| 1 |
#[cfg(any(feature = "alloc", test))] |
| 2 |
use alloc::string::String; |
| 3 |
use core::fmt; |
| 4 |
#[cfg(any(feature = "std", test))] |
| 5 |
use std::error; |
| 6 |
|
| 7 |
#[cfg(any(feature = "alloc", test))] |
| 8 |
use crate::engine::general_purpose::STANDARD; |
| 9 |
use crate::engine::{Config, Engine}; |
| 10 |
use crate::PAD_BYTE; |
| 11 |
|
| 12 |
/// Encode arbitrary octets as base64 using the [`STANDARD` engine](STANDARD). |
| 13 |
/// |
| 14 |
/// See [Engine::encode]. |
| 15 |
#[allow(unused)] |
| 16 |
#[deprecated(since = "0.21.0", note = "Use Engine::encode")] |
| 17 |
#[cfg(any(feature = "alloc", test))] |
| 18 |
pub fn encode<T: AsRef<[u8]>>(input: T) -> String { |
| 19 |
STANDARD.encode(input) |
| 20 |
} |
| 21 |
|
| 22 |
///Encode arbitrary octets as base64 using the provided `Engine` into a new `String`. |
| 23 |
/// |
| 24 |
/// See [Engine::encode]. |
| 25 |
#[allow(unused)] |
| 26 |
#[deprecated(since = "0.21.0", note = "Use Engine::encode")] |
| 27 |
#[cfg(any(feature = "alloc", test))] |
| 28 |
pub fn encode_engine<E: Engine, T: AsRef<[u8]>>(input: T, engine: &E) -> String { |
| 29 |
engine.encode(input) |
| 30 |
} |
| 31 |
|
| 32 |
///Encode arbitrary octets as base64 into a supplied `String`. |
| 33 |
/// |
| 34 |
/// See [Engine::encode_string]. |
| 35 |
#[allow(unused)] |
| 36 |
#[deprecated(since = "0.21.0", note = "Use Engine::encode_string")] |
| 37 |
#[cfg(any(feature = "alloc", test))] |
| 38 |
pub fn encode_engine_string<E: Engine, T: AsRef<[u8]>>( |
| 39 |
input: T, |
| 40 |
output_buf: &mut String, |
| 41 |
engine: &E, |
| 42 |
) { |
| 43 |
engine.encode_string(input, output_buf) |
| 44 |
} |
| 45 |
|
| 46 |
/// Encode arbitrary octets as base64 into a supplied slice. |
| 47 |
/// |
| 48 |
/// See [Engine::encode_slice]. |
| 49 |
#[allow(unused)] |
| 50 |
#[deprecated(since = "0.21.0", note = "Use Engine::encode_slice")] |
| 51 |
pub fn encode_engine_slice<E: Engine, T: AsRef<[u8]>>( |
| 52 |
input: T, |
| 53 |
output_buf: &mut [u8], |
| 54 |
engine: &E, |
| 55 |
) -> Result<usize, EncodeSliceError> { |
| 56 |
engine.encode_slice(input, output_buf) |
| 57 |
} |
| 58 |
|
| 59 |
/// B64-encode and pad (if configured). |
| 60 |
/// |
| 61 |
/// This helper exists to avoid recalculating encoded_size, which is relatively expensive on short |
| 62 |
/// inputs. |
| 63 |
/// |
| 64 |
/// `encoded_size` is the encoded size calculated for `input`. |
| 65 |
/// |
| 66 |
/// `output` must be of size `encoded_size`. |
| 67 |
/// |
| 68 |
/// All bytes in `output` will be written to since it is exactly the size of the output. |
| 69 |
pub(crate) fn encode_with_padding<E: Engine + ?Sized>( |
| 70 |
input: &[u8], |
| 71 |
output: &mut [u8], |
| 72 |
engine: &E, |
| 73 |
expected_encoded_size: usize, |
| 74 |
) { |
| 75 |
debug_assert_eq!(expected_encoded_size, output.len()); |
| 76 |
|
| 77 |
let b64_bytes_written = engine.internal_encode(input, output); |
| 78 |
|
| 79 |
let padding_bytes = if engine.config().encode_padding() { |
| 80 |
add_padding(b64_bytes_written, &mut output[b64_bytes_written..]) |
| 81 |
} else { |
| 82 |
0 |
| 83 |
}; |
| 84 |
|
| 85 |
let encoded_bytes = b64_bytes_written |
| 86 |
.checked_add(padding_bytes) |
| 87 |
.expect("usize overflow when calculating b64 length"); |
| 88 |
|
| 89 |
debug_assert_eq!(expected_encoded_size, encoded_bytes); |
| 90 |
} |
| 91 |
|
| 92 |
/// Calculate the base64 encoded length for a given input length, optionally including any |
| 93 |
/// appropriate padding bytes. |
| 94 |
/// |
| 95 |
/// Returns `None` if the encoded length can't be represented in `usize`. This will happen for |
| 96 |
/// input lengths in approximately the top quarter of the range of `usize`. |
| 97 |
pub const fn encoded_len(bytes_len: usize, padding: bool) -> Option<usize> { |
| 98 |
let rem = bytes_len % 3; |
| 99 |
|
| 100 |
let complete_input_chunks = bytes_len / 3; |
| 101 |
// `?` is disallowed in const, and `let Some(_) = _ else` requires 1.65.0, whereas this |
| 102 |
// messier syntax works on 1.48 |
| 103 |
let complete_chunk_output = |
| 104 |
if let Some(complete_chunk_output) = complete_input_chunks.checked_mul(4) { |
| 105 |
complete_chunk_output |
| 106 |
} else { |
| 107 |
return None; |
| 108 |
}; |
| 109 |
|
| 110 |
if rem > 0 { |
| 111 |
if padding { |
| 112 |
complete_chunk_output.checked_add(4) |
| 113 |
} else { |
| 114 |
let encoded_rem = match rem { |
| 115 |
1 => 2, |
| 116 |
// only other possible remainder is 2 |
| 117 |
// can't use a separate _ => unreachable!() in const fns in ancient rust versions |
| 118 |
_ => 3, |
| 119 |
}; |
| 120 |
complete_chunk_output.checked_add(encoded_rem) |
| 121 |
} |
| 122 |
} else { |
| 123 |
Some(complete_chunk_output) |
| 124 |
} |
| 125 |
} |
| 126 |
|
| 127 |
/// Write padding characters. |
| 128 |
/// `unpadded_output_len` is the size of the unpadded but base64 encoded data. |
| 129 |
/// `output` is the slice where padding should be written, of length at least 2. |
| 130 |
/// |
| 131 |
/// Returns the number of padding bytes written. |
| 132 |
pub(crate) fn add_padding(unpadded_output_len: usize, output: &mut [u8]) -> usize { |
| 133 |
let pad_bytes = (4 - (unpadded_output_len % 4)) % 4; |
| 134 |
// for just a couple bytes, this has better performance than using |
| 135 |
// .fill(), or iterating over mutable refs, which call memset() |
| 136 |
#[allow(clippy::needless_range_loop)] |
| 137 |
for i in 0..pad_bytes { |
| 138 |
output[i] = PAD_BYTE; |
| 139 |
} |
| 140 |
|
| 141 |
pad_bytes |
| 142 |
} |
| 143 |
|
| 144 |
/// Errors that can occur while encoding into a slice. |
| 145 |
#[derive(Clone, Debug, PartialEq, Eq)] |
| 146 |
pub enum EncodeSliceError { |
| 147 |
/// The provided slice is too small. |
| 148 |
OutputSliceTooSmall, |
| 149 |
} |
| 150 |
|
| 151 |
impl fmt::Display for EncodeSliceError { |
| 152 |
fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { |
| 153 |
match self { |
| 154 |
Self::OutputSliceTooSmall => write!(f, "Output slice too small"), |
| 155 |
} |
| 156 |
} |
| 157 |
} |
| 158 |
|
| 159 |
#[cfg(any(feature = "std", test))] |
| 160 |
impl error::Error for EncodeSliceError {} |
| 161 |
|
| 162 |
#[cfg(test)] |
| 163 |
mod tests { |
| 164 |
use super::*; |
| 165 |
|
| 166 |
use crate::{ |
| 167 |
alphabet, |
| 168 |
engine::general_purpose::{GeneralPurpose, NO_PAD, STANDARD}, |
| 169 |
tests::{assert_encode_sanity, random_config, random_engine}, |
| 170 |
}; |
| 171 |
use rand::{ |
| 172 |
distributions::{Distribution, Uniform}, |
| 173 |
Rng, SeedableRng, |
| 174 |
}; |
| 175 |
use std::str; |
| 176 |
|
| 177 |
const URL_SAFE_NO_PAD_ENGINE: GeneralPurpose = GeneralPurpose::new(&alphabet::URL_SAFE, NO_PAD); |
| 178 |
|
| 179 |
#[test] |
| 180 |
fn encoded_size_correct_standard() { |
| 181 |
assert_encoded_length(0, 0, &STANDARD, true); |
| 182 |
|
| 183 |
assert_encoded_length(1, 4, &STANDARD, true); |
| 184 |
assert_encoded_length(2, 4, &STANDARD, true); |
| 185 |
assert_encoded_length(3, 4, &STANDARD, true); |
| 186 |
|
| 187 |
assert_encoded_length(4, 8, &STANDARD, true); |
| 188 |
assert_encoded_length(5, 8, &STANDARD, true); |
| 189 |
assert_encoded_length(6, 8, &STANDARD, true); |
| 190 |
|
| 191 |
assert_encoded_length(7, 12, &STANDARD, true); |
| 192 |
assert_encoded_length(8, 12, &STANDARD, true); |
| 193 |
assert_encoded_length(9, 12, &STANDARD, true); |
| 194 |
|
| 195 |
assert_encoded_length(54, 72, &STANDARD, true); |
| 196 |
|
| 197 |
assert_encoded_length(55, 76, &STANDARD, true); |
| 198 |
assert_encoded_length(56, 76, &STANDARD, true); |
| 199 |
assert_encoded_length(57, 76, &STANDARD, true); |
| 200 |
|
| 201 |
assert_encoded_length(58, 80, &STANDARD, true); |
| 202 |
} |
| 203 |
|
| 204 |
#[test] |
| 205 |
fn encoded_size_correct_no_pad() { |
| 206 |
assert_encoded_length(0, 0, &URL_SAFE_NO_PAD_ENGINE, false); |
| 207 |
|
| 208 |
assert_encoded_length(1, 2, &URL_SAFE_NO_PAD_ENGINE, false); |
| 209 |
assert_encoded_length(2, 3, &URL_SAFE_NO_PAD_ENGINE, false); |
| 210 |
assert_encoded_length(3, 4, &URL_SAFE_NO_PAD_ENGINE, false); |
| 211 |
|
| 212 |
assert_encoded_length(4, 6, &URL_SAFE_NO_PAD_ENGINE, false); |
| 213 |
assert_encoded_length(5, 7, &URL_SAFE_NO_PAD_ENGINE, false); |
| 214 |
assert_encoded_length(6, 8, &URL_SAFE_NO_PAD_ENGINE, false); |
| 215 |
|
| 216 |
assert_encoded_length(7, 10, &URL_SAFE_NO_PAD_ENGINE, false); |
| 217 |
assert_encoded_length(8, 11, &URL_SAFE_NO_PAD_ENGINE, false); |
| 218 |
assert_encoded_length(9, 12, &URL_SAFE_NO_PAD_ENGINE, false); |
| 219 |
|
| 220 |
assert_encoded_length(54, 72, &URL_SAFE_NO_PAD_ENGINE, false); |
| 221 |
|
| 222 |
assert_encoded_length(55, 74, &URL_SAFE_NO_PAD_ENGINE, false); |
| 223 |
assert_encoded_length(56, 75, &URL_SAFE_NO_PAD_ENGINE, false); |
| 224 |
assert_encoded_length(57, 76, &URL_SAFE_NO_PAD_ENGINE, false); |
| 225 |
|
| 226 |
assert_encoded_length(58, 78, &URL_SAFE_NO_PAD_ENGINE, false); |
| 227 |
} |
| 228 |
|
| 229 |
#[test] |
| 230 |
fn encoded_size_overflow() { |
| 231 |
assert_eq!(None, encoded_len(usize::MAX, true)); |
| 232 |
} |
| 233 |
|
| 234 |
#[test] |
| 235 |
fn encode_engine_string_into_nonempty_buffer_doesnt_clobber_prefix() { |
| 236 |
let mut orig_data = Vec::new(); |
| 237 |
let mut prefix = String::new(); |
| 238 |
let mut encoded_data_no_prefix = String::new(); |
| 239 |
let mut encoded_data_with_prefix = String::new(); |
| 240 |
let mut decoded = Vec::new(); |
| 241 |
|
| 242 |
let prefix_len_range = Uniform::new(0, 1000); |
| 243 |
let input_len_range = Uniform::new(0, 1000); |
| 244 |
|
| 245 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 246 |
|
| 247 |
for _ in 0..10_000 { |
| 248 |
orig_data.clear(); |
| 249 |
prefix.clear(); |
| 250 |
encoded_data_no_prefix.clear(); |
| 251 |
encoded_data_with_prefix.clear(); |
| 252 |
decoded.clear(); |
| 253 |
|
| 254 |
let input_len = input_len_range.sample(&mut rng); |
| 255 |
|
| 256 |
for _ in 0..input_len { |
| 257 |
orig_data.push(rng.gen()); |
| 258 |
} |
| 259 |
|
| 260 |
let prefix_len = prefix_len_range.sample(&mut rng); |
| 261 |
for _ in 0..prefix_len { |
| 262 |
// getting convenient random single-byte printable chars that aren't base64 is |
| 263 |
// annoying |
| 264 |
prefix.push('#'); |
| 265 |
} |
| 266 |
encoded_data_with_prefix.push_str(&prefix); |
| 267 |
|
| 268 |
let engine = random_engine(&mut rng); |
| 269 |
engine.encode_string(&orig_data, &mut encoded_data_no_prefix); |
| 270 |
engine.encode_string(&orig_data, &mut encoded_data_with_prefix); |
| 271 |
|
| 272 |
assert_eq!( |
| 273 |
encoded_data_no_prefix.len() + prefix_len, |
| 274 |
encoded_data_with_prefix.len() |
| 275 |
); |
| 276 |
assert_encode_sanity( |
| 277 |
&encoded_data_no_prefix, |
| 278 |
engine.config().encode_padding(), |
| 279 |
input_len, |
| 280 |
); |
| 281 |
assert_encode_sanity( |
| 282 |
&encoded_data_with_prefix[prefix_len..], |
| 283 |
engine.config().encode_padding(), |
| 284 |
input_len, |
| 285 |
); |
| 286 |
|
| 287 |
// append plain encode onto prefix |
| 288 |
prefix.push_str(&encoded_data_no_prefix); |
| 289 |
|
| 290 |
assert_eq!(prefix, encoded_data_with_prefix); |
| 291 |
|
| 292 |
engine |
| 293 |
.decode_vec(&encoded_data_no_prefix, &mut decoded) |
| 294 |
.unwrap(); |
| 295 |
assert_eq!(orig_data, decoded); |
| 296 |
} |
| 297 |
} |
| 298 |
|
| 299 |
#[test] |
| 300 |
fn encode_engine_slice_into_nonempty_buffer_doesnt_clobber_suffix() { |
| 301 |
let mut orig_data = Vec::new(); |
| 302 |
let mut encoded_data = Vec::new(); |
| 303 |
let mut encoded_data_original_state = Vec::new(); |
| 304 |
let mut decoded = Vec::new(); |
| 305 |
|
| 306 |
let input_len_range = Uniform::new(0, 1000); |
| 307 |
|
| 308 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 309 |
|
| 310 |
for _ in 0..10_000 { |
| 311 |
orig_data.clear(); |
| 312 |
encoded_data.clear(); |
| 313 |
encoded_data_original_state.clear(); |
| 314 |
decoded.clear(); |
| 315 |
|
| 316 |
let input_len = input_len_range.sample(&mut rng); |
| 317 |
|
| 318 |
for _ in 0..input_len { |
| 319 |
orig_data.push(rng.gen()); |
| 320 |
} |
| 321 |
|
| 322 |
// plenty of existing garbage in the encoded buffer |
| 323 |
for _ in 0..10 * input_len { |
| 324 |
encoded_data.push(rng.gen()); |
| 325 |
} |
| 326 |
|
| 327 |
encoded_data_original_state.extend_from_slice(&encoded_data); |
| 328 |
|
| 329 |
let engine = random_engine(&mut rng); |
| 330 |
|
| 331 |
let encoded_size = encoded_len(input_len, engine.config().encode_padding()).unwrap(); |
| 332 |
|
| 333 |
assert_eq!( |
| 334 |
encoded_size, |
| 335 |
engine.encode_slice(&orig_data, &mut encoded_data).unwrap() |
| 336 |
); |
| 337 |
|
| 338 |
assert_encode_sanity( |
| 339 |
str::from_utf8(&encoded_data[0..encoded_size]).unwrap(), |
| 340 |
engine.config().encode_padding(), |
| 341 |
input_len, |
| 342 |
); |
| 343 |
|
| 344 |
assert_eq!( |
| 345 |
&encoded_data[encoded_size..], |
| 346 |
&encoded_data_original_state[encoded_size..] |
| 347 |
); |
| 348 |
|
| 349 |
engine |
| 350 |
.decode_vec(&encoded_data[0..encoded_size], &mut decoded) |
| 351 |
.unwrap(); |
| 352 |
assert_eq!(orig_data, decoded); |
| 353 |
} |
| 354 |
} |
| 355 |
|
| 356 |
#[test] |
| 357 |
fn encode_to_slice_random_valid_utf8() { |
| 358 |
let mut input = Vec::new(); |
| 359 |
let mut output = Vec::new(); |
| 360 |
|
| 361 |
let input_len_range = Uniform::new(0, 1000); |
| 362 |
|
| 363 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 364 |
|
| 365 |
for _ in 0..10_000 { |
| 366 |
input.clear(); |
| 367 |
output.clear(); |
| 368 |
|
| 369 |
let input_len = input_len_range.sample(&mut rng); |
| 370 |
|
| 371 |
for _ in 0..input_len { |
| 372 |
input.push(rng.gen()); |
| 373 |
} |
| 374 |
|
| 375 |
let config = random_config(&mut rng); |
| 376 |
let engine = random_engine(&mut rng); |
| 377 |
|
| 378 |
// fill up the output buffer with garbage |
| 379 |
let encoded_size = encoded_len(input_len, config.encode_padding()).unwrap(); |
| 380 |
for _ in 0..encoded_size { |
| 381 |
output.push(rng.gen()); |
| 382 |
} |
| 383 |
|
| 384 |
let orig_output_buf = output.clone(); |
| 385 |
|
| 386 |
let bytes_written = engine.internal_encode(&input, &mut output); |
| 387 |
|
| 388 |
// make sure the part beyond bytes_written is the same garbage it was before |
| 389 |
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]); |
| 390 |
|
| 391 |
// make sure the encoded bytes are UTF-8 |
| 392 |
let _ = str::from_utf8(&output[0..bytes_written]).unwrap(); |
| 393 |
} |
| 394 |
} |
| 395 |
|
| 396 |
#[test] |
| 397 |
fn encode_with_padding_random_valid_utf8() { |
| 398 |
let mut input = Vec::new(); |
| 399 |
let mut output = Vec::new(); |
| 400 |
|
| 401 |
let input_len_range = Uniform::new(0, 1000); |
| 402 |
|
| 403 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 404 |
|
| 405 |
for _ in 0..10_000 { |
| 406 |
input.clear(); |
| 407 |
output.clear(); |
| 408 |
|
| 409 |
let input_len = input_len_range.sample(&mut rng); |
| 410 |
|
| 411 |
for _ in 0..input_len { |
| 412 |
input.push(rng.gen()); |
| 413 |
} |
| 414 |
|
| 415 |
let engine = random_engine(&mut rng); |
| 416 |
|
| 417 |
// fill up the output buffer with garbage |
| 418 |
let encoded_size = encoded_len(input_len, engine.config().encode_padding()).unwrap(); |
| 419 |
for _ in 0..encoded_size + 1000 { |
| 420 |
output.push(rng.gen()); |
| 421 |
} |
| 422 |
|
| 423 |
let orig_output_buf = output.clone(); |
| 424 |
|
| 425 |
encode_with_padding(&input, &mut output[0..encoded_size], &engine, encoded_size); |
| 426 |
|
| 427 |
// make sure the part beyond b64 is the same garbage it was before |
| 428 |
assert_eq!(orig_output_buf[encoded_size..], output[encoded_size..]); |
| 429 |
|
| 430 |
// make sure the encoded bytes are UTF-8 |
| 431 |
let _ = str::from_utf8(&output[0..encoded_size]).unwrap(); |
| 432 |
} |
| 433 |
} |
| 434 |
|
| 435 |
#[test] |
| 436 |
fn add_padding_random_valid_utf8() { |
| 437 |
let mut output = Vec::new(); |
| 438 |
|
| 439 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 440 |
|
| 441 |
// cover our bases for length % 4 |
| 442 |
for unpadded_output_len in 0..20 { |
| 443 |
output.clear(); |
| 444 |
|
| 445 |
// fill output with random |
| 446 |
for _ in 0..100 { |
| 447 |
output.push(rng.gen()); |
| 448 |
} |
| 449 |
|
| 450 |
let orig_output_buf = output.clone(); |
| 451 |
|
| 452 |
let bytes_written = add_padding(unpadded_output_len, &mut output); |
| 453 |
|
| 454 |
// make sure the part beyond bytes_written is the same garbage it was before |
| 455 |
assert_eq!(orig_output_buf[bytes_written..], output[bytes_written..]); |
| 456 |
|
| 457 |
// make sure the encoded bytes are UTF-8 |
| 458 |
let _ = str::from_utf8(&output[0..bytes_written]).unwrap(); |
| 459 |
} |
| 460 |
} |
| 461 |
|
| 462 |
fn assert_encoded_length<E: Engine>( |
| 463 |
input_len: usize, |
| 464 |
enc_len: usize, |
| 465 |
engine: &E, |
| 466 |
padded: bool, |
| 467 |
) { |
| 468 |
assert_eq!(enc_len, encoded_len(input_len, padded).unwrap()); |
| 469 |
|
| 470 |
let mut bytes: Vec<u8> = Vec::new(); |
| 471 |
let mut rng = rand::rngs::SmallRng::from_entropy(); |
| 472 |
|
| 473 |
for _ in 0..input_len { |
| 474 |
bytes.push(rng.gen()); |
| 475 |
} |
| 476 |
|
| 477 |
let encoded = engine.encode(&bytes); |
| 478 |
assert_encode_sanity(&encoded, padded, input_len); |
| 479 |
|
| 480 |
assert_eq!(enc_len, encoded.len()); |
| 481 |
} |
| 482 |
|
| 483 |
#[test] |
| 484 |
fn encode_imap() { |
| 485 |
assert_eq!( |
| 486 |
&GeneralPurpose::new(&alphabet::IMAP_MUTF7, NO_PAD).encode(b"\xFB\xFF"), |
| 487 |
&GeneralPurpose::new(&alphabet::STANDARD, NO_PAD) |
| 488 |
.encode(b"\xFB\xFF") |
| 489 |
.replace('/', ",") |
| 490 |
); |
| 491 |
} |
| 492 |
} |