Show More
@@ -1,421 +1,430 b'' | |||
|
1 | 1 | // Copyright 2019-2020 Georges Racinet <georges.racinet@octobus.net> |
|
2 | 2 | // |
|
3 | 3 | // This software may be used and distributed according to the terms of the |
|
4 | 4 | // GNU General Public License version 2 or any later version. |
|
5 | 5 | |
|
6 | 6 | //! Definitions and utilities for Revision nodes |
|
7 | 7 | //! |
|
8 | 8 | //! In Mercurial code base, it is customary to call "a node" the binary SHA |
|
9 | 9 | //! of a revision. |
|
10 | 10 | |
|
11 | 11 | use crate::errors::HgError; |
|
12 | 12 | use bytes_cast::BytesCast; |
|
13 | 13 | use std::convert::{TryFrom, TryInto}; |
|
14 | 14 | use std::fmt; |
|
15 | 15 | |
|
16 | 16 | /// The length in bytes of a `Node` |
|
17 | 17 | /// |
|
18 | 18 | /// This constant is meant to ease refactors of this module, and |
|
19 | 19 | /// are private so that calling code does not expect all nodes have |
|
20 | 20 | /// the same size, should we support several formats concurrently in |
|
21 | 21 | /// the future. |
|
22 | 22 | pub const NODE_BYTES_LENGTH: usize = 20; |
|
23 | 23 | |
|
24 | 24 | /// Id of the null node. |
|
25 | 25 | /// |
|
26 | 26 | /// Used to indicate the absence of node. |
|
27 | 27 | pub const NULL_NODE_ID: [u8; NODE_BYTES_LENGTH] = [0u8; NODE_BYTES_LENGTH]; |
|
28 | 28 | |
|
29 | 29 | /// The length in bytes of a `Node` |
|
30 | 30 | /// |
|
31 | 31 | /// see also `NODES_BYTES_LENGTH` about it being private. |
|
32 | 32 | const NODE_NYBBLES_LENGTH: usize = 2 * NODE_BYTES_LENGTH; |
|
33 | 33 | |
|
34 | 34 | /// Default for UI presentation |
|
35 | 35 | const SHORT_PREFIX_DEFAULT_NYBBLES_LENGTH: u8 = 12; |
|
36 | 36 | |
|
37 | 37 | /// Private alias for readability and to ease future change |
|
38 | 38 | type NodeData = [u8; NODE_BYTES_LENGTH]; |
|
39 | 39 | |
|
40 | 40 | /// Binary revision SHA |
|
41 | 41 | /// |
|
42 | 42 | /// ## Future changes of hash size |
|
43 | 43 | /// |
|
44 | 44 | /// To accomodate future changes of hash size, Rust callers |
|
45 | 45 | /// should use the conversion methods at the boundaries (FFI, actual |
|
46 | 46 | /// computation of hashes and I/O) only, and only if required. |
|
47 | 47 | /// |
|
48 | 48 | /// All other callers outside of unit tests should just handle `Node` values |
|
49 | 49 | /// and never make any assumption on the actual length, using [`nybbles_len`] |
|
50 | 50 | /// if they need a loop boundary. |
|
51 | 51 | /// |
|
52 | 52 | /// All methods that create a `Node` either take a type that enforces |
|
53 | 53 | /// the size or return an error at runtime. |
|
54 | 54 | /// |
|
55 | 55 | /// [`nybbles_len`]: #method.nybbles_len |
|
56 |
#[derive(Copy, Clone, |
|
|
56 | #[derive(Copy, Clone, PartialEq, BytesCast, derive_more::From)] | |
|
57 | 57 | #[repr(transparent)] |
|
58 | 58 | pub struct Node { |
|
59 | 59 | data: NodeData, |
|
60 | 60 | } |
|
61 | 61 | |
|
62 | impl fmt::Debug for Node { | |
|
63 | fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result { | |
|
64 | let n = format!("{:x?}", self.data); | |
|
65 | // We're using debug_tuple because it makes the output a little | |
|
66 | // more compact without losing data. | |
|
67 | f.debug_tuple("Node").field(&n).finish() | |
|
68 | } | |
|
69 | } | |
|
70 | ||
|
62 | 71 | /// The node value for NULL_REVISION |
|
63 | 72 | pub const NULL_NODE: Node = Node { |
|
64 | 73 | data: [0; NODE_BYTES_LENGTH], |
|
65 | 74 | }; |
|
66 | 75 | |
|
67 | 76 | /// Return an error if the slice has an unexpected length |
|
68 | 77 | impl<'a> TryFrom<&'a [u8]> for &'a Node { |
|
69 | 78 | type Error = (); |
|
70 | 79 | |
|
71 | 80 | #[inline] |
|
72 | 81 | fn try_from(bytes: &'a [u8]) -> Result<Self, Self::Error> { |
|
73 | 82 | match Node::from_bytes(bytes) { |
|
74 | 83 | Ok((node, rest)) if rest.is_empty() => Ok(node), |
|
75 | 84 | _ => Err(()), |
|
76 | 85 | } |
|
77 | 86 | } |
|
78 | 87 | } |
|
79 | 88 | |
|
80 | 89 | /// Return an error if the slice has an unexpected length |
|
81 | 90 | impl TryFrom<&'_ [u8]> for Node { |
|
82 | 91 | type Error = std::array::TryFromSliceError; |
|
83 | 92 | |
|
84 | 93 | #[inline] |
|
85 | 94 | fn try_from(bytes: &'_ [u8]) -> Result<Self, Self::Error> { |
|
86 | 95 | let data = bytes.try_into()?; |
|
87 | 96 | Ok(Self { data }) |
|
88 | 97 | } |
|
89 | 98 | } |
|
90 | 99 | |
|
91 | 100 | impl From<&'_ NodeData> for Node { |
|
92 | 101 | #[inline] |
|
93 | 102 | fn from(data: &'_ NodeData) -> Self { |
|
94 | 103 | Self { data: *data } |
|
95 | 104 | } |
|
96 | 105 | } |
|
97 | 106 | |
|
98 | 107 | impl fmt::LowerHex for Node { |
|
99 | 108 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
|
100 | 109 | for &byte in &self.data { |
|
101 | 110 | write!(f, "{:02x}", byte)? |
|
102 | 111 | } |
|
103 | 112 | Ok(()) |
|
104 | 113 | } |
|
105 | 114 | } |
|
106 | 115 | |
|
107 | 116 | #[derive(Debug)] |
|
108 | 117 | pub struct FromHexError; |
|
109 | 118 | |
|
110 | 119 | /// Low level utility function, also for prefixes |
|
111 | 120 | fn get_nybble(s: &[u8], i: usize) -> u8 { |
|
112 | 121 | if i % 2 == 0 { |
|
113 | 122 | s[i / 2] >> 4 |
|
114 | 123 | } else { |
|
115 | 124 | s[i / 2] & 0x0f |
|
116 | 125 | } |
|
117 | 126 | } |
|
118 | 127 | |
|
119 | 128 | impl Node { |
|
120 | 129 | /// Retrieve the `i`th half-byte of the binary data. |
|
121 | 130 | /// |
|
122 | 131 | /// This is also the `i`th hexadecimal digit in numeric form, |
|
123 | 132 | /// also called a [nybble](https://en.wikipedia.org/wiki/Nibble). |
|
124 | 133 | pub fn get_nybble(&self, i: usize) -> u8 { |
|
125 | 134 | get_nybble(&self.data, i) |
|
126 | 135 | } |
|
127 | 136 | |
|
128 | 137 | /// Length of the data, in nybbles |
|
129 | 138 | pub fn nybbles_len(&self) -> usize { |
|
130 | 139 | // public exposure as an instance method only, so that we can |
|
131 | 140 | // easily support several sizes of hashes if needed in the future. |
|
132 | 141 | NODE_NYBBLES_LENGTH |
|
133 | 142 | } |
|
134 | 143 | |
|
135 | 144 | /// Convert from hexadecimal string representation |
|
136 | 145 | /// |
|
137 | 146 | /// Exact length is required. |
|
138 | 147 | /// |
|
139 | 148 | /// To be used in FFI and I/O only, in order to facilitate future |
|
140 | 149 | /// changes of hash format. |
|
141 | 150 | pub fn from_hex(hex: impl AsRef<[u8]>) -> Result<Node, FromHexError> { |
|
142 | 151 | let prefix = NodePrefix::from_hex(hex)?; |
|
143 | 152 | if prefix.nybbles_len() == NODE_NYBBLES_LENGTH { |
|
144 | 153 | Ok(Self { data: prefix.data }) |
|
145 | 154 | } else { |
|
146 | 155 | Err(FromHexError) |
|
147 | 156 | } |
|
148 | 157 | } |
|
149 | 158 | |
|
150 | 159 | /// `from_hex`, but for input from an internal file of the repository such |
|
151 | 160 | /// as a changelog or manifest entry. |
|
152 | 161 | /// |
|
153 | 162 | /// An error is treated as repository corruption. |
|
154 | 163 | pub fn from_hex_for_repo(hex: impl AsRef<[u8]>) -> Result<Node, HgError> { |
|
155 | 164 | Self::from_hex(hex.as_ref()).map_err(|FromHexError| { |
|
156 | 165 | HgError::CorruptedRepository(format!( |
|
157 | 166 | "Expected a full hexadecimal node ID, found {}", |
|
158 | 167 | String::from_utf8_lossy(hex.as_ref()) |
|
159 | 168 | )) |
|
160 | 169 | }) |
|
161 | 170 | } |
|
162 | 171 | |
|
163 | 172 | /// Provide access to binary data |
|
164 | 173 | /// |
|
165 | 174 | /// This is needed by FFI layers, for instance to return expected |
|
166 | 175 | /// binary values to Python. |
|
167 | 176 | pub fn as_bytes(&self) -> &[u8] { |
|
168 | 177 | &self.data |
|
169 | 178 | } |
|
170 | 179 | |
|
171 | 180 | pub fn short(&self) -> NodePrefix { |
|
172 | 181 | NodePrefix { |
|
173 | 182 | nybbles_len: SHORT_PREFIX_DEFAULT_NYBBLES_LENGTH, |
|
174 | 183 | data: self.data, |
|
175 | 184 | } |
|
176 | 185 | } |
|
177 | 186 | |
|
178 | 187 | pub fn pad_to_256_bits(&self) -> [u8; 32] { |
|
179 | 188 | let mut bits = [0; 32]; |
|
180 | 189 | bits[..NODE_BYTES_LENGTH].copy_from_slice(&self.data); |
|
181 | 190 | bits |
|
182 | 191 | } |
|
183 | 192 | } |
|
184 | 193 | |
|
185 | 194 | /// The beginning of a binary revision SHA. |
|
186 | 195 | /// |
|
187 | 196 | /// Since it can potentially come from an hexadecimal representation with |
|
188 | 197 | /// odd length, it needs to carry around whether the last 4 bits are relevant |
|
189 | 198 | /// or not. |
|
190 | 199 | #[derive(Debug, PartialEq, Copy, Clone)] |
|
191 | 200 | pub struct NodePrefix { |
|
192 | 201 | /// In `1..=NODE_NYBBLES_LENGTH` |
|
193 | 202 | nybbles_len: u8, |
|
194 | 203 | /// The first `4 * length_in_nybbles` bits are used (considering bits |
|
195 | 204 | /// within a bytes in big-endian: most significant first), the rest |
|
196 | 205 | /// are zero. |
|
197 | 206 | data: NodeData, |
|
198 | 207 | } |
|
199 | 208 | |
|
200 | 209 | impl NodePrefix { |
|
201 | 210 | /// Convert from hexadecimal string representation |
|
202 | 211 | /// |
|
203 | 212 | /// Similarly to `hex::decode`, can be used with Unicode string types |
|
204 | 213 | /// (`String`, `&str`) as well as bytes. |
|
205 | 214 | /// |
|
206 | 215 | /// To be used in FFI and I/O only, in order to facilitate future |
|
207 | 216 | /// changes of hash format. |
|
208 | 217 | pub fn from_hex(hex: impl AsRef<[u8]>) -> Result<Self, FromHexError> { |
|
209 | 218 | let hex = hex.as_ref(); |
|
210 | 219 | let len = hex.len(); |
|
211 | 220 | if len > NODE_NYBBLES_LENGTH || len == 0 { |
|
212 | 221 | return Err(FromHexError); |
|
213 | 222 | } |
|
214 | 223 | |
|
215 | 224 | let mut data = [0; NODE_BYTES_LENGTH]; |
|
216 | 225 | let mut nybbles_len = 0; |
|
217 | 226 | for &ascii_byte in hex { |
|
218 | 227 | let nybble = match char::from(ascii_byte).to_digit(16) { |
|
219 | 228 | Some(digit) => digit as u8, |
|
220 | 229 | None => return Err(FromHexError), |
|
221 | 230 | }; |
|
222 | 231 | // Fill in the upper half of a byte first, then the lower half. |
|
223 | 232 | let shift = if nybbles_len % 2 == 0 { 4 } else { 0 }; |
|
224 | 233 | data[nybbles_len as usize / 2] |= nybble << shift; |
|
225 | 234 | nybbles_len += 1; |
|
226 | 235 | } |
|
227 | 236 | Ok(Self { data, nybbles_len }) |
|
228 | 237 | } |
|
229 | 238 | |
|
230 | 239 | pub fn nybbles_len(&self) -> usize { |
|
231 | 240 | self.nybbles_len as _ |
|
232 | 241 | } |
|
233 | 242 | |
|
234 | 243 | pub fn is_prefix_of(&self, node: &Node) -> bool { |
|
235 | 244 | let full_bytes = self.nybbles_len() / 2; |
|
236 | 245 | if self.data[..full_bytes] != node.data[..full_bytes] { |
|
237 | 246 | return false; |
|
238 | 247 | } |
|
239 | 248 | if self.nybbles_len() % 2 == 0 { |
|
240 | 249 | return true; |
|
241 | 250 | } |
|
242 | 251 | let last = self.nybbles_len() - 1; |
|
243 | 252 | self.get_nybble(last) == node.get_nybble(last) |
|
244 | 253 | } |
|
245 | 254 | |
|
246 | 255 | /// Retrieve the `i`th half-byte from the prefix. |
|
247 | 256 | /// |
|
248 | 257 | /// This is also the `i`th hexadecimal digit in numeric form, |
|
249 | 258 | /// also called a [nybble](https://en.wikipedia.org/wiki/Nibble). |
|
250 | 259 | pub fn get_nybble(&self, i: usize) -> u8 { |
|
251 | 260 | assert!(i < self.nybbles_len()); |
|
252 | 261 | get_nybble(&self.data, i) |
|
253 | 262 | } |
|
254 | 263 | |
|
255 | 264 | fn iter_nybbles(&self) -> impl Iterator<Item = u8> + '_ { |
|
256 | 265 | (0..self.nybbles_len()).map(move |i| get_nybble(&self.data, i)) |
|
257 | 266 | } |
|
258 | 267 | |
|
259 | 268 | /// Return the index first nybble that's different from `node` |
|
260 | 269 | /// |
|
261 | 270 | /// If the return value is `None` that means that `self` is |
|
262 | 271 | /// a prefix of `node`, but the current method is a bit slower |
|
263 | 272 | /// than `is_prefix_of`. |
|
264 | 273 | /// |
|
265 | 274 | /// Returned index is as in `get_nybble`, i.e., starting at 0. |
|
266 | 275 | pub fn first_different_nybble(&self, node: &Node) -> Option<usize> { |
|
267 | 276 | self.iter_nybbles() |
|
268 | 277 | .zip(NodePrefix::from(*node).iter_nybbles()) |
|
269 | 278 | .position(|(a, b)| a != b) |
|
270 | 279 | } |
|
271 | 280 | } |
|
272 | 281 | |
|
273 | 282 | impl fmt::LowerHex for NodePrefix { |
|
274 | 283 | fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result { |
|
275 | 284 | let full_bytes = self.nybbles_len() / 2; |
|
276 | 285 | for &byte in &self.data[..full_bytes] { |
|
277 | 286 | write!(f, "{:02x}", byte)? |
|
278 | 287 | } |
|
279 | 288 | if self.nybbles_len() % 2 == 1 { |
|
280 | 289 | let last = self.nybbles_len() - 1; |
|
281 | 290 | write!(f, "{:x}", self.get_nybble(last))? |
|
282 | 291 | } |
|
283 | 292 | Ok(()) |
|
284 | 293 | } |
|
285 | 294 | } |
|
286 | 295 | |
|
287 | 296 | /// A shortcut for full `Node` references |
|
288 | 297 | impl From<&'_ Node> for NodePrefix { |
|
289 | 298 | fn from(node: &'_ Node) -> Self { |
|
290 | 299 | NodePrefix { |
|
291 | 300 | nybbles_len: node.nybbles_len() as _, |
|
292 | 301 | data: node.data, |
|
293 | 302 | } |
|
294 | 303 | } |
|
295 | 304 | } |
|
296 | 305 | |
|
297 | 306 | /// A shortcut for full `Node` references |
|
298 | 307 | impl From<Node> for NodePrefix { |
|
299 | 308 | fn from(node: Node) -> Self { |
|
300 | 309 | NodePrefix { |
|
301 | 310 | nybbles_len: node.nybbles_len() as _, |
|
302 | 311 | data: node.data, |
|
303 | 312 | } |
|
304 | 313 | } |
|
305 | 314 | } |
|
306 | 315 | |
|
307 | 316 | impl PartialEq<Node> for NodePrefix { |
|
308 | 317 | fn eq(&self, other: &Node) -> bool { |
|
309 | 318 | Self::from(*other) == *self |
|
310 | 319 | } |
|
311 | 320 | } |
|
312 | 321 | |
|
313 | 322 | #[cfg(test)] |
|
314 | 323 | mod tests { |
|
315 | 324 | use super::*; |
|
316 | 325 | |
|
317 | 326 | const SAMPLE_NODE_HEX: &str = "0123456789abcdeffedcba9876543210deadbeef"; |
|
318 | 327 | const SAMPLE_NODE: Node = Node { |
|
319 | 328 | data: [ |
|
320 | 329 | 0x01, 0x23, 0x45, 0x67, 0x89, 0xab, 0xcd, 0xef, 0xfe, 0xdc, 0xba, |
|
321 | 330 | 0x98, 0x76, 0x54, 0x32, 0x10, 0xde, 0xad, 0xbe, 0xef, |
|
322 | 331 | ], |
|
323 | 332 | }; |
|
324 | 333 | |
|
325 | 334 | /// Pad an hexadecimal string to reach `NODE_NYBBLES_LENGTH` |
|
326 | 335 | /// The padding is made with zeros. |
|
327 | 336 | pub fn hex_pad_right(hex: &str) -> String { |
|
328 | 337 | let mut res = hex.to_string(); |
|
329 | 338 | while res.len() < NODE_NYBBLES_LENGTH { |
|
330 | 339 | res.push('0'); |
|
331 | 340 | } |
|
332 | 341 | res |
|
333 | 342 | } |
|
334 | 343 | |
|
335 | 344 | #[test] |
|
336 | 345 | fn test_node_from_hex() { |
|
337 | 346 | let not_hex = "012... oops"; |
|
338 | 347 | let too_short = "0123"; |
|
339 | 348 | let too_long = format!("{}0", SAMPLE_NODE_HEX); |
|
340 | 349 | assert_eq!(Node::from_hex(SAMPLE_NODE_HEX).unwrap(), SAMPLE_NODE); |
|
341 | 350 | assert!(Node::from_hex(not_hex).is_err()); |
|
342 | 351 | assert!(Node::from_hex(too_short).is_err()); |
|
343 | 352 | assert!(Node::from_hex(&too_long).is_err()); |
|
344 | 353 | } |
|
345 | 354 | |
|
346 | 355 | #[test] |
|
347 | 356 | fn test_node_encode_hex() { |
|
348 | 357 | assert_eq!(format!("{:x}", SAMPLE_NODE), SAMPLE_NODE_HEX); |
|
349 | 358 | } |
|
350 | 359 | |
|
351 | 360 | #[test] |
|
352 | 361 | fn test_prefix_from_to_hex() -> Result<(), FromHexError> { |
|
353 | 362 | assert_eq!(format!("{:x}", NodePrefix::from_hex("0e1")?), "0e1"); |
|
354 | 363 | assert_eq!(format!("{:x}", NodePrefix::from_hex("0e1a")?), "0e1a"); |
|
355 | 364 | assert_eq!( |
|
356 | 365 | format!("{:x}", NodePrefix::from_hex(SAMPLE_NODE_HEX)?), |
|
357 | 366 | SAMPLE_NODE_HEX |
|
358 | 367 | ); |
|
359 | 368 | Ok(()) |
|
360 | 369 | } |
|
361 | 370 | |
|
362 | 371 | #[test] |
|
363 | 372 | fn test_prefix_from_hex_errors() { |
|
364 | 373 | assert!(NodePrefix::from_hex("testgr").is_err()); |
|
365 | 374 | let mut long = format!("{:x}", NULL_NODE); |
|
366 | 375 | long.push('c'); |
|
367 | 376 | assert!(NodePrefix::from_hex(&long).is_err()) |
|
368 | 377 | } |
|
369 | 378 | |
|
370 | 379 | #[test] |
|
371 | 380 | fn test_is_prefix_of() -> Result<(), FromHexError> { |
|
372 | 381 | let mut node_data = [0; NODE_BYTES_LENGTH]; |
|
373 | 382 | node_data[0] = 0x12; |
|
374 | 383 | node_data[1] = 0xca; |
|
375 | 384 | let node = Node::from(node_data); |
|
376 | 385 | assert!(NodePrefix::from_hex("12")?.is_prefix_of(&node)); |
|
377 | 386 | assert!(!NodePrefix::from_hex("1a")?.is_prefix_of(&node)); |
|
378 | 387 | assert!(NodePrefix::from_hex("12c")?.is_prefix_of(&node)); |
|
379 | 388 | assert!(!NodePrefix::from_hex("12d")?.is_prefix_of(&node)); |
|
380 | 389 | Ok(()) |
|
381 | 390 | } |
|
382 | 391 | |
|
383 | 392 | #[test] |
|
384 | 393 | fn test_get_nybble() -> Result<(), FromHexError> { |
|
385 | 394 | let prefix = NodePrefix::from_hex("dead6789cafe")?; |
|
386 | 395 | assert_eq!(prefix.get_nybble(0), 13); |
|
387 | 396 | assert_eq!(prefix.get_nybble(7), 9); |
|
388 | 397 | Ok(()) |
|
389 | 398 | } |
|
390 | 399 | |
|
391 | 400 | #[test] |
|
392 | 401 | fn test_first_different_nybble_even_prefix() { |
|
393 | 402 | let prefix = NodePrefix::from_hex("12ca").unwrap(); |
|
394 | 403 | let mut node = Node::from([0; NODE_BYTES_LENGTH]); |
|
395 | 404 | assert_eq!(prefix.first_different_nybble(&node), Some(0)); |
|
396 | 405 | node.data[0] = 0x13; |
|
397 | 406 | assert_eq!(prefix.first_different_nybble(&node), Some(1)); |
|
398 | 407 | node.data[0] = 0x12; |
|
399 | 408 | assert_eq!(prefix.first_different_nybble(&node), Some(2)); |
|
400 | 409 | node.data[1] = 0xca; |
|
401 | 410 | // now it is a prefix |
|
402 | 411 | assert_eq!(prefix.first_different_nybble(&node), None); |
|
403 | 412 | } |
|
404 | 413 | |
|
405 | 414 | #[test] |
|
406 | 415 | fn test_first_different_nybble_odd_prefix() { |
|
407 | 416 | let prefix = NodePrefix::from_hex("12c").unwrap(); |
|
408 | 417 | let mut node = Node::from([0; NODE_BYTES_LENGTH]); |
|
409 | 418 | assert_eq!(prefix.first_different_nybble(&node), Some(0)); |
|
410 | 419 | node.data[0] = 0x13; |
|
411 | 420 | assert_eq!(prefix.first_different_nybble(&node), Some(1)); |
|
412 | 421 | node.data[0] = 0x12; |
|
413 | 422 | assert_eq!(prefix.first_different_nybble(&node), Some(2)); |
|
414 | 423 | node.data[1] = 0xca; |
|
415 | 424 | // now it is a prefix |
|
416 | 425 | assert_eq!(prefix.first_different_nybble(&node), None); |
|
417 | 426 | } |
|
418 | 427 | } |
|
419 | 428 | |
|
420 | 429 | #[cfg(test)] |
|
421 | 430 | pub use tests::hex_pad_right; |
General Comments 0
You need to be logged in to leave comments.
Login now