| 1 | //! Frame a stream of bytes based on a length prefix |
| 2 | //! |
| 3 | //! Many protocols delimit their frames by prefacing frame data with a |
| 4 | //! frame head that specifies the length of the frame. The |
| 5 | //! `length_delimited` module provides utilities for handling the length |
| 6 | //! based framing. This allows the consumer to work with entire frames |
| 7 | //! without having to worry about buffering or other framing logic. |
| 8 | //! |
| 9 | //! # Getting started |
| 10 | //! |
| 11 | //! If implementing a protocol from scratch, using length delimited framing |
| 12 | //! is an easy way to get started. [`LengthDelimitedCodec::new()`] will |
| 13 | //! return a length delimited codec using default configuration values. |
| 14 | //! This can then be used to construct a framer to adapt a full-duplex |
| 15 | //! byte stream into a stream of frames. |
| 16 | //! |
| 17 | //! ``` |
| 18 | //! use tokio::io::{AsyncRead, AsyncWrite}; |
| 19 | //! use tokio_util::codec::{Framed, LengthDelimitedCodec}; |
| 20 | //! |
| 21 | //! fn bind_transport<T: AsyncRead + AsyncWrite>(io: T) |
| 22 | //! -> Framed<T, LengthDelimitedCodec> |
| 23 | //! { |
| 24 | //! Framed::new(io, LengthDelimitedCodec::new()) |
| 25 | //! } |
| 26 | //! # pub fn main() {} |
| 27 | //! ``` |
| 28 | //! |
| 29 | //! The returned transport implements `Sink + Stream` for `BytesMut`. It |
| 30 | //! encodes the frame with a big-endian `u32` header denoting the frame |
| 31 | //! payload length: |
| 32 | //! |
| 33 | //! ```text |
| 34 | //! +----------+--------------------------------+ |
| 35 | //! | len: u32 | frame payload | |
| 36 | //! +----------+--------------------------------+ |
| 37 | //! ``` |
| 38 | //! |
| 39 | //! Specifically, given the following: |
| 40 | //! |
| 41 | //! ``` |
| 42 | //! use tokio::io::{AsyncRead, AsyncWrite}; |
| 43 | //! use tokio_util::codec::{Framed, LengthDelimitedCodec}; |
| 44 | //! |
| 45 | //! use futures::SinkExt; |
| 46 | //! use bytes::Bytes; |
| 47 | //! |
| 48 | //! async fn write_frame<T>(io: T) -> Result<(), Box<dyn std::error::Error>> |
| 49 | //! where |
| 50 | //! T: AsyncRead + AsyncWrite + Unpin, |
| 51 | //! { |
| 52 | //! let mut transport = Framed::new(io, LengthDelimitedCodec::new()); |
| 53 | //! let frame = Bytes::from("hello world" ); |
| 54 | //! |
| 55 | //! transport.send(frame).await?; |
| 56 | //! Ok(()) |
| 57 | //! } |
| 58 | //! ``` |
| 59 | //! |
| 60 | //! The encoded frame will look like this: |
| 61 | //! |
| 62 | //! ```text |
| 63 | //! +---- len: u32 ----+---- data ----+ |
| 64 | //! | \x00\x00\x00\x0b | hello world | |
| 65 | //! +------------------+--------------+ |
| 66 | //! ``` |
| 67 | //! |
| 68 | //! # Decoding |
| 69 | //! |
| 70 | //! [`FramedRead`] adapts an [`AsyncRead`] into a `Stream` of [`BytesMut`], |
| 71 | //! such that each yielded [`BytesMut`] value contains the contents of an |
| 72 | //! entire frame. There are many configuration parameters enabling |
| 73 | //! [`FramedRead`] to handle a wide range of protocols. Here are some |
| 74 | //! examples that will cover the various options at a high level. |
| 75 | //! |
| 76 | //! ## Example 1 |
| 77 | //! |
| 78 | //! The following will parse a `u16` length field at offset 0, including the |
| 79 | //! frame head in the yielded `BytesMut`. |
| 80 | //! |
| 81 | //! ``` |
| 82 | //! # use tokio::io::AsyncRead; |
| 83 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 84 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 85 | //! LengthDelimitedCodec::builder() |
| 86 | //! .length_field_offset(0) // default value |
| 87 | //! .length_field_type::<u16>() |
| 88 | //! .length_adjustment(0) // default value |
| 89 | //! .num_skip(0) // Do not strip frame header |
| 90 | //! .new_read(io); |
| 91 | //! # } |
| 92 | //! # pub fn main() {} |
| 93 | //! ``` |
| 94 | //! |
| 95 | //! The following frame will be decoded as such: |
| 96 | //! |
| 97 | //! ```text |
| 98 | //! INPUT DECODED |
| 99 | //! +-- len ---+--- Payload ---+ +-- len ---+--- Payload ---+ |
| 100 | //! | \x00\x0B | Hello world | --> | \x00\x0B | Hello world | |
| 101 | //! +----------+---------------+ +----------+---------------+ |
| 102 | //! ``` |
| 103 | //! |
| 104 | //! The value of the length field is 11 (`\x0B`) which represents the length |
| 105 | //! of the payload, `hello world`. By default, [`FramedRead`] assumes that |
| 106 | //! the length field represents the number of bytes that **follows** the |
| 107 | //! length field. Thus, the entire frame has a length of 13: 2 bytes for the |
| 108 | //! frame head + 11 bytes for the payload. |
| 109 | //! |
| 110 | //! ## Example 2 |
| 111 | //! |
| 112 | //! The following will parse a `u16` length field at offset 0, omitting the |
| 113 | //! frame head in the yielded `BytesMut`. |
| 114 | //! |
| 115 | //! ``` |
| 116 | //! # use tokio::io::AsyncRead; |
| 117 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 118 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 119 | //! LengthDelimitedCodec::builder() |
| 120 | //! .length_field_offset(0) // default value |
| 121 | //! .length_field_type::<u16>() |
| 122 | //! .length_adjustment(0) // default value |
| 123 | //! // `num_skip` is not needed, the default is to skip |
| 124 | //! .new_read(io); |
| 125 | //! # } |
| 126 | //! # pub fn main() {} |
| 127 | //! ``` |
| 128 | //! |
| 129 | //! The following frame will be decoded as such: |
| 130 | //! |
| 131 | //! ```text |
| 132 | //! INPUT DECODED |
| 133 | //! +-- len ---+--- Payload ---+ +--- Payload ---+ |
| 134 | //! | \x00\x0B | Hello world | --> | Hello world | |
| 135 | //! +----------+---------------+ +---------------+ |
| 136 | //! ``` |
| 137 | //! |
| 138 | //! This is similar to the first example, the only difference is that the |
| 139 | //! frame head is **not** included in the yielded `BytesMut` value. |
| 140 | //! |
| 141 | //! ## Example 3 |
| 142 | //! |
| 143 | //! The following will parse a `u16` length field at offset 0, including the |
| 144 | //! frame head in the yielded `BytesMut`. In this case, the length field |
| 145 | //! **includes** the frame head length. |
| 146 | //! |
| 147 | //! ``` |
| 148 | //! # use tokio::io::AsyncRead; |
| 149 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 150 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 151 | //! LengthDelimitedCodec::builder() |
| 152 | //! .length_field_offset(0) // default value |
| 153 | //! .length_field_type::<u16>() |
| 154 | //! .length_adjustment(-2) // size of head |
| 155 | //! .num_skip(0) |
| 156 | //! .new_read(io); |
| 157 | //! # } |
| 158 | //! # pub fn main() {} |
| 159 | //! ``` |
| 160 | //! |
| 161 | //! The following frame will be decoded as such: |
| 162 | //! |
| 163 | //! ```text |
| 164 | //! INPUT DECODED |
| 165 | //! +-- len ---+--- Payload ---+ +-- len ---+--- Payload ---+ |
| 166 | //! | \x00\x0D | Hello world | --> | \x00\x0D | Hello world | |
| 167 | //! +----------+---------------+ +----------+---------------+ |
| 168 | //! ``` |
| 169 | //! |
| 170 | //! In most cases, the length field represents the length of the payload |
| 171 | //! only, as shown in the previous examples. However, in some protocols the |
| 172 | //! length field represents the length of the whole frame, including the |
| 173 | //! head. In such cases, we specify a negative `length_adjustment` to adjust |
| 174 | //! the value provided in the frame head to represent the payload length. |
| 175 | //! |
| 176 | //! ## Example 4 |
| 177 | //! |
| 178 | //! The following will parse a 3 byte length field at offset 0 in a 5 byte |
| 179 | //! frame head, including the frame head in the yielded `BytesMut`. |
| 180 | //! |
| 181 | //! ``` |
| 182 | //! # use tokio::io::AsyncRead; |
| 183 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 184 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 185 | //! LengthDelimitedCodec::builder() |
| 186 | //! .length_field_offset(0) // default value |
| 187 | //! .length_field_length(3) |
| 188 | //! .length_adjustment(2) // remaining head |
| 189 | //! .num_skip(0) |
| 190 | //! .new_read(io); |
| 191 | //! # } |
| 192 | //! # pub fn main() {} |
| 193 | //! ``` |
| 194 | //! |
| 195 | //! The following frame will be decoded as such: |
| 196 | //! |
| 197 | //! ```text |
| 198 | //! INPUT |
| 199 | //! +---- len -----+- head -+--- Payload ---+ |
| 200 | //! | \x00\x00\x0B | \xCAFE | Hello world | |
| 201 | //! +--------------+--------+---------------+ |
| 202 | //! |
| 203 | //! DECODED |
| 204 | //! +---- len -----+- head -+--- Payload ---+ |
| 205 | //! | \x00\x00\x0B | \xCAFE | Hello world | |
| 206 | //! +--------------+--------+---------------+ |
| 207 | //! ``` |
| 208 | //! |
| 209 | //! A more advanced example that shows a case where there is extra frame |
| 210 | //! head data between the length field and the payload. In such cases, it is |
| 211 | //! usually desirable to include the frame head as part of the yielded |
| 212 | //! `BytesMut`. This lets consumers of the length delimited framer to |
| 213 | //! process the frame head as needed. |
| 214 | //! |
| 215 | //! The positive `length_adjustment` value lets `FramedRead` factor in the |
| 216 | //! additional head into the frame length calculation. |
| 217 | //! |
| 218 | //! ## Example 5 |
| 219 | //! |
| 220 | //! The following will parse a `u16` length field at offset 1 of a 4 byte |
| 221 | //! frame head. The first byte and the length field will be omitted from the |
| 222 | //! yielded `BytesMut`, but the trailing 2 bytes of the frame head will be |
| 223 | //! included. |
| 224 | //! |
| 225 | //! ``` |
| 226 | //! # use tokio::io::AsyncRead; |
| 227 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 228 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 229 | //! LengthDelimitedCodec::builder() |
| 230 | //! .length_field_offset(1) // length of hdr1 |
| 231 | //! .length_field_type::<u16>() |
| 232 | //! .length_adjustment(1) // length of hdr2 |
| 233 | //! .num_skip(3) // length of hdr1 + LEN |
| 234 | //! .new_read(io); |
| 235 | //! # } |
| 236 | //! # pub fn main() {} |
| 237 | //! ``` |
| 238 | //! |
| 239 | //! The following frame will be decoded as such: |
| 240 | //! |
| 241 | //! ```text |
| 242 | //! INPUT |
| 243 | //! +- hdr1 -+-- len ---+- hdr2 -+--- Payload ---+ |
| 244 | //! | \xCA | \x00\x0B | \xFE | Hello world | |
| 245 | //! +--------+----------+--------+---------------+ |
| 246 | //! |
| 247 | //! DECODED |
| 248 | //! +- hdr2 -+--- Payload ---+ |
| 249 | //! | \xFE | Hello world | |
| 250 | //! +--------+---------------+ |
| 251 | //! ``` |
| 252 | //! |
| 253 | //! The length field is situated in the middle of the frame head. In this |
| 254 | //! case, the first byte in the frame head could be a version or some other |
| 255 | //! identifier that is not needed for processing. On the other hand, the |
| 256 | //! second half of the head is needed. |
| 257 | //! |
| 258 | //! `length_field_offset` indicates how many bytes to skip before starting |
| 259 | //! to read the length field. `length_adjustment` is the number of bytes to |
| 260 | //! skip starting at the end of the length field. In this case, it is the |
| 261 | //! second half of the head. |
| 262 | //! |
| 263 | //! ## Example 6 |
| 264 | //! |
| 265 | //! The following will parse a `u16` length field at offset 1 of a 4 byte |
| 266 | //! frame head. The first byte and the length field will be omitted from the |
| 267 | //! yielded `BytesMut`, but the trailing 2 bytes of the frame head will be |
| 268 | //! included. In this case, the length field **includes** the frame head |
| 269 | //! length. |
| 270 | //! |
| 271 | //! ``` |
| 272 | //! # use tokio::io::AsyncRead; |
| 273 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 274 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 275 | //! LengthDelimitedCodec::builder() |
| 276 | //! .length_field_offset(1) // length of hdr1 |
| 277 | //! .length_field_type::<u16>() |
| 278 | //! .length_adjustment(-3) // length of hdr1 + LEN, negative |
| 279 | //! .num_skip(3) |
| 280 | //! .new_read(io); |
| 281 | //! # } |
| 282 | //! ``` |
| 283 | //! |
| 284 | //! The following frame will be decoded as such: |
| 285 | //! |
| 286 | //! ```text |
| 287 | //! INPUT |
| 288 | //! +- hdr1 -+-- len ---+- hdr2 -+--- Payload ---+ |
| 289 | //! | \xCA | \x00\x0F | \xFE | Hello world | |
| 290 | //! +--------+----------+--------+---------------+ |
| 291 | //! |
| 292 | //! DECODED |
| 293 | //! +- hdr2 -+--- Payload ---+ |
| 294 | //! | \xFE | Hello world | |
| 295 | //! +--------+---------------+ |
| 296 | //! ``` |
| 297 | //! |
| 298 | //! Similar to the example above, the difference is that the length field |
| 299 | //! represents the length of the entire frame instead of just the payload. |
| 300 | //! The length of `hdr1` and `len` must be counted in `length_adjustment`. |
| 301 | //! Note that the length of `hdr2` does **not** need to be explicitly set |
| 302 | //! anywhere because it already is factored into the total frame length that |
| 303 | //! is read from the byte stream. |
| 304 | //! |
| 305 | //! ## Example 7 |
| 306 | //! |
| 307 | //! The following will parse a 3 byte length field at offset 0 in a 4 byte |
| 308 | //! frame head, excluding the 4th byte from the yielded `BytesMut`. |
| 309 | //! |
| 310 | //! ``` |
| 311 | //! # use tokio::io::AsyncRead; |
| 312 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 313 | //! # fn bind_read<T: AsyncRead>(io: T) { |
| 314 | //! LengthDelimitedCodec::builder() |
| 315 | //! .length_field_offset(0) // default value |
| 316 | //! .length_field_length(3) |
| 317 | //! .length_adjustment(0) // default value |
| 318 | //! .num_skip(4) // skip the first 4 bytes |
| 319 | //! .new_read(io); |
| 320 | //! # } |
| 321 | //! # pub fn main() {} |
| 322 | //! ``` |
| 323 | //! |
| 324 | //! The following frame will be decoded as such: |
| 325 | //! |
| 326 | //! ```text |
| 327 | //! INPUT DECODED |
| 328 | //! +------- len ------+--- Payload ---+ +--- Payload ---+ |
| 329 | //! | \x00\x00\x0B\xFF | Hello world | => | Hello world | |
| 330 | //! +------------------+---------------+ +---------------+ |
| 331 | //! ``` |
| 332 | //! |
| 333 | //! A simple example where there are unused bytes between the length field |
| 334 | //! and the payload. |
| 335 | //! |
| 336 | //! # Encoding |
| 337 | //! |
| 338 | //! [`FramedWrite`] adapts an [`AsyncWrite`] into a `Sink` of [`BytesMut`], |
| 339 | //! such that each submitted [`BytesMut`] is prefaced by a length field. |
| 340 | //! There are fewer configuration options than [`FramedRead`]. Given |
| 341 | //! protocols that have more complex frame heads, an encoder should probably |
| 342 | //! be written by hand using [`Encoder`]. |
| 343 | //! |
| 344 | //! Here is a simple example, given a `FramedWrite` with the following |
| 345 | //! configuration: |
| 346 | //! |
| 347 | //! ``` |
| 348 | //! # use tokio::io::AsyncWrite; |
| 349 | //! # use tokio_util::codec::LengthDelimitedCodec; |
| 350 | //! # fn write_frame<T: AsyncWrite>(io: T) { |
| 351 | //! # let _ = |
| 352 | //! LengthDelimitedCodec::builder() |
| 353 | //! .length_field_type::<u16>() |
| 354 | //! .new_write(io); |
| 355 | //! # } |
| 356 | //! # pub fn main() {} |
| 357 | //! ``` |
| 358 | //! |
| 359 | //! A payload of `hello world` will be encoded as: |
| 360 | //! |
| 361 | //! ```text |
| 362 | //! +- len: u16 -+---- data ----+ |
| 363 | //! | \x00\x0b | hello world | |
| 364 | //! +------------+--------------+ |
| 365 | //! ``` |
| 366 | //! |
| 367 | //! [`LengthDelimitedCodec::new()`]: method@LengthDelimitedCodec::new |
| 368 | //! [`FramedRead`]: struct@FramedRead |
| 369 | //! [`FramedWrite`]: struct@FramedWrite |
| 370 | //! [`AsyncRead`]: trait@tokio::io::AsyncRead |
| 371 | //! [`AsyncWrite`]: trait@tokio::io::AsyncWrite |
| 372 | //! [`Encoder`]: trait@Encoder |
| 373 | //! [`BytesMut`]: bytes::BytesMut |
| 374 | |
| 375 | use crate::codec::{Decoder, Encoder, Framed, FramedRead, FramedWrite}; |
| 376 | |
| 377 | use tokio::io::{AsyncRead, AsyncWrite}; |
| 378 | |
| 379 | use bytes::{Buf, BufMut, Bytes, BytesMut}; |
| 380 | use std::error::Error as StdError; |
| 381 | use std::io::{self, Cursor}; |
| 382 | use std::{cmp, fmt, mem}; |
| 383 | |
| 384 | /// Configure length delimited `LengthDelimitedCodec`s. |
| 385 | /// |
| 386 | /// `Builder` enables constructing configured length delimited codecs. Note |
| 387 | /// that not all configuration settings apply to both encoding and decoding. See |
| 388 | /// the documentation for specific methods for more detail. |
| 389 | #[derive(Debug, Clone, Copy)] |
| 390 | pub struct Builder { |
| 391 | // Maximum frame length |
| 392 | max_frame_len: usize, |
| 393 | |
| 394 | // Number of bytes representing the field length |
| 395 | length_field_len: usize, |
| 396 | |
| 397 | // Number of bytes in the header before the length field |
| 398 | length_field_offset: usize, |
| 399 | |
| 400 | // Adjust the length specified in the header field by this amount |
| 401 | length_adjustment: isize, |
| 402 | |
| 403 | // Total number of bytes to skip before reading the payload, if not set, |
| 404 | // `length_field_len + length_field_offset` |
| 405 | num_skip: Option<usize>, |
| 406 | |
| 407 | // Length field byte order (little or big endian) |
| 408 | length_field_is_big_endian: bool, |
| 409 | } |
| 410 | |
| 411 | /// An error when the number of bytes read is more than max frame length. |
| 412 | pub struct LengthDelimitedCodecError { |
| 413 | _priv: (), |
| 414 | } |
| 415 | |
| 416 | /// A codec for frames delimited by a frame head specifying their lengths. |
| 417 | /// |
| 418 | /// This allows the consumer to work with entire frames without having to worry |
| 419 | /// about buffering or other framing logic. |
| 420 | /// |
| 421 | /// See [module level] documentation for more detail. |
| 422 | /// |
| 423 | /// [module level]: index.html |
| 424 | #[derive(Debug, Clone)] |
| 425 | pub struct LengthDelimitedCodec { |
| 426 | // Configuration values |
| 427 | builder: Builder, |
| 428 | |
| 429 | // Read state |
| 430 | state: DecodeState, |
| 431 | } |
| 432 | |
| 433 | #[derive(Debug, Clone, Copy)] |
| 434 | enum DecodeState { |
| 435 | Head, |
| 436 | Data(usize), |
| 437 | } |
| 438 | |
| 439 | // ===== impl LengthDelimitedCodec ====== |
| 440 | |
| 441 | impl LengthDelimitedCodec { |
| 442 | /// Creates a new `LengthDelimitedCodec` with the default configuration values. |
| 443 | pub fn new() -> Self { |
| 444 | Self { |
| 445 | builder: Builder::new(), |
| 446 | state: DecodeState::Head, |
| 447 | } |
| 448 | } |
| 449 | |
| 450 | /// Creates a new length delimited codec builder with default configuration |
| 451 | /// values. |
| 452 | pub fn builder() -> Builder { |
| 453 | Builder::new() |
| 454 | } |
| 455 | |
| 456 | /// Returns the current max frame setting |
| 457 | /// |
| 458 | /// This is the largest size this codec will accept from the wire. Larger |
| 459 | /// frames will be rejected. |
| 460 | pub fn max_frame_length(&self) -> usize { |
| 461 | self.builder.max_frame_len |
| 462 | } |
| 463 | |
| 464 | /// Updates the max frame setting. |
| 465 | /// |
| 466 | /// The change takes effect the next time a frame is decoded. In other |
| 467 | /// words, if a frame is currently in process of being decoded with a frame |
| 468 | /// size greater than `val` but less than the max frame length in effect |
| 469 | /// before calling this function, then the frame will be allowed. |
| 470 | pub fn set_max_frame_length(&mut self, val: usize) { |
| 471 | self.builder.max_frame_length(val); |
| 472 | } |
| 473 | |
| 474 | fn decode_head(&mut self, src: &mut BytesMut) -> io::Result<Option<usize>> { |
| 475 | let head_len = self.builder.num_head_bytes(); |
| 476 | let field_len = self.builder.length_field_len; |
| 477 | |
| 478 | if src.len() < head_len { |
| 479 | // Not enough data |
| 480 | return Ok(None); |
| 481 | } |
| 482 | |
| 483 | let n = { |
| 484 | let mut src = Cursor::new(&mut *src); |
| 485 | |
| 486 | // Skip the required bytes |
| 487 | src.advance(self.builder.length_field_offset); |
| 488 | |
| 489 | // match endianness |
| 490 | let n = if self.builder.length_field_is_big_endian { |
| 491 | src.get_uint(field_len) |
| 492 | } else { |
| 493 | src.get_uint_le(field_len) |
| 494 | }; |
| 495 | |
| 496 | if n > self.builder.max_frame_len as u64 { |
| 497 | return Err(io::Error::new( |
| 498 | io::ErrorKind::InvalidData, |
| 499 | LengthDelimitedCodecError { _priv: () }, |
| 500 | )); |
| 501 | } |
| 502 | |
| 503 | // The check above ensures there is no overflow |
| 504 | let n = n as usize; |
| 505 | |
| 506 | // Adjust `n` with bounds checking |
| 507 | let n = if self.builder.length_adjustment < 0 { |
| 508 | n.checked_sub(-self.builder.length_adjustment as usize) |
| 509 | } else { |
| 510 | n.checked_add(self.builder.length_adjustment as usize) |
| 511 | }; |
| 512 | |
| 513 | // Error handling |
| 514 | match n { |
| 515 | Some(n) => n, |
| 516 | None => { |
| 517 | return Err(io::Error::new( |
| 518 | io::ErrorKind::InvalidInput, |
| 519 | "provided length would overflow after adjustment" , |
| 520 | )); |
| 521 | } |
| 522 | } |
| 523 | }; |
| 524 | |
| 525 | src.advance(self.builder.get_num_skip()); |
| 526 | |
| 527 | // Ensure that the buffer has enough space to read the incoming |
| 528 | // payload |
| 529 | src.reserve(n.saturating_sub(src.len())); |
| 530 | |
| 531 | Ok(Some(n)) |
| 532 | } |
| 533 | |
| 534 | fn decode_data(&self, n: usize, src: &mut BytesMut) -> Option<BytesMut> { |
| 535 | // At this point, the buffer has already had the required capacity |
| 536 | // reserved. All there is to do is read. |
| 537 | if src.len() < n { |
| 538 | return None; |
| 539 | } |
| 540 | |
| 541 | Some(src.split_to(n)) |
| 542 | } |
| 543 | } |
| 544 | |
| 545 | impl Decoder for LengthDelimitedCodec { |
| 546 | type Item = BytesMut; |
| 547 | type Error = io::Error; |
| 548 | |
| 549 | fn decode(&mut self, src: &mut BytesMut) -> io::Result<Option<BytesMut>> { |
| 550 | let n = match self.state { |
| 551 | DecodeState::Head => match self.decode_head(src)? { |
| 552 | Some(n) => { |
| 553 | self.state = DecodeState::Data(n); |
| 554 | n |
| 555 | } |
| 556 | None => return Ok(None), |
| 557 | }, |
| 558 | DecodeState::Data(n) => n, |
| 559 | }; |
| 560 | |
| 561 | match self.decode_data(n, src) { |
| 562 | Some(data) => { |
| 563 | // Update the decode state |
| 564 | self.state = DecodeState::Head; |
| 565 | |
| 566 | // Make sure the buffer has enough space to read the next head |
| 567 | src.reserve(self.builder.num_head_bytes().saturating_sub(src.len())); |
| 568 | |
| 569 | Ok(Some(data)) |
| 570 | } |
| 571 | None => Ok(None), |
| 572 | } |
| 573 | } |
| 574 | } |
| 575 | |
| 576 | impl Encoder<Bytes> for LengthDelimitedCodec { |
| 577 | type Error = io::Error; |
| 578 | |
| 579 | fn encode(&mut self, data: Bytes, dst: &mut BytesMut) -> Result<(), io::Error> { |
| 580 | let n = data.len(); |
| 581 | |
| 582 | if n > self.builder.max_frame_len { |
| 583 | return Err(io::Error::new( |
| 584 | io::ErrorKind::InvalidInput, |
| 585 | LengthDelimitedCodecError { _priv: () }, |
| 586 | )); |
| 587 | } |
| 588 | |
| 589 | // Adjust `n` with bounds checking |
| 590 | let n = if self.builder.length_adjustment < 0 { |
| 591 | n.checked_add(-self.builder.length_adjustment as usize) |
| 592 | } else { |
| 593 | n.checked_sub(self.builder.length_adjustment as usize) |
| 594 | }; |
| 595 | |
| 596 | let n = n.ok_or_else(|| { |
| 597 | io::Error::new( |
| 598 | io::ErrorKind::InvalidInput, |
| 599 | "provided length would overflow after adjustment" , |
| 600 | ) |
| 601 | })?; |
| 602 | |
| 603 | // Reserve capacity in the destination buffer to fit the frame and |
| 604 | // length field (plus adjustment). |
| 605 | dst.reserve(self.builder.length_field_len + n); |
| 606 | |
| 607 | if self.builder.length_field_is_big_endian { |
| 608 | dst.put_uint(n as u64, self.builder.length_field_len); |
| 609 | } else { |
| 610 | dst.put_uint_le(n as u64, self.builder.length_field_len); |
| 611 | } |
| 612 | |
| 613 | // Write the frame to the buffer |
| 614 | dst.extend_from_slice(&data[..]); |
| 615 | |
| 616 | Ok(()) |
| 617 | } |
| 618 | } |
| 619 | |
| 620 | impl Default for LengthDelimitedCodec { |
| 621 | fn default() -> Self { |
| 622 | Self::new() |
| 623 | } |
| 624 | } |
| 625 | |
| 626 | // ===== impl Builder ===== |
| 627 | |
| 628 | mod builder { |
| 629 | /// Types that can be used with `Builder::length_field_type`. |
| 630 | pub trait LengthFieldType {} |
| 631 | |
| 632 | impl LengthFieldType for u8 {} |
| 633 | impl LengthFieldType for u16 {} |
| 634 | impl LengthFieldType for u32 {} |
| 635 | impl LengthFieldType for u64 {} |
| 636 | |
| 637 | #[cfg (any( |
| 638 | target_pointer_width = "8" , |
| 639 | target_pointer_width = "16" , |
| 640 | target_pointer_width = "32" , |
| 641 | target_pointer_width = "64" , |
| 642 | ))] |
| 643 | impl LengthFieldType for usize {} |
| 644 | } |
| 645 | |
| 646 | impl Builder { |
| 647 | /// Creates a new length delimited codec builder with default configuration |
| 648 | /// values. |
| 649 | /// |
| 650 | /// # Examples |
| 651 | /// |
| 652 | /// ``` |
| 653 | /// # use tokio::io::AsyncRead; |
| 654 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 655 | /// |
| 656 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 657 | /// LengthDelimitedCodec::builder() |
| 658 | /// .length_field_offset(0) |
| 659 | /// .length_field_type::<u16>() |
| 660 | /// .length_adjustment(0) |
| 661 | /// .num_skip(0) |
| 662 | /// .new_read(io); |
| 663 | /// # } |
| 664 | /// # pub fn main() {} |
| 665 | /// ``` |
| 666 | pub fn new() -> Builder { |
| 667 | Builder { |
| 668 | // Default max frame length of 8MB |
| 669 | max_frame_len: 8 * 1_024 * 1_024, |
| 670 | |
| 671 | // Default byte length of 4 |
| 672 | length_field_len: 4, |
| 673 | |
| 674 | // Default to the header field being at the start of the header. |
| 675 | length_field_offset: 0, |
| 676 | |
| 677 | length_adjustment: 0, |
| 678 | |
| 679 | // Total number of bytes to skip before reading the payload, if not set, |
| 680 | // `length_field_len + length_field_offset` |
| 681 | num_skip: None, |
| 682 | |
| 683 | // Default to reading the length field in network (big) endian. |
| 684 | length_field_is_big_endian: true, |
| 685 | } |
| 686 | } |
| 687 | |
| 688 | /// Read the length field as a big endian integer |
| 689 | /// |
| 690 | /// This is the default setting. |
| 691 | /// |
| 692 | /// This configuration option applies to both encoding and decoding. |
| 693 | /// |
| 694 | /// # Examples |
| 695 | /// |
| 696 | /// ``` |
| 697 | /// # use tokio::io::AsyncRead; |
| 698 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 699 | /// |
| 700 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 701 | /// LengthDelimitedCodec::builder() |
| 702 | /// .big_endian() |
| 703 | /// .new_read(io); |
| 704 | /// # } |
| 705 | /// # pub fn main() {} |
| 706 | /// ``` |
| 707 | pub fn big_endian(&mut self) -> &mut Self { |
| 708 | self.length_field_is_big_endian = true; |
| 709 | self |
| 710 | } |
| 711 | |
| 712 | /// Read the length field as a little endian integer |
| 713 | /// |
| 714 | /// The default setting is big endian. |
| 715 | /// |
| 716 | /// This configuration option applies to both encoding and decoding. |
| 717 | /// |
| 718 | /// # Examples |
| 719 | /// |
| 720 | /// ``` |
| 721 | /// # use tokio::io::AsyncRead; |
| 722 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 723 | /// |
| 724 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 725 | /// LengthDelimitedCodec::builder() |
| 726 | /// .little_endian() |
| 727 | /// .new_read(io); |
| 728 | /// # } |
| 729 | /// # pub fn main() {} |
| 730 | /// ``` |
| 731 | pub fn little_endian(&mut self) -> &mut Self { |
| 732 | self.length_field_is_big_endian = false; |
| 733 | self |
| 734 | } |
| 735 | |
| 736 | /// Read the length field as a native endian integer |
| 737 | /// |
| 738 | /// The default setting is big endian. |
| 739 | /// |
| 740 | /// This configuration option applies to both encoding and decoding. |
| 741 | /// |
| 742 | /// # Examples |
| 743 | /// |
| 744 | /// ``` |
| 745 | /// # use tokio::io::AsyncRead; |
| 746 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 747 | /// |
| 748 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 749 | /// LengthDelimitedCodec::builder() |
| 750 | /// .native_endian() |
| 751 | /// .new_read(io); |
| 752 | /// # } |
| 753 | /// # pub fn main() {} |
| 754 | /// ``` |
| 755 | pub fn native_endian(&mut self) -> &mut Self { |
| 756 | if cfg!(target_endian = "big" ) { |
| 757 | self.big_endian() |
| 758 | } else { |
| 759 | self.little_endian() |
| 760 | } |
| 761 | } |
| 762 | |
| 763 | /// Sets the max frame length in bytes |
| 764 | /// |
| 765 | /// This configuration option applies to both encoding and decoding. The |
| 766 | /// default value is 8MB. |
| 767 | /// |
| 768 | /// When decoding, the length field read from the byte stream is checked |
| 769 | /// against this setting **before** any adjustments are applied. When |
| 770 | /// encoding, the length of the submitted payload is checked against this |
| 771 | /// setting. |
| 772 | /// |
| 773 | /// When frames exceed the max length, an `io::Error` with the custom value |
| 774 | /// of the `LengthDelimitedCodecError` type will be returned. |
| 775 | /// |
| 776 | /// # Examples |
| 777 | /// |
| 778 | /// ``` |
| 779 | /// # use tokio::io::AsyncRead; |
| 780 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 781 | /// |
| 782 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 783 | /// LengthDelimitedCodec::builder() |
| 784 | /// .max_frame_length(8 * 1024 * 1024) |
| 785 | /// .new_read(io); |
| 786 | /// # } |
| 787 | /// # pub fn main() {} |
| 788 | /// ``` |
| 789 | pub fn max_frame_length(&mut self, val: usize) -> &mut Self { |
| 790 | self.max_frame_len = val; |
| 791 | self |
| 792 | } |
| 793 | |
| 794 | /// Sets the unsigned integer type used to represent the length field. |
| 795 | /// |
| 796 | /// The default type is [`u32`]. The max type is [`u64`] (or [`usize`] on |
| 797 | /// 64-bit targets). |
| 798 | /// |
| 799 | /// # Examples |
| 800 | /// |
| 801 | /// ``` |
| 802 | /// # use tokio::io::AsyncRead; |
| 803 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 804 | /// |
| 805 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 806 | /// LengthDelimitedCodec::builder() |
| 807 | /// .length_field_type::<u32>() |
| 808 | /// .new_read(io); |
| 809 | /// # } |
| 810 | /// # pub fn main() {} |
| 811 | /// ``` |
| 812 | /// |
| 813 | /// Unlike [`Builder::length_field_length`], this does not fail at runtime |
| 814 | /// and instead produces a compile error: |
| 815 | /// |
| 816 | /// ```compile_fail |
| 817 | /// # use tokio::io::AsyncRead; |
| 818 | /// # use tokio_util::codec::LengthDelimitedCodec; |
| 819 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 820 | /// LengthDelimitedCodec::builder() |
| 821 | /// .length_field_type::<u128>() |
| 822 | /// .new_read(io); |
| 823 | /// # } |
| 824 | /// # pub fn main() {} |
| 825 | /// ``` |
| 826 | pub fn length_field_type<T: builder::LengthFieldType>(&mut self) -> &mut Self { |
| 827 | self.length_field_length(mem::size_of::<T>()) |
| 828 | } |
| 829 | |
| 830 | /// Sets the number of bytes used to represent the length field |
| 831 | /// |
| 832 | /// The default value is `4`. The max value is `8`. |
| 833 | /// |
| 834 | /// This configuration option applies to both encoding and decoding. |
| 835 | /// |
| 836 | /// # Examples |
| 837 | /// |
| 838 | /// ``` |
| 839 | /// # use tokio::io::AsyncRead; |
| 840 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 841 | /// |
| 842 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 843 | /// LengthDelimitedCodec::builder() |
| 844 | /// .length_field_length(4) |
| 845 | /// .new_read(io); |
| 846 | /// # } |
| 847 | /// # pub fn main() {} |
| 848 | /// ``` |
| 849 | pub fn length_field_length(&mut self, val: usize) -> &mut Self { |
| 850 | assert!(val > 0 && val <= 8, "invalid length field length" ); |
| 851 | self.length_field_len = val; |
| 852 | self |
| 853 | } |
| 854 | |
| 855 | /// Sets the number of bytes in the header before the length field |
| 856 | /// |
| 857 | /// This configuration option only applies to decoding. |
| 858 | /// |
| 859 | /// # Examples |
| 860 | /// |
| 861 | /// ``` |
| 862 | /// # use tokio::io::AsyncRead; |
| 863 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 864 | /// |
| 865 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 866 | /// LengthDelimitedCodec::builder() |
| 867 | /// .length_field_offset(1) |
| 868 | /// .new_read(io); |
| 869 | /// # } |
| 870 | /// # pub fn main() {} |
| 871 | /// ``` |
| 872 | pub fn length_field_offset(&mut self, val: usize) -> &mut Self { |
| 873 | self.length_field_offset = val; |
| 874 | self |
| 875 | } |
| 876 | |
| 877 | /// Delta between the payload length specified in the header and the real |
| 878 | /// payload length |
| 879 | /// |
| 880 | /// # Examples |
| 881 | /// |
| 882 | /// ``` |
| 883 | /// # use tokio::io::AsyncRead; |
| 884 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 885 | /// |
| 886 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 887 | /// LengthDelimitedCodec::builder() |
| 888 | /// .length_adjustment(-2) |
| 889 | /// .new_read(io); |
| 890 | /// # } |
| 891 | /// # pub fn main() {} |
| 892 | /// ``` |
| 893 | pub fn length_adjustment(&mut self, val: isize) -> &mut Self { |
| 894 | self.length_adjustment = val; |
| 895 | self |
| 896 | } |
| 897 | |
| 898 | /// Sets the number of bytes to skip before reading the payload |
| 899 | /// |
| 900 | /// Default value is `length_field_len + length_field_offset` |
| 901 | /// |
| 902 | /// This configuration option only applies to decoding |
| 903 | /// |
| 904 | /// # Examples |
| 905 | /// |
| 906 | /// ``` |
| 907 | /// # use tokio::io::AsyncRead; |
| 908 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 909 | /// |
| 910 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 911 | /// LengthDelimitedCodec::builder() |
| 912 | /// .num_skip(4) |
| 913 | /// .new_read(io); |
| 914 | /// # } |
| 915 | /// # pub fn main() {} |
| 916 | /// ``` |
| 917 | pub fn num_skip(&mut self, val: usize) -> &mut Self { |
| 918 | self.num_skip = Some(val); |
| 919 | self |
| 920 | } |
| 921 | |
| 922 | /// Create a configured length delimited `LengthDelimitedCodec` |
| 923 | /// |
| 924 | /// # Examples |
| 925 | /// |
| 926 | /// ``` |
| 927 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 928 | /// # pub fn main() { |
| 929 | /// LengthDelimitedCodec::builder() |
| 930 | /// .length_field_offset(0) |
| 931 | /// .length_field_type::<u16>() |
| 932 | /// .length_adjustment(0) |
| 933 | /// .num_skip(0) |
| 934 | /// .new_codec(); |
| 935 | /// # } |
| 936 | /// ``` |
| 937 | pub fn new_codec(&self) -> LengthDelimitedCodec { |
| 938 | LengthDelimitedCodec { |
| 939 | builder: *self, |
| 940 | state: DecodeState::Head, |
| 941 | } |
| 942 | } |
| 943 | |
| 944 | /// Create a configured length delimited `FramedRead` |
| 945 | /// |
| 946 | /// # Examples |
| 947 | /// |
| 948 | /// ``` |
| 949 | /// # use tokio::io::AsyncRead; |
| 950 | /// use tokio_util::codec::LengthDelimitedCodec; |
| 951 | /// |
| 952 | /// # fn bind_read<T: AsyncRead>(io: T) { |
| 953 | /// LengthDelimitedCodec::builder() |
| 954 | /// .length_field_offset(0) |
| 955 | /// .length_field_type::<u16>() |
| 956 | /// .length_adjustment(0) |
| 957 | /// .num_skip(0) |
| 958 | /// .new_read(io); |
| 959 | /// # } |
| 960 | /// # pub fn main() {} |
| 961 | /// ``` |
| 962 | pub fn new_read<T>(&self, upstream: T) -> FramedRead<T, LengthDelimitedCodec> |
| 963 | where |
| 964 | T: AsyncRead, |
| 965 | { |
| 966 | FramedRead::new(upstream, self.new_codec()) |
| 967 | } |
| 968 | |
| 969 | /// Create a configured length delimited `FramedWrite` |
| 970 | /// |
| 971 | /// # Examples |
| 972 | /// |
| 973 | /// ``` |
| 974 | /// # use tokio::io::AsyncWrite; |
| 975 | /// # use tokio_util::codec::LengthDelimitedCodec; |
| 976 | /// # fn write_frame<T: AsyncWrite>(io: T) { |
| 977 | /// LengthDelimitedCodec::builder() |
| 978 | /// .length_field_type::<u16>() |
| 979 | /// .new_write(io); |
| 980 | /// # } |
| 981 | /// # pub fn main() {} |
| 982 | /// ``` |
| 983 | pub fn new_write<T>(&self, inner: T) -> FramedWrite<T, LengthDelimitedCodec> |
| 984 | where |
| 985 | T: AsyncWrite, |
| 986 | { |
| 987 | FramedWrite::new(inner, self.new_codec()) |
| 988 | } |
| 989 | |
| 990 | /// Create a configured length delimited `Framed` |
| 991 | /// |
| 992 | /// # Examples |
| 993 | /// |
| 994 | /// ``` |
| 995 | /// # use tokio::io::{AsyncRead, AsyncWrite}; |
| 996 | /// # use tokio_util::codec::LengthDelimitedCodec; |
| 997 | /// # fn write_frame<T: AsyncRead + AsyncWrite>(io: T) { |
| 998 | /// # let _ = |
| 999 | /// LengthDelimitedCodec::builder() |
| 1000 | /// .length_field_type::<u16>() |
| 1001 | /// .new_framed(io); |
| 1002 | /// # } |
| 1003 | /// # pub fn main() {} |
| 1004 | /// ``` |
| 1005 | pub fn new_framed<T>(&self, inner: T) -> Framed<T, LengthDelimitedCodec> |
| 1006 | where |
| 1007 | T: AsyncRead + AsyncWrite, |
| 1008 | { |
| 1009 | Framed::new(inner, self.new_codec()) |
| 1010 | } |
| 1011 | |
| 1012 | fn num_head_bytes(&self) -> usize { |
| 1013 | let num = self.length_field_offset + self.length_field_len; |
| 1014 | cmp::max(num, self.num_skip.unwrap_or(0)) |
| 1015 | } |
| 1016 | |
| 1017 | fn get_num_skip(&self) -> usize { |
| 1018 | self.num_skip |
| 1019 | .unwrap_or(self.length_field_offset + self.length_field_len) |
| 1020 | } |
| 1021 | } |
| 1022 | |
| 1023 | impl Default for Builder { |
| 1024 | fn default() -> Self { |
| 1025 | Self::new() |
| 1026 | } |
| 1027 | } |
| 1028 | |
| 1029 | // ===== impl LengthDelimitedCodecError ===== |
| 1030 | |
| 1031 | impl fmt::Debug for LengthDelimitedCodecError { |
| 1032 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 1033 | f.debug_struct("LengthDelimitedCodecError" ).finish() |
| 1034 | } |
| 1035 | } |
| 1036 | |
| 1037 | impl fmt::Display for LengthDelimitedCodecError { |
| 1038 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 1039 | f.write_str("frame size too big" ) |
| 1040 | } |
| 1041 | } |
| 1042 | |
| 1043 | impl StdError for LengthDelimitedCodecError {} |
| 1044 | |