); |
| 384 | /// assert_eq!(parse.parse_peek((stream(&[0b10000000]), 1)), Ok(((stream(&[0b10000000]), 2), false))); |
| 385 | /// ``` |
| 386 | #[doc (alias = "any" )] |
| 387 | pub fn bool<Input, Error: ParserError<(Input, usize)>>( |
| 388 | input: &mut (Input, usize), |
| 389 | ) -> Result<bool, Error> |
| 390 | where |
| 391 | Input: Stream<Token = u8> + StreamIsPartial + Clone, |
| 392 | { |
| 393 | traceimpl Parser<(Input, usize), …, …>(name:"bool" , |input: &mut (Input, usize)| { |
| 394 | let bit: u32 = take(count:1usize).parse_next(input)?; |
| 395 | Ok(bit != 0) |
| 396 | }) |
| 397 | .parse_next(input) |
| 398 | } |
| 399 | |