1 | //! Read and write DWARF's "Little Endian Base 128" (LEB128) variable length |
2 | //! integer encoding. |
3 | //! |
4 | //! The implementation is a direct translation of the psuedocode in the DWARF 4 |
5 | //! standard's appendix C. |
6 | //! |
7 | //! Read and write signed integers: |
8 | //! |
9 | //! ``` |
10 | //! use leb128; |
11 | //! |
12 | //! let mut buf = [0; 1024]; |
13 | //! |
14 | //! // Write to anything that implements `std::io::Write`. |
15 | //! { |
16 | //! let mut writable = &mut buf[..]; |
17 | //! leb128::write::signed(&mut writable, -12345).expect("Should write number" ); |
18 | //! } |
19 | //! |
20 | //! // Read from anything that implements `std::io::Read`. |
21 | //! let mut readable = &buf[..]; |
22 | //! let val = leb128::read::signed(&mut readable).expect("Should read number" ); |
23 | //! assert_eq!(val, -12345); |
24 | //! ``` |
25 | //! |
26 | //! Or read and write unsigned integers: |
27 | //! |
28 | //! ``` |
29 | //! use leb128; |
30 | //! |
31 | //! let mut buf = [0; 1024]; |
32 | //! |
33 | //! { |
34 | //! let mut writable = &mut buf[..]; |
35 | //! leb128::write::unsigned(&mut writable, 98765).expect("Should write number" ); |
36 | //! } |
37 | //! |
38 | //! let mut readable = &buf[..]; |
39 | //! let val = leb128::read::unsigned(&mut readable).expect("Should read number" ); |
40 | //! assert_eq!(val, 98765); |
41 | //! ``` |
42 | |
43 | #![deny (missing_docs)] |
44 | |
45 | #[doc (hidden)] |
46 | pub const CONTINUATION_BIT: u8 = 1 << 7; |
47 | #[doc (hidden)] |
48 | pub const SIGN_BIT: u8 = 1 << 6; |
49 | |
50 | #[doc (hidden)] |
51 | #[inline ] |
52 | pub fn low_bits_of_byte(byte: u8) -> u8 { |
53 | byte & !CONTINUATION_BIT |
54 | } |
55 | |
56 | #[doc (hidden)] |
57 | #[inline ] |
58 | pub fn low_bits_of_u64(val: u64) -> u8 { |
59 | let byte: u64 = val & (std::u8::MAX as u64); |
60 | low_bits_of_byte(byte as u8) |
61 | } |
62 | |
63 | /// A module for reading LEB128-encoded signed and unsigned integers. |
64 | pub mod read { |
65 | use super::{low_bits_of_byte, CONTINUATION_BIT, SIGN_BIT}; |
66 | use std::fmt; |
67 | use std::io; |
68 | |
69 | /// An error type for reading LEB128-encoded values. |
70 | #[derive (Debug)] |
71 | pub enum Error { |
72 | /// There was an underlying IO error. |
73 | IoError(io::Error), |
74 | /// The number being read is larger than can be represented. |
75 | Overflow, |
76 | } |
77 | |
78 | impl From<io::Error> for Error { |
79 | fn from(e: io::Error) -> Self { |
80 | Error::IoError(e) |
81 | } |
82 | } |
83 | |
84 | impl fmt::Display for Error { |
85 | fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> { |
86 | match *self { |
87 | Error::IoError(ref e) => e.fmt(f), |
88 | Error::Overflow => { |
89 | write!(f, "The number being read is larger than can be represented" ) |
90 | } |
91 | } |
92 | } |
93 | } |
94 | |
95 | impl std::error::Error for Error { |
96 | fn source(&self) -> Option<&(dyn std::error::Error + 'static)> { |
97 | match *self { |
98 | Error::IoError(ref e) => Some(e), |
99 | Error::Overflow => None, |
100 | } |
101 | } |
102 | } |
103 | |
104 | /// Read an unsigned LEB128-encoded number from the `std::io::Read` stream |
105 | /// `r`. |
106 | /// |
107 | /// On success, return the number. |
108 | pub fn unsigned<R>(r: &mut R) -> Result<u64, Error> |
109 | where |
110 | R: ?Sized + io::Read, |
111 | { |
112 | let mut result = 0; |
113 | let mut shift = 0; |
114 | |
115 | loop { |
116 | let mut buf = [0]; |
117 | r.read_exact(&mut buf)?; |
118 | |
119 | if shift == 63 && buf[0] != 0x00 && buf[0] != 0x01 { |
120 | while buf[0] & CONTINUATION_BIT != 0 { |
121 | r.read_exact(&mut buf)?; |
122 | } |
123 | return Err(Error::Overflow); |
124 | } |
125 | |
126 | let low_bits = low_bits_of_byte(buf[0]) as u64; |
127 | result |= low_bits << shift; |
128 | |
129 | if buf[0] & CONTINUATION_BIT == 0 { |
130 | return Ok(result); |
131 | } |
132 | |
133 | shift += 7; |
134 | } |
135 | } |
136 | |
137 | /// Read a signed LEB128-encoded number from the `std::io::Read` stream `r`. |
138 | /// |
139 | /// On success, return the number. |
140 | pub fn signed<R>(r: &mut R) -> Result<i64, Error> |
141 | where |
142 | R: ?Sized + io::Read, |
143 | { |
144 | let mut result = 0; |
145 | let mut shift = 0; |
146 | let size = 64; |
147 | let mut byte; |
148 | |
149 | loop { |
150 | let mut buf = [0]; |
151 | r.read_exact(&mut buf)?; |
152 | |
153 | byte = buf[0]; |
154 | if shift == 63 && byte != 0x00 && byte != 0x7f { |
155 | while buf[0] & CONTINUATION_BIT != 0 { |
156 | r.read_exact(&mut buf)?; |
157 | } |
158 | return Err(Error::Overflow); |
159 | } |
160 | |
161 | let low_bits = low_bits_of_byte(byte) as i64; |
162 | result |= low_bits << shift; |
163 | shift += 7; |
164 | |
165 | if byte & CONTINUATION_BIT == 0 { |
166 | break; |
167 | } |
168 | } |
169 | |
170 | if shift < size && (SIGN_BIT & byte) == SIGN_BIT { |
171 | // Sign extend the result. |
172 | result |= !0 << shift; |
173 | } |
174 | |
175 | Ok(result) |
176 | } |
177 | } |
178 | |
179 | /// A module for writing LEB128-encoded signed and unsigned integers. |
180 | pub mod write { |
181 | use super::{low_bits_of_u64, CONTINUATION_BIT}; |
182 | use std::io; |
183 | |
184 | /// Write `val` to the `std::io::Write` stream `w` as an unsigned LEB128 value. |
185 | /// |
186 | /// On success, return the number of bytes written to `w`. |
187 | pub fn unsigned<W>(w: &mut W, mut val: u64) -> Result<usize, io::Error> |
188 | where |
189 | W: ?Sized + io::Write, |
190 | { |
191 | let mut bytes_written = 0; |
192 | loop { |
193 | let mut byte = low_bits_of_u64(val); |
194 | val >>= 7; |
195 | if val != 0 { |
196 | // More bytes to come, so set the continuation bit. |
197 | byte |= CONTINUATION_BIT; |
198 | } |
199 | |
200 | let buf = [byte]; |
201 | w.write_all(&buf)?; |
202 | bytes_written += 1; |
203 | |
204 | if val == 0 { |
205 | return Ok(bytes_written); |
206 | } |
207 | } |
208 | } |
209 | |
210 | /// Write `val` to the `std::io::Write` stream `w` as a signed LEB128 value. |
211 | /// |
212 | /// On success, return the number of bytes written to `w`. |
213 | pub fn signed<W>(w: &mut W, mut val: i64) -> Result<usize, io::Error> |
214 | where |
215 | W: ?Sized + io::Write, |
216 | { |
217 | let mut bytes_written = 0; |
218 | loop { |
219 | let mut byte = val as u8; |
220 | // Keep the sign bit for testing |
221 | val >>= 6; |
222 | let done = val == 0 || val == -1; |
223 | if done { |
224 | byte &= !CONTINUATION_BIT; |
225 | } else { |
226 | // Remove the sign bit |
227 | val >>= 1; |
228 | // More bytes to come, so set the continuation bit. |
229 | byte |= CONTINUATION_BIT; |
230 | } |
231 | |
232 | let buf = [byte]; |
233 | w.write_all(&buf)?; |
234 | bytes_written += 1; |
235 | |
236 | if done { |
237 | return Ok(bytes_written); |
238 | } |
239 | } |
240 | } |
241 | } |
242 | |
243 | #[cfg (test)] |
244 | mod tests { |
245 | use super::*; |
246 | use std; |
247 | use std::io; |
248 | |
249 | #[test ] |
250 | fn test_low_bits_of_byte() { |
251 | for i in 0..127 { |
252 | assert_eq!(i, low_bits_of_byte(i)); |
253 | assert_eq!(i, low_bits_of_byte(i | CONTINUATION_BIT)); |
254 | } |
255 | } |
256 | |
257 | #[test ] |
258 | fn test_low_bits_of_u64() { |
259 | for i in 0u64..127 { |
260 | assert_eq!(i as u8, low_bits_of_u64(1 << 16 | i)); |
261 | assert_eq!( |
262 | i as u8, |
263 | low_bits_of_u64(i << 16 | i | (CONTINUATION_BIT as u64)) |
264 | ); |
265 | } |
266 | } |
267 | |
268 | // Examples from the DWARF 4 standard, section 7.6, figure 22. |
269 | #[test ] |
270 | fn test_read_unsigned() { |
271 | let buf = [2u8]; |
272 | let mut readable = &buf[..]; |
273 | assert_eq!( |
274 | 2, |
275 | read::unsigned(&mut readable).expect("Should read number" ) |
276 | ); |
277 | |
278 | let buf = [127u8]; |
279 | let mut readable = &buf[..]; |
280 | assert_eq!( |
281 | 127, |
282 | read::unsigned(&mut readable).expect("Should read number" ) |
283 | ); |
284 | |
285 | let buf = [CONTINUATION_BIT, 1]; |
286 | let mut readable = &buf[..]; |
287 | assert_eq!( |
288 | 128, |
289 | read::unsigned(&mut readable).expect("Should read number" ) |
290 | ); |
291 | |
292 | let buf = [1u8 | CONTINUATION_BIT, 1]; |
293 | let mut readable = &buf[..]; |
294 | assert_eq!( |
295 | 129, |
296 | read::unsigned(&mut readable).expect("Should read number" ) |
297 | ); |
298 | |
299 | let buf = [2u8 | CONTINUATION_BIT, 1]; |
300 | let mut readable = &buf[..]; |
301 | assert_eq!( |
302 | 130, |
303 | read::unsigned(&mut readable).expect("Should read number" ) |
304 | ); |
305 | |
306 | let buf = [57u8 | CONTINUATION_BIT, 100]; |
307 | let mut readable = &buf[..]; |
308 | assert_eq!( |
309 | 12857, |
310 | read::unsigned(&mut readable).expect("Should read number" ) |
311 | ); |
312 | } |
313 | |
314 | #[test ] |
315 | fn test_read_unsigned_thru_dyn_trait() { |
316 | fn read(r: &mut dyn io::Read) -> u64 { |
317 | read::unsigned(r).expect("Should read number" ) |
318 | } |
319 | |
320 | let buf = [0u8]; |
321 | |
322 | let mut readable = &buf[..]; |
323 | assert_eq!(0, read(&mut readable)); |
324 | |
325 | let mut readable = io::Cursor::new(buf); |
326 | assert_eq!(0, read(&mut readable)); |
327 | } |
328 | |
329 | // Examples from the DWARF 4 standard, section 7.6, figure 23. |
330 | #[test ] |
331 | fn test_read_signed() { |
332 | let buf = [2u8]; |
333 | let mut readable = &buf[..]; |
334 | assert_eq!(2, read::signed(&mut readable).expect("Should read number" )); |
335 | |
336 | let buf = [0x7eu8]; |
337 | let mut readable = &buf[..]; |
338 | assert_eq!(-2, read::signed(&mut readable).expect("Should read number" )); |
339 | |
340 | let buf = [127u8 | CONTINUATION_BIT, 0]; |
341 | let mut readable = &buf[..]; |
342 | assert_eq!( |
343 | 127, |
344 | read::signed(&mut readable).expect("Should read number" ) |
345 | ); |
346 | |
347 | let buf = [1u8 | CONTINUATION_BIT, 0x7f]; |
348 | let mut readable = &buf[..]; |
349 | assert_eq!( |
350 | -127, |
351 | read::signed(&mut readable).expect("Should read number" ) |
352 | ); |
353 | |
354 | let buf = [CONTINUATION_BIT, 1]; |
355 | let mut readable = &buf[..]; |
356 | assert_eq!( |
357 | 128, |
358 | read::signed(&mut readable).expect("Should read number" ) |
359 | ); |
360 | |
361 | let buf = [CONTINUATION_BIT, 0x7f]; |
362 | let mut readable = &buf[..]; |
363 | assert_eq!( |
364 | -128, |
365 | read::signed(&mut readable).expect("Should read number" ) |
366 | ); |
367 | |
368 | let buf = [1u8 | CONTINUATION_BIT, 1]; |
369 | let mut readable = &buf[..]; |
370 | assert_eq!( |
371 | 129, |
372 | read::signed(&mut readable).expect("Should read number" ) |
373 | ); |
374 | |
375 | let buf = [0x7fu8 | CONTINUATION_BIT, 0x7e]; |
376 | let mut readable = &buf[..]; |
377 | assert_eq!( |
378 | -129, |
379 | read::signed(&mut readable).expect("Should read number" ) |
380 | ); |
381 | } |
382 | |
383 | #[test ] |
384 | fn test_read_signed_thru_dyn_trait() { |
385 | fn read(r: &mut dyn io::Read) -> i64 { |
386 | read::signed(r).expect("Should read number" ) |
387 | } |
388 | |
389 | let buf = [0u8]; |
390 | |
391 | let mut readable = &buf[..]; |
392 | assert_eq!(0, read(&mut readable)); |
393 | |
394 | let mut readable = io::Cursor::new(buf); |
395 | assert_eq!(0, read(&mut readable)); |
396 | } |
397 | |
398 | #[test ] |
399 | fn test_read_signed_63_bits() { |
400 | let buf = [ |
401 | CONTINUATION_BIT, |
402 | CONTINUATION_BIT, |
403 | CONTINUATION_BIT, |
404 | CONTINUATION_BIT, |
405 | CONTINUATION_BIT, |
406 | CONTINUATION_BIT, |
407 | CONTINUATION_BIT, |
408 | CONTINUATION_BIT, |
409 | 0x40, |
410 | ]; |
411 | let mut readable = &buf[..]; |
412 | assert_eq!( |
413 | -0x4000000000000000, |
414 | read::signed(&mut readable).expect("Should read number" ) |
415 | ); |
416 | } |
417 | |
418 | #[test ] |
419 | fn test_read_unsigned_not_enough_data() { |
420 | let buf = [CONTINUATION_BIT]; |
421 | let mut readable = &buf[..]; |
422 | match read::unsigned(&mut readable) { |
423 | Err(read::Error::IoError(e)) => assert_eq!(e.kind(), io::ErrorKind::UnexpectedEof), |
424 | otherwise => panic!("Unexpected: {:?}" , otherwise), |
425 | } |
426 | } |
427 | |
428 | #[test ] |
429 | fn test_read_signed_not_enough_data() { |
430 | let buf = [CONTINUATION_BIT]; |
431 | let mut readable = &buf[..]; |
432 | match read::signed(&mut readable) { |
433 | Err(read::Error::IoError(e)) => assert_eq!(e.kind(), io::ErrorKind::UnexpectedEof), |
434 | otherwise => panic!("Unexpected: {:?}" , otherwise), |
435 | } |
436 | } |
437 | |
438 | #[test ] |
439 | fn test_write_unsigned_not_enough_space() { |
440 | let mut buf = [0; 1]; |
441 | let mut writable = &mut buf[..]; |
442 | match write::unsigned(&mut writable, 128) { |
443 | Err(e) => assert_eq!(e.kind(), io::ErrorKind::WriteZero), |
444 | otherwise => panic!("Unexpected: {:?}" , otherwise), |
445 | } |
446 | } |
447 | |
448 | #[test ] |
449 | fn test_write_signed_not_enough_space() { |
450 | let mut buf = [0; 1]; |
451 | let mut writable = &mut buf[..]; |
452 | match write::signed(&mut writable, 128) { |
453 | Err(e) => assert_eq!(e.kind(), io::ErrorKind::WriteZero), |
454 | otherwise => panic!("Unexpected: {:?}" , otherwise), |
455 | } |
456 | } |
457 | |
458 | #[test ] |
459 | fn test_write_unsigned_thru_dyn_trait() { |
460 | fn write(w: &mut dyn io::Write, val: u64) -> usize { |
461 | write::unsigned(w, val).expect("Should write number" ) |
462 | } |
463 | let mut buf = [0u8; 1]; |
464 | |
465 | let mut writable = &mut buf[..]; |
466 | assert_eq!(write(&mut writable, 0), 1); |
467 | assert_eq!(buf[0], 0); |
468 | |
469 | let mut writable = Vec::from(&buf[..]); |
470 | assert_eq!(write(&mut writable, 0), 1); |
471 | assert_eq!(buf[0], 0); |
472 | } |
473 | |
474 | #[test ] |
475 | fn test_write_signed_thru_dyn_trait() { |
476 | fn write(w: &mut dyn io::Write, val: i64) -> usize { |
477 | write::signed(w, val).expect("Should write number" ) |
478 | } |
479 | let mut buf = [0u8; 1]; |
480 | |
481 | let mut writable = &mut buf[..]; |
482 | assert_eq!(write(&mut writable, 0), 1); |
483 | assert_eq!(buf[0], 0); |
484 | |
485 | let mut writable = Vec::from(&buf[..]); |
486 | assert_eq!(write(&mut writable, 0), 1); |
487 | assert_eq!(buf[0], 0); |
488 | } |
489 | |
490 | #[test ] |
491 | fn dogfood_signed() { |
492 | fn inner(i: i64) { |
493 | let mut buf = [0u8; 1024]; |
494 | |
495 | { |
496 | let mut writable = &mut buf[..]; |
497 | write::signed(&mut writable, i).expect("Should write signed number" ); |
498 | } |
499 | |
500 | let mut readable = &buf[..]; |
501 | let result = read::signed(&mut readable).expect("Should be able to read it back again" ); |
502 | assert_eq!(i, result); |
503 | } |
504 | for i in -513..513 { |
505 | inner(i); |
506 | } |
507 | inner(std::i64::MIN); |
508 | } |
509 | |
510 | #[test ] |
511 | fn dogfood_unsigned() { |
512 | for i in 0..1025 { |
513 | let mut buf = [0u8; 1024]; |
514 | |
515 | { |
516 | let mut writable = &mut buf[..]; |
517 | write::unsigned(&mut writable, i).expect("Should write signed number" ); |
518 | } |
519 | |
520 | let mut readable = &buf[..]; |
521 | let result = |
522 | read::unsigned(&mut readable).expect("Should be able to read it back again" ); |
523 | assert_eq!(i, result); |
524 | } |
525 | } |
526 | |
527 | #[test ] |
528 | fn test_read_unsigned_overflow() { |
529 | let buf = [ |
530 | 2u8 | CONTINUATION_BIT, |
531 | 2 | CONTINUATION_BIT, |
532 | 2 | CONTINUATION_BIT, |
533 | 2 | CONTINUATION_BIT, |
534 | 2 | CONTINUATION_BIT, |
535 | 2 | CONTINUATION_BIT, |
536 | 2 | CONTINUATION_BIT, |
537 | 2 | CONTINUATION_BIT, |
538 | 2 | CONTINUATION_BIT, |
539 | 2 | CONTINUATION_BIT, |
540 | 2 | CONTINUATION_BIT, |
541 | 2 | CONTINUATION_BIT, |
542 | 2 | CONTINUATION_BIT, |
543 | 2 | CONTINUATION_BIT, |
544 | 2 | CONTINUATION_BIT, |
545 | 2 | CONTINUATION_BIT, |
546 | 2 | CONTINUATION_BIT, |
547 | 2 | CONTINUATION_BIT, |
548 | 2 | CONTINUATION_BIT, |
549 | 2 | CONTINUATION_BIT, |
550 | 2 | CONTINUATION_BIT, |
551 | 2 | CONTINUATION_BIT, |
552 | 2 | CONTINUATION_BIT, |
553 | 2 | CONTINUATION_BIT, |
554 | 2 | CONTINUATION_BIT, |
555 | 2 | CONTINUATION_BIT, |
556 | 2 | CONTINUATION_BIT, |
557 | 2 | CONTINUATION_BIT, |
558 | 2 | CONTINUATION_BIT, |
559 | 2 | CONTINUATION_BIT, |
560 | 1, |
561 | ]; |
562 | let mut readable = &buf[..]; |
563 | assert!(read::unsigned(&mut readable).is_err()); |
564 | } |
565 | |
566 | #[test ] |
567 | fn test_read_signed_overflow() { |
568 | let buf = [ |
569 | 2u8 | CONTINUATION_BIT, |
570 | 2 | CONTINUATION_BIT, |
571 | 2 | CONTINUATION_BIT, |
572 | 2 | CONTINUATION_BIT, |
573 | 2 | CONTINUATION_BIT, |
574 | 2 | CONTINUATION_BIT, |
575 | 2 | CONTINUATION_BIT, |
576 | 2 | CONTINUATION_BIT, |
577 | 2 | CONTINUATION_BIT, |
578 | 2 | CONTINUATION_BIT, |
579 | 2 | CONTINUATION_BIT, |
580 | 2 | CONTINUATION_BIT, |
581 | 2 | CONTINUATION_BIT, |
582 | 2 | CONTINUATION_BIT, |
583 | 2 | CONTINUATION_BIT, |
584 | 2 | CONTINUATION_BIT, |
585 | 2 | CONTINUATION_BIT, |
586 | 2 | CONTINUATION_BIT, |
587 | 2 | CONTINUATION_BIT, |
588 | 2 | CONTINUATION_BIT, |
589 | 2 | CONTINUATION_BIT, |
590 | 2 | CONTINUATION_BIT, |
591 | 2 | CONTINUATION_BIT, |
592 | 2 | CONTINUATION_BIT, |
593 | 2 | CONTINUATION_BIT, |
594 | 2 | CONTINUATION_BIT, |
595 | 2 | CONTINUATION_BIT, |
596 | 2 | CONTINUATION_BIT, |
597 | 2 | CONTINUATION_BIT, |
598 | 2 | CONTINUATION_BIT, |
599 | 1, |
600 | ]; |
601 | let mut readable = &buf[..]; |
602 | assert!(read::signed(&mut readable).is_err()); |
603 | } |
604 | |
605 | #[test ] |
606 | fn test_read_multiple() { |
607 | let buf = [2u8 | CONTINUATION_BIT, 1u8, 1u8]; |
608 | |
609 | let mut readable = &buf[..]; |
610 | assert_eq!( |
611 | read::unsigned(&mut readable).expect("Should read first number" ), |
612 | 130u64 |
613 | ); |
614 | assert_eq!( |
615 | read::unsigned(&mut readable).expect("Should read first number" ), |
616 | 1u64 |
617 | ); |
618 | } |
619 | |
620 | #[test ] |
621 | fn test_read_multiple_with_overflow() { |
622 | let buf = [ |
623 | 0b1111_1111, |
624 | 0b1111_1111, |
625 | 0b1111_1111, |
626 | 0b1111_1111, |
627 | 0b1111_1111, |
628 | 0b1111_1111, |
629 | 0b1111_1111, |
630 | 0b1111_1111, |
631 | 0b1111_1111, |
632 | 0b1111_1111, |
633 | 0b0111_1111, // Overflow! |
634 | 0b1110_0100, |
635 | 0b1110_0000, |
636 | 0b0000_0010, // 45156 |
637 | ]; |
638 | let mut readable = &buf[..]; |
639 | |
640 | assert!(if let read::Error::Overflow = |
641 | read::unsigned(&mut readable).expect_err("Should fail with Error::Overflow" ) |
642 | { |
643 | true |
644 | } else { |
645 | false |
646 | }); |
647 | assert_eq!( |
648 | read::unsigned(&mut readable).expect("Should succeed with correct value" ), |
649 | 45156 |
650 | ); |
651 | } |
652 | } |
653 | |