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