Show More
@@ -1,917 +1,917 b'' | |||
|
1 | 1 | # copies.py - copy detection for Mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008 Matt Mackall <mpm@selenic.com> |
|
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 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import collections |
|
11 | 11 | import heapq |
|
12 | 12 | import os |
|
13 | 13 | |
|
14 | 14 | from .i18n import _ |
|
15 | 15 | |
|
16 | 16 | from . import ( |
|
17 | 17 | match as matchmod, |
|
18 | 18 | node, |
|
19 | 19 | pathutil, |
|
20 | 20 | scmutil, |
|
21 | 21 | util, |
|
22 | 22 | ) |
|
23 | 23 | from .utils import ( |
|
24 | 24 | stringutil, |
|
25 | 25 | ) |
|
26 | 26 | |
|
27 | 27 | def _findlimit(repo, ctxa, ctxb): |
|
28 | 28 | """ |
|
29 | 29 | Find the last revision that needs to be checked to ensure that a full |
|
30 | 30 | transitive closure for file copies can be properly calculated. |
|
31 | 31 | Generally, this means finding the earliest revision number that's an |
|
32 | 32 | ancestor of a or b but not both, except when a or b is a direct descendent |
|
33 | 33 | of the other, in which case we can return the minimum revnum of a and b. |
|
34 | 34 | """ |
|
35 | 35 | |
|
36 | 36 | # basic idea: |
|
37 | 37 | # - mark a and b with different sides |
|
38 | 38 | # - if a parent's children are all on the same side, the parent is |
|
39 | 39 | # on that side, otherwise it is on no side |
|
40 | 40 | # - walk the graph in topological order with the help of a heap; |
|
41 | 41 | # - add unseen parents to side map |
|
42 | 42 | # - clear side of any parent that has children on different sides |
|
43 | 43 | # - track number of interesting revs that might still be on a side |
|
44 | 44 | # - track the lowest interesting rev seen |
|
45 | 45 | # - quit when interesting revs is zero |
|
46 | 46 | |
|
47 | 47 | cl = repo.changelog |
|
48 | 48 | wdirparents = None |
|
49 | 49 | a = ctxa.rev() |
|
50 | 50 | b = ctxb.rev() |
|
51 | 51 | if a is None: |
|
52 | 52 | wdirparents = (ctxa.p1(), ctxa.p2()) |
|
53 | 53 | a = node.wdirrev |
|
54 | 54 | if b is None: |
|
55 | 55 | assert not wdirparents |
|
56 | 56 | wdirparents = (ctxb.p1(), ctxb.p2()) |
|
57 | 57 | b = node.wdirrev |
|
58 | 58 | |
|
59 | 59 | side = {a: -1, b: 1} |
|
60 | 60 | visit = [-a, -b] |
|
61 | 61 | heapq.heapify(visit) |
|
62 | 62 | interesting = len(visit) |
|
63 | 63 | limit = node.wdirrev |
|
64 | 64 | |
|
65 | 65 | while interesting: |
|
66 | 66 | r = -heapq.heappop(visit) |
|
67 | 67 | if r == node.wdirrev: |
|
68 | 68 | parents = [pctx.rev() for pctx in wdirparents] |
|
69 | 69 | else: |
|
70 | 70 | parents = cl.parentrevs(r) |
|
71 | 71 | if parents[1] == node.nullrev: |
|
72 | 72 | parents = parents[:1] |
|
73 | 73 | for p in parents: |
|
74 | 74 | if p not in side: |
|
75 | 75 | # first time we see p; add it to visit |
|
76 | 76 | side[p] = side[r] |
|
77 | 77 | if side[p]: |
|
78 | 78 | interesting += 1 |
|
79 | 79 | heapq.heappush(visit, -p) |
|
80 | 80 | elif side[p] and side[p] != side[r]: |
|
81 | 81 | # p was interesting but now we know better |
|
82 | 82 | side[p] = 0 |
|
83 | 83 | interesting -= 1 |
|
84 | 84 | if side[r]: |
|
85 | 85 | limit = r # lowest rev visited |
|
86 | 86 | interesting -= 1 |
|
87 | 87 | |
|
88 | 88 | # Consider the following flow (see test-commit-amend.t under issue4405): |
|
89 | 89 | # 1/ File 'a0' committed |
|
90 | 90 | # 2/ File renamed from 'a0' to 'a1' in a new commit (call it 'a1') |
|
91 | 91 | # 3/ Move back to first commit |
|
92 | 92 | # 4/ Create a new commit via revert to contents of 'a1' (call it 'a1-amend') |
|
93 | 93 | # 5/ Rename file from 'a1' to 'a2' and commit --amend 'a1-msg' |
|
94 | 94 | # |
|
95 | 95 | # During the amend in step five, we will be in this state: |
|
96 | 96 | # |
|
97 | 97 | # @ 3 temporary amend commit for a1-amend |
|
98 | 98 | # | |
|
99 | 99 | # o 2 a1-amend |
|
100 | 100 | # | |
|
101 | 101 | # | o 1 a1 |
|
102 | 102 | # |/ |
|
103 | 103 | # o 0 a0 |
|
104 | 104 | # |
|
105 | 105 | # When _findlimit is called, a and b are revs 3 and 0, so limit will be 2, |
|
106 | 106 | # yet the filelog has the copy information in rev 1 and we will not look |
|
107 | 107 | # back far enough unless we also look at the a and b as candidates. |
|
108 | 108 | # This only occurs when a is a descendent of b or visa-versa. |
|
109 | 109 | return min(limit, a, b) |
|
110 | 110 | |
|
111 | 111 | def _chain(src, dst, a, b): |
|
112 | 112 | """chain two sets of copies a->b""" |
|
113 | 113 | t = a.copy() |
|
114 | 114 | for k, v in b.iteritems(): |
|
115 | 115 | if v in t: |
|
116 | 116 | # found a chain |
|
117 | 117 | if t[v] != k: |
|
118 | 118 | # file wasn't renamed back to itself |
|
119 | 119 | t[k] = t[v] |
|
120 | 120 | if v not in dst: |
|
121 | 121 | # chain was a rename, not a copy |
|
122 | 122 | del t[v] |
|
123 | 123 | if v in src: |
|
124 | 124 | # file is a copy of an existing file |
|
125 | 125 | t[k] = v |
|
126 | 126 | |
|
127 | 127 | # remove criss-crossed copies |
|
128 | 128 | for k, v in list(t.items()): |
|
129 | 129 | if k in src and v in dst: |
|
130 | 130 | del t[k] |
|
131 | 131 | |
|
132 | 132 | return t |
|
133 | 133 | |
|
134 | 134 | def _tracefile(fctx, am, limit=node.nullrev): |
|
135 | 135 | """return file context that is the ancestor of fctx present in ancestor |
|
136 | 136 | manifest am, stopping after the first ancestor lower than limit""" |
|
137 | 137 | |
|
138 | 138 | for f in fctx.ancestors(): |
|
139 | 139 | if am.get(f.path(), None) == f.filenode(): |
|
140 | 140 | return f |
|
141 | 141 | if limit >= 0 and not f.isintroducedafter(limit): |
|
142 | 142 | return None |
|
143 | 143 | |
|
144 | 144 | def _dirstatecopies(d, match=None): |
|
145 | 145 | ds = d._repo.dirstate |
|
146 | 146 | c = ds.copies().copy() |
|
147 | 147 | for k in list(c): |
|
148 | 148 | if ds[k] not in 'anm' or (match and not match(k)): |
|
149 | 149 | del c[k] |
|
150 | 150 | return c |
|
151 | 151 | |
|
152 | 152 | def _computeforwardmissing(a, b, match=None): |
|
153 | 153 | """Computes which files are in b but not a. |
|
154 | 154 | This is its own function so extensions can easily wrap this call to see what |
|
155 | 155 | files _forwardcopies is about to process. |
|
156 | 156 | """ |
|
157 | 157 | ma = a.manifest() |
|
158 | 158 | mb = b.manifest() |
|
159 | 159 | return mb.filesnotin(ma, match=match) |
|
160 | 160 | |
|
161 | 161 | def _committedforwardcopies(a, b, match): |
|
162 | 162 | """Like _forwardcopies(), but b.rev() cannot be None (working copy)""" |
|
163 | 163 | # files might have to be traced back to the fctx parent of the last |
|
164 | 164 | # one-side-only changeset, but not further back than that |
|
165 | 165 | repo = a._repo |
|
166 | 166 | debug = repo.ui.debugflag and repo.ui.configbool('devel', 'debug.copies') |
|
167 | 167 | dbg = repo.ui.debug |
|
168 | 168 | if debug: |
|
169 | 169 | dbg('debug.copies: looking into rename from %s to %s\n' |
|
170 | 170 | % (a, b)) |
|
171 | 171 | limit = _findlimit(repo, a, b) |
|
172 | 172 | if debug: |
|
173 | 173 | dbg('debug.copies: search limit: %d\n' % limit) |
|
174 | 174 | am = a.manifest() |
|
175 | 175 | |
|
176 | 176 | # find where new files came from |
|
177 | 177 | # we currently don't try to find where old files went, too expensive |
|
178 | 178 | # this means we can miss a case like 'hg rm b; hg cp a b' |
|
179 | 179 | cm = {} |
|
180 | 180 | |
|
181 | 181 | # Computing the forward missing is quite expensive on large manifests, since |
|
182 | 182 | # it compares the entire manifests. We can optimize it in the common use |
|
183 | 183 | # case of computing what copies are in a commit versus its parent (like |
|
184 | 184 | # during a rebase or histedit). Note, we exclude merge commits from this |
|
185 | 185 | # optimization, since the ctx.files() for a merge commit is not correct for |
|
186 | 186 | # this comparison. |
|
187 | 187 | forwardmissingmatch = match |
|
188 | 188 | if b.p1() == a and b.p2().node() == node.nullid: |
|
189 | 189 | filesmatcher = scmutil.matchfiles(a._repo, b.files()) |
|
190 | 190 | forwardmissingmatch = matchmod.intersectmatchers(match, filesmatcher) |
|
191 | 191 | missing = _computeforwardmissing(a, b, match=forwardmissingmatch) |
|
192 | 192 | |
|
193 | 193 | ancestrycontext = a._repo.changelog.ancestors([b.rev()], inclusive=True) |
|
194 | 194 | |
|
195 | 195 | if debug: |
|
196 | 196 | dbg('debug.copies: missing file to search: %d\n' % len(missing)) |
|
197 | 197 | |
|
198 | 198 | for f in missing: |
|
199 | 199 | if debug: |
|
200 | 200 | dbg('debug.copies: tracing file: %s\n' % f) |
|
201 | 201 | fctx = b[f] |
|
202 | 202 | fctx._ancestrycontext = ancestrycontext |
|
203 | 203 | |
|
204 | 204 | if debug: |
|
205 | 205 | start = util.timer() |
|
206 | 206 | ofctx = _tracefile(fctx, am, limit) |
|
207 | 207 | if ofctx: |
|
208 | 208 | if debug: |
|
209 | 209 | dbg('debug.copies: rename of: %s\n' % ofctx._path) |
|
210 | 210 | cm[f] = ofctx.path() |
|
211 | 211 | if debug: |
|
212 | 212 | dbg('debug.copies: time: %f seconds\n' |
|
213 | 213 | % (util.timer() - start)) |
|
214 | 214 | return cm |
|
215 | 215 | |
|
216 | 216 | def _forwardcopies(a, b, match=None): |
|
217 | 217 | """find {dst@b: src@a} copy mapping where a is an ancestor of b""" |
|
218 | 218 | |
|
219 | 219 | match = a.repo().narrowmatch(match) |
|
220 | 220 | # check for working copy |
|
221 | 221 | if b.rev() is None: |
|
222 | 222 | if a == b.p1(): |
|
223 | 223 | # short-circuit to avoid issues with merge states |
|
224 | 224 | return _dirstatecopies(b, match) |
|
225 | 225 | |
|
226 | 226 | cm = _committedforwardcopies(a, b.p1(), match) |
|
227 | 227 | # combine copies from dirstate if necessary |
|
228 | 228 | return _chain(a, b, cm, _dirstatecopies(b, match)) |
|
229 | 229 | return _committedforwardcopies(a, b, match) |
|
230 | 230 | |
|
231 | 231 | def _backwardrenames(a, b): |
|
232 | 232 | if a._repo.ui.config('experimental', 'copytrace') == 'off': |
|
233 | 233 | return {} |
|
234 | 234 | |
|
235 | 235 | # Even though we're not taking copies into account, 1:n rename situations |
|
236 | 236 | # can still exist (e.g. hg cp a b; hg mv a c). In those cases we |
|
237 | 237 | # arbitrarily pick one of the renames. |
|
238 | 238 | f = _forwardcopies(b, a) |
|
239 | 239 | r = {} |
|
240 | 240 | for k, v in sorted(f.iteritems()): |
|
241 | 241 | # remove copies |
|
242 | 242 | if v in a: |
|
243 | 243 | continue |
|
244 | 244 | r[v] = k |
|
245 | 245 | return r |
|
246 | 246 | |
|
247 | 247 | def pathcopies(x, y, match=None): |
|
248 | 248 | """find {dst@y: src@x} copy mapping for directed compare""" |
|
249 | 249 | repo = x._repo |
|
250 | 250 | debug = repo.ui.debugflag and repo.ui.configbool('devel', 'debug.copies') |
|
251 | 251 | if debug: |
|
252 | 252 | repo.ui.debug('debug.copies: searching copies from %s to %s\n' |
|
253 | 253 | % (x, y)) |
|
254 | 254 | if x == y or not x or not y: |
|
255 | 255 | return {} |
|
256 | 256 | a = y.ancestor(x) |
|
257 | 257 | if a == x: |
|
258 | 258 | if debug: |
|
259 | 259 | repo.ui.debug('debug.copies: search mode: forward\n') |
|
260 | 260 | return _forwardcopies(x, y, match=match) |
|
261 | 261 | if a == y: |
|
262 | 262 | if debug: |
|
263 | 263 | repo.ui.debug('debug.copies: search mode: backward\n') |
|
264 | 264 | return _backwardrenames(x, y) |
|
265 | 265 | if debug: |
|
266 | 266 | repo.ui.debug('debug.copies: search mode: combined\n') |
|
267 | 267 | return _chain(x, y, _backwardrenames(x, a), |
|
268 | 268 | _forwardcopies(a, y, match=match)) |
|
269 | 269 | |
|
270 | 270 | def _computenonoverlap(repo, c1, c2, addedinm1, addedinm2, baselabel=''): |
|
271 | 271 | """Computes, based on addedinm1 and addedinm2, the files exclusive to c1 |
|
272 | 272 | and c2. This is its own function so extensions can easily wrap this call |
|
273 | 273 | to see what files mergecopies is about to process. |
|
274 | 274 | |
|
275 | 275 | Even though c1 and c2 are not used in this function, they are useful in |
|
276 | 276 | other extensions for being able to read the file nodes of the changed files. |
|
277 | 277 | |
|
278 | 278 | "baselabel" can be passed to help distinguish the multiple computations |
|
279 | 279 | done in the graft case. |
|
280 | 280 | """ |
|
281 | 281 | u1 = sorted(addedinm1 - addedinm2) |
|
282 | 282 | u2 = sorted(addedinm2 - addedinm1) |
|
283 | 283 | |
|
284 | 284 | header = " unmatched files in %s" |
|
285 | 285 | if baselabel: |
|
286 | 286 | header += ' (from %s)' % baselabel |
|
287 | 287 | if u1: |
|
288 | 288 | repo.ui.debug("%s:\n %s\n" % (header % 'local', "\n ".join(u1))) |
|
289 | 289 | if u2: |
|
290 | 290 | repo.ui.debug("%s:\n %s\n" % (header % 'other', "\n ".join(u2))) |
|
291 | 291 | |
|
292 | 292 | return u1, u2 |
|
293 | 293 | |
|
294 | 294 | def _makegetfctx(ctx): |
|
295 | 295 | """return a 'getfctx' function suitable for _checkcopies usage |
|
296 | 296 | |
|
297 | 297 | We have to re-setup the function building 'filectx' for each |
|
298 | 298 | '_checkcopies' to ensure the linkrev adjustment is properly setup for |
|
299 | 299 | each. Linkrev adjustment is important to avoid bug in rename |
|
300 | 300 | detection. Moreover, having a proper '_ancestrycontext' setup ensures |
|
301 | 301 | the performance impact of this adjustment is kept limited. Without it, |
|
302 | 302 | each file could do a full dag traversal making the time complexity of |
|
303 | 303 | the operation explode (see issue4537). |
|
304 | 304 | |
|
305 | 305 | This function exists here mostly to limit the impact on stable. Feel |
|
306 | 306 | free to refactor on default. |
|
307 | 307 | """ |
|
308 | 308 | rev = ctx.rev() |
|
309 | 309 | repo = ctx._repo |
|
310 | 310 | ac = getattr(ctx, '_ancestrycontext', None) |
|
311 | 311 | if ac is None: |
|
312 | 312 | revs = [rev] |
|
313 | 313 | if rev is None: |
|
314 | 314 | revs = [p.rev() for p in ctx.parents()] |
|
315 | 315 | ac = repo.changelog.ancestors(revs, inclusive=True) |
|
316 | 316 | ctx._ancestrycontext = ac |
|
317 | 317 | def makectx(f, n): |
|
318 | 318 | if n in node.wdirfilenodeids: # in a working context? |
|
319 | 319 | if ctx.rev() is None: |
|
320 | 320 | return ctx.filectx(f) |
|
321 | 321 | return repo[None][f] |
|
322 | 322 | fctx = repo.filectx(f, fileid=n) |
|
323 | 323 | # setup only needed for filectx not create from a changectx |
|
324 | 324 | fctx._ancestrycontext = ac |
|
325 | 325 | fctx._descendantrev = rev |
|
326 | 326 | return fctx |
|
327 | 327 | return util.lrucachefunc(makectx) |
|
328 | 328 | |
|
329 | 329 | def _combinecopies(copyfrom, copyto, finalcopy, diverge, incompletediverge): |
|
330 | 330 | """combine partial copy paths""" |
|
331 | 331 | remainder = {} |
|
332 | 332 | for f in copyfrom: |
|
333 | 333 | if f in copyto: |
|
334 | 334 | finalcopy[copyto[f]] = copyfrom[f] |
|
335 | 335 | del copyto[f] |
|
336 | 336 | for f in incompletediverge: |
|
337 | 337 | assert f not in diverge |
|
338 | 338 | ic = incompletediverge[f] |
|
339 | 339 | if ic[0] in copyto: |
|
340 | 340 | diverge[f] = [copyto[ic[0]], ic[1]] |
|
341 | 341 | else: |
|
342 | 342 | remainder[f] = ic |
|
343 | 343 | return remainder |
|
344 | 344 | |
|
345 | 345 | def mergecopies(repo, c1, c2, base): |
|
346 | 346 | """ |
|
347 | 347 | The function calling different copytracing algorithms on the basis of config |
|
348 | 348 | which find moves and copies between context c1 and c2 that are relevant for |
|
349 | 349 | merging. 'base' will be used as the merge base. |
|
350 | 350 | |
|
351 | 351 | Copytracing is used in commands like rebase, merge, unshelve, etc to merge |
|
352 | 352 | files that were moved/ copied in one merge parent and modified in another. |
|
353 | 353 | For example: |
|
354 | 354 | |
|
355 | 355 | o ---> 4 another commit |
|
356 | 356 | | |
|
357 | 357 | | o ---> 3 commit that modifies a.txt |
|
358 | 358 | | / |
|
359 | 359 | o / ---> 2 commit that moves a.txt to b.txt |
|
360 | 360 | |/ |
|
361 | 361 | o ---> 1 merge base |
|
362 | 362 | |
|
363 | 363 | If we try to rebase revision 3 on revision 4, since there is no a.txt in |
|
364 | 364 | revision 4, and if user have copytrace disabled, we prints the following |
|
365 | 365 | message: |
|
366 | 366 | |
|
367 | 367 | ```other changed <file> which local deleted``` |
|
368 | 368 | |
|
369 | 369 | Returns five dicts: "copy", "movewithdir", "diverge", "renamedelete" and |
|
370 | 370 | "dirmove". |
|
371 | 371 | |
|
372 | 372 | "copy" is a mapping from destination name -> source name, |
|
373 | 373 | where source is in c1 and destination is in c2 or vice-versa. |
|
374 | 374 | |
|
375 | 375 | "movewithdir" is a mapping from source name -> destination name, |
|
376 | 376 | where the file at source present in one context but not the other |
|
377 | 377 | needs to be moved to destination by the merge process, because the |
|
378 | 378 | other context moved the directory it is in. |
|
379 | 379 | |
|
380 | 380 | "diverge" is a mapping of source name -> list of destination names |
|
381 | 381 | for divergent renames. |
|
382 | 382 | |
|
383 | 383 | "renamedelete" is a mapping of source name -> list of destination |
|
384 | 384 | names for files deleted in c1 that were renamed in c2 or vice-versa. |
|
385 | 385 | |
|
386 | 386 | "dirmove" is a mapping of detected source dir -> destination dir renames. |
|
387 | 387 | This is needed for handling changes to new files previously grafted into |
|
388 | 388 | renamed directories. |
|
389 | 389 | """ |
|
390 | 390 | # avoid silly behavior for update from empty dir |
|
391 | 391 | if not c1 or not c2 or c1 == c2: |
|
392 | 392 | return {}, {}, {}, {}, {} |
|
393 | 393 | |
|
394 | 394 | # avoid silly behavior for parent -> working dir |
|
395 | 395 | if c2.node() is None and c1.node() == repo.dirstate.p1(): |
|
396 | 396 | return repo.dirstate.copies(), {}, {}, {}, {} |
|
397 | 397 | |
|
398 | 398 | copytracing = repo.ui.config('experimental', 'copytrace') |
|
399 | 399 | boolctrace = stringutil.parsebool(copytracing) |
|
400 | 400 | |
|
401 | 401 | # Copy trace disabling is explicitly below the node == p1 logic above |
|
402 | 402 | # because the logic above is required for a simple copy to be kept across a |
|
403 | 403 | # rebase. |
|
404 | 404 | if copytracing == 'heuristics': |
|
405 | 405 | # Do full copytracing if only non-public revisions are involved as |
|
406 | 406 | # that will be fast enough and will also cover the copies which could |
|
407 | 407 | # be missed by heuristics |
|
408 | 408 | if _isfullcopytraceable(repo, c1, base): |
|
409 | 409 | return _fullcopytracing(repo, c1, c2, base) |
|
410 | 410 | return _heuristicscopytracing(repo, c1, c2, base) |
|
411 | 411 | elif boolctrace is False: |
|
412 | 412 | # stringutil.parsebool() returns None when it is unable to parse the |
|
413 | 413 | # value, so we should rely on making sure copytracing is on such cases |
|
414 | 414 | return {}, {}, {}, {}, {} |
|
415 | 415 | else: |
|
416 | 416 | return _fullcopytracing(repo, c1, c2, base) |
|
417 | 417 | |
|
418 | 418 | def _isfullcopytraceable(repo, c1, base): |
|
419 | 419 | """ Checks that if base, source and destination are all no-public branches, |
|
420 | 420 | if yes let's use the full copytrace algorithm for increased capabilities |
|
421 | 421 | since it will be fast enough. |
|
422 | 422 | |
|
423 | 423 | `experimental.copytrace.sourcecommitlimit` can be used to set a limit for |
|
424 | 424 | number of changesets from c1 to base such that if number of changesets are |
|
425 | 425 | more than the limit, full copytracing algorithm won't be used. |
|
426 | 426 | """ |
|
427 | 427 | if c1.rev() is None: |
|
428 | 428 | c1 = c1.p1() |
|
429 | 429 | if c1.mutable() and base.mutable(): |
|
430 | 430 | sourcecommitlimit = repo.ui.configint('experimental', |
|
431 | 431 | 'copytrace.sourcecommitlimit') |
|
432 | 432 | commits = len(repo.revs('%d::%d', base.rev(), c1.rev())) |
|
433 | 433 | return commits < sourcecommitlimit |
|
434 | 434 | return False |
|
435 | 435 | |
|
436 | 436 | def _fullcopytracing(repo, c1, c2, base): |
|
437 | 437 | """ The full copytracing algorithm which finds all the new files that were |
|
438 | 438 | added from merge base up to the top commit and for each file it checks if |
|
439 | 439 | this file was copied from another file. |
|
440 | 440 | |
|
441 | 441 | This is pretty slow when a lot of changesets are involved but will track all |
|
442 | 442 | the copies. |
|
443 | 443 | """ |
|
444 | 444 | # In certain scenarios (e.g. graft, update or rebase), base can be |
|
445 | 445 | # overridden We still need to know a real common ancestor in this case We |
|
446 | 446 | # can't just compute _c1.ancestor(_c2) and compare it to ca, because there |
|
447 | 447 | # can be multiple common ancestors, e.g. in case of bidmerge. Because our |
|
448 | 448 | # caller may not know if the revision passed in lieu of the CA is a genuine |
|
449 | 449 | # common ancestor or not without explicitly checking it, it's better to |
|
450 | 450 | # determine that here. |
|
451 | 451 | # |
|
452 | 452 | # base.isancestorof(wc) is False, work around that |
|
453 | 453 | _c1 = c1.p1() if c1.rev() is None else c1 |
|
454 | 454 | _c2 = c2.p1() if c2.rev() is None else c2 |
|
455 | 455 | # an endpoint is "dirty" if it isn't a descendant of the merge base |
|
456 | 456 | # if we have a dirty endpoint, we need to trigger graft logic, and also |
|
457 | 457 | # keep track of which endpoint is dirty |
|
458 | 458 | dirtyc1 = not base.isancestorof(_c1) |
|
459 | 459 | dirtyc2 = not base.isancestorof(_c2) |
|
460 | 460 | graft = dirtyc1 or dirtyc2 |
|
461 | 461 | tca = base |
|
462 | 462 | if graft: |
|
463 | 463 | tca = _c1.ancestor(_c2) |
|
464 | 464 | |
|
465 | 465 | limit = _findlimit(repo, c1, c2) |
|
466 | 466 | repo.ui.debug(" searching for copies back to rev %d\n" % limit) |
|
467 | 467 | |
|
468 | 468 | m1 = c1.manifest() |
|
469 | 469 | m2 = c2.manifest() |
|
470 | 470 | mb = base.manifest() |
|
471 | 471 | |
|
472 | 472 | # gather data from _checkcopies: |
|
473 | 473 | # - diverge = record all diverges in this dict |
|
474 | 474 | # - copy = record all non-divergent copies in this dict |
|
475 | 475 | # - fullcopy = record all copies in this dict |
|
476 | 476 | # - incomplete = record non-divergent partial copies here |
|
477 | 477 | # - incompletediverge = record divergent partial copies here |
|
478 | 478 | diverge = {} # divergence data is shared |
|
479 | 479 | incompletediverge = {} |
|
480 | 480 | data1 = {'copy': {}, |
|
481 | 481 | 'fullcopy': {}, |
|
482 | 482 | 'incomplete': {}, |
|
483 | 483 | 'diverge': diverge, |
|
484 | 484 | 'incompletediverge': incompletediverge, |
|
485 | 485 | } |
|
486 | 486 | data2 = {'copy': {}, |
|
487 | 487 | 'fullcopy': {}, |
|
488 | 488 | 'incomplete': {}, |
|
489 | 489 | 'diverge': diverge, |
|
490 | 490 | 'incompletediverge': incompletediverge, |
|
491 | 491 | } |
|
492 | 492 | |
|
493 | 493 | # find interesting file sets from manifests |
|
494 | 494 | addedinm1 = m1.filesnotin(mb, repo.narrowmatch()) |
|
495 | 495 | addedinm2 = m2.filesnotin(mb, repo.narrowmatch()) |
|
496 | 496 | bothnew = sorted(addedinm1 & addedinm2) |
|
497 | 497 | if tca == base: |
|
498 | 498 | # unmatched file from base |
|
499 | 499 | u1r, u2r = _computenonoverlap(repo, c1, c2, addedinm1, addedinm2) |
|
500 | 500 | u1u, u2u = u1r, u2r |
|
501 | 501 | else: |
|
502 | 502 | # unmatched file from base (DAG rotation in the graft case) |
|
503 | 503 | u1r, u2r = _computenonoverlap(repo, c1, c2, addedinm1, addedinm2, |
|
504 | 504 | baselabel='base') |
|
505 | 505 | # unmatched file from topological common ancestors (no DAG rotation) |
|
506 | 506 | # need to recompute this for directory move handling when grafting |
|
507 | 507 | mta = tca.manifest() |
|
508 | 508 | u1u, u2u = _computenonoverlap(repo, c1, c2, |
|
509 | 509 | m1.filesnotin(mta, repo.narrowmatch()), |
|
510 | 510 | m2.filesnotin(mta, repo.narrowmatch()), |
|
511 | 511 | baselabel='topological common ancestor') |
|
512 | 512 | |
|
513 | 513 | for f in u1u: |
|
514 | 514 | _checkcopies(c1, c2, f, base, tca, dirtyc1, limit, data1) |
|
515 | 515 | |
|
516 | 516 | for f in u2u: |
|
517 | 517 | _checkcopies(c2, c1, f, base, tca, dirtyc2, limit, data2) |
|
518 | 518 | |
|
519 | 519 | copy = dict(data1['copy']) |
|
520 | 520 | copy.update(data2['copy']) |
|
521 | 521 | fullcopy = dict(data1['fullcopy']) |
|
522 | 522 | fullcopy.update(data2['fullcopy']) |
|
523 | 523 | |
|
524 | 524 | if dirtyc1: |
|
525 | 525 | _combinecopies(data2['incomplete'], data1['incomplete'], copy, diverge, |
|
526 | 526 | incompletediverge) |
|
527 | 527 | else: |
|
528 | 528 | _combinecopies(data1['incomplete'], data2['incomplete'], copy, diverge, |
|
529 | 529 | incompletediverge) |
|
530 | 530 | |
|
531 | 531 | renamedelete = {} |
|
532 | 532 | renamedeleteset = set() |
|
533 | 533 | divergeset = set() |
|
534 | 534 | for of, fl in list(diverge.items()): |
|
535 | 535 | if len(fl) == 1 or of in c1 or of in c2: |
|
536 | 536 | del diverge[of] # not actually divergent, or not a rename |
|
537 | 537 | if of not in c1 and of not in c2: |
|
538 | 538 | # renamed on one side, deleted on the other side, but filter |
|
539 | 539 | # out files that have been renamed and then deleted |
|
540 | 540 | renamedelete[of] = [f for f in fl if f in c1 or f in c2] |
|
541 | 541 | renamedeleteset.update(fl) # reverse map for below |
|
542 | 542 | else: |
|
543 | 543 | divergeset.update(fl) # reverse map for below |
|
544 | 544 | |
|
545 | 545 | if bothnew: |
|
546 | 546 | repo.ui.debug(" unmatched files new in both:\n %s\n" |
|
547 | 547 | % "\n ".join(bothnew)) |
|
548 | 548 | bothdiverge = {} |
|
549 | 549 | bothincompletediverge = {} |
|
550 | 550 | remainder = {} |
|
551 | 551 | both1 = {'copy': {}, |
|
552 | 552 | 'fullcopy': {}, |
|
553 | 553 | 'incomplete': {}, |
|
554 | 554 | 'diverge': bothdiverge, |
|
555 | 555 | 'incompletediverge': bothincompletediverge |
|
556 | 556 | } |
|
557 | 557 | both2 = {'copy': {}, |
|
558 | 558 | 'fullcopy': {}, |
|
559 | 559 | 'incomplete': {}, |
|
560 | 560 | 'diverge': bothdiverge, |
|
561 | 561 | 'incompletediverge': bothincompletediverge |
|
562 | 562 | } |
|
563 | 563 | for f in bothnew: |
|
564 | 564 | _checkcopies(c1, c2, f, base, tca, dirtyc1, limit, both1) |
|
565 | 565 | _checkcopies(c2, c1, f, base, tca, dirtyc2, limit, both2) |
|
566 | 566 | if dirtyc1: |
|
567 | 567 | # incomplete copies may only be found on the "dirty" side for bothnew |
|
568 | 568 | assert not both2['incomplete'] |
|
569 | 569 | remainder = _combinecopies({}, both1['incomplete'], copy, bothdiverge, |
|
570 | 570 | bothincompletediverge) |
|
571 | 571 | elif dirtyc2: |
|
572 | 572 | assert not both1['incomplete'] |
|
573 | 573 | remainder = _combinecopies({}, both2['incomplete'], copy, bothdiverge, |
|
574 | 574 | bothincompletediverge) |
|
575 | 575 | else: |
|
576 | 576 | # incomplete copies and divergences can't happen outside grafts |
|
577 | 577 | assert not both1['incomplete'] |
|
578 | 578 | assert not both2['incomplete'] |
|
579 | 579 | assert not bothincompletediverge |
|
580 | 580 | for f in remainder: |
|
581 | 581 | assert f not in bothdiverge |
|
582 | 582 | ic = remainder[f] |
|
583 | 583 | if ic[0] in (m1 if dirtyc1 else m2): |
|
584 | 584 | # backed-out rename on one side, but watch out for deleted files |
|
585 | 585 | bothdiverge[f] = ic |
|
586 | 586 | for of, fl in bothdiverge.items(): |
|
587 | 587 | if len(fl) == 2 and fl[0] == fl[1]: |
|
588 | 588 | copy[fl[0]] = of # not actually divergent, just matching renames |
|
589 | 589 | |
|
590 | 590 | if fullcopy and repo.ui.debugflag: |
|
591 | 591 | repo.ui.debug(" all copies found (* = to merge, ! = divergent, " |
|
592 | 592 | "% = renamed and deleted):\n") |
|
593 | 593 | for f in sorted(fullcopy): |
|
594 | 594 | note = "" |
|
595 | 595 | if f in copy: |
|
596 | 596 | note += "*" |
|
597 | 597 | if f in divergeset: |
|
598 | 598 | note += "!" |
|
599 | 599 | if f in renamedeleteset: |
|
600 | 600 | note += "%" |
|
601 | 601 | repo.ui.debug(" src: '%s' -> dst: '%s' %s\n" % (fullcopy[f], f, |
|
602 | 602 | note)) |
|
603 | 603 | del divergeset |
|
604 | 604 | |
|
605 | 605 | if not fullcopy: |
|
606 | 606 | return copy, {}, diverge, renamedelete, {} |
|
607 | 607 | |
|
608 | 608 | repo.ui.debug(" checking for directory renames\n") |
|
609 | 609 | |
|
610 | 610 | # generate a directory move map |
|
611 | 611 | d1, d2 = c1.dirs(), c2.dirs() |
|
612 | 612 | # Hack for adding '', which is not otherwise added, to d1 and d2 |
|
613 | 613 | d1.addpath('/') |
|
614 | 614 | d2.addpath('/') |
|
615 | 615 | invalid = set() |
|
616 | 616 | dirmove = {} |
|
617 | 617 | |
|
618 | 618 | # examine each file copy for a potential directory move, which is |
|
619 | 619 | # when all the files in a directory are moved to a new directory |
|
620 | 620 | for dst, src in fullcopy.iteritems(): |
|
621 | 621 | dsrc, ddst = pathutil.dirname(src), pathutil.dirname(dst) |
|
622 | 622 | if dsrc in invalid: |
|
623 | 623 | # already seen to be uninteresting |
|
624 | 624 | continue |
|
625 | 625 | elif dsrc in d1 and ddst in d1: |
|
626 | 626 | # directory wasn't entirely moved locally |
|
627 | 627 | invalid.add(dsrc) |
|
628 | 628 | elif dsrc in d2 and ddst in d2: |
|
629 | 629 | # directory wasn't entirely moved remotely |
|
630 | 630 | invalid.add(dsrc) |
|
631 | 631 | elif dsrc in dirmove and dirmove[dsrc] != ddst: |
|
632 | 632 | # files from the same directory moved to two different places |
|
633 | 633 | invalid.add(dsrc) |
|
634 | 634 | else: |
|
635 | 635 | # looks good so far |
|
636 | 636 | dirmove[dsrc] = ddst |
|
637 | 637 | |
|
638 | 638 | for i in invalid: |
|
639 | 639 | if i in dirmove: |
|
640 | 640 | del dirmove[i] |
|
641 | 641 | del d1, d2, invalid |
|
642 | 642 | |
|
643 | 643 | if not dirmove: |
|
644 | 644 | return copy, {}, diverge, renamedelete, {} |
|
645 | 645 | |
|
646 | 646 | dirmove = {k + "/": v + "/" for k, v in dirmove.iteritems()} |
|
647 | 647 | |
|
648 | 648 | for d in dirmove: |
|
649 | 649 | repo.ui.debug(" discovered dir src: '%s' -> dst: '%s'\n" % |
|
650 | 650 | (d, dirmove[d])) |
|
651 | 651 | |
|
652 | 652 | movewithdir = {} |
|
653 | 653 | # check unaccounted nonoverlapping files against directory moves |
|
654 | 654 | for f in u1r + u2r: |
|
655 | 655 | if f not in fullcopy: |
|
656 | 656 | for d in dirmove: |
|
657 | 657 | if f.startswith(d): |
|
658 | 658 | # new file added in a directory that was moved, move it |
|
659 | 659 | df = dirmove[d] + f[len(d):] |
|
660 | 660 | if df not in copy: |
|
661 | 661 | movewithdir[f] = df |
|
662 | 662 | repo.ui.debug((" pending file src: '%s' -> " |
|
663 | 663 | "dst: '%s'\n") % (f, df)) |
|
664 | 664 | break |
|
665 | 665 | |
|
666 | 666 | return copy, movewithdir, diverge, renamedelete, dirmove |
|
667 | 667 | |
|
668 | 668 | def _heuristicscopytracing(repo, c1, c2, base): |
|
669 | 669 | """ Fast copytracing using filename heuristics |
|
670 | 670 | |
|
671 | 671 | Assumes that moves or renames are of following two types: |
|
672 | 672 | |
|
673 | 673 | 1) Inside a directory only (same directory name but different filenames) |
|
674 | 674 | 2) Move from one directory to another |
|
675 | 675 | (same filenames but different directory names) |
|
676 | 676 | |
|
677 | 677 | Works only when there are no merge commits in the "source branch". |
|
678 | 678 | Source branch is commits from base up to c2 not including base. |
|
679 | 679 | |
|
680 | 680 | If merge is involved it fallbacks to _fullcopytracing(). |
|
681 | 681 | |
|
682 | 682 | Can be used by setting the following config: |
|
683 | 683 | |
|
684 | 684 | [experimental] |
|
685 | 685 | copytrace = heuristics |
|
686 | 686 | |
|
687 | 687 | In some cases the copy/move candidates found by heuristics can be very large |
|
688 | 688 | in number and that will make the algorithm slow. The number of possible |
|
689 | 689 | candidates to check can be limited by using the config |
|
690 | 690 | `experimental.copytrace.movecandidateslimit` which defaults to 100. |
|
691 | 691 | """ |
|
692 | 692 | |
|
693 | 693 | if c1.rev() is None: |
|
694 | 694 | c1 = c1.p1() |
|
695 | 695 | if c2.rev() is None: |
|
696 | 696 | c2 = c2.p1() |
|
697 | 697 | |
|
698 | 698 | copies = {} |
|
699 | 699 | |
|
700 | 700 | changedfiles = set() |
|
701 | 701 | m1 = c1.manifest() |
|
702 | 702 | if not repo.revs('%d::%d', base.rev(), c2.rev()): |
|
703 | 703 | # If base is not in c2 branch, we switch to fullcopytracing |
|
704 | 704 | repo.ui.debug("switching to full copytracing as base is not " |
|
705 | 705 | "an ancestor of c2\n") |
|
706 | 706 | return _fullcopytracing(repo, c1, c2, base) |
|
707 | 707 | |
|
708 | 708 | ctx = c2 |
|
709 | 709 | while ctx != base: |
|
710 | 710 | if len(ctx.parents()) == 2: |
|
711 | 711 | # To keep things simple let's not handle merges |
|
712 | 712 | repo.ui.debug("switching to full copytracing because of merges\n") |
|
713 | 713 | return _fullcopytracing(repo, c1, c2, base) |
|
714 | 714 | changedfiles.update(ctx.files()) |
|
715 | 715 | ctx = ctx.p1() |
|
716 | 716 | |
|
717 | 717 | cp = _forwardcopies(base, c2) |
|
718 | 718 | for dst, src in cp.iteritems(): |
|
719 | 719 | if src in m1: |
|
720 | 720 | copies[dst] = src |
|
721 | 721 | |
|
722 | 722 | # file is missing if it isn't present in the destination, but is present in |
|
723 | 723 | # the base and present in the source. |
|
724 | 724 | # Presence in the base is important to exclude added files, presence in the |
|
725 | 725 | # source is important to exclude removed files. |
|
726 | 726 | filt = lambda f: f not in m1 and f in base and f in c2 |
|
727 | 727 | missingfiles = [f for f in changedfiles if filt(f)] |
|
728 | 728 | |
|
729 | 729 | if missingfiles: |
|
730 | 730 | basenametofilename = collections.defaultdict(list) |
|
731 | 731 | dirnametofilename = collections.defaultdict(list) |
|
732 | 732 | |
|
733 | 733 | for f in m1.filesnotin(base.manifest()): |
|
734 | 734 | basename = os.path.basename(f) |
|
735 | 735 | dirname = os.path.dirname(f) |
|
736 | 736 | basenametofilename[basename].append(f) |
|
737 | 737 | dirnametofilename[dirname].append(f) |
|
738 | 738 | |
|
739 | 739 | for f in missingfiles: |
|
740 | 740 | basename = os.path.basename(f) |
|
741 | 741 | dirname = os.path.dirname(f) |
|
742 | 742 | samebasename = basenametofilename[basename] |
|
743 | 743 | samedirname = dirnametofilename[dirname] |
|
744 | 744 | movecandidates = samebasename + samedirname |
|
745 | 745 | # f is guaranteed to be present in c2, that's why |
|
746 | 746 | # c2.filectx(f) won't fail |
|
747 | 747 | f2 = c2.filectx(f) |
|
748 | 748 | # we can have a lot of candidates which can slow down the heuristics |
|
749 | 749 | # config value to limit the number of candidates moves to check |
|
750 | 750 | maxcandidates = repo.ui.configint('experimental', |
|
751 | 751 | 'copytrace.movecandidateslimit') |
|
752 | 752 | |
|
753 | 753 | if len(movecandidates) > maxcandidates: |
|
754 | 754 | repo.ui.status(_("skipping copytracing for '%s', more " |
|
755 | 755 | "candidates than the limit: %d\n") |
|
756 | 756 | % (f, len(movecandidates))) |
|
757 | 757 | continue |
|
758 | 758 | |
|
759 | 759 | for candidate in movecandidates: |
|
760 | 760 | f1 = c1.filectx(candidate) |
|
761 | 761 | if _related(f1, f2): |
|
762 | 762 | # if there are a few related copies then we'll merge |
|
763 | 763 | # changes into all of them. This matches the behaviour |
|
764 | 764 | # of upstream copytracing |
|
765 | 765 | copies[candidate] = f |
|
766 | 766 | |
|
767 | 767 | return copies, {}, {}, {}, {} |
|
768 | 768 | |
|
769 | 769 | def _related(f1, f2): |
|
770 | 770 | """return True if f1 and f2 filectx have a common ancestor |
|
771 | 771 | |
|
772 | 772 | Walk back to common ancestor to see if the two files originate |
|
773 | 773 | from the same file. Since workingfilectx's rev() is None it messes |
|
774 | 774 | up the integer comparison logic, hence the pre-step check for |
|
775 | 775 | None (f1 and f2 can only be workingfilectx's initially). |
|
776 | 776 | """ |
|
777 | 777 | |
|
778 | 778 | if f1 == f2: |
|
779 |
return |
|
|
779 | return True # a match | |
|
780 | 780 | |
|
781 | 781 | g1, g2 = f1.ancestors(), f2.ancestors() |
|
782 | 782 | try: |
|
783 | 783 | f1r, f2r = f1.linkrev(), f2.linkrev() |
|
784 | 784 | |
|
785 | 785 | if f1r is None: |
|
786 | 786 | f1 = next(g1) |
|
787 | 787 | if f2r is None: |
|
788 | 788 | f2 = next(g2) |
|
789 | 789 | |
|
790 | 790 | while True: |
|
791 | 791 | f1r, f2r = f1.linkrev(), f2.linkrev() |
|
792 | 792 | if f1r > f2r: |
|
793 | 793 | f1 = next(g1) |
|
794 | 794 | elif f2r > f1r: |
|
795 | 795 | f2 = next(g2) |
|
796 | 796 | else: # f1 and f2 point to files in the same linkrev |
|
797 | 797 | return f1 == f2 # true if they point to the same file |
|
798 | 798 | except StopIteration: |
|
799 | 799 | return False |
|
800 | 800 | |
|
801 | 801 | def _checkcopies(srcctx, dstctx, f, base, tca, remotebase, limit, data): |
|
802 | 802 | """ |
|
803 | 803 | check possible copies of f from msrc to mdst |
|
804 | 804 | |
|
805 | 805 | srcctx = starting context for f in msrc |
|
806 | 806 | dstctx = destination context for f in mdst |
|
807 | 807 | f = the filename to check (as in msrc) |
|
808 | 808 | base = the changectx used as a merge base |
|
809 | 809 | tca = topological common ancestor for graft-like scenarios |
|
810 | 810 | remotebase = True if base is outside tca::srcctx, False otherwise |
|
811 | 811 | limit = the rev number to not search beyond |
|
812 | 812 | data = dictionary of dictionary to store copy data. (see mergecopies) |
|
813 | 813 | |
|
814 | 814 | note: limit is only an optimization, and provides no guarantee that |
|
815 | 815 | irrelevant revisions will not be visited |
|
816 | 816 | there is no easy way to make this algorithm stop in a guaranteed way |
|
817 | 817 | once it "goes behind a certain revision". |
|
818 | 818 | """ |
|
819 | 819 | |
|
820 | 820 | msrc = srcctx.manifest() |
|
821 | 821 | mdst = dstctx.manifest() |
|
822 | 822 | mb = base.manifest() |
|
823 | 823 | mta = tca.manifest() |
|
824 | 824 | # Might be true if this call is about finding backward renames, |
|
825 | 825 | # This happens in the case of grafts because the DAG is then rotated. |
|
826 | 826 | # If the file exists in both the base and the source, we are not looking |
|
827 | 827 | # for a rename on the source side, but on the part of the DAG that is |
|
828 | 828 | # traversed backwards. |
|
829 | 829 | # |
|
830 | 830 | # In the case there is both backward and forward renames (before and after |
|
831 | 831 | # the base) this is more complicated as we must detect a divergence. |
|
832 | 832 | # We use 'backwards = False' in that case. |
|
833 | 833 | backwards = not remotebase and base != tca and f in mb |
|
834 | 834 | getsrcfctx = _makegetfctx(srcctx) |
|
835 | 835 | getdstfctx = _makegetfctx(dstctx) |
|
836 | 836 | |
|
837 | 837 | if msrc[f] == mb.get(f) and not remotebase: |
|
838 | 838 | # Nothing to merge |
|
839 | 839 | return |
|
840 | 840 | |
|
841 | 841 | of = None |
|
842 | 842 | seen = {f} |
|
843 | 843 | for oc in getsrcfctx(f, msrc[f]).ancestors(): |
|
844 | 844 | of = oc.path() |
|
845 | 845 | if of in seen: |
|
846 | 846 | # check limit late - grab last rename before |
|
847 | 847 | if oc.linkrev() < limit: |
|
848 | 848 | break |
|
849 | 849 | continue |
|
850 | 850 | seen.add(of) |
|
851 | 851 | |
|
852 | 852 | # remember for dir rename detection |
|
853 | 853 | if backwards: |
|
854 | 854 | data['fullcopy'][of] = f # grafting backwards through renames |
|
855 | 855 | else: |
|
856 | 856 | data['fullcopy'][f] = of |
|
857 | 857 | if of not in mdst: |
|
858 | 858 | continue # no match, keep looking |
|
859 | 859 | if mdst[of] == mb.get(of): |
|
860 | 860 | return # no merge needed, quit early |
|
861 | 861 | c2 = getdstfctx(of, mdst[of]) |
|
862 | 862 | # c2 might be a plain new file on added on destination side that is |
|
863 | 863 | # unrelated to the droids we are looking for. |
|
864 | 864 | cr = _related(oc, c2) |
|
865 | 865 | if cr and (of == f or of == c2.path()): # non-divergent |
|
866 | 866 | if backwards: |
|
867 | 867 | data['copy'][of] = f |
|
868 | 868 | elif of in mb: |
|
869 | 869 | data['copy'][f] = of |
|
870 | 870 | elif remotebase: # special case: a <- b <- a -> b "ping-pong" rename |
|
871 | 871 | data['copy'][of] = f |
|
872 | 872 | del data['fullcopy'][f] |
|
873 | 873 | data['fullcopy'][of] = f |
|
874 | 874 | else: # divergence w.r.t. graft CA on one side of topological CA |
|
875 | 875 | for sf in seen: |
|
876 | 876 | if sf in mb: |
|
877 | 877 | assert sf not in data['diverge'] |
|
878 | 878 | data['diverge'][sf] = [f, of] |
|
879 | 879 | break |
|
880 | 880 | return |
|
881 | 881 | |
|
882 | 882 | if of in mta: |
|
883 | 883 | if backwards or remotebase: |
|
884 | 884 | data['incomplete'][of] = f |
|
885 | 885 | else: |
|
886 | 886 | for sf in seen: |
|
887 | 887 | if sf in mb: |
|
888 | 888 | if tca == base: |
|
889 | 889 | data['diverge'].setdefault(sf, []).append(f) |
|
890 | 890 | else: |
|
891 | 891 | data['incompletediverge'][sf] = [of, f] |
|
892 | 892 | return |
|
893 | 893 | |
|
894 | 894 | def duplicatecopies(repo, wctx, rev, fromrev, skiprev=None): |
|
895 | 895 | """reproduce copies from fromrev to rev in the dirstate |
|
896 | 896 | |
|
897 | 897 | If skiprev is specified, it's a revision that should be used to |
|
898 | 898 | filter copy records. Any copies that occur between fromrev and |
|
899 | 899 | skiprev will not be duplicated, even if they appear in the set of |
|
900 | 900 | copies between fromrev and rev. |
|
901 | 901 | """ |
|
902 | 902 | exclude = {} |
|
903 | 903 | ctraceconfig = repo.ui.config('experimental', 'copytrace') |
|
904 | 904 | bctrace = stringutil.parsebool(ctraceconfig) |
|
905 | 905 | if (skiprev is not None and |
|
906 | 906 | (ctraceconfig == 'heuristics' or bctrace or bctrace is None)): |
|
907 | 907 | # copytrace='off' skips this line, but not the entire function because |
|
908 | 908 | # the line below is O(size of the repo) during a rebase, while the rest |
|
909 | 909 | # of the function is much faster (and is required for carrying copy |
|
910 | 910 | # metadata across the rebase anyway). |
|
911 | 911 | exclude = pathcopies(repo[fromrev], repo[skiprev]) |
|
912 | 912 | for dst, src in pathcopies(repo[fromrev], repo[rev]).iteritems(): |
|
913 | 913 | # copies.pathcopies returns backward renames, so dst might not |
|
914 | 914 | # actually be in the dirstate |
|
915 | 915 | if dst in exclude: |
|
916 | 916 | continue |
|
917 | 917 | wctx[dst].markcopied(src) |
General Comments 0
You need to be logged in to leave comments.
Login now