Show More
@@ -1,709 +1,694 b'' | |||
|
1 | 1 | # dirstatemap.py |
|
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 | |
|
7 | 7 | import errno |
|
8 | 8 | |
|
9 | 9 | from .i18n import _ |
|
10 | 10 | |
|
11 | 11 | from . import ( |
|
12 | 12 | error, |
|
13 | 13 | pathutil, |
|
14 | 14 | policy, |
|
15 | 15 | txnutil, |
|
16 | 16 | util, |
|
17 | 17 | ) |
|
18 | 18 | |
|
19 | 19 | from .dirstateutils import ( |
|
20 | 20 | docket as docketmod, |
|
21 | 21 | v2, |
|
22 | 22 | ) |
|
23 | 23 | |
|
24 | 24 | parsers = policy.importmod('parsers') |
|
25 | 25 | rustmod = policy.importrust('dirstate') |
|
26 | 26 | |
|
27 | 27 | propertycache = util.propertycache |
|
28 | 28 | |
|
29 | 29 | if rustmod is None: |
|
30 | 30 | DirstateItem = parsers.DirstateItem |
|
31 | 31 | else: |
|
32 | 32 | DirstateItem = rustmod.DirstateItem |
|
33 | 33 | |
|
34 | 34 | rangemask = 0x7FFFFFFF |
|
35 | 35 | |
|
36 | 36 | |
|
37 | 37 | class _dirstatemapcommon: |
|
38 | 38 | """ |
|
39 | 39 | Methods that are identical for both implementations of the dirstatemap |
|
40 | 40 | class, with and without Rust extensions enabled. |
|
41 | 41 | """ |
|
42 | 42 | |
|
43 | 43 | # please pytype |
|
44 | 44 | |
|
45 | 45 | _map = None |
|
46 | 46 | copymap = None |
|
47 | 47 | |
|
48 | 48 | def __init__(self, ui, opener, root, nodeconstants, use_dirstate_v2): |
|
49 | 49 | self._use_dirstate_v2 = use_dirstate_v2 |
|
50 | 50 | self._nodeconstants = nodeconstants |
|
51 | 51 | self._ui = ui |
|
52 | 52 | self._opener = opener |
|
53 | 53 | self._root = root |
|
54 | 54 | self._filename = b'dirstate' |
|
55 | 55 | self._nodelen = 20 # Also update Rust code when changing this! |
|
56 | 56 | self._parents = None |
|
57 | 57 | self._dirtyparents = False |
|
58 | 58 | self._docket = None |
|
59 | 59 | |
|
60 | 60 | # for consistent view between _pl() and _read() invocations |
|
61 | 61 | self._pendingmode = None |
|
62 | 62 | |
|
63 | 63 | def preload(self): |
|
64 | 64 | """Loads the underlying data, if it's not already loaded""" |
|
65 | 65 | self._map |
|
66 | 66 | |
|
67 | 67 | def get(self, key, default=None): |
|
68 | 68 | return self._map.get(key, default) |
|
69 | 69 | |
|
70 | 70 | def __len__(self): |
|
71 | 71 | return len(self._map) |
|
72 | 72 | |
|
73 | 73 | def __iter__(self): |
|
74 | 74 | return iter(self._map) |
|
75 | 75 | |
|
76 | 76 | def __contains__(self, key): |
|
77 | 77 | return key in self._map |
|
78 | 78 | |
|
79 | 79 | def __getitem__(self, item): |
|
80 | 80 | return self._map[item] |
|
81 | 81 | |
|
82 | 82 | ### disk interaction |
|
83 | 83 | |
|
84 | 84 | def _opendirstatefile(self): |
|
85 | 85 | fp, mode = txnutil.trypending(self._root, self._opener, self._filename) |
|
86 | 86 | if self._pendingmode is not None and self._pendingmode != mode: |
|
87 | 87 | fp.close() |
|
88 | 88 | raise error.Abort( |
|
89 | 89 | _(b'working directory state may be changed parallelly') |
|
90 | 90 | ) |
|
91 | 91 | self._pendingmode = mode |
|
92 | 92 | return fp |
|
93 | 93 | |
|
94 | 94 | def _readdirstatefile(self, size=-1): |
|
95 | 95 | try: |
|
96 | 96 | with self._opendirstatefile() as fp: |
|
97 | 97 | return fp.read(size) |
|
98 | 98 | except IOError as err: |
|
99 | 99 | if err.errno != errno.ENOENT: |
|
100 | 100 | raise |
|
101 | 101 | # File doesn't exist, so the current state is empty |
|
102 | 102 | return b'' |
|
103 | 103 | |
|
104 | 104 | @property |
|
105 | 105 | def docket(self): |
|
106 | 106 | if not self._docket: |
|
107 | 107 | if not self._use_dirstate_v2: |
|
108 | 108 | raise error.ProgrammingError( |
|
109 | 109 | b'dirstate only has a docket in v2 format' |
|
110 | 110 | ) |
|
111 | 111 | self._docket = docketmod.DirstateDocket.parse( |
|
112 | 112 | self._readdirstatefile(), self._nodeconstants |
|
113 | 113 | ) |
|
114 | 114 | return self._docket |
|
115 | 115 | |
|
116 | 116 | def write_v2_no_append(self, tr, st, meta, packed): |
|
117 | 117 | old_docket = self.docket |
|
118 | 118 | new_docket = docketmod.DirstateDocket.with_new_uuid( |
|
119 | 119 | self.parents(), len(packed), meta |
|
120 | 120 | ) |
|
121 | 121 | data_filename = new_docket.data_filename() |
|
122 | 122 | if tr: |
|
123 | 123 | tr.add(data_filename, 0) |
|
124 | 124 | self._opener.write(data_filename, packed) |
|
125 | 125 | # Write the new docket after the new data file has been |
|
126 | 126 | # written. Because `st` was opened with `atomictemp=True`, |
|
127 | 127 | # the actual `.hg/dirstate` file is only affected on close. |
|
128 | 128 | st.write(new_docket.serialize()) |
|
129 | 129 | st.close() |
|
130 | 130 | # Remove the old data file after the new docket pointing to |
|
131 | 131 | # the new data file was written. |
|
132 | 132 | if old_docket.uuid: |
|
133 | 133 | data_filename = old_docket.data_filename() |
|
134 | 134 | unlink = lambda _tr=None: self._opener.unlink(data_filename) |
|
135 | 135 | if tr: |
|
136 | 136 | category = b"dirstate-v2-clean-" + old_docket.uuid |
|
137 | 137 | tr.addpostclose(category, unlink) |
|
138 | 138 | else: |
|
139 | 139 | unlink() |
|
140 | 140 | self._docket = new_docket |
|
141 | 141 | |
|
142 | 142 | ### reading/setting parents |
|
143 | 143 | |
|
144 | 144 | def parents(self): |
|
145 | 145 | if not self._parents: |
|
146 | 146 | if self._use_dirstate_v2: |
|
147 | 147 | self._parents = self.docket.parents |
|
148 | 148 | else: |
|
149 | 149 | read_len = self._nodelen * 2 |
|
150 | 150 | st = self._readdirstatefile(read_len) |
|
151 | 151 | l = len(st) |
|
152 | 152 | if l == read_len: |
|
153 | 153 | self._parents = ( |
|
154 | 154 | st[: self._nodelen], |
|
155 | 155 | st[self._nodelen : 2 * self._nodelen], |
|
156 | 156 | ) |
|
157 | 157 | elif l == 0: |
|
158 | 158 | self._parents = ( |
|
159 | 159 | self._nodeconstants.nullid, |
|
160 | 160 | self._nodeconstants.nullid, |
|
161 | 161 | ) |
|
162 | 162 | else: |
|
163 | 163 | raise error.Abort( |
|
164 | 164 | _(b'working directory state appears damaged!') |
|
165 | 165 | ) |
|
166 | 166 | |
|
167 | 167 | return self._parents |
|
168 | 168 | |
|
169 | 169 | |
|
170 | 170 | class dirstatemap(_dirstatemapcommon): |
|
171 | 171 | """Map encapsulating the dirstate's contents. |
|
172 | 172 | |
|
173 | 173 | The dirstate contains the following state: |
|
174 | 174 | |
|
175 | 175 | - `identity` is the identity of the dirstate file, which can be used to |
|
176 | 176 | detect when changes have occurred to the dirstate file. |
|
177 | 177 | |
|
178 | 178 | - `parents` is a pair containing the parents of the working copy. The |
|
179 | 179 | parents are updated by calling `setparents`. |
|
180 | 180 | |
|
181 | 181 | - the state map maps filenames to tuples of (state, mode, size, mtime), |
|
182 | 182 | where state is a single character representing 'normal', 'added', |
|
183 | 183 | 'removed', or 'merged'. It is read by treating the dirstate as a |
|
184 | 184 | dict. File state is updated by calling various methods (see each |
|
185 | 185 | documentation for details): |
|
186 | 186 | |
|
187 | 187 | - `reset_state`, |
|
188 | 188 | - `set_tracked` |
|
189 | 189 | - `set_untracked` |
|
190 | 190 | - `set_clean` |
|
191 | 191 | - `set_possibly_dirty` |
|
192 | 192 | |
|
193 | 193 | - `copymap` maps destination filenames to their source filename. |
|
194 | 194 | |
|
195 | 195 | The dirstate also provides the following views onto the state: |
|
196 | 196 | |
|
197 | 197 | - `filefoldmap` is a dict mapping normalized filenames to the denormalized |
|
198 | 198 | form that they appear as in the dirstate. |
|
199 | 199 | |
|
200 | 200 | - `dirfoldmap` is a dict mapping normalized directory names to the |
|
201 | 201 | denormalized form that they appear as in the dirstate. |
|
202 | 202 | """ |
|
203 | 203 | |
|
204 | 204 | ### Core data storage and access |
|
205 | 205 | |
|
206 | 206 | @propertycache |
|
207 | 207 | def _map(self): |
|
208 | 208 | self._map = {} |
|
209 | 209 | self.read() |
|
210 | 210 | return self._map |
|
211 | 211 | |
|
212 | 212 | @propertycache |
|
213 | 213 | def copymap(self): |
|
214 | 214 | self.copymap = {} |
|
215 | 215 | self._map |
|
216 | 216 | return self.copymap |
|
217 | 217 | |
|
218 | 218 | def clear(self): |
|
219 | 219 | self._map.clear() |
|
220 | 220 | self.copymap.clear() |
|
221 | 221 | self.setparents(self._nodeconstants.nullid, self._nodeconstants.nullid) |
|
222 | 222 | util.clearcachedproperty(self, b"_dirs") |
|
223 | 223 | util.clearcachedproperty(self, b"_alldirs") |
|
224 | 224 | util.clearcachedproperty(self, b"filefoldmap") |
|
225 | 225 | util.clearcachedproperty(self, b"dirfoldmap") |
|
226 | 226 | |
|
227 | 227 | def items(self): |
|
228 | 228 | return self._map.items() |
|
229 | 229 | |
|
230 | 230 | # forward for python2,3 compat |
|
231 | 231 | iteritems = items |
|
232 | 232 | |
|
233 | 233 | def debug_iter(self, all): |
|
234 | 234 | """ |
|
235 | 235 | Return an iterator of (filename, state, mode, size, mtime) tuples |
|
236 | 236 | |
|
237 | 237 | `all` is unused when Rust is not enabled |
|
238 | 238 | """ |
|
239 | 239 | for (filename, item) in self.items(): |
|
240 | 240 | yield (filename, item.state, item.mode, item.size, item.mtime) |
|
241 | 241 | |
|
242 | 242 | def keys(self): |
|
243 | 243 | return self._map.keys() |
|
244 | 244 | |
|
245 | 245 | ### reading/setting parents |
|
246 | 246 | |
|
247 | 247 | def setparents(self, p1, p2, fold_p2=False): |
|
248 | 248 | self._parents = (p1, p2) |
|
249 | 249 | self._dirtyparents = True |
|
250 | 250 | copies = {} |
|
251 | 251 | if fold_p2: |
|
252 | 252 | for f, s in self._map.items(): |
|
253 | 253 | # Discard "merged" markers when moving away from a merge state |
|
254 | 254 | if s.p2_info: |
|
255 | 255 | source = self.copymap.pop(f, None) |
|
256 | 256 | if source: |
|
257 | 257 | copies[f] = source |
|
258 | 258 | s.drop_merge_data() |
|
259 | 259 | return copies |
|
260 | 260 | |
|
261 | 261 | ### disk interaction |
|
262 | 262 | |
|
263 | 263 | def read(self): |
|
264 | 264 | # ignore HG_PENDING because identity is used only for writing |
|
265 | 265 | self.identity = util.filestat.frompath( |
|
266 | 266 | self._opener.join(self._filename) |
|
267 | 267 | ) |
|
268 | 268 | |
|
269 | 269 | if self._use_dirstate_v2: |
|
270 | 270 | if not self.docket.uuid: |
|
271 | 271 | return |
|
272 | 272 | st = self._opener.read(self.docket.data_filename()) |
|
273 | 273 | else: |
|
274 | 274 | st = self._readdirstatefile() |
|
275 | 275 | |
|
276 | 276 | if not st: |
|
277 | 277 | return |
|
278 | 278 | |
|
279 | 279 | # TODO: adjust this estimate for dirstate-v2 |
|
280 | 280 | if util.safehasattr(parsers, b'dict_new_presized'): |
|
281 | 281 | # Make an estimate of the number of files in the dirstate based on |
|
282 | 282 | # its size. This trades wasting some memory for avoiding costly |
|
283 | 283 | # resizes. Each entry have a prefix of 17 bytes followed by one or |
|
284 | 284 | # two path names. Studies on various large-scale real-world repositories |
|
285 | 285 | # found 54 bytes a reasonable upper limit for the average path names. |
|
286 | 286 | # Copy entries are ignored for the sake of this estimate. |
|
287 | 287 | self._map = parsers.dict_new_presized(len(st) // 71) |
|
288 | 288 | |
|
289 | 289 | # Python's garbage collector triggers a GC each time a certain number |
|
290 | 290 | # of container objects (the number being defined by |
|
291 | 291 | # gc.get_threshold()) are allocated. parse_dirstate creates a tuple |
|
292 | 292 | # for each file in the dirstate. The C version then immediately marks |
|
293 | 293 | # them as not to be tracked by the collector. However, this has no |
|
294 | 294 | # effect on when GCs are triggered, only on what objects the GC looks |
|
295 | 295 | # into. This means that O(number of files) GCs are unavoidable. |
|
296 | 296 | # Depending on when in the process's lifetime the dirstate is parsed, |
|
297 | 297 | # this can get very expensive. As a workaround, disable GC while |
|
298 | 298 | # parsing the dirstate. |
|
299 | 299 | # |
|
300 | 300 | # (we cannot decorate the function directly since it is in a C module) |
|
301 | 301 | if self._use_dirstate_v2: |
|
302 | 302 | p = self.docket.parents |
|
303 | 303 | meta = self.docket.tree_metadata |
|
304 | 304 | parse_dirstate = util.nogc(v2.parse_dirstate) |
|
305 | 305 | parse_dirstate(self._map, self.copymap, st, meta) |
|
306 | 306 | else: |
|
307 | 307 | parse_dirstate = util.nogc(parsers.parse_dirstate) |
|
308 | 308 | p = parse_dirstate(self._map, self.copymap, st) |
|
309 | 309 | if not self._dirtyparents: |
|
310 | 310 | self.setparents(*p) |
|
311 | 311 | |
|
312 | 312 | # Avoid excess attribute lookups by fast pathing certain checks |
|
313 | 313 | self.__contains__ = self._map.__contains__ |
|
314 | 314 | self.__getitem__ = self._map.__getitem__ |
|
315 | 315 | self.get = self._map.get |
|
316 | 316 | |
|
317 | 317 | def write(self, tr, st): |
|
318 | 318 | if self._use_dirstate_v2: |
|
319 | 319 | packed, meta = v2.pack_dirstate(self._map, self.copymap) |
|
320 | 320 | self.write_v2_no_append(tr, st, meta, packed) |
|
321 | 321 | else: |
|
322 | 322 | packed = parsers.pack_dirstate( |
|
323 | 323 | self._map, self.copymap, self.parents() |
|
324 | 324 | ) |
|
325 | 325 | st.write(packed) |
|
326 | 326 | st.close() |
|
327 | 327 | self._dirtyparents = False |
|
328 | 328 | |
|
329 | 329 | @propertycache |
|
330 | 330 | def identity(self): |
|
331 | 331 | self._map |
|
332 | 332 | return self.identity |
|
333 | 333 | |
|
334 | 334 | ### code related to maintaining and accessing "extra" property |
|
335 | 335 | # (e.g. "has_dir") |
|
336 | 336 | |
|
337 | 337 | def _dirs_incr(self, filename, old_entry=None): |
|
338 | 338 | """increment the dirstate counter if applicable""" |
|
339 | 339 | if ( |
|
340 | 340 | old_entry is None or old_entry.removed |
|
341 | 341 | ) and "_dirs" in self.__dict__: |
|
342 | 342 | self._dirs.addpath(filename) |
|
343 | 343 | if old_entry is None and "_alldirs" in self.__dict__: |
|
344 | 344 | self._alldirs.addpath(filename) |
|
345 | 345 | |
|
346 | 346 | def _dirs_decr(self, filename, old_entry=None, remove_variant=False): |
|
347 | 347 | """decrement the dirstate counter if applicable""" |
|
348 | 348 | if old_entry is not None: |
|
349 | 349 | if "_dirs" in self.__dict__ and not old_entry.removed: |
|
350 | 350 | self._dirs.delpath(filename) |
|
351 | 351 | if "_alldirs" in self.__dict__ and not remove_variant: |
|
352 | 352 | self._alldirs.delpath(filename) |
|
353 | 353 | elif remove_variant and "_alldirs" in self.__dict__: |
|
354 | 354 | self._alldirs.addpath(filename) |
|
355 | 355 | if "filefoldmap" in self.__dict__: |
|
356 | 356 | normed = util.normcase(filename) |
|
357 | 357 | self.filefoldmap.pop(normed, None) |
|
358 | 358 | |
|
359 | 359 | @propertycache |
|
360 | 360 | def filefoldmap(self): |
|
361 | 361 | """Returns a dictionary mapping normalized case paths to their |
|
362 | 362 | non-normalized versions. |
|
363 | 363 | """ |
|
364 | 364 | try: |
|
365 | 365 | makefilefoldmap = parsers.make_file_foldmap |
|
366 | 366 | except AttributeError: |
|
367 | 367 | pass |
|
368 | 368 | else: |
|
369 | 369 | return makefilefoldmap( |
|
370 | 370 | self._map, util.normcasespec, util.normcasefallback |
|
371 | 371 | ) |
|
372 | 372 | |
|
373 | 373 | f = {} |
|
374 | 374 | normcase = util.normcase |
|
375 | 375 | for name, s in self._map.items(): |
|
376 | 376 | if not s.removed: |
|
377 | 377 | f[normcase(name)] = name |
|
378 | 378 | f[b'.'] = b'.' # prevents useless util.fspath() invocation |
|
379 | 379 | return f |
|
380 | 380 | |
|
381 | 381 | @propertycache |
|
382 | 382 | def dirfoldmap(self): |
|
383 | 383 | f = {} |
|
384 | 384 | normcase = util.normcase |
|
385 | 385 | for name in self._dirs: |
|
386 | 386 | f[normcase(name)] = name |
|
387 | 387 | return f |
|
388 | 388 | |
|
389 | 389 | def hastrackeddir(self, d): |
|
390 | 390 | """ |
|
391 | 391 | Returns True if the dirstate contains a tracked (not removed) file |
|
392 | 392 | in this directory. |
|
393 | 393 | """ |
|
394 | 394 | return d in self._dirs |
|
395 | 395 | |
|
396 | 396 | def hasdir(self, d): |
|
397 | 397 | """ |
|
398 | 398 | Returns True if the dirstate contains a file (tracked or removed) |
|
399 | 399 | in this directory. |
|
400 | 400 | """ |
|
401 | 401 | return d in self._alldirs |
|
402 | 402 | |
|
403 | 403 | @propertycache |
|
404 | 404 | def _dirs(self): |
|
405 | 405 | return pathutil.dirs(self._map, only_tracked=True) |
|
406 | 406 | |
|
407 | 407 | @propertycache |
|
408 | 408 | def _alldirs(self): |
|
409 | 409 | return pathutil.dirs(self._map) |
|
410 | 410 | |
|
411 | 411 | ### code related to manipulation of entries and copy-sources |
|
412 | 412 | |
|
413 | 413 | def reset_state( |
|
414 | 414 | self, |
|
415 | 415 | filename, |
|
416 | 416 | wc_tracked=False, |
|
417 | 417 | p1_tracked=False, |
|
418 | 418 | p2_info=False, |
|
419 | 419 | has_meaningful_mtime=True, |
|
420 | 420 | parentfiledata=None, |
|
421 | 421 | ): |
|
422 | 422 | """Set a entry to a given state, diregarding all previous state |
|
423 | 423 | |
|
424 | 424 | This is to be used by the part of the dirstate API dedicated to |
|
425 | 425 | adjusting the dirstate after a update/merge. |
|
426 | 426 | |
|
427 | 427 | note: calling this might result to no entry existing at all if the |
|
428 | 428 | dirstate map does not see any point at having one for this file |
|
429 | 429 | anymore. |
|
430 | 430 | """ |
|
431 | 431 | # copy information are now outdated |
|
432 | 432 | # (maybe new information should be in directly passed to this function) |
|
433 | 433 | self.copymap.pop(filename, None) |
|
434 | 434 | |
|
435 | 435 | if not (p1_tracked or p2_info or wc_tracked): |
|
436 | 436 | old_entry = self._map.get(filename) |
|
437 | 437 | self._drop_entry(filename) |
|
438 | 438 | self._dirs_decr(filename, old_entry=old_entry) |
|
439 | 439 | return |
|
440 | 440 | |
|
441 | 441 | old_entry = self._map.get(filename) |
|
442 | 442 | self._dirs_incr(filename, old_entry) |
|
443 | 443 | entry = DirstateItem( |
|
444 | 444 | wc_tracked=wc_tracked, |
|
445 | 445 | p1_tracked=p1_tracked, |
|
446 | 446 | p2_info=p2_info, |
|
447 | 447 | has_meaningful_mtime=has_meaningful_mtime, |
|
448 | 448 | parentfiledata=parentfiledata, |
|
449 | 449 | ) |
|
450 | 450 | self._map[filename] = entry |
|
451 | 451 | |
|
452 | 452 | def set_tracked(self, filename): |
|
453 | 453 | new = False |
|
454 | 454 | entry = self.get(filename) |
|
455 | 455 | if entry is None: |
|
456 | 456 | self._dirs_incr(filename) |
|
457 | 457 | entry = DirstateItem( |
|
458 | 458 | wc_tracked=True, |
|
459 | 459 | ) |
|
460 | 460 | |
|
461 | 461 | self._map[filename] = entry |
|
462 | 462 | new = True |
|
463 | 463 | elif not entry.tracked: |
|
464 | 464 | self._dirs_incr(filename, entry) |
|
465 | 465 | entry.set_tracked() |
|
466 | 466 | self._refresh_entry(filename, entry) |
|
467 | 467 | new = True |
|
468 | 468 | else: |
|
469 | 469 | # XXX This is probably overkill for more case, but we need this to |
|
470 | 470 | # fully replace the `normallookup` call with `set_tracked` one. |
|
471 | 471 | # Consider smoothing this in the future. |
|
472 | 472 | entry.set_possibly_dirty() |
|
473 | 473 | self._refresh_entry(filename, entry) |
|
474 | 474 | return new |
|
475 | 475 | |
|
476 | 476 | def set_untracked(self, f): |
|
477 | 477 | """Mark a file as no longer tracked in the dirstate map""" |
|
478 | 478 | entry = self.get(f) |
|
479 | 479 | if entry is None: |
|
480 | 480 | return False |
|
481 | 481 | else: |
|
482 | 482 | self._dirs_decr(f, old_entry=entry, remove_variant=not entry.added) |
|
483 | 483 | if not entry.p2_info: |
|
484 | 484 | self.copymap.pop(f, None) |
|
485 | 485 | entry.set_untracked() |
|
486 | 486 | self._refresh_entry(f, entry) |
|
487 | 487 | return True |
|
488 | 488 | |
|
489 | 489 | def set_clean(self, filename, mode, size, mtime): |
|
490 | 490 | """mark a file as back to a clean state""" |
|
491 | 491 | entry = self[filename] |
|
492 | 492 | size = size & rangemask |
|
493 | 493 | entry.set_clean(mode, size, mtime) |
|
494 | 494 | self._refresh_entry(filename, entry) |
|
495 | 495 | self.copymap.pop(filename, None) |
|
496 | 496 | |
|
497 | 497 | def set_possibly_dirty(self, filename): |
|
498 | 498 | """record that the current state of the file on disk is unknown""" |
|
499 | 499 | entry = self[filename] |
|
500 | 500 | entry.set_possibly_dirty() |
|
501 | 501 | self._refresh_entry(filename, entry) |
|
502 | 502 | |
|
503 | 503 | def _refresh_entry(self, f, entry): |
|
504 | 504 | """record updated state of an entry""" |
|
505 | 505 | if not entry.any_tracked: |
|
506 | 506 | self._map.pop(f, None) |
|
507 | 507 | |
|
508 | 508 | def _drop_entry(self, f): |
|
509 | 509 | """remove any entry for file f |
|
510 | 510 | |
|
511 | 511 | This should also drop associated copy information |
|
512 | 512 | |
|
513 | 513 | The fact we actually need to drop it is the responsability of the caller""" |
|
514 | 514 | self._map.pop(f, None) |
|
515 | 515 | self.copymap.pop(f, None) |
|
516 | 516 | |
|
517 | 517 | |
|
518 | 518 | if rustmod is not None: |
|
519 | 519 | |
|
520 | 520 | class dirstatemap(_dirstatemapcommon): |
|
521 | 521 | |
|
522 | 522 | ### Core data storage and access |
|
523 | 523 | |
|
524 | 524 | @propertycache |
|
525 | 525 | def _map(self): |
|
526 | 526 | """ |
|
527 | 527 | Fills the Dirstatemap when called. |
|
528 | 528 | """ |
|
529 | 529 | # ignore HG_PENDING because identity is used only for writing |
|
530 | 530 | self.identity = util.filestat.frompath( |
|
531 | 531 | self._opener.join(self._filename) |
|
532 | 532 | ) |
|
533 | 533 | |
|
534 | 534 | if self._use_dirstate_v2: |
|
535 | 535 | if self.docket.uuid: |
|
536 | 536 | # TODO: use mmap when possible |
|
537 | 537 | data = self._opener.read(self.docket.data_filename()) |
|
538 | 538 | else: |
|
539 | 539 | data = b'' |
|
540 | 540 | self._map = rustmod.DirstateMap.new_v2( |
|
541 | 541 | data, self.docket.data_size, self.docket.tree_metadata |
|
542 | 542 | ) |
|
543 | 543 | parents = self.docket.parents |
|
544 | 544 | else: |
|
545 | 545 | self._map, parents = rustmod.DirstateMap.new_v1( |
|
546 | 546 | self._readdirstatefile() |
|
547 | 547 | ) |
|
548 | 548 | |
|
549 | 549 | if parents and not self._dirtyparents: |
|
550 | 550 | self.setparents(*parents) |
|
551 | 551 | |
|
552 | 552 | self.__contains__ = self._map.__contains__ |
|
553 | 553 | self.__getitem__ = self._map.__getitem__ |
|
554 | 554 | self.get = self._map.get |
|
555 | 555 | return self._map |
|
556 | 556 | |
|
557 | 557 | @property |
|
558 | 558 | def copymap(self): |
|
559 | 559 | return self._map.copymap() |
|
560 | 560 | |
|
561 | 561 | def debug_iter(self, all): |
|
562 | 562 | """ |
|
563 | 563 | Return an iterator of (filename, state, mode, size, mtime) tuples |
|
564 | 564 | |
|
565 | 565 | `all`: also include with `state == b' '` dirstate tree nodes that |
|
566 | 566 | don't have an associated `DirstateItem`. |
|
567 | 567 | |
|
568 | 568 | """ |
|
569 | 569 | return self._map.debug_iter(all) |
|
570 | 570 | |
|
571 | 571 | def clear(self): |
|
572 | 572 | self._map.clear() |
|
573 | 573 | self.setparents( |
|
574 | 574 | self._nodeconstants.nullid, self._nodeconstants.nullid |
|
575 | 575 | ) |
|
576 | 576 | util.clearcachedproperty(self, b"_dirs") |
|
577 | 577 | util.clearcachedproperty(self, b"_alldirs") |
|
578 | 578 | util.clearcachedproperty(self, b"dirfoldmap") |
|
579 | 579 | |
|
580 | 580 | def items(self): |
|
581 | 581 | return self._map.items() |
|
582 | 582 | |
|
583 | 583 | # forward for python2,3 compat |
|
584 | 584 | iteritems = items |
|
585 | 585 | |
|
586 | 586 | def keys(self): |
|
587 | 587 | return iter(self._map) |
|
588 | 588 | |
|
589 | 589 | ### reading/setting parents |
|
590 | 590 | |
|
591 | 591 | def setparents(self, p1, p2, fold_p2=False): |
|
592 | 592 | self._parents = (p1, p2) |
|
593 | 593 | self._dirtyparents = True |
|
594 | 594 | copies = {} |
|
595 | 595 | if fold_p2: |
|
596 | # Collect into an intermediate list to avoid a `RuntimeError` | |
|
597 | # exception due to mutation during iteration. | |
|
598 | # TODO: move this the whole loop to Rust where `iter_mut` | |
|
599 | # enables in-place mutation of elements of a collection while | |
|
600 | # iterating it, without mutating the collection itself. | |
|
601 | files_with_p2_info = [ | |
|
602 | f for f, s in self._map.items() if s.p2_info | |
|
603 | ] | |
|
604 | rust_map = self._map | |
|
605 | for f in files_with_p2_info: | |
|
606 | e = rust_map.get(f) | |
|
607 | source = self.copymap.pop(f, None) | |
|
608 | if source: | |
|
609 | copies[f] = source | |
|
610 | e.drop_merge_data() | |
|
611 | rust_map.set_dirstate_item(f, e) | |
|
596 | copies = self._map.setparents_fixup() | |
|
612 | 597 | return copies |
|
613 | 598 | |
|
614 | 599 | ### disk interaction |
|
615 | 600 | |
|
616 | 601 | @propertycache |
|
617 | 602 | def identity(self): |
|
618 | 603 | self._map |
|
619 | 604 | return self.identity |
|
620 | 605 | |
|
621 | 606 | def write(self, tr, st): |
|
622 | 607 | if not self._use_dirstate_v2: |
|
623 | 608 | p1, p2 = self.parents() |
|
624 | 609 | packed = self._map.write_v1(p1, p2) |
|
625 | 610 | st.write(packed) |
|
626 | 611 | st.close() |
|
627 | 612 | self._dirtyparents = False |
|
628 | 613 | return |
|
629 | 614 | |
|
630 | 615 | # We can only append to an existing data file if there is one |
|
631 | 616 | can_append = self.docket.uuid is not None |
|
632 | 617 | packed, meta, append = self._map.write_v2(can_append) |
|
633 | 618 | if append: |
|
634 | 619 | docket = self.docket |
|
635 | 620 | data_filename = docket.data_filename() |
|
636 | 621 | if tr: |
|
637 | 622 | tr.add(data_filename, docket.data_size) |
|
638 | 623 | with self._opener(data_filename, b'r+b') as fp: |
|
639 | 624 | fp.seek(docket.data_size) |
|
640 | 625 | assert fp.tell() == docket.data_size |
|
641 | 626 | written = fp.write(packed) |
|
642 | 627 | if written is not None: # py2 may return None |
|
643 | 628 | assert written == len(packed), (written, len(packed)) |
|
644 | 629 | docket.data_size += len(packed) |
|
645 | 630 | docket.parents = self.parents() |
|
646 | 631 | docket.tree_metadata = meta |
|
647 | 632 | st.write(docket.serialize()) |
|
648 | 633 | st.close() |
|
649 | 634 | else: |
|
650 | 635 | self.write_v2_no_append(tr, st, meta, packed) |
|
651 | 636 | # Reload from the newly-written file |
|
652 | 637 | util.clearcachedproperty(self, b"_map") |
|
653 | 638 | self._dirtyparents = False |
|
654 | 639 | |
|
655 | 640 | ### code related to maintaining and accessing "extra" property |
|
656 | 641 | # (e.g. "has_dir") |
|
657 | 642 | |
|
658 | 643 | @propertycache |
|
659 | 644 | def filefoldmap(self): |
|
660 | 645 | """Returns a dictionary mapping normalized case paths to their |
|
661 | 646 | non-normalized versions. |
|
662 | 647 | """ |
|
663 | 648 | return self._map.filefoldmapasdict() |
|
664 | 649 | |
|
665 | 650 | def hastrackeddir(self, d): |
|
666 | 651 | return self._map.hastrackeddir(d) |
|
667 | 652 | |
|
668 | 653 | def hasdir(self, d): |
|
669 | 654 | return self._map.hasdir(d) |
|
670 | 655 | |
|
671 | 656 | @propertycache |
|
672 | 657 | def dirfoldmap(self): |
|
673 | 658 | f = {} |
|
674 | 659 | normcase = util.normcase |
|
675 | 660 | for name in self._map.tracked_dirs(): |
|
676 | 661 | f[normcase(name)] = name |
|
677 | 662 | return f |
|
678 | 663 | |
|
679 | 664 | ### code related to manipulation of entries and copy-sources |
|
680 | 665 | |
|
681 | 666 | def set_tracked(self, f): |
|
682 | 667 | return self._map.set_tracked(f) |
|
683 | 668 | |
|
684 | 669 | def set_untracked(self, f): |
|
685 | 670 | return self._map.set_untracked(f) |
|
686 | 671 | |
|
687 | 672 | def set_clean(self, filename, mode, size, mtime): |
|
688 | 673 | self._map.set_clean(filename, mode, size, mtime) |
|
689 | 674 | |
|
690 | 675 | def set_possibly_dirty(self, f): |
|
691 | 676 | self._map.set_possibly_dirty(f) |
|
692 | 677 | |
|
693 | 678 | def reset_state( |
|
694 | 679 | self, |
|
695 | 680 | filename, |
|
696 | 681 | wc_tracked=False, |
|
697 | 682 | p1_tracked=False, |
|
698 | 683 | p2_info=False, |
|
699 | 684 | has_meaningful_mtime=True, |
|
700 | 685 | parentfiledata=None, |
|
701 | 686 | ): |
|
702 | 687 | return self._map.reset_state( |
|
703 | 688 | filename, |
|
704 | 689 | wc_tracked, |
|
705 | 690 | p1_tracked, |
|
706 | 691 | p2_info, |
|
707 | 692 | has_meaningful_mtime, |
|
708 | 693 | parentfiledata, |
|
709 | 694 | ) |
@@ -1,1404 +1,1439 b'' | |||
|
1 | 1 | use bytes_cast::BytesCast; |
|
2 | 2 | use micro_timer::timed; |
|
3 | 3 | use std::borrow::Cow; |
|
4 | 4 | use std::path::PathBuf; |
|
5 | 5 | |
|
6 | 6 | use super::on_disk; |
|
7 | 7 | use super::on_disk::DirstateV2ParseError; |
|
8 | 8 | use super::owning::OwningDirstateMap; |
|
9 | 9 | use super::path_with_basename::WithBasename; |
|
10 | 10 | use crate::dirstate::parsers::pack_entry; |
|
11 | 11 | use crate::dirstate::parsers::packed_entry_size; |
|
12 | 12 | use crate::dirstate::parsers::parse_dirstate_entries; |
|
13 | 13 | use crate::dirstate::CopyMapIter; |
|
14 | 14 | use crate::dirstate::DirstateV2Data; |
|
15 | 15 | use crate::dirstate::ParentFileData; |
|
16 | 16 | use crate::dirstate::StateMapIter; |
|
17 | 17 | use crate::dirstate::TruncatedTimestamp; |
|
18 | 18 | use crate::matchers::Matcher; |
|
19 | 19 | use crate::utils::hg_path::{HgPath, HgPathBuf}; |
|
20 | 20 | use crate::DirstateEntry; |
|
21 | 21 | use crate::DirstateError; |
|
22 | 22 | use crate::DirstateMapError; |
|
23 | 23 | use crate::DirstateParents; |
|
24 | 24 | use crate::DirstateStatus; |
|
25 | 25 | use crate::EntryState; |
|
26 | 26 | use crate::FastHashbrownMap as FastHashMap; |
|
27 | 27 | use crate::PatternFileWarning; |
|
28 | 28 | use crate::StatusError; |
|
29 | 29 | use crate::StatusOptions; |
|
30 | 30 | |
|
31 | 31 | /// Append to an existing data file if the amount of unreachable data (not used |
|
32 | 32 | /// anymore) is less than this fraction of the total amount of existing data. |
|
33 | 33 | const ACCEPTABLE_UNREACHABLE_BYTES_RATIO: f32 = 0.5; |
|
34 | 34 | |
|
35 | 35 | pub struct DirstateMap<'on_disk> { |
|
36 | 36 | /// Contents of the `.hg/dirstate` file |
|
37 | 37 | pub(super) on_disk: &'on_disk [u8], |
|
38 | 38 | |
|
39 | 39 | pub(super) root: ChildNodes<'on_disk>, |
|
40 | 40 | |
|
41 | 41 | /// Number of nodes anywhere in the tree that have `.entry.is_some()`. |
|
42 | 42 | pub(super) nodes_with_entry_count: u32, |
|
43 | 43 | |
|
44 | 44 | /// Number of nodes anywhere in the tree that have |
|
45 | 45 | /// `.copy_source.is_some()`. |
|
46 | 46 | pub(super) nodes_with_copy_source_count: u32, |
|
47 | 47 | |
|
48 | 48 | /// See on_disk::Header |
|
49 | 49 | pub(super) ignore_patterns_hash: on_disk::IgnorePatternsHash, |
|
50 | 50 | |
|
51 | 51 | /// How many bytes of `on_disk` are not used anymore |
|
52 | 52 | pub(super) unreachable_bytes: u32, |
|
53 | 53 | } |
|
54 | 54 | |
|
55 | 55 | /// Using a plain `HgPathBuf` of the full path from the repository root as a |
|
56 | 56 | /// map key would also work: all paths in a given map have the same parent |
|
57 | 57 | /// path, so comparing full paths gives the same result as comparing base |
|
58 | 58 | /// names. However `HashMap` would waste time always re-hashing the same |
|
59 | 59 | /// string prefix. |
|
60 | 60 | pub(super) type NodeKey<'on_disk> = WithBasename<Cow<'on_disk, HgPath>>; |
|
61 | 61 | |
|
62 | 62 | /// Similar to `&'tree Cow<'on_disk, HgPath>`, but can also be returned |
|
63 | 63 | /// for on-disk nodes that don’t actually have a `Cow` to borrow. |
|
64 | 64 | pub(super) enum BorrowedPath<'tree, 'on_disk> { |
|
65 | 65 | InMemory(&'tree HgPathBuf), |
|
66 | 66 | OnDisk(&'on_disk HgPath), |
|
67 | 67 | } |
|
68 | 68 | |
|
69 | 69 | pub(super) enum ChildNodes<'on_disk> { |
|
70 | 70 | InMemory(FastHashMap<NodeKey<'on_disk>, Node<'on_disk>>), |
|
71 | 71 | OnDisk(&'on_disk [on_disk::Node]), |
|
72 | 72 | } |
|
73 | 73 | |
|
74 | 74 | pub(super) enum ChildNodesRef<'tree, 'on_disk> { |
|
75 | 75 | InMemory(&'tree FastHashMap<NodeKey<'on_disk>, Node<'on_disk>>), |
|
76 | 76 | OnDisk(&'on_disk [on_disk::Node]), |
|
77 | 77 | } |
|
78 | 78 | |
|
79 | 79 | pub(super) enum NodeRef<'tree, 'on_disk> { |
|
80 | 80 | InMemory(&'tree NodeKey<'on_disk>, &'tree Node<'on_disk>), |
|
81 | 81 | OnDisk(&'on_disk on_disk::Node), |
|
82 | 82 | } |
|
83 | 83 | |
|
84 | 84 | impl<'tree, 'on_disk> BorrowedPath<'tree, 'on_disk> { |
|
85 | 85 | pub fn detach_from_tree(&self) -> Cow<'on_disk, HgPath> { |
|
86 | 86 | match *self { |
|
87 | 87 | BorrowedPath::InMemory(in_memory) => Cow::Owned(in_memory.clone()), |
|
88 | 88 | BorrowedPath::OnDisk(on_disk) => Cow::Borrowed(on_disk), |
|
89 | 89 | } |
|
90 | 90 | } |
|
91 | 91 | } |
|
92 | 92 | |
|
93 | 93 | impl<'tree, 'on_disk> std::ops::Deref for BorrowedPath<'tree, 'on_disk> { |
|
94 | 94 | type Target = HgPath; |
|
95 | 95 | |
|
96 | 96 | fn deref(&self) -> &HgPath { |
|
97 | 97 | match *self { |
|
98 | 98 | BorrowedPath::InMemory(in_memory) => in_memory, |
|
99 | 99 | BorrowedPath::OnDisk(on_disk) => on_disk, |
|
100 | 100 | } |
|
101 | 101 | } |
|
102 | 102 | } |
|
103 | 103 | |
|
104 | 104 | impl Default for ChildNodes<'_> { |
|
105 | 105 | fn default() -> Self { |
|
106 | 106 | ChildNodes::InMemory(Default::default()) |
|
107 | 107 | } |
|
108 | 108 | } |
|
109 | 109 | |
|
110 | 110 | impl<'on_disk> ChildNodes<'on_disk> { |
|
111 | 111 | pub(super) fn as_ref<'tree>( |
|
112 | 112 | &'tree self, |
|
113 | 113 | ) -> ChildNodesRef<'tree, 'on_disk> { |
|
114 | 114 | match self { |
|
115 | 115 | ChildNodes::InMemory(nodes) => ChildNodesRef::InMemory(nodes), |
|
116 | 116 | ChildNodes::OnDisk(nodes) => ChildNodesRef::OnDisk(nodes), |
|
117 | 117 | } |
|
118 | 118 | } |
|
119 | 119 | |
|
120 | 120 | pub(super) fn is_empty(&self) -> bool { |
|
121 | 121 | match self { |
|
122 | 122 | ChildNodes::InMemory(nodes) => nodes.is_empty(), |
|
123 | 123 | ChildNodes::OnDisk(nodes) => nodes.is_empty(), |
|
124 | 124 | } |
|
125 | 125 | } |
|
126 | 126 | |
|
127 | 127 | fn make_mut( |
|
128 | 128 | &mut self, |
|
129 | 129 | on_disk: &'on_disk [u8], |
|
130 | 130 | unreachable_bytes: &mut u32, |
|
131 | 131 | ) -> Result< |
|
132 | 132 | &mut FastHashMap<NodeKey<'on_disk>, Node<'on_disk>>, |
|
133 | 133 | DirstateV2ParseError, |
|
134 | 134 | > { |
|
135 | 135 | match self { |
|
136 | 136 | ChildNodes::InMemory(nodes) => Ok(nodes), |
|
137 | 137 | ChildNodes::OnDisk(nodes) => { |
|
138 | 138 | *unreachable_bytes += |
|
139 | 139 | std::mem::size_of_val::<[on_disk::Node]>(nodes) as u32; |
|
140 | 140 | let nodes = nodes |
|
141 | 141 | .iter() |
|
142 | 142 | .map(|node| { |
|
143 | 143 | Ok(( |
|
144 | 144 | node.path(on_disk)?, |
|
145 | 145 | node.to_in_memory_node(on_disk)?, |
|
146 | 146 | )) |
|
147 | 147 | }) |
|
148 | 148 | .collect::<Result<_, _>>()?; |
|
149 | 149 | *self = ChildNodes::InMemory(nodes); |
|
150 | 150 | match self { |
|
151 | 151 | ChildNodes::InMemory(nodes) => Ok(nodes), |
|
152 | 152 | ChildNodes::OnDisk(_) => unreachable!(), |
|
153 | 153 | } |
|
154 | 154 | } |
|
155 | 155 | } |
|
156 | 156 | } |
|
157 | 157 | } |
|
158 | 158 | |
|
159 | 159 | impl<'tree, 'on_disk> ChildNodesRef<'tree, 'on_disk> { |
|
160 | 160 | pub(super) fn get( |
|
161 | 161 | &self, |
|
162 | 162 | base_name: &HgPath, |
|
163 | 163 | on_disk: &'on_disk [u8], |
|
164 | 164 | ) -> Result<Option<NodeRef<'tree, 'on_disk>>, DirstateV2ParseError> { |
|
165 | 165 | match self { |
|
166 | 166 | ChildNodesRef::InMemory(nodes) => Ok(nodes |
|
167 | 167 | .get_key_value(base_name) |
|
168 | 168 | .map(|(k, v)| NodeRef::InMemory(k, v))), |
|
169 | 169 | ChildNodesRef::OnDisk(nodes) => { |
|
170 | 170 | let mut parse_result = Ok(()); |
|
171 | 171 | let search_result = nodes.binary_search_by(|node| { |
|
172 | 172 | match node.base_name(on_disk) { |
|
173 | 173 | Ok(node_base_name) => node_base_name.cmp(base_name), |
|
174 | 174 | Err(e) => { |
|
175 | 175 | parse_result = Err(e); |
|
176 | 176 | // Dummy comparison result, `search_result` won’t |
|
177 | 177 | // be used since `parse_result` is an error |
|
178 | 178 | std::cmp::Ordering::Equal |
|
179 | 179 | } |
|
180 | 180 | } |
|
181 | 181 | }); |
|
182 | 182 | parse_result.map(|()| { |
|
183 | 183 | search_result.ok().map(|i| NodeRef::OnDisk(&nodes[i])) |
|
184 | 184 | }) |
|
185 | 185 | } |
|
186 | 186 | } |
|
187 | 187 | } |
|
188 | 188 | |
|
189 | 189 | /// Iterate in undefined order |
|
190 | 190 | pub(super) fn iter( |
|
191 | 191 | &self, |
|
192 | 192 | ) -> impl Iterator<Item = NodeRef<'tree, 'on_disk>> { |
|
193 | 193 | match self { |
|
194 | 194 | ChildNodesRef::InMemory(nodes) => itertools::Either::Left( |
|
195 | 195 | nodes.iter().map(|(k, v)| NodeRef::InMemory(k, v)), |
|
196 | 196 | ), |
|
197 | 197 | ChildNodesRef::OnDisk(nodes) => { |
|
198 | 198 | itertools::Either::Right(nodes.iter().map(NodeRef::OnDisk)) |
|
199 | 199 | } |
|
200 | 200 | } |
|
201 | 201 | } |
|
202 | 202 | |
|
203 | 203 | /// Iterate in parallel in undefined order |
|
204 | 204 | pub(super) fn par_iter( |
|
205 | 205 | &self, |
|
206 | 206 | ) -> impl rayon::iter::ParallelIterator<Item = NodeRef<'tree, 'on_disk>> |
|
207 | 207 | { |
|
208 | 208 | use rayon::prelude::*; |
|
209 | 209 | match self { |
|
210 | 210 | ChildNodesRef::InMemory(nodes) => rayon::iter::Either::Left( |
|
211 | 211 | nodes.par_iter().map(|(k, v)| NodeRef::InMemory(k, v)), |
|
212 | 212 | ), |
|
213 | 213 | ChildNodesRef::OnDisk(nodes) => rayon::iter::Either::Right( |
|
214 | 214 | nodes.par_iter().map(NodeRef::OnDisk), |
|
215 | 215 | ), |
|
216 | 216 | } |
|
217 | 217 | } |
|
218 | 218 | |
|
219 | 219 | pub(super) fn sorted(&self) -> Vec<NodeRef<'tree, 'on_disk>> { |
|
220 | 220 | match self { |
|
221 | 221 | ChildNodesRef::InMemory(nodes) => { |
|
222 | 222 | let mut vec: Vec<_> = nodes |
|
223 | 223 | .iter() |
|
224 | 224 | .map(|(k, v)| NodeRef::InMemory(k, v)) |
|
225 | 225 | .collect(); |
|
226 | 226 | fn sort_key<'a>(node: &'a NodeRef) -> &'a HgPath { |
|
227 | 227 | match node { |
|
228 | 228 | NodeRef::InMemory(path, _node) => path.base_name(), |
|
229 | 229 | NodeRef::OnDisk(_) => unreachable!(), |
|
230 | 230 | } |
|
231 | 231 | } |
|
232 | 232 | // `sort_unstable_by_key` doesn’t allow keys borrowing from the |
|
233 | 233 | // value: https://github.com/rust-lang/rust/issues/34162 |
|
234 | 234 | vec.sort_unstable_by(|a, b| sort_key(a).cmp(sort_key(b))); |
|
235 | 235 | vec |
|
236 | 236 | } |
|
237 | 237 | ChildNodesRef::OnDisk(nodes) => { |
|
238 | 238 | // Nodes on disk are already sorted |
|
239 | 239 | nodes.iter().map(NodeRef::OnDisk).collect() |
|
240 | 240 | } |
|
241 | 241 | } |
|
242 | 242 | } |
|
243 | 243 | } |
|
244 | 244 | |
|
245 | 245 | impl<'tree, 'on_disk> NodeRef<'tree, 'on_disk> { |
|
246 | 246 | pub(super) fn full_path( |
|
247 | 247 | &self, |
|
248 | 248 | on_disk: &'on_disk [u8], |
|
249 | 249 | ) -> Result<&'tree HgPath, DirstateV2ParseError> { |
|
250 | 250 | match self { |
|
251 | 251 | NodeRef::InMemory(path, _node) => Ok(path.full_path()), |
|
252 | 252 | NodeRef::OnDisk(node) => node.full_path(on_disk), |
|
253 | 253 | } |
|
254 | 254 | } |
|
255 | 255 | |
|
256 | 256 | /// Returns a `BorrowedPath`, which can be turned into a `Cow<'on_disk, |
|
257 | 257 | /// HgPath>` detached from `'tree` |
|
258 | 258 | pub(super) fn full_path_borrowed( |
|
259 | 259 | &self, |
|
260 | 260 | on_disk: &'on_disk [u8], |
|
261 | 261 | ) -> Result<BorrowedPath<'tree, 'on_disk>, DirstateV2ParseError> { |
|
262 | 262 | match self { |
|
263 | 263 | NodeRef::InMemory(path, _node) => match path.full_path() { |
|
264 | 264 | Cow::Borrowed(on_disk) => Ok(BorrowedPath::OnDisk(on_disk)), |
|
265 | 265 | Cow::Owned(in_memory) => Ok(BorrowedPath::InMemory(in_memory)), |
|
266 | 266 | }, |
|
267 | 267 | NodeRef::OnDisk(node) => { |
|
268 | 268 | Ok(BorrowedPath::OnDisk(node.full_path(on_disk)?)) |
|
269 | 269 | } |
|
270 | 270 | } |
|
271 | 271 | } |
|
272 | 272 | |
|
273 | 273 | pub(super) fn base_name( |
|
274 | 274 | &self, |
|
275 | 275 | on_disk: &'on_disk [u8], |
|
276 | 276 | ) -> Result<&'tree HgPath, DirstateV2ParseError> { |
|
277 | 277 | match self { |
|
278 | 278 | NodeRef::InMemory(path, _node) => Ok(path.base_name()), |
|
279 | 279 | NodeRef::OnDisk(node) => node.base_name(on_disk), |
|
280 | 280 | } |
|
281 | 281 | } |
|
282 | 282 | |
|
283 | 283 | pub(super) fn children( |
|
284 | 284 | &self, |
|
285 | 285 | on_disk: &'on_disk [u8], |
|
286 | 286 | ) -> Result<ChildNodesRef<'tree, 'on_disk>, DirstateV2ParseError> { |
|
287 | 287 | match self { |
|
288 | 288 | NodeRef::InMemory(_path, node) => Ok(node.children.as_ref()), |
|
289 | 289 | NodeRef::OnDisk(node) => { |
|
290 | 290 | Ok(ChildNodesRef::OnDisk(node.children(on_disk)?)) |
|
291 | 291 | } |
|
292 | 292 | } |
|
293 | 293 | } |
|
294 | 294 | |
|
295 | 295 | pub(super) fn has_copy_source(&self) -> bool { |
|
296 | 296 | match self { |
|
297 | 297 | NodeRef::InMemory(_path, node) => node.copy_source.is_some(), |
|
298 | 298 | NodeRef::OnDisk(node) => node.has_copy_source(), |
|
299 | 299 | } |
|
300 | 300 | } |
|
301 | 301 | |
|
302 | 302 | pub(super) fn copy_source( |
|
303 | 303 | &self, |
|
304 | 304 | on_disk: &'on_disk [u8], |
|
305 | 305 | ) -> Result<Option<&'tree HgPath>, DirstateV2ParseError> { |
|
306 | 306 | match self { |
|
307 | 307 | NodeRef::InMemory(_path, node) => { |
|
308 | 308 | Ok(node.copy_source.as_ref().map(|s| &**s)) |
|
309 | 309 | } |
|
310 | 310 | NodeRef::OnDisk(node) => node.copy_source(on_disk), |
|
311 | 311 | } |
|
312 | 312 | } |
|
313 | 313 | /// Returns a `BorrowedPath`, which can be turned into a `Cow<'on_disk, |
|
314 | 314 | /// HgPath>` detached from `'tree` |
|
315 | 315 | pub(super) fn copy_source_borrowed( |
|
316 | 316 | &self, |
|
317 | 317 | on_disk: &'on_disk [u8], |
|
318 | 318 | ) -> Result<Option<BorrowedPath<'tree, 'on_disk>>, DirstateV2ParseError> |
|
319 | 319 | { |
|
320 | 320 | Ok(match self { |
|
321 | 321 | NodeRef::InMemory(_path, node) => { |
|
322 | 322 | node.copy_source.as_ref().map(|source| match source { |
|
323 | 323 | Cow::Borrowed(on_disk) => BorrowedPath::OnDisk(on_disk), |
|
324 | 324 | Cow::Owned(in_memory) => BorrowedPath::InMemory(in_memory), |
|
325 | 325 | }) |
|
326 | 326 | } |
|
327 | 327 | NodeRef::OnDisk(node) => node |
|
328 | 328 | .copy_source(on_disk)? |
|
329 | 329 | .map(|source| BorrowedPath::OnDisk(source)), |
|
330 | 330 | }) |
|
331 | 331 | } |
|
332 | 332 | |
|
333 | 333 | pub(super) fn entry( |
|
334 | 334 | &self, |
|
335 | 335 | ) -> Result<Option<DirstateEntry>, DirstateV2ParseError> { |
|
336 | 336 | match self { |
|
337 | 337 | NodeRef::InMemory(_path, node) => { |
|
338 | 338 | Ok(node.data.as_entry().copied()) |
|
339 | 339 | } |
|
340 | 340 | NodeRef::OnDisk(node) => node.entry(), |
|
341 | 341 | } |
|
342 | 342 | } |
|
343 | 343 | |
|
344 | 344 | pub(super) fn state( |
|
345 | 345 | &self, |
|
346 | 346 | ) -> Result<Option<EntryState>, DirstateV2ParseError> { |
|
347 | 347 | Ok(self.entry()?.and_then(|e| { |
|
348 | 348 | if e.any_tracked() { |
|
349 | 349 | Some(e.state()) |
|
350 | 350 | } else { |
|
351 | 351 | None |
|
352 | 352 | } |
|
353 | 353 | })) |
|
354 | 354 | } |
|
355 | 355 | |
|
356 | 356 | pub(super) fn cached_directory_mtime( |
|
357 | 357 | &self, |
|
358 | 358 | ) -> Result<Option<TruncatedTimestamp>, DirstateV2ParseError> { |
|
359 | 359 | match self { |
|
360 | 360 | NodeRef::InMemory(_path, node) => Ok(match node.data { |
|
361 | 361 | NodeData::CachedDirectory { mtime } => Some(mtime), |
|
362 | 362 | _ => None, |
|
363 | 363 | }), |
|
364 | 364 | NodeRef::OnDisk(node) => node.cached_directory_mtime(), |
|
365 | 365 | } |
|
366 | 366 | } |
|
367 | 367 | |
|
368 | 368 | pub(super) fn descendants_with_entry_count(&self) -> u32 { |
|
369 | 369 | match self { |
|
370 | 370 | NodeRef::InMemory(_path, node) => { |
|
371 | 371 | node.descendants_with_entry_count |
|
372 | 372 | } |
|
373 | 373 | NodeRef::OnDisk(node) => node.descendants_with_entry_count.get(), |
|
374 | 374 | } |
|
375 | 375 | } |
|
376 | 376 | |
|
377 | 377 | pub(super) fn tracked_descendants_count(&self) -> u32 { |
|
378 | 378 | match self { |
|
379 | 379 | NodeRef::InMemory(_path, node) => node.tracked_descendants_count, |
|
380 | 380 | NodeRef::OnDisk(node) => node.tracked_descendants_count.get(), |
|
381 | 381 | } |
|
382 | 382 | } |
|
383 | 383 | } |
|
384 | 384 | |
|
385 | 385 | /// Represents a file or a directory |
|
386 | 386 | #[derive(Default)] |
|
387 | 387 | pub(super) struct Node<'on_disk> { |
|
388 | 388 | pub(super) data: NodeData, |
|
389 | 389 | |
|
390 | 390 | pub(super) copy_source: Option<Cow<'on_disk, HgPath>>, |
|
391 | 391 | |
|
392 | 392 | pub(super) children: ChildNodes<'on_disk>, |
|
393 | 393 | |
|
394 | 394 | /// How many (non-inclusive) descendants of this node have an entry. |
|
395 | 395 | pub(super) descendants_with_entry_count: u32, |
|
396 | 396 | |
|
397 | 397 | /// How many (non-inclusive) descendants of this node have an entry whose |
|
398 | 398 | /// state is "tracked". |
|
399 | 399 | pub(super) tracked_descendants_count: u32, |
|
400 | 400 | } |
|
401 | 401 | |
|
402 | 402 | pub(super) enum NodeData { |
|
403 | 403 | Entry(DirstateEntry), |
|
404 | 404 | CachedDirectory { mtime: TruncatedTimestamp }, |
|
405 | 405 | None, |
|
406 | 406 | } |
|
407 | 407 | |
|
408 | 408 | impl Default for NodeData { |
|
409 | 409 | fn default() -> Self { |
|
410 | 410 | NodeData::None |
|
411 | 411 | } |
|
412 | 412 | } |
|
413 | 413 | |
|
414 | 414 | impl NodeData { |
|
415 | 415 | fn has_entry(&self) -> bool { |
|
416 | 416 | match self { |
|
417 | 417 | NodeData::Entry(_) => true, |
|
418 | 418 | _ => false, |
|
419 | 419 | } |
|
420 | 420 | } |
|
421 | 421 | |
|
422 | 422 | fn as_entry(&self) -> Option<&DirstateEntry> { |
|
423 | 423 | match self { |
|
424 | 424 | NodeData::Entry(entry) => Some(entry), |
|
425 | 425 | _ => None, |
|
426 | 426 | } |
|
427 | 427 | } |
|
428 | 428 | |
|
429 | 429 | fn as_entry_mut(&mut self) -> Option<&mut DirstateEntry> { |
|
430 | 430 | match self { |
|
431 | 431 | NodeData::Entry(entry) => Some(entry), |
|
432 | 432 | _ => None, |
|
433 | 433 | } |
|
434 | 434 | } |
|
435 | 435 | } |
|
436 | 436 | |
|
437 | 437 | impl<'on_disk> DirstateMap<'on_disk> { |
|
438 | 438 | pub(super) fn empty(on_disk: &'on_disk [u8]) -> Self { |
|
439 | 439 | Self { |
|
440 | 440 | on_disk, |
|
441 | 441 | root: ChildNodes::default(), |
|
442 | 442 | nodes_with_entry_count: 0, |
|
443 | 443 | nodes_with_copy_source_count: 0, |
|
444 | 444 | ignore_patterns_hash: [0; on_disk::IGNORE_PATTERNS_HASH_LEN], |
|
445 | 445 | unreachable_bytes: 0, |
|
446 | 446 | } |
|
447 | 447 | } |
|
448 | 448 | |
|
449 | 449 | #[timed] |
|
450 | 450 | pub fn new_v2( |
|
451 | 451 | on_disk: &'on_disk [u8], |
|
452 | 452 | data_size: usize, |
|
453 | 453 | metadata: &[u8], |
|
454 | 454 | ) -> Result<Self, DirstateError> { |
|
455 | 455 | if let Some(data) = on_disk.get(..data_size) { |
|
456 | 456 | Ok(on_disk::read(data, metadata)?) |
|
457 | 457 | } else { |
|
458 | 458 | Err(DirstateV2ParseError.into()) |
|
459 | 459 | } |
|
460 | 460 | } |
|
461 | 461 | |
|
462 | 462 | #[timed] |
|
463 | 463 | pub fn new_v1( |
|
464 | 464 | on_disk: &'on_disk [u8], |
|
465 | 465 | ) -> Result<(Self, Option<DirstateParents>), DirstateError> { |
|
466 | 466 | let mut map = Self::empty(on_disk); |
|
467 | 467 | if map.on_disk.is_empty() { |
|
468 | 468 | return Ok((map, None)); |
|
469 | 469 | } |
|
470 | 470 | |
|
471 | 471 | let parents = parse_dirstate_entries( |
|
472 | 472 | map.on_disk, |
|
473 | 473 | |path, entry, copy_source| { |
|
474 | 474 | let tracked = entry.state().is_tracked(); |
|
475 | 475 | let node = Self::get_or_insert_node( |
|
476 | 476 | map.on_disk, |
|
477 | 477 | &mut map.unreachable_bytes, |
|
478 | 478 | &mut map.root, |
|
479 | 479 | path, |
|
480 | 480 | WithBasename::to_cow_borrowed, |
|
481 | 481 | |ancestor| { |
|
482 | 482 | if tracked { |
|
483 | 483 | ancestor.tracked_descendants_count += 1 |
|
484 | 484 | } |
|
485 | 485 | ancestor.descendants_with_entry_count += 1 |
|
486 | 486 | }, |
|
487 | 487 | )?; |
|
488 | 488 | assert!( |
|
489 | 489 | !node.data.has_entry(), |
|
490 | 490 | "duplicate dirstate entry in read" |
|
491 | 491 | ); |
|
492 | 492 | assert!( |
|
493 | 493 | node.copy_source.is_none(), |
|
494 | 494 | "duplicate dirstate entry in read" |
|
495 | 495 | ); |
|
496 | 496 | node.data = NodeData::Entry(*entry); |
|
497 | 497 | node.copy_source = copy_source.map(Cow::Borrowed); |
|
498 | 498 | map.nodes_with_entry_count += 1; |
|
499 | 499 | if copy_source.is_some() { |
|
500 | 500 | map.nodes_with_copy_source_count += 1 |
|
501 | 501 | } |
|
502 | 502 | Ok(()) |
|
503 | 503 | }, |
|
504 | 504 | )?; |
|
505 | 505 | let parents = Some(parents.clone()); |
|
506 | 506 | |
|
507 | 507 | Ok((map, parents)) |
|
508 | 508 | } |
|
509 | 509 | |
|
510 | 510 | /// Assuming dirstate-v2 format, returns whether the next write should |
|
511 | 511 | /// append to the existing data file that contains `self.on_disk` (true), |
|
512 | 512 | /// or create a new data file from scratch (false). |
|
513 | 513 | pub(super) fn write_should_append(&self) -> bool { |
|
514 | 514 | let ratio = self.unreachable_bytes as f32 / self.on_disk.len() as f32; |
|
515 | 515 | ratio < ACCEPTABLE_UNREACHABLE_BYTES_RATIO |
|
516 | 516 | } |
|
517 | 517 | |
|
518 | 518 | fn get_node<'tree>( |
|
519 | 519 | &'tree self, |
|
520 | 520 | path: &HgPath, |
|
521 | 521 | ) -> Result<Option<NodeRef<'tree, 'on_disk>>, DirstateV2ParseError> { |
|
522 | 522 | let mut children = self.root.as_ref(); |
|
523 | 523 | let mut components = path.components(); |
|
524 | 524 | let mut component = |
|
525 | 525 | components.next().expect("expected at least one components"); |
|
526 | 526 | loop { |
|
527 | 527 | if let Some(child) = children.get(component, self.on_disk)? { |
|
528 | 528 | if let Some(next_component) = components.next() { |
|
529 | 529 | component = next_component; |
|
530 | 530 | children = child.children(self.on_disk)?; |
|
531 | 531 | } else { |
|
532 | 532 | return Ok(Some(child)); |
|
533 | 533 | } |
|
534 | 534 | } else { |
|
535 | 535 | return Ok(None); |
|
536 | 536 | } |
|
537 | 537 | } |
|
538 | 538 | } |
|
539 | 539 | |
|
540 | 540 | /// Returns a mutable reference to the node at `path` if it exists |
|
541 | 541 | /// |
|
542 | 542 | /// This takes `root` instead of `&mut self` so that callers can mutate |
|
543 | 543 | /// other fields while the returned borrow is still valid |
|
544 | 544 | fn get_node_mut<'tree>( |
|
545 | 545 | on_disk: &'on_disk [u8], |
|
546 | 546 | unreachable_bytes: &mut u32, |
|
547 | 547 | root: &'tree mut ChildNodes<'on_disk>, |
|
548 | 548 | path: &HgPath, |
|
549 | 549 | ) -> Result<Option<&'tree mut Node<'on_disk>>, DirstateV2ParseError> { |
|
550 | 550 | let mut children = root; |
|
551 | 551 | let mut components = path.components(); |
|
552 | 552 | let mut component = |
|
553 | 553 | components.next().expect("expected at least one components"); |
|
554 | 554 | loop { |
|
555 | 555 | if let Some(child) = children |
|
556 | 556 | .make_mut(on_disk, unreachable_bytes)? |
|
557 | 557 | .get_mut(component) |
|
558 | 558 | { |
|
559 | 559 | if let Some(next_component) = components.next() { |
|
560 | 560 | component = next_component; |
|
561 | 561 | children = &mut child.children; |
|
562 | 562 | } else { |
|
563 | 563 | return Ok(Some(child)); |
|
564 | 564 | } |
|
565 | 565 | } else { |
|
566 | 566 | return Ok(None); |
|
567 | 567 | } |
|
568 | 568 | } |
|
569 | 569 | } |
|
570 | 570 | |
|
571 | 571 | pub(super) fn get_or_insert<'tree, 'path>( |
|
572 | 572 | &'tree mut self, |
|
573 | 573 | path: &HgPath, |
|
574 | 574 | ) -> Result<&'tree mut Node<'on_disk>, DirstateV2ParseError> { |
|
575 | 575 | Self::get_or_insert_node( |
|
576 | 576 | self.on_disk, |
|
577 | 577 | &mut self.unreachable_bytes, |
|
578 | 578 | &mut self.root, |
|
579 | 579 | path, |
|
580 | 580 | WithBasename::to_cow_owned, |
|
581 | 581 | |_| {}, |
|
582 | 582 | ) |
|
583 | 583 | } |
|
584 | 584 | |
|
585 | 585 | fn get_or_insert_node<'tree, 'path>( |
|
586 | 586 | on_disk: &'on_disk [u8], |
|
587 | 587 | unreachable_bytes: &mut u32, |
|
588 | 588 | root: &'tree mut ChildNodes<'on_disk>, |
|
589 | 589 | path: &'path HgPath, |
|
590 | 590 | to_cow: impl Fn( |
|
591 | 591 | WithBasename<&'path HgPath>, |
|
592 | 592 | ) -> WithBasename<Cow<'on_disk, HgPath>>, |
|
593 | 593 | mut each_ancestor: impl FnMut(&mut Node), |
|
594 | 594 | ) -> Result<&'tree mut Node<'on_disk>, DirstateV2ParseError> { |
|
595 | 595 | let mut child_nodes = root; |
|
596 | 596 | let mut inclusive_ancestor_paths = |
|
597 | 597 | WithBasename::inclusive_ancestors_of(path); |
|
598 | 598 | let mut ancestor_path = inclusive_ancestor_paths |
|
599 | 599 | .next() |
|
600 | 600 | .expect("expected at least one inclusive ancestor"); |
|
601 | 601 | loop { |
|
602 | 602 | let (_, child_node) = child_nodes |
|
603 | 603 | .make_mut(on_disk, unreachable_bytes)? |
|
604 | 604 | .raw_entry_mut() |
|
605 | 605 | .from_key(ancestor_path.base_name()) |
|
606 | 606 | .or_insert_with(|| (to_cow(ancestor_path), Node::default())); |
|
607 | 607 | if let Some(next) = inclusive_ancestor_paths.next() { |
|
608 | 608 | each_ancestor(child_node); |
|
609 | 609 | ancestor_path = next; |
|
610 | 610 | child_nodes = &mut child_node.children; |
|
611 | 611 | } else { |
|
612 | 612 | return Ok(child_node); |
|
613 | 613 | } |
|
614 | 614 | } |
|
615 | 615 | } |
|
616 | 616 | |
|
617 | 617 | fn reset_state( |
|
618 | 618 | &mut self, |
|
619 | 619 | filename: &HgPath, |
|
620 | 620 | old_entry_opt: Option<DirstateEntry>, |
|
621 | 621 | wc_tracked: bool, |
|
622 | 622 | p1_tracked: bool, |
|
623 | 623 | p2_info: bool, |
|
624 | 624 | has_meaningful_mtime: bool, |
|
625 | 625 | parent_file_data_opt: Option<ParentFileData>, |
|
626 | 626 | ) -> Result<(), DirstateError> { |
|
627 | 627 | let (had_entry, was_tracked) = match old_entry_opt { |
|
628 | 628 | Some(old_entry) => (true, old_entry.tracked()), |
|
629 | 629 | None => (false, false), |
|
630 | 630 | }; |
|
631 | 631 | let node = Self::get_or_insert_node( |
|
632 | 632 | self.on_disk, |
|
633 | 633 | &mut self.unreachable_bytes, |
|
634 | 634 | &mut self.root, |
|
635 | 635 | filename, |
|
636 | 636 | WithBasename::to_cow_owned, |
|
637 | 637 | |ancestor| { |
|
638 | 638 | if !had_entry { |
|
639 | 639 | ancestor.descendants_with_entry_count += 1; |
|
640 | 640 | } |
|
641 | 641 | if was_tracked { |
|
642 | 642 | if !wc_tracked { |
|
643 | 643 | ancestor.tracked_descendants_count = ancestor |
|
644 | 644 | .tracked_descendants_count |
|
645 | 645 | .checked_sub(1) |
|
646 | 646 | .expect("tracked count to be >= 0"); |
|
647 | 647 | } |
|
648 | 648 | } else { |
|
649 | 649 | if wc_tracked { |
|
650 | 650 | ancestor.tracked_descendants_count += 1; |
|
651 | 651 | } |
|
652 | 652 | } |
|
653 | 653 | }, |
|
654 | 654 | )?; |
|
655 | 655 | |
|
656 | 656 | let v2_data = if let Some(parent_file_data) = parent_file_data_opt { |
|
657 | 657 | DirstateV2Data { |
|
658 | 658 | wc_tracked, |
|
659 | 659 | p1_tracked, |
|
660 | 660 | p2_info, |
|
661 | 661 | mode_size: parent_file_data.mode_size, |
|
662 | 662 | mtime: if has_meaningful_mtime { |
|
663 | 663 | parent_file_data.mtime |
|
664 | 664 | } else { |
|
665 | 665 | None |
|
666 | 666 | }, |
|
667 | 667 | ..Default::default() |
|
668 | 668 | } |
|
669 | 669 | } else { |
|
670 | 670 | DirstateV2Data { |
|
671 | 671 | wc_tracked, |
|
672 | 672 | p1_tracked, |
|
673 | 673 | p2_info, |
|
674 | 674 | ..Default::default() |
|
675 | 675 | } |
|
676 | 676 | }; |
|
677 | 677 | if !had_entry { |
|
678 | 678 | self.nodes_with_entry_count += 1; |
|
679 | 679 | } |
|
680 | 680 | node.data = NodeData::Entry(DirstateEntry::from_v2_data(v2_data)); |
|
681 | 681 | Ok(()) |
|
682 | 682 | } |
|
683 | 683 | |
|
684 | 684 | fn set_tracked( |
|
685 | 685 | &mut self, |
|
686 | 686 | filename: &HgPath, |
|
687 | 687 | old_entry_opt: Option<DirstateEntry>, |
|
688 | 688 | ) -> Result<bool, DirstateV2ParseError> { |
|
689 | 689 | let was_tracked = old_entry_opt.map_or(false, |e| e.tracked()); |
|
690 | 690 | let had_entry = old_entry_opt.is_some(); |
|
691 | 691 | let tracked_count_increment = if was_tracked { 0 } else { 1 }; |
|
692 | 692 | let mut new = false; |
|
693 | 693 | |
|
694 | 694 | let node = Self::get_or_insert_node( |
|
695 | 695 | self.on_disk, |
|
696 | 696 | &mut self.unreachable_bytes, |
|
697 | 697 | &mut self.root, |
|
698 | 698 | filename, |
|
699 | 699 | WithBasename::to_cow_owned, |
|
700 | 700 | |ancestor| { |
|
701 | 701 | if !had_entry { |
|
702 | 702 | ancestor.descendants_with_entry_count += 1; |
|
703 | 703 | } |
|
704 | 704 | |
|
705 | 705 | ancestor.tracked_descendants_count += tracked_count_increment; |
|
706 | 706 | }, |
|
707 | 707 | )?; |
|
708 | 708 | let new_entry = if let Some(old_entry) = old_entry_opt { |
|
709 | 709 | let mut e = old_entry.clone(); |
|
710 | 710 | if e.tracked() { |
|
711 | 711 | // XXX |
|
712 | 712 | // This is probably overkill for more case, but we need this to |
|
713 | 713 | // fully replace the `normallookup` call with `set_tracked` |
|
714 | 714 | // one. Consider smoothing this in the future. |
|
715 | 715 | e.set_possibly_dirty(); |
|
716 | 716 | } else { |
|
717 | 717 | new = true; |
|
718 | 718 | e.set_tracked(); |
|
719 | 719 | } |
|
720 | 720 | e |
|
721 | 721 | } else { |
|
722 | 722 | self.nodes_with_entry_count += 1; |
|
723 | 723 | new = true; |
|
724 | 724 | DirstateEntry::new_tracked() |
|
725 | 725 | }; |
|
726 | 726 | node.data = NodeData::Entry(new_entry); |
|
727 | 727 | Ok(new) |
|
728 | 728 | } |
|
729 | 729 | |
|
730 | 730 | /// It is the responsibility of the caller to know that there was an entry |
|
731 | 731 | /// there before. Does not handle the removal of copy source |
|
732 | 732 | fn set_untracked( |
|
733 | 733 | &mut self, |
|
734 | 734 | filename: &HgPath, |
|
735 | 735 | old_entry: DirstateEntry, |
|
736 | 736 | ) -> Result<(), DirstateV2ParseError> { |
|
737 | 737 | let node = Self::get_or_insert_node( |
|
738 | 738 | self.on_disk, |
|
739 | 739 | &mut self.unreachable_bytes, |
|
740 | 740 | &mut self.root, |
|
741 | 741 | filename, |
|
742 | 742 | WithBasename::to_cow_owned, |
|
743 | 743 | |ancestor| { |
|
744 | 744 | ancestor.tracked_descendants_count = ancestor |
|
745 | 745 | .tracked_descendants_count |
|
746 | 746 | .checked_sub(1) |
|
747 | 747 | .expect("tracked_descendants_count should be >= 0"); |
|
748 | 748 | }, |
|
749 | 749 | )?; |
|
750 | 750 | let mut new_entry = old_entry.clone(); |
|
751 | 751 | new_entry.set_untracked(); |
|
752 | 752 | node.data = NodeData::Entry(new_entry); |
|
753 | 753 | Ok(()) |
|
754 | 754 | } |
|
755 | 755 | |
|
756 | 756 | fn set_clean( |
|
757 | 757 | &mut self, |
|
758 | 758 | filename: &HgPath, |
|
759 | 759 | old_entry: DirstateEntry, |
|
760 | 760 | mode: u32, |
|
761 | 761 | size: u32, |
|
762 | 762 | mtime: TruncatedTimestamp, |
|
763 | 763 | ) -> Result<(), DirstateError> { |
|
764 | 764 | let node = Self::get_or_insert_node( |
|
765 | 765 | self.on_disk, |
|
766 | 766 | &mut self.unreachable_bytes, |
|
767 | 767 | &mut self.root, |
|
768 | 768 | filename, |
|
769 | 769 | WithBasename::to_cow_owned, |
|
770 | 770 | |ancestor| { |
|
771 | 771 | if !old_entry.tracked() { |
|
772 | 772 | ancestor.tracked_descendants_count += 1; |
|
773 | 773 | } |
|
774 | 774 | }, |
|
775 | 775 | )?; |
|
776 | 776 | let mut new_entry = old_entry.clone(); |
|
777 | 777 | new_entry.set_clean(mode, size, mtime); |
|
778 | 778 | node.data = NodeData::Entry(new_entry); |
|
779 | 779 | Ok(()) |
|
780 | 780 | } |
|
781 | 781 | |
|
782 | 782 | fn set_possibly_dirty( |
|
783 | 783 | &mut self, |
|
784 | 784 | filename: &HgPath, |
|
785 | 785 | ) -> Result<(), DirstateError> { |
|
786 | 786 | let node = Self::get_or_insert_node( |
|
787 | 787 | self.on_disk, |
|
788 | 788 | &mut self.unreachable_bytes, |
|
789 | 789 | &mut self.root, |
|
790 | 790 | filename, |
|
791 | 791 | WithBasename::to_cow_owned, |
|
792 | 792 | |_ancestor| {}, |
|
793 | 793 | )?; |
|
794 | 794 | let entry = node.data.as_entry_mut().expect("entry should exist"); |
|
795 | 795 | entry.set_possibly_dirty(); |
|
796 | 796 | node.data = NodeData::Entry(*entry); |
|
797 | 797 | Ok(()) |
|
798 | 798 | } |
|
799 | 799 | |
|
800 | 800 | fn iter_nodes<'tree>( |
|
801 | 801 | &'tree self, |
|
802 | 802 | ) -> impl Iterator< |
|
803 | 803 | Item = Result<NodeRef<'tree, 'on_disk>, DirstateV2ParseError>, |
|
804 | 804 | > + 'tree { |
|
805 | 805 | // Depth first tree traversal. |
|
806 | 806 | // |
|
807 | 807 | // If we could afford internal iteration and recursion, |
|
808 | 808 | // this would look like: |
|
809 | 809 | // |
|
810 | 810 | // ``` |
|
811 | 811 | // fn traverse_children( |
|
812 | 812 | // children: &ChildNodes, |
|
813 | 813 | // each: &mut impl FnMut(&Node), |
|
814 | 814 | // ) { |
|
815 | 815 | // for child in children.values() { |
|
816 | 816 | // traverse_children(&child.children, each); |
|
817 | 817 | // each(child); |
|
818 | 818 | // } |
|
819 | 819 | // } |
|
820 | 820 | // ``` |
|
821 | 821 | // |
|
822 | 822 | // However we want an external iterator and therefore can’t use the |
|
823 | 823 | // call stack. Use an explicit stack instead: |
|
824 | 824 | let mut stack = Vec::new(); |
|
825 | 825 | let mut iter = self.root.as_ref().iter(); |
|
826 | 826 | std::iter::from_fn(move || { |
|
827 | 827 | while let Some(child_node) = iter.next() { |
|
828 | 828 | let children = match child_node.children(self.on_disk) { |
|
829 | 829 | Ok(children) => children, |
|
830 | 830 | Err(error) => return Some(Err(error)), |
|
831 | 831 | }; |
|
832 | 832 | // Pseudo-recursion |
|
833 | 833 | let new_iter = children.iter(); |
|
834 | 834 | let old_iter = std::mem::replace(&mut iter, new_iter); |
|
835 | 835 | stack.push((child_node, old_iter)); |
|
836 | 836 | } |
|
837 | 837 | // Found the end of a `children.iter()` iterator. |
|
838 | 838 | if let Some((child_node, next_iter)) = stack.pop() { |
|
839 | 839 | // "Return" from pseudo-recursion by restoring state from the |
|
840 | 840 | // explicit stack |
|
841 | 841 | iter = next_iter; |
|
842 | 842 | |
|
843 | 843 | Some(Ok(child_node)) |
|
844 | 844 | } else { |
|
845 | 845 | // Reached the bottom of the stack, we’re done |
|
846 | 846 | None |
|
847 | 847 | } |
|
848 | 848 | }) |
|
849 | 849 | } |
|
850 | 850 | |
|
851 | 851 | fn count_dropped_path(unreachable_bytes: &mut u32, path: &Cow<HgPath>) { |
|
852 | 852 | if let Cow::Borrowed(path) = path { |
|
853 | 853 | *unreachable_bytes += path.len() as u32 |
|
854 | 854 | } |
|
855 | 855 | } |
|
856 | 856 | } |
|
857 | 857 | |
|
858 | 858 | /// Like `Iterator::filter_map`, but over a fallible iterator of `Result`s. |
|
859 | 859 | /// |
|
860 | 860 | /// The callback is only called for incoming `Ok` values. Errors are passed |
|
861 | 861 | /// through as-is. In order to let it use the `?` operator the callback is |
|
862 | 862 | /// expected to return a `Result` of `Option`, instead of an `Option` of |
|
863 | 863 | /// `Result`. |
|
864 | 864 | fn filter_map_results<'a, I, F, A, B, E>( |
|
865 | 865 | iter: I, |
|
866 | 866 | f: F, |
|
867 | 867 | ) -> impl Iterator<Item = Result<B, E>> + 'a |
|
868 | 868 | where |
|
869 | 869 | I: Iterator<Item = Result<A, E>> + 'a, |
|
870 | 870 | F: Fn(A) -> Result<Option<B>, E> + 'a, |
|
871 | 871 | { |
|
872 | 872 | iter.filter_map(move |result| match result { |
|
873 | 873 | Ok(node) => f(node).transpose(), |
|
874 | 874 | Err(e) => Some(Err(e)), |
|
875 | 875 | }) |
|
876 | 876 | } |
|
877 | 877 | |
|
878 | 878 | impl OwningDirstateMap { |
|
879 | 879 | pub fn clear(&mut self) { |
|
880 | 880 | self.with_dmap_mut(|map| { |
|
881 | 881 | map.root = Default::default(); |
|
882 | 882 | map.nodes_with_entry_count = 0; |
|
883 | 883 | map.nodes_with_copy_source_count = 0; |
|
884 | 884 | }); |
|
885 | 885 | } |
|
886 | 886 | |
|
887 | 887 | pub fn set_entry( |
|
888 | 888 | &mut self, |
|
889 | 889 | filename: &HgPath, |
|
890 | 890 | entry: DirstateEntry, |
|
891 | 891 | ) -> Result<(), DirstateV2ParseError> { |
|
892 | 892 | self.with_dmap_mut(|map| { |
|
893 | 893 | map.get_or_insert(&filename)?.data = NodeData::Entry(entry); |
|
894 | 894 | Ok(()) |
|
895 | 895 | }) |
|
896 | 896 | } |
|
897 | 897 | |
|
898 | 898 | pub fn set_tracked( |
|
899 | 899 | &mut self, |
|
900 | 900 | filename: &HgPath, |
|
901 | 901 | ) -> Result<bool, DirstateV2ParseError> { |
|
902 | 902 | let old_entry_opt = self.get(filename)?; |
|
903 | 903 | self.with_dmap_mut(|map| map.set_tracked(filename, old_entry_opt)) |
|
904 | 904 | } |
|
905 | 905 | |
|
906 | 906 | pub fn set_untracked( |
|
907 | 907 | &mut self, |
|
908 | 908 | filename: &HgPath, |
|
909 | 909 | ) -> Result<bool, DirstateError> { |
|
910 | 910 | let old_entry_opt = self.get(filename)?; |
|
911 | 911 | match old_entry_opt { |
|
912 | 912 | None => Ok(false), |
|
913 | 913 | Some(old_entry) => { |
|
914 | 914 | if !old_entry.tracked() { |
|
915 | 915 | // `DirstateMap::set_untracked` is not a noop if |
|
916 | 916 | // already not tracked as it will decrement the |
|
917 | 917 | // tracked counters while going down. |
|
918 | 918 | return Ok(true); |
|
919 | 919 | } |
|
920 | 920 | if old_entry.added() { |
|
921 | 921 | // Untracking an "added" entry will just result in a |
|
922 | 922 | // worthless entry (and other parts of the code will |
|
923 | 923 | // complain about it), just drop it entirely. |
|
924 | 924 | self.drop_entry_and_copy_source(filename)?; |
|
925 | 925 | return Ok(true); |
|
926 | 926 | } |
|
927 | 927 | if !old_entry.p2_info() { |
|
928 | 928 | self.copy_map_remove(filename)?; |
|
929 | 929 | } |
|
930 | 930 | |
|
931 | 931 | self.with_dmap_mut(|map| { |
|
932 | 932 | map.set_untracked(filename, old_entry)?; |
|
933 | 933 | Ok(true) |
|
934 | 934 | }) |
|
935 | 935 | } |
|
936 | 936 | } |
|
937 | 937 | } |
|
938 | 938 | |
|
939 | 939 | pub fn set_clean( |
|
940 | 940 | &mut self, |
|
941 | 941 | filename: &HgPath, |
|
942 | 942 | mode: u32, |
|
943 | 943 | size: u32, |
|
944 | 944 | mtime: TruncatedTimestamp, |
|
945 | 945 | ) -> Result<(), DirstateError> { |
|
946 | 946 | let old_entry = match self.get(filename)? { |
|
947 | 947 | None => { |
|
948 | 948 | return Err( |
|
949 | 949 | DirstateMapError::PathNotFound(filename.into()).into() |
|
950 | 950 | ) |
|
951 | 951 | } |
|
952 | 952 | Some(e) => e, |
|
953 | 953 | }; |
|
954 | 954 | self.copy_map_remove(filename)?; |
|
955 | 955 | self.with_dmap_mut(|map| { |
|
956 | 956 | map.set_clean(filename, old_entry, mode, size, mtime) |
|
957 | 957 | }) |
|
958 | 958 | } |
|
959 | 959 | |
|
960 | 960 | pub fn set_possibly_dirty( |
|
961 | 961 | &mut self, |
|
962 | 962 | filename: &HgPath, |
|
963 | 963 | ) -> Result<(), DirstateError> { |
|
964 | 964 | if self.get(filename)?.is_none() { |
|
965 | 965 | return Err(DirstateMapError::PathNotFound(filename.into()).into()); |
|
966 | 966 | } |
|
967 | 967 | self.with_dmap_mut(|map| map.set_possibly_dirty(filename)) |
|
968 | 968 | } |
|
969 | 969 | |
|
970 | 970 | pub fn reset_state( |
|
971 | 971 | &mut self, |
|
972 | 972 | filename: &HgPath, |
|
973 | 973 | wc_tracked: bool, |
|
974 | 974 | p1_tracked: bool, |
|
975 | 975 | p2_info: bool, |
|
976 | 976 | has_meaningful_mtime: bool, |
|
977 | 977 | parent_file_data_opt: Option<ParentFileData>, |
|
978 | 978 | ) -> Result<(), DirstateError> { |
|
979 | 979 | if !(p1_tracked || p2_info || wc_tracked) { |
|
980 | 980 | self.drop_entry_and_copy_source(filename)?; |
|
981 | 981 | return Ok(()); |
|
982 | 982 | } |
|
983 | 983 | self.copy_map_remove(filename)?; |
|
984 | 984 | let old_entry_opt = self.get(filename)?; |
|
985 | 985 | self.with_dmap_mut(|map| { |
|
986 | 986 | map.reset_state( |
|
987 | 987 | filename, |
|
988 | 988 | old_entry_opt, |
|
989 | 989 | wc_tracked, |
|
990 | 990 | p1_tracked, |
|
991 | 991 | p2_info, |
|
992 | 992 | has_meaningful_mtime, |
|
993 | 993 | parent_file_data_opt, |
|
994 | 994 | ) |
|
995 | 995 | }) |
|
996 | 996 | } |
|
997 | 997 | |
|
998 | 998 | pub fn drop_entry_and_copy_source( |
|
999 | 999 | &mut self, |
|
1000 | 1000 | filename: &HgPath, |
|
1001 | 1001 | ) -> Result<(), DirstateError> { |
|
1002 | 1002 | let was_tracked = self |
|
1003 | 1003 | .get(filename)? |
|
1004 | 1004 | .map_or(false, |e| e.state().is_tracked()); |
|
1005 | 1005 | struct Dropped { |
|
1006 | 1006 | was_tracked: bool, |
|
1007 | 1007 | had_entry: bool, |
|
1008 | 1008 | had_copy_source: bool, |
|
1009 | 1009 | } |
|
1010 | 1010 | |
|
1011 | 1011 | /// If this returns `Ok(Some((dropped, removed)))`, then |
|
1012 | 1012 | /// |
|
1013 | 1013 | /// * `dropped` is about the leaf node that was at `filename` |
|
1014 | 1014 | /// * `removed` is whether this particular level of recursion just |
|
1015 | 1015 | /// removed a node in `nodes`. |
|
1016 | 1016 | fn recur<'on_disk>( |
|
1017 | 1017 | on_disk: &'on_disk [u8], |
|
1018 | 1018 | unreachable_bytes: &mut u32, |
|
1019 | 1019 | nodes: &mut ChildNodes<'on_disk>, |
|
1020 | 1020 | path: &HgPath, |
|
1021 | 1021 | ) -> Result<Option<(Dropped, bool)>, DirstateV2ParseError> { |
|
1022 | 1022 | let (first_path_component, rest_of_path) = |
|
1023 | 1023 | path.split_first_component(); |
|
1024 | 1024 | let nodes = nodes.make_mut(on_disk, unreachable_bytes)?; |
|
1025 | 1025 | let node = if let Some(node) = nodes.get_mut(first_path_component) |
|
1026 | 1026 | { |
|
1027 | 1027 | node |
|
1028 | 1028 | } else { |
|
1029 | 1029 | return Ok(None); |
|
1030 | 1030 | }; |
|
1031 | 1031 | let dropped; |
|
1032 | 1032 | if let Some(rest) = rest_of_path { |
|
1033 | 1033 | if let Some((d, removed)) = recur( |
|
1034 | 1034 | on_disk, |
|
1035 | 1035 | unreachable_bytes, |
|
1036 | 1036 | &mut node.children, |
|
1037 | 1037 | rest, |
|
1038 | 1038 | )? { |
|
1039 | 1039 | dropped = d; |
|
1040 | 1040 | if dropped.had_entry { |
|
1041 | 1041 | node.descendants_with_entry_count = node |
|
1042 | 1042 | .descendants_with_entry_count |
|
1043 | 1043 | .checked_sub(1) |
|
1044 | 1044 | .expect( |
|
1045 | 1045 | "descendants_with_entry_count should be >= 0", |
|
1046 | 1046 | ); |
|
1047 | 1047 | } |
|
1048 | 1048 | if dropped.was_tracked { |
|
1049 | 1049 | node.tracked_descendants_count = node |
|
1050 | 1050 | .tracked_descendants_count |
|
1051 | 1051 | .checked_sub(1) |
|
1052 | 1052 | .expect( |
|
1053 | 1053 | "tracked_descendants_count should be >= 0", |
|
1054 | 1054 | ); |
|
1055 | 1055 | } |
|
1056 | 1056 | |
|
1057 | 1057 | // Directory caches must be invalidated when removing a |
|
1058 | 1058 | // child node |
|
1059 | 1059 | if removed { |
|
1060 | 1060 | if let NodeData::CachedDirectory { .. } = &node.data { |
|
1061 | 1061 | node.data = NodeData::None |
|
1062 | 1062 | } |
|
1063 | 1063 | } |
|
1064 | 1064 | } else { |
|
1065 | 1065 | return Ok(None); |
|
1066 | 1066 | } |
|
1067 | 1067 | } else { |
|
1068 | 1068 | let entry = node.data.as_entry(); |
|
1069 | 1069 | let was_tracked = entry.map_or(false, |entry| entry.tracked()); |
|
1070 | 1070 | let had_entry = entry.is_some(); |
|
1071 | 1071 | if had_entry { |
|
1072 | 1072 | node.data = NodeData::None |
|
1073 | 1073 | } |
|
1074 | 1074 | let mut had_copy_source = false; |
|
1075 | 1075 | if let Some(source) = &node.copy_source { |
|
1076 | 1076 | DirstateMap::count_dropped_path(unreachable_bytes, source); |
|
1077 | 1077 | had_copy_source = true; |
|
1078 | 1078 | node.copy_source = None |
|
1079 | 1079 | } |
|
1080 | 1080 | dropped = Dropped { |
|
1081 | 1081 | was_tracked, |
|
1082 | 1082 | had_entry, |
|
1083 | 1083 | had_copy_source, |
|
1084 | 1084 | }; |
|
1085 | 1085 | } |
|
1086 | 1086 | // After recursion, for both leaf (rest_of_path is None) nodes and |
|
1087 | 1087 | // parent nodes, remove a node if it just became empty. |
|
1088 | 1088 | let remove = !node.data.has_entry() |
|
1089 | 1089 | && node.copy_source.is_none() |
|
1090 | 1090 | && node.children.is_empty(); |
|
1091 | 1091 | if remove { |
|
1092 | 1092 | let (key, _) = |
|
1093 | 1093 | nodes.remove_entry(first_path_component).unwrap(); |
|
1094 | 1094 | DirstateMap::count_dropped_path( |
|
1095 | 1095 | unreachable_bytes, |
|
1096 | 1096 | key.full_path(), |
|
1097 | 1097 | ) |
|
1098 | 1098 | } |
|
1099 | 1099 | Ok(Some((dropped, remove))) |
|
1100 | 1100 | } |
|
1101 | 1101 | |
|
1102 | 1102 | self.with_dmap_mut(|map| { |
|
1103 | 1103 | if let Some((dropped, _removed)) = recur( |
|
1104 | 1104 | map.on_disk, |
|
1105 | 1105 | &mut map.unreachable_bytes, |
|
1106 | 1106 | &mut map.root, |
|
1107 | 1107 | filename, |
|
1108 | 1108 | )? { |
|
1109 | 1109 | if dropped.had_entry { |
|
1110 | 1110 | map.nodes_with_entry_count = map |
|
1111 | 1111 | .nodes_with_entry_count |
|
1112 | 1112 | .checked_sub(1) |
|
1113 | 1113 | .expect("nodes_with_entry_count should be >= 0"); |
|
1114 | 1114 | } |
|
1115 | 1115 | if dropped.had_copy_source { |
|
1116 | 1116 | map.nodes_with_copy_source_count = map |
|
1117 | 1117 | .nodes_with_copy_source_count |
|
1118 | 1118 | .checked_sub(1) |
|
1119 | 1119 | .expect("nodes_with_copy_source_count should be >= 0"); |
|
1120 | 1120 | } |
|
1121 | 1121 | } else { |
|
1122 | 1122 | debug_assert!(!was_tracked); |
|
1123 | 1123 | } |
|
1124 | 1124 | Ok(()) |
|
1125 | 1125 | }) |
|
1126 | 1126 | } |
|
1127 | 1127 | |
|
1128 | 1128 | pub fn has_tracked_dir( |
|
1129 | 1129 | &mut self, |
|
1130 | 1130 | directory: &HgPath, |
|
1131 | 1131 | ) -> Result<bool, DirstateError> { |
|
1132 | 1132 | self.with_dmap_mut(|map| { |
|
1133 | 1133 | if let Some(node) = map.get_node(directory)? { |
|
1134 | 1134 | // A node without a `DirstateEntry` was created to hold child |
|
1135 | 1135 | // nodes, and is therefore a directory. |
|
1136 | 1136 | let state = node.state()?; |
|
1137 | 1137 | Ok(state.is_none() && node.tracked_descendants_count() > 0) |
|
1138 | 1138 | } else { |
|
1139 | 1139 | Ok(false) |
|
1140 | 1140 | } |
|
1141 | 1141 | }) |
|
1142 | 1142 | } |
|
1143 | 1143 | |
|
1144 | 1144 | pub fn has_dir( |
|
1145 | 1145 | &mut self, |
|
1146 | 1146 | directory: &HgPath, |
|
1147 | 1147 | ) -> Result<bool, DirstateError> { |
|
1148 | 1148 | self.with_dmap_mut(|map| { |
|
1149 | 1149 | if let Some(node) = map.get_node(directory)? { |
|
1150 | 1150 | // A node without a `DirstateEntry` was created to hold child |
|
1151 | 1151 | // nodes, and is therefore a directory. |
|
1152 | 1152 | let state = node.state()?; |
|
1153 | 1153 | Ok(state.is_none() && node.descendants_with_entry_count() > 0) |
|
1154 | 1154 | } else { |
|
1155 | 1155 | Ok(false) |
|
1156 | 1156 | } |
|
1157 | 1157 | }) |
|
1158 | 1158 | } |
|
1159 | 1159 | |
|
1160 | 1160 | #[timed] |
|
1161 | 1161 | pub fn pack_v1( |
|
1162 | 1162 | &self, |
|
1163 | 1163 | parents: DirstateParents, |
|
1164 | 1164 | ) -> Result<Vec<u8>, DirstateError> { |
|
1165 | 1165 | let map = self.get_map(); |
|
1166 | 1166 | // Optizimation (to be measured?): pre-compute size to avoid `Vec` |
|
1167 | 1167 | // reallocations |
|
1168 | 1168 | let mut size = parents.as_bytes().len(); |
|
1169 | 1169 | for node in map.iter_nodes() { |
|
1170 | 1170 | let node = node?; |
|
1171 | 1171 | if node.entry()?.is_some() { |
|
1172 | 1172 | size += packed_entry_size( |
|
1173 | 1173 | node.full_path(map.on_disk)?, |
|
1174 | 1174 | node.copy_source(map.on_disk)?, |
|
1175 | 1175 | ); |
|
1176 | 1176 | } |
|
1177 | 1177 | } |
|
1178 | 1178 | |
|
1179 | 1179 | let mut packed = Vec::with_capacity(size); |
|
1180 | 1180 | packed.extend(parents.as_bytes()); |
|
1181 | 1181 | |
|
1182 | 1182 | for node in map.iter_nodes() { |
|
1183 | 1183 | let node = node?; |
|
1184 | 1184 | if let Some(entry) = node.entry()? { |
|
1185 | 1185 | pack_entry( |
|
1186 | 1186 | node.full_path(map.on_disk)?, |
|
1187 | 1187 | &entry, |
|
1188 | 1188 | node.copy_source(map.on_disk)?, |
|
1189 | 1189 | &mut packed, |
|
1190 | 1190 | ); |
|
1191 | 1191 | } |
|
1192 | 1192 | } |
|
1193 | 1193 | Ok(packed) |
|
1194 | 1194 | } |
|
1195 | 1195 | |
|
1196 | 1196 | /// Returns new data and metadata together with whether that data should be |
|
1197 | 1197 | /// appended to the existing data file whose content is at |
|
1198 | 1198 | /// `map.on_disk` (true), instead of written to a new data file |
|
1199 | 1199 | /// (false). |
|
1200 | 1200 | #[timed] |
|
1201 | 1201 | pub fn pack_v2( |
|
1202 | 1202 | &self, |
|
1203 | 1203 | can_append: bool, |
|
1204 | 1204 | ) -> Result<(Vec<u8>, on_disk::TreeMetadata, bool), DirstateError> { |
|
1205 | 1205 | let map = self.get_map(); |
|
1206 | 1206 | on_disk::write(map, can_append) |
|
1207 | 1207 | } |
|
1208 | 1208 | |
|
1209 | 1209 | /// `callback` allows the caller to process and do something with the |
|
1210 | 1210 | /// results of the status. This is needed to do so efficiently (i.e. |
|
1211 | 1211 | /// without cloning the `DirstateStatus` object with its paths) because |
|
1212 | 1212 | /// we need to borrow from `Self`. |
|
1213 | 1213 | pub fn with_status<R>( |
|
1214 | 1214 | &mut self, |
|
1215 | 1215 | matcher: &(dyn Matcher + Sync), |
|
1216 | 1216 | root_dir: PathBuf, |
|
1217 | 1217 | ignore_files: Vec<PathBuf>, |
|
1218 | 1218 | options: StatusOptions, |
|
1219 | 1219 | callback: impl for<'r> FnOnce( |
|
1220 | 1220 | Result<(DirstateStatus<'r>, Vec<PatternFileWarning>), StatusError>, |
|
1221 | 1221 | ) -> R, |
|
1222 | 1222 | ) -> R { |
|
1223 | 1223 | self.with_dmap_mut(|map| { |
|
1224 | 1224 | callback(super::status::status( |
|
1225 | 1225 | map, |
|
1226 | 1226 | matcher, |
|
1227 | 1227 | root_dir, |
|
1228 | 1228 | ignore_files, |
|
1229 | 1229 | options, |
|
1230 | 1230 | )) |
|
1231 | 1231 | }) |
|
1232 | 1232 | } |
|
1233 | 1233 | |
|
1234 | 1234 | pub fn copy_map_len(&self) -> usize { |
|
1235 | 1235 | let map = self.get_map(); |
|
1236 | 1236 | map.nodes_with_copy_source_count as usize |
|
1237 | 1237 | } |
|
1238 | 1238 | |
|
1239 | 1239 | pub fn copy_map_iter(&self) -> CopyMapIter<'_> { |
|
1240 | 1240 | let map = self.get_map(); |
|
1241 | 1241 | Box::new(filter_map_results(map.iter_nodes(), move |node| { |
|
1242 | 1242 | Ok(if let Some(source) = node.copy_source(map.on_disk)? { |
|
1243 | 1243 | Some((node.full_path(map.on_disk)?, source)) |
|
1244 | 1244 | } else { |
|
1245 | 1245 | None |
|
1246 | 1246 | }) |
|
1247 | 1247 | })) |
|
1248 | 1248 | } |
|
1249 | 1249 | |
|
1250 | 1250 | pub fn copy_map_contains_key( |
|
1251 | 1251 | &self, |
|
1252 | 1252 | key: &HgPath, |
|
1253 | 1253 | ) -> Result<bool, DirstateV2ParseError> { |
|
1254 | 1254 | let map = self.get_map(); |
|
1255 | 1255 | Ok(if let Some(node) = map.get_node(key)? { |
|
1256 | 1256 | node.has_copy_source() |
|
1257 | 1257 | } else { |
|
1258 | 1258 | false |
|
1259 | 1259 | }) |
|
1260 | 1260 | } |
|
1261 | 1261 | |
|
1262 | 1262 | pub fn copy_map_get( |
|
1263 | 1263 | &self, |
|
1264 | 1264 | key: &HgPath, |
|
1265 | 1265 | ) -> Result<Option<&HgPath>, DirstateV2ParseError> { |
|
1266 | 1266 | let map = self.get_map(); |
|
1267 | 1267 | if let Some(node) = map.get_node(key)? { |
|
1268 | 1268 | if let Some(source) = node.copy_source(map.on_disk)? { |
|
1269 | 1269 | return Ok(Some(source)); |
|
1270 | 1270 | } |
|
1271 | 1271 | } |
|
1272 | 1272 | Ok(None) |
|
1273 | 1273 | } |
|
1274 | 1274 | |
|
1275 | 1275 | pub fn copy_map_remove( |
|
1276 | 1276 | &mut self, |
|
1277 | 1277 | key: &HgPath, |
|
1278 | 1278 | ) -> Result<Option<HgPathBuf>, DirstateV2ParseError> { |
|
1279 | 1279 | self.with_dmap_mut(|map| { |
|
1280 | 1280 | let count = &mut map.nodes_with_copy_source_count; |
|
1281 | 1281 | let unreachable_bytes = &mut map.unreachable_bytes; |
|
1282 | 1282 | Ok(DirstateMap::get_node_mut( |
|
1283 | 1283 | map.on_disk, |
|
1284 | 1284 | unreachable_bytes, |
|
1285 | 1285 | &mut map.root, |
|
1286 | 1286 | key, |
|
1287 | 1287 | )? |
|
1288 | 1288 | .and_then(|node| { |
|
1289 | 1289 | if let Some(source) = &node.copy_source { |
|
1290 | 1290 | *count -= 1; |
|
1291 | 1291 | DirstateMap::count_dropped_path(unreachable_bytes, source); |
|
1292 | 1292 | } |
|
1293 | 1293 | node.copy_source.take().map(Cow::into_owned) |
|
1294 | 1294 | })) |
|
1295 | 1295 | }) |
|
1296 | 1296 | } |
|
1297 | 1297 | |
|
1298 | 1298 | pub fn copy_map_insert( |
|
1299 | 1299 | &mut self, |
|
1300 | 1300 | key: HgPathBuf, |
|
1301 | 1301 | value: HgPathBuf, |
|
1302 | 1302 | ) -> Result<Option<HgPathBuf>, DirstateV2ParseError> { |
|
1303 | 1303 | self.with_dmap_mut(|map| { |
|
1304 | 1304 | let node = DirstateMap::get_or_insert_node( |
|
1305 | 1305 | map.on_disk, |
|
1306 | 1306 | &mut map.unreachable_bytes, |
|
1307 | 1307 | &mut map.root, |
|
1308 | 1308 | &key, |
|
1309 | 1309 | WithBasename::to_cow_owned, |
|
1310 | 1310 | |_ancestor| {}, |
|
1311 | 1311 | )?; |
|
1312 | 1312 | if node.copy_source.is_none() { |
|
1313 | 1313 | map.nodes_with_copy_source_count += 1 |
|
1314 | 1314 | } |
|
1315 | 1315 | Ok(node.copy_source.replace(value.into()).map(Cow::into_owned)) |
|
1316 | 1316 | }) |
|
1317 | 1317 | } |
|
1318 | 1318 | |
|
1319 | 1319 | pub fn len(&self) -> usize { |
|
1320 | 1320 | let map = self.get_map(); |
|
1321 | 1321 | map.nodes_with_entry_count as usize |
|
1322 | 1322 | } |
|
1323 | 1323 | |
|
1324 | 1324 | pub fn contains_key( |
|
1325 | 1325 | &self, |
|
1326 | 1326 | key: &HgPath, |
|
1327 | 1327 | ) -> Result<bool, DirstateV2ParseError> { |
|
1328 | 1328 | Ok(self.get(key)?.is_some()) |
|
1329 | 1329 | } |
|
1330 | 1330 | |
|
1331 | 1331 | pub fn get( |
|
1332 | 1332 | &self, |
|
1333 | 1333 | key: &HgPath, |
|
1334 | 1334 | ) -> Result<Option<DirstateEntry>, DirstateV2ParseError> { |
|
1335 | 1335 | let map = self.get_map(); |
|
1336 | 1336 | Ok(if let Some(node) = map.get_node(key)? { |
|
1337 | 1337 | node.entry()? |
|
1338 | 1338 | } else { |
|
1339 | 1339 | None |
|
1340 | 1340 | }) |
|
1341 | 1341 | } |
|
1342 | 1342 | |
|
1343 | 1343 | pub fn iter(&self) -> StateMapIter<'_> { |
|
1344 | 1344 | let map = self.get_map(); |
|
1345 | 1345 | Box::new(filter_map_results(map.iter_nodes(), move |node| { |
|
1346 | 1346 | Ok(if let Some(entry) = node.entry()? { |
|
1347 | 1347 | Some((node.full_path(map.on_disk)?, entry)) |
|
1348 | 1348 | } else { |
|
1349 | 1349 | None |
|
1350 | 1350 | }) |
|
1351 | 1351 | })) |
|
1352 | 1352 | } |
|
1353 | 1353 | |
|
1354 | 1354 | pub fn iter_tracked_dirs( |
|
1355 | 1355 | &mut self, |
|
1356 | 1356 | ) -> Result< |
|
1357 | 1357 | Box< |
|
1358 | 1358 | dyn Iterator<Item = Result<&HgPath, DirstateV2ParseError>> |
|
1359 | 1359 | + Send |
|
1360 | 1360 | + '_, |
|
1361 | 1361 | >, |
|
1362 | 1362 | DirstateError, |
|
1363 | 1363 | > { |
|
1364 | 1364 | let map = self.get_map(); |
|
1365 | 1365 | let on_disk = map.on_disk; |
|
1366 | 1366 | Ok(Box::new(filter_map_results( |
|
1367 | 1367 | map.iter_nodes(), |
|
1368 | 1368 | move |node| { |
|
1369 | 1369 | Ok(if node.tracked_descendants_count() > 0 { |
|
1370 | 1370 | Some(node.full_path(on_disk)?) |
|
1371 | 1371 | } else { |
|
1372 | 1372 | None |
|
1373 | 1373 | }) |
|
1374 | 1374 | }, |
|
1375 | 1375 | ))) |
|
1376 | 1376 | } |
|
1377 | 1377 | |
|
1378 | /// Only public because it needs to be exposed to the Python layer. | |
|
1379 | /// It is not the full `setparents` logic, only the parts that mutate the | |
|
1380 | /// entries. | |
|
1381 | pub fn setparents_fixup( | |
|
1382 | &mut self, | |
|
1383 | ) -> Result<Vec<(HgPathBuf, HgPathBuf)>, DirstateV2ParseError> { | |
|
1384 | // XXX | |
|
1385 | // All the copying and re-querying is quite inefficient, but this is | |
|
1386 | // still a lot better than doing it from Python. | |
|
1387 | // | |
|
1388 | // The better solution is to develop a mechanism for `iter_mut`, | |
|
1389 | // which will be a lot more involved: we're dealing with a lazy, | |
|
1390 | // append-mostly, tree-like data structure. This will do for now. | |
|
1391 | let mut copies = vec![]; | |
|
1392 | let mut files_with_p2_info = vec![]; | |
|
1393 | for res in self.iter() { | |
|
1394 | let (path, entry) = res?; | |
|
1395 | if entry.p2_info() { | |
|
1396 | files_with_p2_info.push(path.to_owned()) | |
|
1397 | } | |
|
1398 | } | |
|
1399 | self.with_dmap_mut(|map| { | |
|
1400 | for path in files_with_p2_info.iter() { | |
|
1401 | let node = map.get_or_insert(path)?; | |
|
1402 | let entry = | |
|
1403 | node.data.as_entry_mut().expect("entry should exist"); | |
|
1404 | entry.drop_merge_data(); | |
|
1405 | if let Some(source) = node.copy_source.take().as_deref() { | |
|
1406 | copies.push((path.to_owned(), source.to_owned())); | |
|
1407 | } | |
|
1408 | } | |
|
1409 | Ok(copies) | |
|
1410 | }) | |
|
1411 | } | |
|
1412 | ||
|
1378 | 1413 | pub fn debug_iter( |
|
1379 | 1414 | &self, |
|
1380 | 1415 | all: bool, |
|
1381 | 1416 | ) -> Box< |
|
1382 | 1417 | dyn Iterator< |
|
1383 | 1418 | Item = Result< |
|
1384 | 1419 | (&HgPath, (u8, i32, i32, i32)), |
|
1385 | 1420 | DirstateV2ParseError, |
|
1386 | 1421 | >, |
|
1387 | 1422 | > + Send |
|
1388 | 1423 | + '_, |
|
1389 | 1424 | > { |
|
1390 | 1425 | let map = self.get_map(); |
|
1391 | 1426 | Box::new(filter_map_results(map.iter_nodes(), move |node| { |
|
1392 | 1427 | let debug_tuple = if let Some(entry) = node.entry()? { |
|
1393 | 1428 | entry.debug_tuple() |
|
1394 | 1429 | } else if !all { |
|
1395 | 1430 | return Ok(None); |
|
1396 | 1431 | } else if let Some(mtime) = node.cached_directory_mtime()? { |
|
1397 | 1432 | (b' ', 0, -1, mtime.truncated_seconds() as i32) |
|
1398 | 1433 | } else { |
|
1399 | 1434 | (b' ', 0, -1, -1) |
|
1400 | 1435 | }; |
|
1401 | 1436 | Ok(Some((node.full_path(map.on_disk)?, debug_tuple))) |
|
1402 | 1437 | })) |
|
1403 | 1438 | } |
|
1404 | 1439 | } |
@@ -1,559 +1,572 b'' | |||
|
1 | 1 | // dirstate_map.rs |
|
2 | 2 | // |
|
3 | 3 | // Copyright 2019 Raphaël Gomès <rgomes@octobus.net> |
|
4 | 4 | // |
|
5 | 5 | // This software may be used and distributed according to the terms of the |
|
6 | 6 | // GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | //! Bindings for the `hg::dirstate::dirstate_map` file provided by the |
|
9 | 9 | //! `hg-core` package. |
|
10 | 10 | |
|
11 | 11 | use std::cell::{RefCell, RefMut}; |
|
12 | 12 | use std::convert::TryInto; |
|
13 | 13 | |
|
14 | 14 | use cpython::{ |
|
15 | 15 | exc, PyBool, PyBytes, PyClone, PyDict, PyErr, PyList, PyNone, PyObject, |
|
16 | 16 | PyResult, Python, PythonObject, ToPyObject, UnsafePyLeaked, |
|
17 | 17 | }; |
|
18 | 18 | use hg::dirstate::{ParentFileData, TruncatedTimestamp}; |
|
19 | 19 | |
|
20 | 20 | use crate::{ |
|
21 | 21 | dirstate::copymap::{CopyMap, CopyMapItemsIterator, CopyMapKeysIterator}, |
|
22 | 22 | dirstate::item::DirstateItem, |
|
23 | 23 | pybytes_deref::PyBytesDeref, |
|
24 | 24 | }; |
|
25 | 25 | use hg::{ |
|
26 | 26 | dirstate::StateMapIter, |
|
27 | 27 | dirstate_tree::on_disk::DirstateV2ParseError, |
|
28 | 28 | dirstate_tree::owning::OwningDirstateMap, |
|
29 | 29 | revlog::Node, |
|
30 | 30 | utils::files::normalize_case, |
|
31 | 31 | utils::hg_path::{HgPath, HgPathBuf}, |
|
32 | 32 | DirstateEntry, DirstateError, DirstateParents, EntryState, |
|
33 | 33 | }; |
|
34 | 34 | |
|
35 | 35 | // TODO |
|
36 | 36 | // This object needs to share references to multiple members of its Rust |
|
37 | 37 | // inner struct, namely `copy_map`, `dirs` and `all_dirs`. |
|
38 | 38 | // Right now `CopyMap` is done, but it needs to have an explicit reference |
|
39 | 39 | // to `RustDirstateMap` which itself needs to have an encapsulation for |
|
40 | 40 | // every method in `CopyMap` (copymapcopy, etc.). |
|
41 | 41 | // This is ugly and hard to maintain. |
|
42 | 42 | // The same logic applies to `dirs` and `all_dirs`, however the `Dirs` |
|
43 | 43 | // `py_class!` is already implemented and does not mention |
|
44 | 44 | // `RustDirstateMap`, rightfully so. |
|
45 | 45 | // All attributes also have to have a separate refcount data attribute for |
|
46 | 46 | // leaks, with all methods that go along for reference sharing. |
|
47 | 47 | py_class!(pub class DirstateMap |py| { |
|
48 | 48 | @shared data inner: OwningDirstateMap; |
|
49 | 49 | |
|
50 | 50 | /// Returns a `(dirstate_map, parents)` tuple |
|
51 | 51 | @staticmethod |
|
52 | 52 | def new_v1( |
|
53 | 53 | on_disk: PyBytes, |
|
54 | 54 | ) -> PyResult<PyObject> { |
|
55 | 55 | let on_disk = PyBytesDeref::new(py, on_disk); |
|
56 | 56 | let (map, parents) = OwningDirstateMap::new_v1(on_disk) |
|
57 | 57 | .map_err(|e| dirstate_error(py, e))?; |
|
58 | 58 | let map = Self::create_instance(py, map)?; |
|
59 | 59 | let p1 = PyBytes::new(py, parents.p1.as_bytes()); |
|
60 | 60 | let p2 = PyBytes::new(py, parents.p2.as_bytes()); |
|
61 | 61 | let parents = (p1, p2); |
|
62 | 62 | Ok((map, parents).to_py_object(py).into_object()) |
|
63 | 63 | } |
|
64 | 64 | |
|
65 | 65 | /// Returns a DirstateMap |
|
66 | 66 | @staticmethod |
|
67 | 67 | def new_v2( |
|
68 | 68 | on_disk: PyBytes, |
|
69 | 69 | data_size: usize, |
|
70 | 70 | tree_metadata: PyBytes, |
|
71 | 71 | ) -> PyResult<PyObject> { |
|
72 | 72 | let dirstate_error = |e: DirstateError| { |
|
73 | 73 | PyErr::new::<exc::OSError, _>(py, format!("Dirstate error: {:?}", e)) |
|
74 | 74 | }; |
|
75 | 75 | let on_disk = PyBytesDeref::new(py, on_disk); |
|
76 | 76 | let map = OwningDirstateMap::new_v2( |
|
77 | 77 | on_disk, data_size, tree_metadata.data(py), |
|
78 | 78 | ).map_err(dirstate_error)?; |
|
79 | 79 | let map = Self::create_instance(py, map)?; |
|
80 | 80 | Ok(map.into_object()) |
|
81 | 81 | } |
|
82 | 82 | |
|
83 | 83 | def clear(&self) -> PyResult<PyObject> { |
|
84 | 84 | self.inner(py).borrow_mut().clear(); |
|
85 | 85 | Ok(py.None()) |
|
86 | 86 | } |
|
87 | 87 | |
|
88 | 88 | def get( |
|
89 | 89 | &self, |
|
90 | 90 | key: PyObject, |
|
91 | 91 | default: Option<PyObject> = None |
|
92 | 92 | ) -> PyResult<Option<PyObject>> { |
|
93 | 93 | let key = key.extract::<PyBytes>(py)?; |
|
94 | 94 | match self |
|
95 | 95 | .inner(py) |
|
96 | 96 | .borrow() |
|
97 | 97 | .get(HgPath::new(key.data(py))) |
|
98 | 98 | .map_err(|e| v2_error(py, e))? |
|
99 | 99 | { |
|
100 | 100 | Some(entry) => { |
|
101 | 101 | Ok(Some(DirstateItem::new_as_pyobject(py, entry)?)) |
|
102 | 102 | }, |
|
103 | 103 | None => Ok(default) |
|
104 | 104 | } |
|
105 | 105 | } |
|
106 | 106 | |
|
107 | 107 | def set_dirstate_item( |
|
108 | 108 | &self, |
|
109 | 109 | path: PyObject, |
|
110 | 110 | item: DirstateItem |
|
111 | 111 | ) -> PyResult<PyObject> { |
|
112 | 112 | let f = path.extract::<PyBytes>(py)?; |
|
113 | 113 | let filename = HgPath::new(f.data(py)); |
|
114 | 114 | self.inner(py) |
|
115 | 115 | .borrow_mut() |
|
116 | 116 | .set_entry(filename, item.get_entry(py)) |
|
117 | 117 | .map_err(|e| v2_error(py, e))?; |
|
118 | 118 | Ok(py.None()) |
|
119 | 119 | } |
|
120 | 120 | |
|
121 | 121 | def set_tracked(&self, f: PyObject) -> PyResult<PyBool> { |
|
122 | 122 | let bytes = f.extract::<PyBytes>(py)?; |
|
123 | 123 | let path = HgPath::new(bytes.data(py)); |
|
124 | 124 | let res = self.inner(py).borrow_mut().set_tracked(path); |
|
125 | 125 | let was_tracked = res.or_else(|_| { |
|
126 | 126 | Err(PyErr::new::<exc::OSError, _>(py, "Dirstate error".to_string())) |
|
127 | 127 | })?; |
|
128 | 128 | Ok(was_tracked.to_py_object(py)) |
|
129 | 129 | } |
|
130 | 130 | |
|
131 | 131 | def set_untracked(&self, f: PyObject) -> PyResult<PyBool> { |
|
132 | 132 | let bytes = f.extract::<PyBytes>(py)?; |
|
133 | 133 | let path = HgPath::new(bytes.data(py)); |
|
134 | 134 | let res = self.inner(py).borrow_mut().set_untracked(path); |
|
135 | 135 | let was_tracked = res.or_else(|_| { |
|
136 | 136 | Err(PyErr::new::<exc::OSError, _>(py, "Dirstate error".to_string())) |
|
137 | 137 | })?; |
|
138 | 138 | Ok(was_tracked.to_py_object(py)) |
|
139 | 139 | } |
|
140 | 140 | |
|
141 | 141 | def set_clean( |
|
142 | 142 | &self, |
|
143 | 143 | f: PyObject, |
|
144 | 144 | mode: u32, |
|
145 | 145 | size: u32, |
|
146 | 146 | mtime: (i64, u32, bool) |
|
147 | 147 | ) -> PyResult<PyNone> { |
|
148 | 148 | let (mtime_s, mtime_ns, second_ambiguous) = mtime; |
|
149 | 149 | let timestamp = TruncatedTimestamp::new_truncate( |
|
150 | 150 | mtime_s, mtime_ns, second_ambiguous |
|
151 | 151 | ); |
|
152 | 152 | let bytes = f.extract::<PyBytes>(py)?; |
|
153 | 153 | let path = HgPath::new(bytes.data(py)); |
|
154 | 154 | let res = self.inner(py).borrow_mut().set_clean( |
|
155 | 155 | path, mode, size, timestamp, |
|
156 | 156 | ); |
|
157 | 157 | res.or_else(|_| { |
|
158 | 158 | Err(PyErr::new::<exc::OSError, _>(py, "Dirstate error".to_string())) |
|
159 | 159 | })?; |
|
160 | 160 | Ok(PyNone) |
|
161 | 161 | } |
|
162 | 162 | |
|
163 | 163 | def set_possibly_dirty(&self, f: PyObject) -> PyResult<PyNone> { |
|
164 | 164 | let bytes = f.extract::<PyBytes>(py)?; |
|
165 | 165 | let path = HgPath::new(bytes.data(py)); |
|
166 | 166 | let res = self.inner(py).borrow_mut().set_possibly_dirty(path); |
|
167 | 167 | res.or_else(|_| { |
|
168 | 168 | Err(PyErr::new::<exc::OSError, _>(py, "Dirstate error".to_string())) |
|
169 | 169 | })?; |
|
170 | 170 | Ok(PyNone) |
|
171 | 171 | } |
|
172 | 172 | |
|
173 | 173 | def reset_state( |
|
174 | 174 | &self, |
|
175 | 175 | f: PyObject, |
|
176 | 176 | wc_tracked: bool, |
|
177 | 177 | p1_tracked: bool, |
|
178 | 178 | p2_info: bool, |
|
179 | 179 | has_meaningful_mtime: bool, |
|
180 | 180 | parentfiledata: Option<(u32, u32, Option<(i64, u32, bool)>)>, |
|
181 | 181 | ) -> PyResult<PyNone> { |
|
182 | 182 | let mut has_meaningful_mtime = has_meaningful_mtime; |
|
183 | 183 | let parent_file_data = match parentfiledata { |
|
184 | 184 | None => { |
|
185 | 185 | has_meaningful_mtime = false; |
|
186 | 186 | None |
|
187 | 187 | }, |
|
188 | 188 | Some(data) => { |
|
189 | 189 | let (mode, size, mtime_info) = data; |
|
190 | 190 | let mtime = if let Some(mtime_info) = mtime_info { |
|
191 | 191 | let (mtime_s, mtime_ns, second_ambiguous) = mtime_info; |
|
192 | 192 | let timestamp = TruncatedTimestamp::new_truncate( |
|
193 | 193 | mtime_s, mtime_ns, second_ambiguous |
|
194 | 194 | ); |
|
195 | 195 | Some(timestamp) |
|
196 | 196 | } else { |
|
197 | 197 | has_meaningful_mtime = false; |
|
198 | 198 | None |
|
199 | 199 | }; |
|
200 | 200 | Some(ParentFileData { |
|
201 | 201 | mode_size: Some((mode, size)), |
|
202 | 202 | mtime, |
|
203 | 203 | }) |
|
204 | 204 | } |
|
205 | 205 | }; |
|
206 | 206 | let bytes = f.extract::<PyBytes>(py)?; |
|
207 | 207 | let path = HgPath::new(bytes.data(py)); |
|
208 | 208 | let res = self.inner(py).borrow_mut().reset_state( |
|
209 | 209 | path, |
|
210 | 210 | wc_tracked, |
|
211 | 211 | p1_tracked, |
|
212 | 212 | p2_info, |
|
213 | 213 | has_meaningful_mtime, |
|
214 | 214 | parent_file_data, |
|
215 | 215 | ); |
|
216 | 216 | res.or_else(|_| { |
|
217 | 217 | Err(PyErr::new::<exc::OSError, _>(py, "Dirstate error".to_string())) |
|
218 | 218 | })?; |
|
219 | 219 | Ok(PyNone) |
|
220 | 220 | } |
|
221 | 221 | |
|
222 | 222 | def drop_item_and_copy_source( |
|
223 | 223 | &self, |
|
224 | 224 | f: PyBytes, |
|
225 | 225 | ) -> PyResult<PyNone> { |
|
226 | 226 | self.inner(py) |
|
227 | 227 | .borrow_mut() |
|
228 | 228 | .drop_entry_and_copy_source(HgPath::new(f.data(py))) |
|
229 | 229 | .map_err(|e |dirstate_error(py, e))?; |
|
230 | 230 | Ok(PyNone) |
|
231 | 231 | } |
|
232 | 232 | |
|
233 | 233 | def hastrackeddir(&self, d: PyObject) -> PyResult<PyBool> { |
|
234 | 234 | let d = d.extract::<PyBytes>(py)?; |
|
235 | 235 | Ok(self.inner(py).borrow_mut() |
|
236 | 236 | .has_tracked_dir(HgPath::new(d.data(py))) |
|
237 | 237 | .map_err(|e| { |
|
238 | 238 | PyErr::new::<exc::ValueError, _>(py, e.to_string()) |
|
239 | 239 | })? |
|
240 | 240 | .to_py_object(py)) |
|
241 | 241 | } |
|
242 | 242 | |
|
243 | 243 | def hasdir(&self, d: PyObject) -> PyResult<PyBool> { |
|
244 | 244 | let d = d.extract::<PyBytes>(py)?; |
|
245 | 245 | Ok(self.inner(py).borrow_mut() |
|
246 | 246 | .has_dir(HgPath::new(d.data(py))) |
|
247 | 247 | .map_err(|e| { |
|
248 | 248 | PyErr::new::<exc::ValueError, _>(py, e.to_string()) |
|
249 | 249 | })? |
|
250 | 250 | .to_py_object(py)) |
|
251 | 251 | } |
|
252 | 252 | |
|
253 | 253 | def write_v1( |
|
254 | 254 | &self, |
|
255 | 255 | p1: PyObject, |
|
256 | 256 | p2: PyObject, |
|
257 | 257 | ) -> PyResult<PyBytes> { |
|
258 | 258 | let inner = self.inner(py).borrow(); |
|
259 | 259 | let parents = DirstateParents { |
|
260 | 260 | p1: extract_node_id(py, &p1)?, |
|
261 | 261 | p2: extract_node_id(py, &p2)?, |
|
262 | 262 | }; |
|
263 | 263 | let result = inner.pack_v1(parents); |
|
264 | 264 | match result { |
|
265 | 265 | Ok(packed) => Ok(PyBytes::new(py, &packed)), |
|
266 | 266 | Err(_) => Err(PyErr::new::<exc::OSError, _>( |
|
267 | 267 | py, |
|
268 | 268 | "Dirstate error".to_string(), |
|
269 | 269 | )), |
|
270 | 270 | } |
|
271 | 271 | } |
|
272 | 272 | |
|
273 | 273 | /// Returns new data together with whether that data should be appended to |
|
274 | 274 | /// the existing data file whose content is at `self.on_disk` (True), |
|
275 | 275 | /// instead of written to a new data file (False). |
|
276 | 276 | def write_v2( |
|
277 | 277 | &self, |
|
278 | 278 | can_append: bool, |
|
279 | 279 | ) -> PyResult<PyObject> { |
|
280 | 280 | let inner = self.inner(py).borrow(); |
|
281 | 281 | let result = inner.pack_v2(can_append); |
|
282 | 282 | match result { |
|
283 | 283 | Ok((packed, tree_metadata, append)) => { |
|
284 | 284 | let packed = PyBytes::new(py, &packed); |
|
285 | 285 | let tree_metadata = PyBytes::new(py, tree_metadata.as_bytes()); |
|
286 | 286 | let tuple = (packed, tree_metadata, append); |
|
287 | 287 | Ok(tuple.to_py_object(py).into_object()) |
|
288 | 288 | }, |
|
289 | 289 | Err(_) => Err(PyErr::new::<exc::OSError, _>( |
|
290 | 290 | py, |
|
291 | 291 | "Dirstate error".to_string(), |
|
292 | 292 | )), |
|
293 | 293 | } |
|
294 | 294 | } |
|
295 | 295 | |
|
296 | 296 | def filefoldmapasdict(&self) -> PyResult<PyDict> { |
|
297 | 297 | let dict = PyDict::new(py); |
|
298 | 298 | for item in self.inner(py).borrow_mut().iter() { |
|
299 | 299 | let (path, entry) = item.map_err(|e| v2_error(py, e))?; |
|
300 | 300 | if entry.state() != EntryState::Removed { |
|
301 | 301 | let key = normalize_case(path); |
|
302 | 302 | let value = path; |
|
303 | 303 | dict.set_item( |
|
304 | 304 | py, |
|
305 | 305 | PyBytes::new(py, key.as_bytes()).into_object(), |
|
306 | 306 | PyBytes::new(py, value.as_bytes()).into_object(), |
|
307 | 307 | )?; |
|
308 | 308 | } |
|
309 | 309 | } |
|
310 | 310 | Ok(dict) |
|
311 | 311 | } |
|
312 | 312 | |
|
313 | 313 | def __len__(&self) -> PyResult<usize> { |
|
314 | 314 | Ok(self.inner(py).borrow().len()) |
|
315 | 315 | } |
|
316 | 316 | |
|
317 | 317 | def __contains__(&self, key: PyObject) -> PyResult<bool> { |
|
318 | 318 | let key = key.extract::<PyBytes>(py)?; |
|
319 | 319 | self.inner(py) |
|
320 | 320 | .borrow() |
|
321 | 321 | .contains_key(HgPath::new(key.data(py))) |
|
322 | 322 | .map_err(|e| v2_error(py, e)) |
|
323 | 323 | } |
|
324 | 324 | |
|
325 | 325 | def __getitem__(&self, key: PyObject) -> PyResult<PyObject> { |
|
326 | 326 | let key = key.extract::<PyBytes>(py)?; |
|
327 | 327 | let key = HgPath::new(key.data(py)); |
|
328 | 328 | match self |
|
329 | 329 | .inner(py) |
|
330 | 330 | .borrow() |
|
331 | 331 | .get(key) |
|
332 | 332 | .map_err(|e| v2_error(py, e))? |
|
333 | 333 | { |
|
334 | 334 | Some(entry) => { |
|
335 | 335 | Ok(DirstateItem::new_as_pyobject(py, entry)?) |
|
336 | 336 | }, |
|
337 | 337 | None => Err(PyErr::new::<exc::KeyError, _>( |
|
338 | 338 | py, |
|
339 | 339 | String::from_utf8_lossy(key.as_bytes()), |
|
340 | 340 | )), |
|
341 | 341 | } |
|
342 | 342 | } |
|
343 | 343 | |
|
344 | 344 | def keys(&self) -> PyResult<DirstateMapKeysIterator> { |
|
345 | 345 | let leaked_ref = self.inner(py).leak_immutable(); |
|
346 | 346 | DirstateMapKeysIterator::from_inner( |
|
347 | 347 | py, |
|
348 | 348 | unsafe { leaked_ref.map(py, |o| o.iter()) }, |
|
349 | 349 | ) |
|
350 | 350 | } |
|
351 | 351 | |
|
352 | 352 | def items(&self) -> PyResult<DirstateMapItemsIterator> { |
|
353 | 353 | let leaked_ref = self.inner(py).leak_immutable(); |
|
354 | 354 | DirstateMapItemsIterator::from_inner( |
|
355 | 355 | py, |
|
356 | 356 | unsafe { leaked_ref.map(py, |o| o.iter()) }, |
|
357 | 357 | ) |
|
358 | 358 | } |
|
359 | 359 | |
|
360 | 360 | def __iter__(&self) -> PyResult<DirstateMapKeysIterator> { |
|
361 | 361 | let leaked_ref = self.inner(py).leak_immutable(); |
|
362 | 362 | DirstateMapKeysIterator::from_inner( |
|
363 | 363 | py, |
|
364 | 364 | unsafe { leaked_ref.map(py, |o| o.iter()) }, |
|
365 | 365 | ) |
|
366 | 366 | } |
|
367 | 367 | |
|
368 | 368 | // TODO all copymap* methods, see docstring above |
|
369 | 369 | def copymapcopy(&self) -> PyResult<PyDict> { |
|
370 | 370 | let dict = PyDict::new(py); |
|
371 | 371 | for item in self.inner(py).borrow().copy_map_iter() { |
|
372 | 372 | let (key, value) = item.map_err(|e| v2_error(py, e))?; |
|
373 | 373 | dict.set_item( |
|
374 | 374 | py, |
|
375 | 375 | PyBytes::new(py, key.as_bytes()), |
|
376 | 376 | PyBytes::new(py, value.as_bytes()), |
|
377 | 377 | )?; |
|
378 | 378 | } |
|
379 | 379 | Ok(dict) |
|
380 | 380 | } |
|
381 | 381 | |
|
382 | 382 | def copymapgetitem(&self, key: PyObject) -> PyResult<PyBytes> { |
|
383 | 383 | let key = key.extract::<PyBytes>(py)?; |
|
384 | 384 | match self |
|
385 | 385 | .inner(py) |
|
386 | 386 | .borrow() |
|
387 | 387 | .copy_map_get(HgPath::new(key.data(py))) |
|
388 | 388 | .map_err(|e| v2_error(py, e))? |
|
389 | 389 | { |
|
390 | 390 | Some(copy) => Ok(PyBytes::new(py, copy.as_bytes())), |
|
391 | 391 | None => Err(PyErr::new::<exc::KeyError, _>( |
|
392 | 392 | py, |
|
393 | 393 | String::from_utf8_lossy(key.data(py)), |
|
394 | 394 | )), |
|
395 | 395 | } |
|
396 | 396 | } |
|
397 | 397 | def copymap(&self) -> PyResult<CopyMap> { |
|
398 | 398 | CopyMap::from_inner(py, self.clone_ref(py)) |
|
399 | 399 | } |
|
400 | 400 | |
|
401 | 401 | def copymaplen(&self) -> PyResult<usize> { |
|
402 | 402 | Ok(self.inner(py).borrow().copy_map_len()) |
|
403 | 403 | } |
|
404 | 404 | def copymapcontains(&self, key: PyObject) -> PyResult<bool> { |
|
405 | 405 | let key = key.extract::<PyBytes>(py)?; |
|
406 | 406 | self.inner(py) |
|
407 | 407 | .borrow() |
|
408 | 408 | .copy_map_contains_key(HgPath::new(key.data(py))) |
|
409 | 409 | .map_err(|e| v2_error(py, e)) |
|
410 | 410 | } |
|
411 | 411 | def copymapget( |
|
412 | 412 | &self, |
|
413 | 413 | key: PyObject, |
|
414 | 414 | default: Option<PyObject> |
|
415 | 415 | ) -> PyResult<Option<PyObject>> { |
|
416 | 416 | let key = key.extract::<PyBytes>(py)?; |
|
417 | 417 | match self |
|
418 | 418 | .inner(py) |
|
419 | 419 | .borrow() |
|
420 | 420 | .copy_map_get(HgPath::new(key.data(py))) |
|
421 | 421 | .map_err(|e| v2_error(py, e))? |
|
422 | 422 | { |
|
423 | 423 | Some(copy) => Ok(Some( |
|
424 | 424 | PyBytes::new(py, copy.as_bytes()).into_object(), |
|
425 | 425 | )), |
|
426 | 426 | None => Ok(default), |
|
427 | 427 | } |
|
428 | 428 | } |
|
429 | 429 | def copymapsetitem( |
|
430 | 430 | &self, |
|
431 | 431 | key: PyObject, |
|
432 | 432 | value: PyObject |
|
433 | 433 | ) -> PyResult<PyObject> { |
|
434 | 434 | let key = key.extract::<PyBytes>(py)?; |
|
435 | 435 | let value = value.extract::<PyBytes>(py)?; |
|
436 | 436 | self.inner(py) |
|
437 | 437 | .borrow_mut() |
|
438 | 438 | .copy_map_insert( |
|
439 | 439 | HgPathBuf::from_bytes(key.data(py)), |
|
440 | 440 | HgPathBuf::from_bytes(value.data(py)), |
|
441 | 441 | ) |
|
442 | 442 | .map_err(|e| v2_error(py, e))?; |
|
443 | 443 | Ok(py.None()) |
|
444 | 444 | } |
|
445 | 445 | def copymappop( |
|
446 | 446 | &self, |
|
447 | 447 | key: PyObject, |
|
448 | 448 | default: Option<PyObject> |
|
449 | 449 | ) -> PyResult<Option<PyObject>> { |
|
450 | 450 | let key = key.extract::<PyBytes>(py)?; |
|
451 | 451 | match self |
|
452 | 452 | .inner(py) |
|
453 | 453 | .borrow_mut() |
|
454 | 454 | .copy_map_remove(HgPath::new(key.data(py))) |
|
455 | 455 | .map_err(|e| v2_error(py, e))? |
|
456 | 456 | { |
|
457 | 457 | Some(copy) => Ok(Some( |
|
458 | 458 | PyBytes::new(py, copy.as_bytes()).into_object(), |
|
459 | 459 | )), |
|
460 | 460 | None => Ok(default), |
|
461 | 461 | } |
|
462 | 462 | } |
|
463 | 463 | |
|
464 | 464 | def copymapiter(&self) -> PyResult<CopyMapKeysIterator> { |
|
465 | 465 | let leaked_ref = self.inner(py).leak_immutable(); |
|
466 | 466 | CopyMapKeysIterator::from_inner( |
|
467 | 467 | py, |
|
468 | 468 | unsafe { leaked_ref.map(py, |o| o.copy_map_iter()) }, |
|
469 | 469 | ) |
|
470 | 470 | } |
|
471 | 471 | |
|
472 | 472 | def copymapitemsiter(&self) -> PyResult<CopyMapItemsIterator> { |
|
473 | 473 | let leaked_ref = self.inner(py).leak_immutable(); |
|
474 | 474 | CopyMapItemsIterator::from_inner( |
|
475 | 475 | py, |
|
476 | 476 | unsafe { leaked_ref.map(py, |o| o.copy_map_iter()) }, |
|
477 | 477 | ) |
|
478 | 478 | } |
|
479 | 479 | |
|
480 | 480 | def tracked_dirs(&self) -> PyResult<PyList> { |
|
481 | 481 | let dirs = PyList::new(py, &[]); |
|
482 | 482 | for path in self.inner(py).borrow_mut().iter_tracked_dirs() |
|
483 | 483 | .map_err(|e |dirstate_error(py, e))? |
|
484 | 484 | { |
|
485 | 485 | let path = path.map_err(|e| v2_error(py, e))?; |
|
486 | 486 | let path = PyBytes::new(py, path.as_bytes()); |
|
487 | 487 | dirs.append(py, path.into_object()) |
|
488 | 488 | } |
|
489 | 489 | Ok(dirs) |
|
490 | 490 | } |
|
491 | 491 | |
|
492 | def setparents_fixup(&self) -> PyResult<PyDict> { | |
|
493 | let dict = PyDict::new(py); | |
|
494 | let copies = self.inner(py).borrow_mut().setparents_fixup(); | |
|
495 | for (key, value) in copies.map_err(|e| v2_error(py, e))? { | |
|
496 | dict.set_item( | |
|
497 | py, | |
|
498 | PyBytes::new(py, key.as_bytes()), | |
|
499 | PyBytes::new(py, value.as_bytes()), | |
|
500 | )?; | |
|
501 | } | |
|
502 | Ok(dict) | |
|
503 | } | |
|
504 | ||
|
492 | 505 | def debug_iter(&self, all: bool) -> PyResult<PyList> { |
|
493 | 506 | let dirs = PyList::new(py, &[]); |
|
494 | 507 | for item in self.inner(py).borrow().debug_iter(all) { |
|
495 | 508 | let (path, (state, mode, size, mtime)) = |
|
496 | 509 | item.map_err(|e| v2_error(py, e))?; |
|
497 | 510 | let path = PyBytes::new(py, path.as_bytes()); |
|
498 | 511 | let item = (path, state, mode, size, mtime); |
|
499 | 512 | dirs.append(py, item.to_py_object(py).into_object()) |
|
500 | 513 | } |
|
501 | 514 | Ok(dirs) |
|
502 | 515 | } |
|
503 | 516 | }); |
|
504 | 517 | |
|
505 | 518 | impl DirstateMap { |
|
506 | 519 | pub fn get_inner_mut<'a>( |
|
507 | 520 | &'a self, |
|
508 | 521 | py: Python<'a>, |
|
509 | 522 | ) -> RefMut<'a, OwningDirstateMap> { |
|
510 | 523 | self.inner(py).borrow_mut() |
|
511 | 524 | } |
|
512 | 525 | fn translate_key( |
|
513 | 526 | py: Python, |
|
514 | 527 | res: Result<(&HgPath, DirstateEntry), DirstateV2ParseError>, |
|
515 | 528 | ) -> PyResult<Option<PyBytes>> { |
|
516 | 529 | let (f, _entry) = res.map_err(|e| v2_error(py, e))?; |
|
517 | 530 | Ok(Some(PyBytes::new(py, f.as_bytes()))) |
|
518 | 531 | } |
|
519 | 532 | fn translate_key_value( |
|
520 | 533 | py: Python, |
|
521 | 534 | res: Result<(&HgPath, DirstateEntry), DirstateV2ParseError>, |
|
522 | 535 | ) -> PyResult<Option<(PyBytes, PyObject)>> { |
|
523 | 536 | let (f, entry) = res.map_err(|e| v2_error(py, e))?; |
|
524 | 537 | Ok(Some(( |
|
525 | 538 | PyBytes::new(py, f.as_bytes()), |
|
526 | 539 | DirstateItem::new_as_pyobject(py, entry)?, |
|
527 | 540 | ))) |
|
528 | 541 | } |
|
529 | 542 | } |
|
530 | 543 | |
|
531 | 544 | py_shared_iterator!( |
|
532 | 545 | DirstateMapKeysIterator, |
|
533 | 546 | UnsafePyLeaked<StateMapIter<'static>>, |
|
534 | 547 | DirstateMap::translate_key, |
|
535 | 548 | Option<PyBytes> |
|
536 | 549 | ); |
|
537 | 550 | |
|
538 | 551 | py_shared_iterator!( |
|
539 | 552 | DirstateMapItemsIterator, |
|
540 | 553 | UnsafePyLeaked<StateMapIter<'static>>, |
|
541 | 554 | DirstateMap::translate_key_value, |
|
542 | 555 | Option<(PyBytes, PyObject)> |
|
543 | 556 | ); |
|
544 | 557 | |
|
545 | 558 | fn extract_node_id(py: Python, obj: &PyObject) -> PyResult<Node> { |
|
546 | 559 | let bytes = obj.extract::<PyBytes>(py)?; |
|
547 | 560 | match bytes.data(py).try_into() { |
|
548 | 561 | Ok(s) => Ok(s), |
|
549 | 562 | Err(e) => Err(PyErr::new::<exc::ValueError, _>(py, e.to_string())), |
|
550 | 563 | } |
|
551 | 564 | } |
|
552 | 565 | |
|
553 | 566 | pub(super) fn v2_error(py: Python<'_>, _: DirstateV2ParseError) -> PyErr { |
|
554 | 567 | PyErr::new::<exc::ValueError, _>(py, "corrupted dirstate-v2") |
|
555 | 568 | } |
|
556 | 569 | |
|
557 | 570 | fn dirstate_error(py: Python<'_>, e: DirstateError) -> PyErr { |
|
558 | 571 | PyErr::new::<exc::OSError, _>(py, format!("Dirstate error: {:?}", e)) |
|
559 | 572 | } |
General Comments 0
You need to be logged in to leave comments.
Login now