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]
low_bits_of_byte(byte: u8) -> u852 pub fn low_bits_of_byte(byte: u8) -> u8 {
53     byte & !CONTINUATION_BIT
54 }
55 
56 #[doc(hidden)]
57 #[inline]
low_bits_of_u64(val: u64) -> u858 pub fn low_bits_of_u64(val: u64) -> u8 {
59     let byte = val & (std::u8::MAX as u64);
60     low_bits_of_byte(byte as u8)
61 }
62 
63 /// A module for reading signed and unsigned integers that have been LEB128
64 /// encoded.
65 pub mod read {
66     use super::{low_bits_of_byte, CONTINUATION_BIT, SIGN_BIT};
67     use std::fmt;
68     use std::io;
69 
70     /// An enumeration of the possible errors that can occur when reading a
71     /// number encoded with LEB128.
72     #[derive(Debug)]
73     pub enum Error {
74         /// There was an underlying IO error.
75         IoError(io::Error),
76         /// The number being read is larger than can be represented.
77         Overflow,
78     }
79 
80     impl From<io::Error> for Error {
from(e: io::Error) -> Self81         fn from(e: io::Error) -> Self {
82             Error::IoError(e)
83         }
84     }
85 
86     impl fmt::Display for Error {
fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error>87         fn fmt(&self, f: &mut fmt::Formatter) -> Result<(), fmt::Error> {
88             write!(
89                 f,
90                 "leb128::read::Error: {}",
91                 ::std::error::Error::description(self)
92             )
93         }
94     }
95 
96     impl ::std::error::Error for Error {
description(&self) -> &str97         fn description(&self) -> &str {
98             match *self {
99                 Error::IoError(ref e) => e.description(),
100                 Error::Overflow => "The number being read is larger than can be represented",
101             }
102         }
103 
cause(&self) -> Option<&::std::error::Error>104         fn cause(&self) -> Option<&::std::error::Error> {
105             match *self {
106                 Error::IoError(ref e) => Some(e),
107                 Error::Overflow => None,
108             }
109         }
110     }
111 
112     /// Read an unsigned LEB128 number from the given `std::io::Read`able and
113     /// return it or an error if reading failed.
unsigned<R>(r: &mut R) -> Result<u64, Error> where R: io::Read,114     pub fn unsigned<R>(r: &mut R) -> Result<u64, Error>
115     where
116         R: io::Read,
117     {
118         let mut result = 0;
119         let mut shift = 0;
120 
121         loop {
122             let mut buf = [0];
123             try!(r.read_exact(&mut buf));
124 
125             if shift == 63 && buf[0] != 0x00 && buf[0] != 0x01 {
126                 return Err(Error::Overflow);
127             }
128 
129             let low_bits = low_bits_of_byte(buf[0]) as u64;
130             result |= low_bits << shift;
131 
132             if buf[0] & CONTINUATION_BIT == 0 {
133                 return Ok(result);
134             }
135 
136             shift += 7;
137         }
138     }
139 
140     /// Read a signed LEB128 number from the given `std::io::Read`able and
141     /// return it or an error if reading failed.
signed<R>(r: &mut R) -> Result<i64, Error> where R: io::Read,142     pub fn signed<R>(r: &mut R) -> Result<i64, Error>
143     where
144         R: io::Read,
145     {
146         let mut result = 0;
147         let mut shift = 0;
148         let size = 64;
149         let mut byte;
150 
151         loop {
152             let mut buf = [0];
153             try!(r.read_exact(&mut buf));
154 
155             byte = buf[0];
156             if shift == 63 && byte != 0x00 && byte != 0x7f {
157                 return Err(Error::Overflow);
158             }
159 
160             let low_bits = low_bits_of_byte(byte) as i64;
161             result |= low_bits << shift;
162             shift += 7;
163 
164             if byte & CONTINUATION_BIT == 0 {
165                 break;
166             }
167         }
168 
169         if shift < size && (SIGN_BIT & byte) == SIGN_BIT {
170             // Sign extend the result.
171             result |= !0 << shift;
172         }
173 
174         Ok(result)
175     }
176 }
177 
178 /// A module for writing integers encoded as LEB128.
179 pub mod write {
180     use super::{low_bits_of_u64, CONTINUATION_BIT};
181     use std::io;
182 
183     /// Write the given unsigned number using the LEB128 encoding to the given
184     /// `std::io::Write`able. Returns the number of bytes written to `w`, or an
185     /// error if writing failed.
unsigned<W>(w: &mut W, mut val: u64) -> Result<usize, io::Error> where W: ?Sized + io::Write,186     pub fn unsigned<W>(w: &mut W, mut val: u64) -> Result<usize, io::Error>
187     where
188         W: ?Sized + io::Write,
189     {
190         let mut bytes_written = 0;
191         loop {
192             let mut byte = low_bits_of_u64(val);
193             val >>= 7;
194             if val != 0 {
195                 // More bytes to come, so set the continuation bit.
196                 byte |= CONTINUATION_BIT;
197             }
198 
199             let buf = [byte];
200             try!(w.write_all(&buf));
201             bytes_written += 1;
202 
203             if val == 0 {
204                 return Ok(bytes_written);
205             }
206         }
207     }
208 
209     /// Write the given signed number using the LEB128 encoding to the given
210     /// `std::io::Write`able. Returns the number of bytes written to `w`, or an
211     /// error if writing failed.
signed<W>(w: &mut W, mut val: i64) -> Result<usize, io::Error> where W: ?Sized + io::Write,212     pub fn signed<W>(w: &mut W, mut val: i64) -> Result<usize, io::Error>
213     where
214         W: ?Sized + io::Write,
215     {
216         let mut bytes_written = 0;
217         loop {
218             let mut byte = val as u8;
219             // Keep the sign bit for testing
220             val >>= 6;
221             let done = val == 0 || val == -1;
222             if done {
223                 byte &= !CONTINUATION_BIT;
224             } else {
225                 // Remove the sign bit
226                 val >>= 1;
227                 // More bytes to come, so set the continuation bit.
228                 byte |= CONTINUATION_BIT;
229             }
230 
231             let buf = [byte];
232             try!(w.write_all(&buf));
233             bytes_written += 1;
234 
235             if done {
236                 return Ok(bytes_written);
237             }
238         }
239     }
240 }
241 
242 #[cfg(test)]
243 mod tests {
244     use super::*;
245     use std;
246     use std::io;
247 
248     #[test]
test_low_bits_of_byte()249     fn test_low_bits_of_byte() {
250         for i in 0..127 {
251             assert_eq!(i, low_bits_of_byte(i));
252             assert_eq!(i, low_bits_of_byte(i | CONTINUATION_BIT));
253         }
254     }
255 
256     #[test]
test_low_bits_of_u64()257     fn test_low_bits_of_u64() {
258         for i in 0u64..127 {
259             assert_eq!(i as u8, low_bits_of_u64(1 << 16 | i));
260             assert_eq!(
261                 i as u8,
262                 low_bits_of_u64(i << 16 | i | (CONTINUATION_BIT as u64))
263             );
264         }
265     }
266 
267     // Examples from the DWARF 4 standard, section 7.6, figure 22.
268     #[test]
test_read_unsigned()269     fn test_read_unsigned() {
270         let buf = [2u8];
271         let mut readable = &buf[..];
272         assert_eq!(
273             2,
274             read::unsigned(&mut readable).expect("Should read number")
275         );
276 
277         let buf = [127u8];
278         let mut readable = &buf[..];
279         assert_eq!(
280             127,
281             read::unsigned(&mut readable).expect("Should read number")
282         );
283 
284         let buf = [CONTINUATION_BIT, 1];
285         let mut readable = &buf[..];
286         assert_eq!(
287             128,
288             read::unsigned(&mut readable).expect("Should read number")
289         );
290 
291         let buf = [1u8 | CONTINUATION_BIT, 1];
292         let mut readable = &buf[..];
293         assert_eq!(
294             129,
295             read::unsigned(&mut readable).expect("Should read number")
296         );
297 
298         let buf = [2u8 | CONTINUATION_BIT, 1];
299         let mut readable = &buf[..];
300         assert_eq!(
301             130,
302             read::unsigned(&mut readable).expect("Should read number")
303         );
304 
305         let buf = [57u8 | CONTINUATION_BIT, 100];
306         let mut readable = &buf[..];
307         assert_eq!(
308             12857,
309             read::unsigned(&mut readable).expect("Should read number")
310         );
311     }
312 
313     // Examples from the DWARF 4 standard, section 7.6, figure 23.
314     #[test]
test_read_signed()315     fn test_read_signed() {
316         let buf = [2u8];
317         let mut readable = &buf[..];
318         assert_eq!(2, read::signed(&mut readable).expect("Should read number"));
319 
320         let buf = [0x7eu8];
321         let mut readable = &buf[..];
322         assert_eq!(-2, read::signed(&mut readable).expect("Should read number"));
323 
324         let buf = [127u8 | CONTINUATION_BIT, 0];
325         let mut readable = &buf[..];
326         assert_eq!(
327             127,
328             read::signed(&mut readable).expect("Should read number")
329         );
330 
331         let buf = [1u8 | CONTINUATION_BIT, 0x7f];
332         let mut readable = &buf[..];
333         assert_eq!(
334             -127,
335             read::signed(&mut readable).expect("Should read number")
336         );
337 
338         let buf = [CONTINUATION_BIT, 1];
339         let mut readable = &buf[..];
340         assert_eq!(
341             128,
342             read::signed(&mut readable).expect("Should read number")
343         );
344 
345         let buf = [CONTINUATION_BIT, 0x7f];
346         let mut readable = &buf[..];
347         assert_eq!(
348             -128,
349             read::signed(&mut readable).expect("Should read number")
350         );
351 
352         let buf = [1u8 | CONTINUATION_BIT, 1];
353         let mut readable = &buf[..];
354         assert_eq!(
355             129,
356             read::signed(&mut readable).expect("Should read number")
357         );
358 
359         let buf = [0x7fu8 | CONTINUATION_BIT, 0x7e];
360         let mut readable = &buf[..];
361         assert_eq!(
362             -129,
363             read::signed(&mut readable).expect("Should read number")
364         );
365     }
366 
367     #[test]
test_read_signed_63_bits()368     fn test_read_signed_63_bits() {
369         let buf = [
370             CONTINUATION_BIT,
371             CONTINUATION_BIT,
372             CONTINUATION_BIT,
373             CONTINUATION_BIT,
374             CONTINUATION_BIT,
375             CONTINUATION_BIT,
376             CONTINUATION_BIT,
377             CONTINUATION_BIT,
378             0x40,
379         ];
380         let mut readable = &buf[..];
381         assert_eq!(
382             -0x4000000000000000,
383             read::signed(&mut readable).expect("Should read number")
384         );
385     }
386 
387     #[test]
test_read_unsigned_not_enough_data()388     fn test_read_unsigned_not_enough_data() {
389         let buf = [CONTINUATION_BIT];
390         let mut readable = &buf[..];
391         match read::unsigned(&mut readable) {
392             Err(read::Error::IoError(e)) => assert_eq!(e.kind(), io::ErrorKind::UnexpectedEof),
393             otherwise => panic!("Unexpected: {:?}", otherwise),
394         }
395     }
396 
397     #[test]
test_read_signed_not_enough_data()398     fn test_read_signed_not_enough_data() {
399         let buf = [CONTINUATION_BIT];
400         let mut readable = &buf[..];
401         match read::signed(&mut readable) {
402             Err(read::Error::IoError(e)) => assert_eq!(e.kind(), io::ErrorKind::UnexpectedEof),
403             otherwise => panic!("Unexpected: {:?}", otherwise),
404         }
405     }
406 
407     #[test]
test_write_unsigned_not_enough_space()408     fn test_write_unsigned_not_enough_space() {
409         let mut buf = [0; 1];
410         let mut writable = &mut buf[..];
411         match write::unsigned(&mut writable, 128) {
412             Err(e) => assert_eq!(e.kind(), io::ErrorKind::WriteZero),
413             otherwise => panic!("Unexpected: {:?}", otherwise),
414         }
415     }
416 
417     #[test]
test_write_signed_not_enough_space()418     fn test_write_signed_not_enough_space() {
419         let mut buf = [0; 1];
420         let mut writable = &mut buf[..];
421         match write::signed(&mut writable, 128) {
422             Err(e) => assert_eq!(e.kind(), io::ErrorKind::WriteZero),
423             otherwise => panic!("Unexpected: {:?}", otherwise),
424         }
425     }
426 
427     #[test]
dogfood_signed()428     fn dogfood_signed() {
429         fn inner(i: i64) {
430             let mut buf = [0u8; 1024];
431 
432             {
433                 let mut writable = &mut buf[..];
434                 write::signed(&mut writable, i).expect("Should write signed number");
435             }
436 
437             let mut readable = &buf[..];
438             let result = read::signed(&mut readable).expect("Should be able to read it back again");
439             assert_eq!(i, result);
440         }
441         for i in -513..513 {
442             inner(i);
443         }
444         inner(std::i64::MIN);
445     }
446 
447     #[test]
dogfood_unsigned()448     fn dogfood_unsigned() {
449         for i in 0..1025 {
450             let mut buf = [0u8; 1024];
451 
452             {
453                 let mut writable = &mut buf[..];
454                 write::unsigned(&mut writable, i).expect("Should write signed number");
455             }
456 
457             let mut readable = &buf[..];
458             let result =
459                 read::unsigned(&mut readable).expect("Should be able to read it back again");
460             assert_eq!(i, result);
461         }
462     }
463 
464     #[test]
test_read_unsigned_overflow()465     fn test_read_unsigned_overflow() {
466         let buf = [
467             2u8 | CONTINUATION_BIT,
468             2 | CONTINUATION_BIT,
469             2 | CONTINUATION_BIT,
470             2 | CONTINUATION_BIT,
471             2 | CONTINUATION_BIT,
472             2 | CONTINUATION_BIT,
473             2 | CONTINUATION_BIT,
474             2 | CONTINUATION_BIT,
475             2 | CONTINUATION_BIT,
476             2 | CONTINUATION_BIT,
477             2 | CONTINUATION_BIT,
478             2 | CONTINUATION_BIT,
479             2 | CONTINUATION_BIT,
480             2 | CONTINUATION_BIT,
481             2 | CONTINUATION_BIT,
482             2 | CONTINUATION_BIT,
483             2 | CONTINUATION_BIT,
484             2 | CONTINUATION_BIT,
485             2 | CONTINUATION_BIT,
486             2 | CONTINUATION_BIT,
487             2 | CONTINUATION_BIT,
488             2 | CONTINUATION_BIT,
489             2 | CONTINUATION_BIT,
490             2 | CONTINUATION_BIT,
491             2 | CONTINUATION_BIT,
492             2 | CONTINUATION_BIT,
493             2 | CONTINUATION_BIT,
494             2 | CONTINUATION_BIT,
495             2 | CONTINUATION_BIT,
496             2 | CONTINUATION_BIT,
497             1,
498         ];
499         let mut readable = &buf[..];
500         assert!(read::unsigned(&mut readable).is_err());
501     }
502 
503     #[test]
test_read_signed_overflow()504     fn test_read_signed_overflow() {
505         let buf = [
506             2u8 | CONTINUATION_BIT,
507             2 | CONTINUATION_BIT,
508             2 | CONTINUATION_BIT,
509             2 | CONTINUATION_BIT,
510             2 | CONTINUATION_BIT,
511             2 | CONTINUATION_BIT,
512             2 | CONTINUATION_BIT,
513             2 | CONTINUATION_BIT,
514             2 | CONTINUATION_BIT,
515             2 | CONTINUATION_BIT,
516             2 | CONTINUATION_BIT,
517             2 | CONTINUATION_BIT,
518             2 | CONTINUATION_BIT,
519             2 | CONTINUATION_BIT,
520             2 | CONTINUATION_BIT,
521             2 | CONTINUATION_BIT,
522             2 | CONTINUATION_BIT,
523             2 | CONTINUATION_BIT,
524             2 | CONTINUATION_BIT,
525             2 | CONTINUATION_BIT,
526             2 | CONTINUATION_BIT,
527             2 | CONTINUATION_BIT,
528             2 | CONTINUATION_BIT,
529             2 | CONTINUATION_BIT,
530             2 | CONTINUATION_BIT,
531             2 | CONTINUATION_BIT,
532             2 | CONTINUATION_BIT,
533             2 | CONTINUATION_BIT,
534             2 | CONTINUATION_BIT,
535             2 | CONTINUATION_BIT,
536             1,
537         ];
538         let mut readable = &buf[..];
539         assert!(read::signed(&mut readable).is_err());
540     }
541 
542     #[test]
test_read_multiple()543     fn test_read_multiple() {
544         let buf = [2u8 | CONTINUATION_BIT, 1u8, 1u8];
545 
546         let mut readable = &buf[..];
547         assert_eq!(
548             read::unsigned(&mut readable).expect("Should read first number"),
549             130u64
550         );
551         assert_eq!(
552             read::unsigned(&mut readable).expect("Should read first number"),
553             1u64
554         );
555     }
556 }
557