Show More
@@ -1,1924 +1,1932 b'' | |||
|
1 | 1 | # Copyright 2009-2010 Gregory P. Ward |
|
2 | 2 | # Copyright 2009-2010 Intelerad Medical Systems Incorporated |
|
3 | 3 | # Copyright 2010-2011 Fog Creek Software |
|
4 | 4 | # Copyright 2010-2011 Unity Technologies |
|
5 | 5 | # |
|
6 | 6 | # This software may be used and distributed according to the terms of the |
|
7 | 7 | # GNU General Public License version 2 or any later version. |
|
8 | 8 | |
|
9 | 9 | '''Overridden Mercurial commands and functions for the largefiles extension''' |
|
10 | 10 | |
|
11 | 11 | import contextlib |
|
12 | 12 | import copy |
|
13 | 13 | import os |
|
14 | 14 | |
|
15 | 15 | from mercurial.i18n import _ |
|
16 | 16 | |
|
17 | 17 | from mercurial.pycompat import open |
|
18 | 18 | |
|
19 | 19 | from mercurial.hgweb import webcommands |
|
20 | 20 | |
|
21 | 21 | from mercurial import ( |
|
22 | 22 | archival, |
|
23 | 23 | cmdutil, |
|
24 | 24 | copies as copiesmod, |
|
25 | 25 | dirstate, |
|
26 | 26 | error, |
|
27 | 27 | exchange, |
|
28 | 28 | extensions, |
|
29 | 29 | exthelper, |
|
30 | 30 | filemerge, |
|
31 | 31 | hg, |
|
32 | 32 | logcmdutil, |
|
33 | 33 | match as matchmod, |
|
34 | 34 | merge, |
|
35 | 35 | mergestate as mergestatemod, |
|
36 | 36 | pathutil, |
|
37 | 37 | pycompat, |
|
38 | 38 | scmutil, |
|
39 | 39 | smartset, |
|
40 | 40 | subrepo, |
|
41 | 41 | url as urlmod, |
|
42 | 42 | util, |
|
43 | 43 | ) |
|
44 | 44 | |
|
45 | 45 | from mercurial.upgrade_utils import ( |
|
46 | 46 | actions as upgrade_actions, |
|
47 | 47 | ) |
|
48 | 48 | |
|
49 | 49 | from . import ( |
|
50 | 50 | lfcommands, |
|
51 | 51 | lfutil, |
|
52 | 52 | storefactory, |
|
53 | 53 | ) |
|
54 | 54 | |
|
55 | 55 | ACTION_ADD = mergestatemod.ACTION_ADD |
|
56 | 56 | ACTION_DELETED_CHANGED = mergestatemod.ACTION_DELETED_CHANGED |
|
57 | 57 | ACTION_GET = mergestatemod.ACTION_GET |
|
58 | 58 | ACTION_KEEP = mergestatemod.ACTION_KEEP |
|
59 | 59 | ACTION_REMOVE = mergestatemod.ACTION_REMOVE |
|
60 | 60 | |
|
61 | 61 | eh = exthelper.exthelper() |
|
62 | 62 | |
|
63 | 63 | lfstatus = lfutil.lfstatus |
|
64 | 64 | |
|
65 | 65 | MERGE_ACTION_LARGEFILE_MARK_REMOVED = mergestatemod.MergeAction('lfmr') |
|
66 | 66 | |
|
67 | 67 | # -- Utility functions: commonly/repeatedly needed functionality --------------- |
|
68 | 68 | |
|
69 | 69 | |
|
70 | 70 | def composelargefilematcher(match, manifest): |
|
71 | 71 | """create a matcher that matches only the largefiles in the original |
|
72 | 72 | matcher""" |
|
73 | 73 | m = copy.copy(match) |
|
74 | m._was_tampered_with = True | |
|
74 | 75 | lfile = lambda f: lfutil.standin(f) in manifest |
|
75 | 76 | m._files = [lf for lf in m._files if lfile(lf)] |
|
76 | 77 | m._fileset = set(m._files) |
|
77 | 78 | m.always = lambda: False |
|
78 | 79 | origmatchfn = m.matchfn |
|
79 | 80 | m.matchfn = lambda f: lfile(f) and origmatchfn(f) |
|
80 | 81 | return m |
|
81 | 82 | |
|
82 | 83 | |
|
83 | 84 | def composenormalfilematcher(match, manifest, exclude=None): |
|
84 | 85 | excluded = set() |
|
85 | 86 | if exclude is not None: |
|
86 | 87 | excluded.update(exclude) |
|
87 | 88 | |
|
88 | 89 | m = copy.copy(match) |
|
90 | m._was_tampered_with = True | |
|
89 | 91 | notlfile = lambda f: not ( |
|
90 | 92 | lfutil.isstandin(f) or lfutil.standin(f) in manifest or f in excluded |
|
91 | 93 | ) |
|
92 | 94 | m._files = [lf for lf in m._files if notlfile(lf)] |
|
93 | 95 | m._fileset = set(m._files) |
|
94 | 96 | m.always = lambda: False |
|
95 | 97 | origmatchfn = m.matchfn |
|
96 | 98 | m.matchfn = lambda f: notlfile(f) and origmatchfn(f) |
|
97 | 99 | return m |
|
98 | 100 | |
|
99 | 101 | |
|
100 | 102 | def addlargefiles(ui, repo, isaddremove, matcher, uipathfn, **opts): |
|
101 | 103 | large = opts.get('large') |
|
102 | 104 | lfsize = lfutil.getminsize( |
|
103 | 105 | ui, lfutil.islfilesrepo(repo), opts.get('lfsize') |
|
104 | 106 | ) |
|
105 | 107 | |
|
106 | 108 | lfmatcher = None |
|
107 | 109 | if lfutil.islfilesrepo(repo): |
|
108 | 110 | lfpats = ui.configlist(lfutil.longname, b'patterns') |
|
109 | 111 | if lfpats: |
|
110 | 112 | lfmatcher = matchmod.match(repo.root, b'', list(lfpats)) |
|
111 | 113 | |
|
112 | 114 | lfnames = [] |
|
113 | 115 | m = matcher |
|
114 | 116 | |
|
115 | 117 | wctx = repo[None] |
|
116 | 118 | for f in wctx.walk(matchmod.badmatch(m, lambda x, y: None)): |
|
117 | 119 | exact = m.exact(f) |
|
118 | 120 | lfile = lfutil.standin(f) in wctx |
|
119 | 121 | nfile = f in wctx |
|
120 | 122 | exists = lfile or nfile |
|
121 | 123 | |
|
122 | 124 | # Don't warn the user when they attempt to add a normal tracked file. |
|
123 | 125 | # The normal add code will do that for us. |
|
124 | 126 | if exact and exists: |
|
125 | 127 | if lfile: |
|
126 | 128 | ui.warn(_(b'%s already a largefile\n') % uipathfn(f)) |
|
127 | 129 | continue |
|
128 | 130 | |
|
129 | 131 | if (exact or not exists) and not lfutil.isstandin(f): |
|
130 | 132 | # In case the file was removed previously, but not committed |
|
131 | 133 | # (issue3507) |
|
132 | 134 | if not repo.wvfs.exists(f): |
|
133 | 135 | continue |
|
134 | 136 | |
|
135 | 137 | abovemin = ( |
|
136 | 138 | lfsize and repo.wvfs.lstat(f).st_size >= lfsize * 1024 * 1024 |
|
137 | 139 | ) |
|
138 | 140 | if large or abovemin or (lfmatcher and lfmatcher(f)): |
|
139 | 141 | lfnames.append(f) |
|
140 | 142 | if ui.verbose or not exact: |
|
141 | 143 | ui.status(_(b'adding %s as a largefile\n') % uipathfn(f)) |
|
142 | 144 | |
|
143 | 145 | bad = [] |
|
144 | 146 | |
|
145 | 147 | # Need to lock, otherwise there could be a race condition between |
|
146 | 148 | # when standins are created and added to the repo. |
|
147 | 149 | with repo.wlock(): |
|
148 | 150 | if not opts.get('dry_run'): |
|
149 | 151 | standins = [] |
|
150 | 152 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
151 | 153 | for f in lfnames: |
|
152 | 154 | standinname = lfutil.standin(f) |
|
153 | 155 | lfutil.writestandin( |
|
154 | 156 | repo, |
|
155 | 157 | standinname, |
|
156 | 158 | hash=b'', |
|
157 | 159 | executable=lfutil.getexecutable(repo.wjoin(f)), |
|
158 | 160 | ) |
|
159 | 161 | standins.append(standinname) |
|
160 | 162 | lfdirstate.set_tracked(f) |
|
161 | 163 | lfdirstate.write(repo.currenttransaction()) |
|
162 | 164 | bad += [ |
|
163 | 165 | lfutil.splitstandin(f) |
|
164 | 166 | for f in repo[None].add(standins) |
|
165 | 167 | if f in m.files() |
|
166 | 168 | ] |
|
167 | 169 | |
|
168 | 170 | added = [f for f in lfnames if f not in bad] |
|
169 | 171 | return added, bad |
|
170 | 172 | |
|
171 | 173 | |
|
172 | 174 | def removelargefiles(ui, repo, isaddremove, matcher, uipathfn, dryrun, **opts): |
|
173 | 175 | after = opts.get('after') |
|
174 | 176 | m = composelargefilematcher(matcher, repo[None].manifest()) |
|
175 | 177 | with lfstatus(repo): |
|
176 | 178 | s = repo.status(match=m, clean=not isaddremove) |
|
177 | 179 | manifest = repo[None].manifest() |
|
178 | 180 | modified, added, deleted, clean = [ |
|
179 | 181 | [f for f in list if lfutil.standin(f) in manifest] |
|
180 | 182 | for list in (s.modified, s.added, s.deleted, s.clean) |
|
181 | 183 | ] |
|
182 | 184 | |
|
183 | 185 | def warn(files, msg): |
|
184 | 186 | for f in files: |
|
185 | 187 | ui.warn(msg % uipathfn(f)) |
|
186 | 188 | return int(len(files) > 0) |
|
187 | 189 | |
|
188 | 190 | if after: |
|
189 | 191 | remove = deleted |
|
190 | 192 | result = warn( |
|
191 | 193 | modified + added + clean, _(b'not removing %s: file still exists\n') |
|
192 | 194 | ) |
|
193 | 195 | else: |
|
194 | 196 | remove = deleted + clean |
|
195 | 197 | result = warn( |
|
196 | 198 | modified, |
|
197 | 199 | _( |
|
198 | 200 | b'not removing %s: file is modified (use -f' |
|
199 | 201 | b' to force removal)\n' |
|
200 | 202 | ), |
|
201 | 203 | ) |
|
202 | 204 | result = ( |
|
203 | 205 | warn( |
|
204 | 206 | added, |
|
205 | 207 | _( |
|
206 | 208 | b'not removing %s: file has been marked for add' |
|
207 | 209 | b' (use forget to undo)\n' |
|
208 | 210 | ), |
|
209 | 211 | ) |
|
210 | 212 | or result |
|
211 | 213 | ) |
|
212 | 214 | |
|
213 | 215 | # Need to lock because standin files are deleted then removed from the |
|
214 | 216 | # repository and we could race in-between. |
|
215 | 217 | with repo.wlock(): |
|
216 | 218 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
217 | 219 | for f in sorted(remove): |
|
218 | 220 | if ui.verbose or not m.exact(f): |
|
219 | 221 | ui.status(_(b'removing %s\n') % uipathfn(f)) |
|
220 | 222 | |
|
221 | 223 | if not dryrun: |
|
222 | 224 | if not after: |
|
223 | 225 | repo.wvfs.unlinkpath(f, ignoremissing=True) |
|
224 | 226 | |
|
225 | 227 | if dryrun: |
|
226 | 228 | return result |
|
227 | 229 | |
|
228 | 230 | remove = [lfutil.standin(f) for f in remove] |
|
229 | 231 | # If this is being called by addremove, let the original addremove |
|
230 | 232 | # function handle this. |
|
231 | 233 | if not isaddremove: |
|
232 | 234 | for f in remove: |
|
233 | 235 | repo.wvfs.unlinkpath(f, ignoremissing=True) |
|
234 | 236 | repo[None].forget(remove) |
|
235 | 237 | |
|
236 | 238 | for f in remove: |
|
237 | 239 | lfdirstate.set_untracked(lfutil.splitstandin(f)) |
|
238 | 240 | |
|
239 | 241 | lfdirstate.write(repo.currenttransaction()) |
|
240 | 242 | |
|
241 | 243 | return result |
|
242 | 244 | |
|
243 | 245 | |
|
244 | 246 | # For overriding mercurial.hgweb.webcommands so that largefiles will |
|
245 | 247 | # appear at their right place in the manifests. |
|
246 | 248 | @eh.wrapfunction(webcommands, 'decodepath') |
|
247 | 249 | def decodepath(orig, path): |
|
248 | 250 | return lfutil.splitstandin(path) or path |
|
249 | 251 | |
|
250 | 252 | |
|
251 | 253 | # -- Wrappers: modify existing commands -------------------------------- |
|
252 | 254 | |
|
253 | 255 | |
|
254 | 256 | @eh.wrapcommand( |
|
255 | 257 | b'add', |
|
256 | 258 | opts=[ |
|
257 | 259 | (b'', b'large', None, _(b'add as largefile')), |
|
258 | 260 | (b'', b'normal', None, _(b'add as normal file')), |
|
259 | 261 | ( |
|
260 | 262 | b'', |
|
261 | 263 | b'lfsize', |
|
262 | 264 | b'', |
|
263 | 265 | _( |
|
264 | 266 | b'add all files above this size (in megabytes) ' |
|
265 | 267 | b'as largefiles (default: 10)' |
|
266 | 268 | ), |
|
267 | 269 | ), |
|
268 | 270 | ], |
|
269 | 271 | ) |
|
270 | 272 | def overrideadd(orig, ui, repo, *pats, **opts): |
|
271 | 273 | if opts.get('normal') and opts.get('large'): |
|
272 | 274 | raise error.Abort(_(b'--normal cannot be used with --large')) |
|
273 | 275 | return orig(ui, repo, *pats, **opts) |
|
274 | 276 | |
|
275 | 277 | |
|
276 | 278 | @eh.wrapfunction(cmdutil, 'add') |
|
277 | 279 | def cmdutiladd(orig, ui, repo, matcher, prefix, uipathfn, explicitonly, **opts): |
|
278 | 280 | # The --normal flag short circuits this override |
|
279 | 281 | if opts.get('normal'): |
|
280 | 282 | return orig(ui, repo, matcher, prefix, uipathfn, explicitonly, **opts) |
|
281 | 283 | |
|
282 | 284 | ladded, lbad = addlargefiles(ui, repo, False, matcher, uipathfn, **opts) |
|
283 | 285 | normalmatcher = composenormalfilematcher( |
|
284 | 286 | matcher, repo[None].manifest(), ladded |
|
285 | 287 | ) |
|
286 | 288 | bad = orig(ui, repo, normalmatcher, prefix, uipathfn, explicitonly, **opts) |
|
287 | 289 | |
|
288 | 290 | bad.extend(f for f in lbad) |
|
289 | 291 | return bad |
|
290 | 292 | |
|
291 | 293 | |
|
292 | 294 | @eh.wrapfunction(cmdutil, 'remove') |
|
293 | 295 | def cmdutilremove( |
|
294 | 296 | orig, ui, repo, matcher, prefix, uipathfn, after, force, subrepos, dryrun |
|
295 | 297 | ): |
|
296 | 298 | normalmatcher = composenormalfilematcher(matcher, repo[None].manifest()) |
|
297 | 299 | result = orig( |
|
298 | 300 | ui, |
|
299 | 301 | repo, |
|
300 | 302 | normalmatcher, |
|
301 | 303 | prefix, |
|
302 | 304 | uipathfn, |
|
303 | 305 | after, |
|
304 | 306 | force, |
|
305 | 307 | subrepos, |
|
306 | 308 | dryrun, |
|
307 | 309 | ) |
|
308 | 310 | return ( |
|
309 | 311 | removelargefiles( |
|
310 | 312 | ui, repo, False, matcher, uipathfn, dryrun, after=after, force=force |
|
311 | 313 | ) |
|
312 | 314 | or result |
|
313 | 315 | ) |
|
314 | 316 | |
|
315 | 317 | |
|
316 | 318 | @eh.wrapfunction(dirstate.dirstate, '_changing') |
|
317 | 319 | @contextlib.contextmanager |
|
318 | 320 | def _changing(orig, self, repo, change_type): |
|
319 | 321 | pre = sub_dirstate = getattr(self, '_sub_dirstate', None) |
|
320 | 322 | try: |
|
321 | 323 | lfd = getattr(self, '_large_file_dirstate', False) |
|
322 | 324 | if sub_dirstate is None and not lfd: |
|
323 | 325 | sub_dirstate = lfutil.openlfdirstate(repo.ui, repo) |
|
324 | 326 | self._sub_dirstate = sub_dirstate |
|
325 | 327 | if not lfd: |
|
326 | 328 | assert self._sub_dirstate is not None |
|
327 | 329 | with orig(self, repo, change_type): |
|
328 | 330 | if sub_dirstate is None: |
|
329 | 331 | yield |
|
330 | 332 | else: |
|
331 | 333 | with sub_dirstate._changing(repo, change_type): |
|
332 | 334 | yield |
|
333 | 335 | finally: |
|
334 | 336 | self._sub_dirstate = pre |
|
335 | 337 | |
|
336 | 338 | |
|
337 | 339 | @eh.wrapfunction(dirstate.dirstate, 'running_status') |
|
338 | 340 | @contextlib.contextmanager |
|
339 | 341 | def running_status(orig, self, repo): |
|
340 | 342 | pre = sub_dirstate = getattr(self, '_sub_dirstate', None) |
|
341 | 343 | try: |
|
342 | 344 | lfd = getattr(self, '_large_file_dirstate', False) |
|
343 | 345 | if sub_dirstate is None and not lfd: |
|
344 | 346 | sub_dirstate = lfutil.openlfdirstate(repo.ui, repo) |
|
345 | 347 | self._sub_dirstate = sub_dirstate |
|
346 | 348 | if not lfd: |
|
347 | 349 | assert self._sub_dirstate is not None |
|
348 | 350 | with orig(self, repo): |
|
349 | 351 | if sub_dirstate is None: |
|
350 | 352 | yield |
|
351 | 353 | else: |
|
352 | 354 | with sub_dirstate.running_status(repo): |
|
353 | 355 | yield |
|
354 | 356 | finally: |
|
355 | 357 | self._sub_dirstate = pre |
|
356 | 358 | |
|
357 | 359 | |
|
358 | 360 | @eh.wrapfunction(subrepo.hgsubrepo, 'status') |
|
359 | 361 | def overridestatusfn(orig, repo, rev2, **opts): |
|
360 | 362 | with lfstatus(repo._repo): |
|
361 | 363 | return orig(repo, rev2, **opts) |
|
362 | 364 | |
|
363 | 365 | |
|
364 | 366 | @eh.wrapcommand(b'status') |
|
365 | 367 | def overridestatus(orig, ui, repo, *pats, **opts): |
|
366 | 368 | with lfstatus(repo): |
|
367 | 369 | return orig(ui, repo, *pats, **opts) |
|
368 | 370 | |
|
369 | 371 | |
|
370 | 372 | @eh.wrapfunction(subrepo.hgsubrepo, 'dirty') |
|
371 | 373 | def overridedirty(orig, repo, ignoreupdate=False, missing=False): |
|
372 | 374 | with lfstatus(repo._repo): |
|
373 | 375 | return orig(repo, ignoreupdate=ignoreupdate, missing=missing) |
|
374 | 376 | |
|
375 | 377 | |
|
376 | 378 | @eh.wrapcommand(b'log') |
|
377 | 379 | def overridelog(orig, ui, repo, *pats, **opts): |
|
378 | 380 | def overridematchandpats( |
|
379 | 381 | orig, |
|
380 | 382 | ctx, |
|
381 | 383 | pats=(), |
|
382 | 384 | opts=None, |
|
383 | 385 | globbed=False, |
|
384 | 386 | default=b'relpath', |
|
385 | 387 | badfn=None, |
|
386 | 388 | ): |
|
387 | 389 | """Matcher that merges root directory with .hglf, suitable for log. |
|
388 | 390 | It is still possible to match .hglf directly. |
|
389 | 391 | For any listed files run log on the standin too. |
|
390 | 392 | matchfn tries both the given filename and with .hglf stripped. |
|
391 | 393 | """ |
|
392 | 394 | if opts is None: |
|
393 | 395 | opts = {} |
|
394 | 396 | matchandpats = orig(ctx, pats, opts, globbed, default, badfn=badfn) |
|
395 | 397 | m, p = copy.copy(matchandpats) |
|
396 | 398 | |
|
397 | 399 | if m.always(): |
|
398 | 400 | # We want to match everything anyway, so there's no benefit trying |
|
399 | 401 | # to add standins. |
|
400 | 402 | return matchandpats |
|
401 | 403 | |
|
402 | 404 | pats = set(p) |
|
403 | 405 | |
|
404 | 406 | def fixpats(pat, tostandin=lfutil.standin): |
|
405 | 407 | if pat.startswith(b'set:'): |
|
406 | 408 | return pat |
|
407 | 409 | |
|
408 | 410 | kindpat = matchmod._patsplit(pat, None) |
|
409 | 411 | |
|
410 | 412 | if kindpat[0] is not None: |
|
411 | 413 | return kindpat[0] + b':' + tostandin(kindpat[1]) |
|
412 | 414 | return tostandin(kindpat[1]) |
|
413 | 415 | |
|
414 | 416 | cwd = repo.getcwd() |
|
415 | 417 | if cwd: |
|
416 | 418 | hglf = lfutil.shortname |
|
417 | 419 | back = util.pconvert(repo.pathto(hglf)[: -len(hglf)]) |
|
418 | 420 | |
|
419 | 421 | def tostandin(f): |
|
420 | 422 | # The file may already be a standin, so truncate the back |
|
421 | 423 | # prefix and test before mangling it. This avoids turning |
|
422 | 424 | # 'glob:../.hglf/foo*' into 'glob:../.hglf/../.hglf/foo*'. |
|
423 | 425 | if f.startswith(back) and lfutil.splitstandin(f[len(back) :]): |
|
424 | 426 | return f |
|
425 | 427 | |
|
426 | 428 | # An absolute path is from outside the repo, so truncate the |
|
427 | 429 | # path to the root before building the standin. Otherwise cwd |
|
428 | 430 | # is somewhere in the repo, relative to root, and needs to be |
|
429 | 431 | # prepended before building the standin. |
|
430 | 432 | if os.path.isabs(cwd): |
|
431 | 433 | f = f[len(back) :] |
|
432 | 434 | else: |
|
433 | 435 | f = cwd + b'/' + f |
|
434 | 436 | return back + lfutil.standin(f) |
|
435 | 437 | |
|
436 | 438 | else: |
|
437 | 439 | |
|
438 | 440 | def tostandin(f): |
|
439 | 441 | if lfutil.isstandin(f): |
|
440 | 442 | return f |
|
441 | 443 | return lfutil.standin(f) |
|
442 | 444 | |
|
443 | 445 | pats.update(fixpats(f, tostandin) for f in p) |
|
444 | 446 | |
|
447 | m._was_tampered_with = True | |
|
448 | ||
|
445 | 449 | for i in range(0, len(m._files)): |
|
446 | 450 | # Don't add '.hglf' to m.files, since that is already covered by '.' |
|
447 | 451 | if m._files[i] == b'.': |
|
448 | 452 | continue |
|
449 | 453 | standin = lfutil.standin(m._files[i]) |
|
450 | 454 | # If the "standin" is a directory, append instead of replace to |
|
451 | 455 | # support naming a directory on the command line with only |
|
452 | 456 | # largefiles. The original directory is kept to support normal |
|
453 | 457 | # files. |
|
454 | 458 | if standin in ctx: |
|
455 | 459 | m._files[i] = standin |
|
456 | 460 | elif m._files[i] not in ctx and repo.wvfs.isdir(standin): |
|
457 | 461 | m._files.append(standin) |
|
458 | 462 | |
|
459 | 463 | m._fileset = set(m._files) |
|
460 | 464 | m.always = lambda: False |
|
461 | 465 | origmatchfn = m.matchfn |
|
462 | 466 | |
|
463 | 467 | def lfmatchfn(f): |
|
464 | 468 | lf = lfutil.splitstandin(f) |
|
465 | 469 | if lf is not None and origmatchfn(lf): |
|
466 | 470 | return True |
|
467 | 471 | r = origmatchfn(f) |
|
468 | 472 | return r |
|
469 | 473 | |
|
470 | 474 | m.matchfn = lfmatchfn |
|
471 | 475 | |
|
472 | 476 | ui.debug(b'updated patterns: %s\n' % b', '.join(sorted(pats))) |
|
473 | 477 | return m, pats |
|
474 | 478 | |
|
475 | 479 | # For hg log --patch, the match object is used in two different senses: |
|
476 | 480 | # (1) to determine what revisions should be printed out, and |
|
477 | 481 | # (2) to determine what files to print out diffs for. |
|
478 | 482 | # The magic matchandpats override should be used for case (1) but not for |
|
479 | 483 | # case (2). |
|
480 | 484 | oldmatchandpats = scmutil.matchandpats |
|
481 | 485 | |
|
482 | 486 | def overridemakefilematcher(orig, repo, pats, opts, badfn=None): |
|
483 | 487 | wctx = repo[None] |
|
484 | 488 | match, pats = oldmatchandpats(wctx, pats, opts, badfn=badfn) |
|
485 | 489 | return lambda ctx: match |
|
486 | 490 | |
|
487 | 491 | wrappedmatchandpats = extensions.wrappedfunction( |
|
488 | 492 | scmutil, 'matchandpats', overridematchandpats |
|
489 | 493 | ) |
|
490 | 494 | wrappedmakefilematcher = extensions.wrappedfunction( |
|
491 | 495 | logcmdutil, '_makenofollowfilematcher', overridemakefilematcher |
|
492 | 496 | ) |
|
493 | 497 | with wrappedmatchandpats, wrappedmakefilematcher: |
|
494 | 498 | return orig(ui, repo, *pats, **opts) |
|
495 | 499 | |
|
496 | 500 | |
|
497 | 501 | @eh.wrapcommand( |
|
498 | 502 | b'verify', |
|
499 | 503 | opts=[ |
|
500 | 504 | ( |
|
501 | 505 | b'', |
|
502 | 506 | b'large', |
|
503 | 507 | None, |
|
504 | 508 | _(b'verify that all largefiles in current revision exists'), |
|
505 | 509 | ), |
|
506 | 510 | ( |
|
507 | 511 | b'', |
|
508 | 512 | b'lfa', |
|
509 | 513 | None, |
|
510 | 514 | _(b'verify largefiles in all revisions, not just current'), |
|
511 | 515 | ), |
|
512 | 516 | ( |
|
513 | 517 | b'', |
|
514 | 518 | b'lfc', |
|
515 | 519 | None, |
|
516 | 520 | _(b'verify local largefile contents, not just existence'), |
|
517 | 521 | ), |
|
518 | 522 | ], |
|
519 | 523 | ) |
|
520 | 524 | def overrideverify(orig, ui, repo, *pats, **opts): |
|
521 | 525 | large = opts.pop('large', False) |
|
522 | 526 | all = opts.pop('lfa', False) |
|
523 | 527 | contents = opts.pop('lfc', False) |
|
524 | 528 | |
|
525 | 529 | result = orig(ui, repo, *pats, **opts) |
|
526 | 530 | if large or all or contents: |
|
527 | 531 | result = result or lfcommands.verifylfiles(ui, repo, all, contents) |
|
528 | 532 | return result |
|
529 | 533 | |
|
530 | 534 | |
|
531 | 535 | @eh.wrapcommand( |
|
532 | 536 | b'debugstate', |
|
533 | 537 | opts=[(b'', b'large', None, _(b'display largefiles dirstate'))], |
|
534 | 538 | ) |
|
535 | 539 | def overridedebugstate(orig, ui, repo, *pats, **opts): |
|
536 | 540 | large = opts.pop('large', False) |
|
537 | 541 | if large: |
|
538 | 542 | |
|
539 | 543 | class fakerepo: |
|
540 | 544 | dirstate = lfutil.openlfdirstate(ui, repo) |
|
541 | 545 | |
|
542 | 546 | orig(ui, fakerepo, *pats, **opts) |
|
543 | 547 | else: |
|
544 | 548 | orig(ui, repo, *pats, **opts) |
|
545 | 549 | |
|
546 | 550 | |
|
547 | 551 | # Before starting the manifest merge, merge.updates will call |
|
548 | 552 | # _checkunknownfile to check if there are any files in the merged-in |
|
549 | 553 | # changeset that collide with unknown files in the working copy. |
|
550 | 554 | # |
|
551 | 555 | # The largefiles are seen as unknown, so this prevents us from merging |
|
552 | 556 | # in a file 'foo' if we already have a largefile with the same name. |
|
553 | 557 | # |
|
554 | 558 | # The overridden function filters the unknown files by removing any |
|
555 | 559 | # largefiles. This makes the merge proceed and we can then handle this |
|
556 | 560 | # case further in the overridden calculateupdates function below. |
|
557 | 561 | @eh.wrapfunction(merge, '_checkunknownfile') |
|
558 | 562 | def overridecheckunknownfile( |
|
559 | 563 | origfn, dirstate, wvfs, dircache, wctx, mctx, f, f2=None |
|
560 | 564 | ): |
|
561 | 565 | if lfutil.standin(dirstate.normalize(f)) in wctx: |
|
562 | 566 | return False |
|
563 | 567 | return origfn(dirstate, wvfs, dircache, wctx, mctx, f, f2) |
|
564 | 568 | |
|
565 | 569 | |
|
566 | 570 | # The manifest merge handles conflicts on the manifest level. We want |
|
567 | 571 | # to handle changes in largefile-ness of files at this level too. |
|
568 | 572 | # |
|
569 | 573 | # The strategy is to run the original calculateupdates and then process |
|
570 | 574 | # the action list it outputs. There are two cases we need to deal with: |
|
571 | 575 | # |
|
572 | 576 | # 1. Normal file in p1, largefile in p2. Here the largefile is |
|
573 | 577 | # detected via its standin file, which will enter the working copy |
|
574 | 578 | # with a "get" action. It is not "merge" since the standin is all |
|
575 | 579 | # Mercurial is concerned with at this level -- the link to the |
|
576 | 580 | # existing normal file is not relevant here. |
|
577 | 581 | # |
|
578 | 582 | # 2. Largefile in p1, normal file in p2. Here we get a "merge" action |
|
579 | 583 | # since the largefile will be present in the working copy and |
|
580 | 584 | # different from the normal file in p2. Mercurial therefore |
|
581 | 585 | # triggers a merge action. |
|
582 | 586 | # |
|
583 | 587 | # In both cases, we prompt the user and emit new actions to either |
|
584 | 588 | # remove the standin (if the normal file was kept) or to remove the |
|
585 | 589 | # normal file and get the standin (if the largefile was kept). The |
|
586 | 590 | # default prompt answer is to use the largefile version since it was |
|
587 | 591 | # presumably changed on purpose. |
|
588 | 592 | # |
|
589 | 593 | # Finally, the merge.applyupdates function will then take care of |
|
590 | 594 | # writing the files into the working copy and lfcommands.updatelfiles |
|
591 | 595 | # will update the largefiles. |
|
592 | 596 | @eh.wrapfunction(merge, 'calculateupdates') |
|
593 | 597 | def overridecalculateupdates( |
|
594 | 598 | origfn, repo, p1, p2, pas, branchmerge, force, acceptremote, *args, **kwargs |
|
595 | 599 | ): |
|
596 | 600 | overwrite = force and not branchmerge |
|
597 | 601 | mresult = origfn( |
|
598 | 602 | repo, p1, p2, pas, branchmerge, force, acceptremote, *args, **kwargs |
|
599 | 603 | ) |
|
600 | 604 | |
|
601 | 605 | if overwrite: |
|
602 | 606 | return mresult |
|
603 | 607 | |
|
604 | 608 | # Convert to dictionary with filename as key and action as value. |
|
605 | 609 | lfiles = set() |
|
606 | 610 | for f in mresult.files(): |
|
607 | 611 | splitstandin = lfutil.splitstandin(f) |
|
608 | 612 | if splitstandin is not None and splitstandin in p1: |
|
609 | 613 | lfiles.add(splitstandin) |
|
610 | 614 | elif lfutil.standin(f) in p1: |
|
611 | 615 | lfiles.add(f) |
|
612 | 616 | |
|
613 | 617 | for lfile in sorted(lfiles): |
|
614 | 618 | standin = lfutil.standin(lfile) |
|
615 | 619 | (lm, largs, lmsg) = mresult.getfile(lfile, (None, None, None)) |
|
616 | 620 | (sm, sargs, smsg) = mresult.getfile(standin, (None, None, None)) |
|
617 | 621 | |
|
618 | 622 | if sm in (ACTION_GET, ACTION_DELETED_CHANGED) and lm != ACTION_REMOVE: |
|
619 | 623 | if sm == ACTION_DELETED_CHANGED: |
|
620 | 624 | f1, f2, fa, move, anc = sargs |
|
621 | 625 | sargs = (p2[f2].flags(), False) |
|
622 | 626 | # Case 1: normal file in the working copy, largefile in |
|
623 | 627 | # the second parent |
|
624 | 628 | usermsg = ( |
|
625 | 629 | _( |
|
626 | 630 | b'remote turned local normal file %s into a largefile\n' |
|
627 | 631 | b'use (l)argefile or keep (n)ormal file?' |
|
628 | 632 | b'$$ &Largefile $$ &Normal file' |
|
629 | 633 | ) |
|
630 | 634 | % lfile |
|
631 | 635 | ) |
|
632 | 636 | if repo.ui.promptchoice(usermsg, 0) == 0: # pick remote largefile |
|
633 | 637 | mresult.addfile( |
|
634 | 638 | lfile, ACTION_REMOVE, None, b'replaced by standin' |
|
635 | 639 | ) |
|
636 | 640 | mresult.addfile(standin, ACTION_GET, sargs, b'replaces standin') |
|
637 | 641 | else: # keep local normal file |
|
638 | 642 | mresult.addfile(lfile, ACTION_KEEP, None, b'replaces standin') |
|
639 | 643 | if branchmerge: |
|
640 | 644 | mresult.addfile( |
|
641 | 645 | standin, |
|
642 | 646 | ACTION_KEEP, |
|
643 | 647 | None, |
|
644 | 648 | b'replaced by non-standin', |
|
645 | 649 | ) |
|
646 | 650 | else: |
|
647 | 651 | mresult.addfile( |
|
648 | 652 | standin, |
|
649 | 653 | ACTION_REMOVE, |
|
650 | 654 | None, |
|
651 | 655 | b'replaced by non-standin', |
|
652 | 656 | ) |
|
653 | 657 | if lm in (ACTION_GET, ACTION_DELETED_CHANGED) and sm != ACTION_REMOVE: |
|
654 | 658 | if lm == ACTION_DELETED_CHANGED: |
|
655 | 659 | f1, f2, fa, move, anc = largs |
|
656 | 660 | largs = (p2[f2].flags(), False) |
|
657 | 661 | # Case 2: largefile in the working copy, normal file in |
|
658 | 662 | # the second parent |
|
659 | 663 | usermsg = ( |
|
660 | 664 | _( |
|
661 | 665 | b'remote turned local largefile %s into a normal file\n' |
|
662 | 666 | b'keep (l)argefile or use (n)ormal file?' |
|
663 | 667 | b'$$ &Largefile $$ &Normal file' |
|
664 | 668 | ) |
|
665 | 669 | % lfile |
|
666 | 670 | ) |
|
667 | 671 | if repo.ui.promptchoice(usermsg, 0) == 0: # keep local largefile |
|
668 | 672 | if branchmerge: |
|
669 | 673 | # largefile can be restored from standin safely |
|
670 | 674 | mresult.addfile( |
|
671 | 675 | lfile, |
|
672 | 676 | ACTION_KEEP, |
|
673 | 677 | None, |
|
674 | 678 | b'replaced by standin', |
|
675 | 679 | ) |
|
676 | 680 | mresult.addfile( |
|
677 | 681 | standin, ACTION_KEEP, None, b'replaces standin' |
|
678 | 682 | ) |
|
679 | 683 | else: |
|
680 | 684 | # "lfile" should be marked as "removed" without |
|
681 | 685 | # removal of itself |
|
682 | 686 | mresult.addfile( |
|
683 | 687 | lfile, |
|
684 | 688 | MERGE_ACTION_LARGEFILE_MARK_REMOVED, |
|
685 | 689 | None, |
|
686 | 690 | b'forget non-standin largefile', |
|
687 | 691 | ) |
|
688 | 692 | |
|
689 | 693 | # linear-merge should treat this largefile as 're-added' |
|
690 | 694 | mresult.addfile(standin, ACTION_ADD, None, b'keep standin') |
|
691 | 695 | else: # pick remote normal file |
|
692 | 696 | mresult.addfile(lfile, ACTION_GET, largs, b'replaces standin') |
|
693 | 697 | mresult.addfile( |
|
694 | 698 | standin, |
|
695 | 699 | ACTION_REMOVE, |
|
696 | 700 | None, |
|
697 | 701 | b'replaced by non-standin', |
|
698 | 702 | ) |
|
699 | 703 | |
|
700 | 704 | return mresult |
|
701 | 705 | |
|
702 | 706 | |
|
703 | 707 | @eh.wrapfunction(mergestatemod, 'recordupdates') |
|
704 | 708 | def mergerecordupdates(orig, repo, actions, branchmerge, getfiledata): |
|
705 | 709 | if MERGE_ACTION_LARGEFILE_MARK_REMOVED in actions: |
|
706 | 710 | lfdirstate = lfutil.openlfdirstate(repo.ui, repo) |
|
707 | 711 | for lfile, args, msg in actions[MERGE_ACTION_LARGEFILE_MARK_REMOVED]: |
|
708 | 712 | # this should be executed before 'orig', to execute 'remove' |
|
709 | 713 | # before all other actions |
|
710 | 714 | repo.dirstate.update_file(lfile, p1_tracked=True, wc_tracked=False) |
|
711 | 715 | # make sure lfile doesn't get synclfdirstate'd as normal |
|
712 | 716 | lfdirstate.update_file(lfile, p1_tracked=False, wc_tracked=True) |
|
713 | 717 | |
|
714 | 718 | return orig(repo, actions, branchmerge, getfiledata) |
|
715 | 719 | |
|
716 | 720 | |
|
717 | 721 | # Override filemerge to prompt the user about how they wish to merge |
|
718 | 722 | # largefiles. This will handle identical edits without prompting the user. |
|
719 | 723 | @eh.wrapfunction(filemerge, 'filemerge') |
|
720 | 724 | def overridefilemerge( |
|
721 | 725 | origfn, repo, wctx, mynode, orig, fcd, fco, fca, labels=None |
|
722 | 726 | ): |
|
723 | 727 | if not lfutil.isstandin(orig) or fcd.isabsent() or fco.isabsent(): |
|
724 | 728 | return origfn(repo, wctx, mynode, orig, fcd, fco, fca, labels=labels) |
|
725 | 729 | |
|
726 | 730 | ahash = lfutil.readasstandin(fca).lower() |
|
727 | 731 | dhash = lfutil.readasstandin(fcd).lower() |
|
728 | 732 | ohash = lfutil.readasstandin(fco).lower() |
|
729 | 733 | if ( |
|
730 | 734 | ohash != ahash |
|
731 | 735 | and ohash != dhash |
|
732 | 736 | and ( |
|
733 | 737 | dhash == ahash |
|
734 | 738 | or repo.ui.promptchoice( |
|
735 | 739 | _( |
|
736 | 740 | b'largefile %s has a merge conflict\nancestor was %s\n' |
|
737 | 741 | b'you can keep (l)ocal %s or take (o)ther %s.\n' |
|
738 | 742 | b'what do you want to do?' |
|
739 | 743 | b'$$ &Local $$ &Other' |
|
740 | 744 | ) |
|
741 | 745 | % (lfutil.splitstandin(orig), ahash, dhash, ohash), |
|
742 | 746 | 0, |
|
743 | 747 | ) |
|
744 | 748 | == 1 |
|
745 | 749 | ) |
|
746 | 750 | ): |
|
747 | 751 | repo.wwrite(fcd.path(), fco.data(), fco.flags()) |
|
748 | 752 | return 0, False |
|
749 | 753 | |
|
750 | 754 | |
|
751 | 755 | @eh.wrapfunction(copiesmod, 'pathcopies') |
|
752 | 756 | def copiespathcopies(orig, ctx1, ctx2, match=None): |
|
753 | 757 | copies = orig(ctx1, ctx2, match=match) |
|
754 | 758 | updated = {} |
|
755 | 759 | |
|
756 | 760 | for k, v in copies.items(): |
|
757 | 761 | updated[lfutil.splitstandin(k) or k] = lfutil.splitstandin(v) or v |
|
758 | 762 | |
|
759 | 763 | return updated |
|
760 | 764 | |
|
761 | 765 | |
|
762 | 766 | # Copy first changes the matchers to match standins instead of |
|
763 | 767 | # largefiles. Then it overrides util.copyfile in that function it |
|
764 | 768 | # checks if the destination largefile already exists. It also keeps a |
|
765 | 769 | # list of copied files so that the largefiles can be copied and the |
|
766 | 770 | # dirstate updated. |
|
767 | 771 | @eh.wrapfunction(cmdutil, 'copy') |
|
768 | 772 | def overridecopy(orig, ui, repo, pats, opts, rename=False): |
|
769 | 773 | # doesn't remove largefile on rename |
|
770 | 774 | if len(pats) < 2: |
|
771 | 775 | # this isn't legal, let the original function deal with it |
|
772 | 776 | return orig(ui, repo, pats, opts, rename) |
|
773 | 777 | |
|
774 | 778 | # This could copy both lfiles and normal files in one command, |
|
775 | 779 | # but we don't want to do that. First replace their matcher to |
|
776 | 780 | # only match normal files and run it, then replace it to just |
|
777 | 781 | # match largefiles and run it again. |
|
778 | 782 | nonormalfiles = False |
|
779 | 783 | nolfiles = False |
|
780 | 784 | manifest = repo[None].manifest() |
|
781 | 785 | |
|
782 | 786 | def normalfilesmatchfn( |
|
783 | 787 | orig, |
|
784 | 788 | ctx, |
|
785 | 789 | pats=(), |
|
786 | 790 | opts=None, |
|
787 | 791 | globbed=False, |
|
788 | 792 | default=b'relpath', |
|
789 | 793 | badfn=None, |
|
790 | 794 | ): |
|
791 | 795 | if opts is None: |
|
792 | 796 | opts = {} |
|
793 | 797 | match = orig(ctx, pats, opts, globbed, default, badfn=badfn) |
|
794 | 798 | return composenormalfilematcher(match, manifest) |
|
795 | 799 | |
|
796 | 800 | with extensions.wrappedfunction(scmutil, 'match', normalfilesmatchfn): |
|
797 | 801 | try: |
|
798 | 802 | result = orig(ui, repo, pats, opts, rename) |
|
799 | 803 | except error.Abort as e: |
|
800 | 804 | if e.message != _(b'no files to copy'): |
|
801 | 805 | raise e |
|
802 | 806 | else: |
|
803 | 807 | nonormalfiles = True |
|
804 | 808 | result = 0 |
|
805 | 809 | |
|
806 | 810 | # The first rename can cause our current working directory to be removed. |
|
807 | 811 | # In that case there is nothing left to copy/rename so just quit. |
|
808 | 812 | try: |
|
809 | 813 | repo.getcwd() |
|
810 | 814 | except OSError: |
|
811 | 815 | return result |
|
812 | 816 | |
|
813 | 817 | def makestandin(relpath): |
|
814 | 818 | path = pathutil.canonpath(repo.root, repo.getcwd(), relpath) |
|
815 | 819 | return repo.wvfs.join(lfutil.standin(path)) |
|
816 | 820 | |
|
817 | 821 | fullpats = scmutil.expandpats(pats) |
|
818 | 822 | dest = fullpats[-1] |
|
819 | 823 | |
|
820 | 824 | if os.path.isdir(dest): |
|
821 | 825 | if not os.path.isdir(makestandin(dest)): |
|
822 | 826 | os.makedirs(makestandin(dest)) |
|
823 | 827 | |
|
824 | 828 | try: |
|
825 | 829 | # When we call orig below it creates the standins but we don't add |
|
826 | 830 | # them to the dir state until later so lock during that time. |
|
827 | 831 | wlock = repo.wlock() |
|
828 | 832 | |
|
829 | 833 | manifest = repo[None].manifest() |
|
830 | 834 | |
|
831 | 835 | def overridematch( |
|
832 | 836 | orig, |
|
833 | 837 | ctx, |
|
834 | 838 | pats=(), |
|
835 | 839 | opts=None, |
|
836 | 840 | globbed=False, |
|
837 | 841 | default=b'relpath', |
|
838 | 842 | badfn=None, |
|
839 | 843 | ): |
|
840 | 844 | if opts is None: |
|
841 | 845 | opts = {} |
|
842 | 846 | newpats = [] |
|
843 | 847 | # The patterns were previously mangled to add the standin |
|
844 | 848 | # directory; we need to remove that now |
|
845 | 849 | for pat in pats: |
|
846 | 850 | if matchmod.patkind(pat) is None and lfutil.shortname in pat: |
|
847 | 851 | newpats.append(pat.replace(lfutil.shortname, b'')) |
|
848 | 852 | else: |
|
849 | 853 | newpats.append(pat) |
|
850 | 854 | match = orig(ctx, newpats, opts, globbed, default, badfn=badfn) |
|
851 | 855 | m = copy.copy(match) |
|
856 | m._was_tampered_with = True | |
|
852 | 857 | lfile = lambda f: lfutil.standin(f) in manifest |
|
853 | 858 | m._files = [lfutil.standin(f) for f in m._files if lfile(f)] |
|
854 | 859 | m._fileset = set(m._files) |
|
855 | 860 | origmatchfn = m.matchfn |
|
856 | 861 | |
|
857 | 862 | def matchfn(f): |
|
858 | 863 | lfile = lfutil.splitstandin(f) |
|
859 | 864 | return ( |
|
860 | 865 | lfile is not None |
|
861 | 866 | and (f in manifest) |
|
862 | 867 | and origmatchfn(lfile) |
|
863 | 868 | or None |
|
864 | 869 | ) |
|
865 | 870 | |
|
866 | 871 | m.matchfn = matchfn |
|
867 | 872 | return m |
|
868 | 873 | |
|
869 | 874 | listpats = [] |
|
870 | 875 | for pat in pats: |
|
871 | 876 | if matchmod.patkind(pat) is not None: |
|
872 | 877 | listpats.append(pat) |
|
873 | 878 | else: |
|
874 | 879 | listpats.append(makestandin(pat)) |
|
875 | 880 | |
|
876 | 881 | copiedfiles = [] |
|
877 | 882 | |
|
878 | 883 | def overridecopyfile(orig, src, dest, *args, **kwargs): |
|
879 | 884 | if lfutil.shortname in src and dest.startswith( |
|
880 | 885 | repo.wjoin(lfutil.shortname) |
|
881 | 886 | ): |
|
882 | 887 | destlfile = dest.replace(lfutil.shortname, b'') |
|
883 | 888 | if not opts[b'force'] and os.path.exists(destlfile): |
|
884 | 889 | raise IOError( |
|
885 | 890 | b'', _(b'destination largefile already exists') |
|
886 | 891 | ) |
|
887 | 892 | copiedfiles.append((src, dest)) |
|
888 | 893 | orig(src, dest, *args, **kwargs) |
|
889 | 894 | |
|
890 | 895 | with extensions.wrappedfunction(util, 'copyfile', overridecopyfile): |
|
891 | 896 | with extensions.wrappedfunction(scmutil, 'match', overridematch): |
|
892 | 897 | result += orig(ui, repo, listpats, opts, rename) |
|
893 | 898 | |
|
894 | 899 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
895 | 900 | for (src, dest) in copiedfiles: |
|
896 | 901 | if lfutil.shortname in src and dest.startswith( |
|
897 | 902 | repo.wjoin(lfutil.shortname) |
|
898 | 903 | ): |
|
899 | 904 | srclfile = src.replace(repo.wjoin(lfutil.standin(b'')), b'') |
|
900 | 905 | destlfile = dest.replace(repo.wjoin(lfutil.standin(b'')), b'') |
|
901 | 906 | destlfiledir = repo.wvfs.dirname(repo.wjoin(destlfile)) or b'.' |
|
902 | 907 | if not os.path.isdir(destlfiledir): |
|
903 | 908 | os.makedirs(destlfiledir) |
|
904 | 909 | if rename: |
|
905 | 910 | os.rename(repo.wjoin(srclfile), repo.wjoin(destlfile)) |
|
906 | 911 | |
|
907 | 912 | # The file is gone, but this deletes any empty parent |
|
908 | 913 | # directories as a side-effect. |
|
909 | 914 | repo.wvfs.unlinkpath(srclfile, ignoremissing=True) |
|
910 | 915 | lfdirstate.set_untracked(srclfile) |
|
911 | 916 | else: |
|
912 | 917 | util.copyfile(repo.wjoin(srclfile), repo.wjoin(destlfile)) |
|
913 | 918 | |
|
914 | 919 | lfdirstate.set_tracked(destlfile) |
|
915 | 920 | lfdirstate.write(repo.currenttransaction()) |
|
916 | 921 | except error.Abort as e: |
|
917 | 922 | if e.message != _(b'no files to copy'): |
|
918 | 923 | raise e |
|
919 | 924 | else: |
|
920 | 925 | nolfiles = True |
|
921 | 926 | finally: |
|
922 | 927 | wlock.release() |
|
923 | 928 | |
|
924 | 929 | if nolfiles and nonormalfiles: |
|
925 | 930 | raise error.Abort(_(b'no files to copy')) |
|
926 | 931 | |
|
927 | 932 | return result |
|
928 | 933 | |
|
929 | 934 | |
|
930 | 935 | # When the user calls revert, we have to be careful to not revert any |
|
931 | 936 | # changes to other largefiles accidentally. This means we have to keep |
|
932 | 937 | # track of the largefiles that are being reverted so we only pull down |
|
933 | 938 | # the necessary largefiles. |
|
934 | 939 | # |
|
935 | 940 | # Standins are only updated (to match the hash of largefiles) before |
|
936 | 941 | # commits. Update the standins then run the original revert, changing |
|
937 | 942 | # the matcher to hit standins instead of largefiles. Based on the |
|
938 | 943 | # resulting standins update the largefiles. |
|
939 | 944 | @eh.wrapfunction(cmdutil, 'revert') |
|
940 | 945 | def overriderevert(orig, ui, repo, ctx, *pats, **opts): |
|
941 | 946 | # Because we put the standins in a bad state (by updating them) |
|
942 | 947 | # and then return them to a correct state we need to lock to |
|
943 | 948 | # prevent others from changing them in their incorrect state. |
|
944 | 949 | with repo.wlock(), repo.dirstate.running_status(repo): |
|
945 | 950 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
946 | 951 | s = lfutil.lfdirstatestatus(lfdirstate, repo) |
|
947 | 952 | lfdirstate.write(repo.currenttransaction()) |
|
948 | 953 | for lfile in s.modified: |
|
949 | 954 | lfutil.updatestandin(repo, lfile, lfutil.standin(lfile)) |
|
950 | 955 | for lfile in s.deleted: |
|
951 | 956 | fstandin = lfutil.standin(lfile) |
|
952 | 957 | if repo.wvfs.exists(fstandin): |
|
953 | 958 | repo.wvfs.unlink(fstandin) |
|
954 | 959 | |
|
955 | 960 | oldstandins = lfutil.getstandinsstate(repo) |
|
956 | 961 | |
|
957 | 962 | def overridematch( |
|
958 | 963 | orig, |
|
959 | 964 | mctx, |
|
960 | 965 | pats=(), |
|
961 | 966 | opts=None, |
|
962 | 967 | globbed=False, |
|
963 | 968 | default=b'relpath', |
|
964 | 969 | badfn=None, |
|
965 | 970 | ): |
|
966 | 971 | if opts is None: |
|
967 | 972 | opts = {} |
|
968 | 973 | match = orig(mctx, pats, opts, globbed, default, badfn=badfn) |
|
969 | 974 | m = copy.copy(match) |
|
975 | m._was_tampered_with = True | |
|
970 | 976 | |
|
971 | 977 | # revert supports recursing into subrepos, and though largefiles |
|
972 | 978 | # currently doesn't work correctly in that case, this match is |
|
973 | 979 | # called, so the lfdirstate above may not be the correct one for |
|
974 | 980 | # this invocation of match. |
|
975 | 981 | lfdirstate = lfutil.openlfdirstate( |
|
976 | 982 | mctx.repo().ui, mctx.repo(), False |
|
977 | 983 | ) |
|
978 | 984 | |
|
979 | 985 | wctx = repo[None] |
|
980 | 986 | matchfiles = [] |
|
981 | 987 | for f in m._files: |
|
982 | 988 | standin = lfutil.standin(f) |
|
983 | 989 | if standin in ctx or standin in mctx: |
|
984 | 990 | matchfiles.append(standin) |
|
985 | 991 | elif standin in wctx or lfdirstate.get_entry(f).removed: |
|
986 | 992 | continue |
|
987 | 993 | else: |
|
988 | 994 | matchfiles.append(f) |
|
989 | 995 | m._files = matchfiles |
|
990 | 996 | m._fileset = set(m._files) |
|
991 | 997 | origmatchfn = m.matchfn |
|
992 | 998 | |
|
993 | 999 | def matchfn(f): |
|
994 | 1000 | lfile = lfutil.splitstandin(f) |
|
995 | 1001 | if lfile is not None: |
|
996 | 1002 | return origmatchfn(lfile) and (f in ctx or f in mctx) |
|
997 | 1003 | return origmatchfn(f) |
|
998 | 1004 | |
|
999 | 1005 | m.matchfn = matchfn |
|
1000 | 1006 | return m |
|
1001 | 1007 | |
|
1002 | 1008 | with extensions.wrappedfunction(scmutil, 'match', overridematch): |
|
1003 | 1009 | orig(ui, repo, ctx, *pats, **opts) |
|
1004 | 1010 | |
|
1005 | 1011 | newstandins = lfutil.getstandinsstate(repo) |
|
1006 | 1012 | filelist = lfutil.getlfilestoupdate(oldstandins, newstandins) |
|
1007 | 1013 | # lfdirstate should be 'normallookup'-ed for updated files, |
|
1008 | 1014 | # because reverting doesn't touch dirstate for 'normal' files |
|
1009 | 1015 | # when target revision is explicitly specified: in such case, |
|
1010 | 1016 | # 'n' and valid timestamp in dirstate doesn't ensure 'clean' |
|
1011 | 1017 | # of target (standin) file. |
|
1012 | 1018 | lfcommands.updatelfiles( |
|
1013 | 1019 | ui, repo, filelist, printmessage=False, normallookup=True |
|
1014 | 1020 | ) |
|
1015 | 1021 | |
|
1016 | 1022 | |
|
1017 | 1023 | # after pulling changesets, we need to take some extra care to get |
|
1018 | 1024 | # largefiles updated remotely |
|
1019 | 1025 | @eh.wrapcommand( |
|
1020 | 1026 | b'pull', |
|
1021 | 1027 | opts=[ |
|
1022 | 1028 | ( |
|
1023 | 1029 | b'', |
|
1024 | 1030 | b'all-largefiles', |
|
1025 | 1031 | None, |
|
1026 | 1032 | _(b'download all pulled versions of largefiles (DEPRECATED)'), |
|
1027 | 1033 | ), |
|
1028 | 1034 | ( |
|
1029 | 1035 | b'', |
|
1030 | 1036 | b'lfrev', |
|
1031 | 1037 | [], |
|
1032 | 1038 | _(b'download largefiles for these revisions'), |
|
1033 | 1039 | _(b'REV'), |
|
1034 | 1040 | ), |
|
1035 | 1041 | ], |
|
1036 | 1042 | ) |
|
1037 | 1043 | def overridepull(orig, ui, repo, source=None, **opts): |
|
1038 | 1044 | revsprepull = len(repo) |
|
1039 | 1045 | if not source: |
|
1040 | 1046 | source = b'default' |
|
1041 | 1047 | repo.lfpullsource = source |
|
1042 | 1048 | result = orig(ui, repo, source, **opts) |
|
1043 | 1049 | revspostpull = len(repo) |
|
1044 | 1050 | lfrevs = opts.get('lfrev', []) |
|
1045 | 1051 | if opts.get('all_largefiles'): |
|
1046 | 1052 | lfrevs.append(b'pulled()') |
|
1047 | 1053 | if lfrevs and revspostpull > revsprepull: |
|
1048 | 1054 | numcached = 0 |
|
1049 | 1055 | repo.firstpulled = revsprepull # for pulled() revset expression |
|
1050 | 1056 | try: |
|
1051 | 1057 | for rev in logcmdutil.revrange(repo, lfrevs): |
|
1052 | 1058 | ui.note(_(b'pulling largefiles for revision %d\n') % rev) |
|
1053 | 1059 | (cached, missing) = lfcommands.cachelfiles(ui, repo, rev) |
|
1054 | 1060 | numcached += len(cached) |
|
1055 | 1061 | finally: |
|
1056 | 1062 | del repo.firstpulled |
|
1057 | 1063 | ui.status(_(b"%d largefiles cached\n") % numcached) |
|
1058 | 1064 | return result |
|
1059 | 1065 | |
|
1060 | 1066 | |
|
1061 | 1067 | @eh.wrapcommand( |
|
1062 | 1068 | b'push', |
|
1063 | 1069 | opts=[ |
|
1064 | 1070 | ( |
|
1065 | 1071 | b'', |
|
1066 | 1072 | b'lfrev', |
|
1067 | 1073 | [], |
|
1068 | 1074 | _(b'upload largefiles for these revisions'), |
|
1069 | 1075 | _(b'REV'), |
|
1070 | 1076 | ) |
|
1071 | 1077 | ], |
|
1072 | 1078 | ) |
|
1073 | 1079 | def overridepush(orig, ui, repo, *args, **kwargs): |
|
1074 | 1080 | """Override push command and store --lfrev parameters in opargs""" |
|
1075 | 1081 | lfrevs = kwargs.pop('lfrev', None) |
|
1076 | 1082 | if lfrevs: |
|
1077 | 1083 | opargs = kwargs.setdefault('opargs', {}) |
|
1078 | 1084 | opargs[b'lfrevs'] = logcmdutil.revrange(repo, lfrevs) |
|
1079 | 1085 | return orig(ui, repo, *args, **kwargs) |
|
1080 | 1086 | |
|
1081 | 1087 | |
|
1082 | 1088 | @eh.wrapfunction(exchange, 'pushoperation') |
|
1083 | 1089 | def exchangepushoperation(orig, *args, **kwargs): |
|
1084 | 1090 | """Override pushoperation constructor and store lfrevs parameter""" |
|
1085 | 1091 | lfrevs = kwargs.pop('lfrevs', None) |
|
1086 | 1092 | pushop = orig(*args, **kwargs) |
|
1087 | 1093 | pushop.lfrevs = lfrevs |
|
1088 | 1094 | return pushop |
|
1089 | 1095 | |
|
1090 | 1096 | |
|
1091 | 1097 | @eh.revsetpredicate(b'pulled()') |
|
1092 | 1098 | def pulledrevsetsymbol(repo, subset, x): |
|
1093 | 1099 | """Changesets that just has been pulled. |
|
1094 | 1100 | |
|
1095 | 1101 | Only available with largefiles from pull --lfrev expressions. |
|
1096 | 1102 | |
|
1097 | 1103 | .. container:: verbose |
|
1098 | 1104 | |
|
1099 | 1105 | Some examples: |
|
1100 | 1106 | |
|
1101 | 1107 | - pull largefiles for all new changesets:: |
|
1102 | 1108 | |
|
1103 | 1109 | hg pull -lfrev "pulled()" |
|
1104 | 1110 | |
|
1105 | 1111 | - pull largefiles for all new branch heads:: |
|
1106 | 1112 | |
|
1107 | 1113 | hg pull -lfrev "head(pulled()) and not closed()" |
|
1108 | 1114 | |
|
1109 | 1115 | """ |
|
1110 | 1116 | |
|
1111 | 1117 | try: |
|
1112 | 1118 | firstpulled = repo.firstpulled |
|
1113 | 1119 | except AttributeError: |
|
1114 | 1120 | raise error.Abort(_(b"pulled() only available in --lfrev")) |
|
1115 | 1121 | return smartset.baseset([r for r in subset if r >= firstpulled]) |
|
1116 | 1122 | |
|
1117 | 1123 | |
|
1118 | 1124 | @eh.wrapcommand( |
|
1119 | 1125 | b'clone', |
|
1120 | 1126 | opts=[ |
|
1121 | 1127 | ( |
|
1122 | 1128 | b'', |
|
1123 | 1129 | b'all-largefiles', |
|
1124 | 1130 | None, |
|
1125 | 1131 | _(b'download all versions of all largefiles'), |
|
1126 | 1132 | ) |
|
1127 | 1133 | ], |
|
1128 | 1134 | ) |
|
1129 | 1135 | def overrideclone(orig, ui, source, dest=None, **opts): |
|
1130 | 1136 | d = dest |
|
1131 | 1137 | if d is None: |
|
1132 | 1138 | d = hg.defaultdest(source) |
|
1133 | 1139 | if opts.get('all_largefiles') and not hg.islocal(d): |
|
1134 | 1140 | raise error.Abort( |
|
1135 | 1141 | _(b'--all-largefiles is incompatible with non-local destination %s') |
|
1136 | 1142 | % d |
|
1137 | 1143 | ) |
|
1138 | 1144 | |
|
1139 | 1145 | return orig(ui, source, dest, **opts) |
|
1140 | 1146 | |
|
1141 | 1147 | |
|
1142 | 1148 | @eh.wrapfunction(hg, 'clone') |
|
1143 | 1149 | def hgclone(orig, ui, opts, *args, **kwargs): |
|
1144 | 1150 | result = orig(ui, opts, *args, **kwargs) |
|
1145 | 1151 | |
|
1146 | 1152 | if result is not None: |
|
1147 | 1153 | sourcerepo, destrepo = result |
|
1148 | 1154 | repo = destrepo.local() |
|
1149 | 1155 | |
|
1150 | 1156 | # When cloning to a remote repo (like through SSH), no repo is available |
|
1151 | 1157 | # from the peer. Therefore the largefiles can't be downloaded and the |
|
1152 | 1158 | # hgrc can't be updated. |
|
1153 | 1159 | if not repo: |
|
1154 | 1160 | return result |
|
1155 | 1161 | |
|
1156 | 1162 | # Caching is implicitly limited to 'rev' option, since the dest repo was |
|
1157 | 1163 | # truncated at that point. The user may expect a download count with |
|
1158 | 1164 | # this option, so attempt whether or not this is a largefile repo. |
|
1159 | 1165 | if opts.get(b'all_largefiles'): |
|
1160 | 1166 | success, missing = lfcommands.downloadlfiles(ui, repo) |
|
1161 | 1167 | |
|
1162 | 1168 | if missing != 0: |
|
1163 | 1169 | return None |
|
1164 | 1170 | |
|
1165 | 1171 | return result |
|
1166 | 1172 | |
|
1167 | 1173 | |
|
1168 | 1174 | @eh.wrapcommand(b'rebase', extension=b'rebase') |
|
1169 | 1175 | def overriderebasecmd(orig, ui, repo, **opts): |
|
1170 | 1176 | if not hasattr(repo, '_largefilesenabled'): |
|
1171 | 1177 | return orig(ui, repo, **opts) |
|
1172 | 1178 | |
|
1173 | 1179 | resuming = opts.get('continue') |
|
1174 | 1180 | repo._lfcommithooks.append(lfutil.automatedcommithook(resuming)) |
|
1175 | 1181 | repo._lfstatuswriters.append(lambda *msg, **opts: None) |
|
1176 | 1182 | try: |
|
1177 | 1183 | with ui.configoverride( |
|
1178 | 1184 | {(b'rebase', b'experimental.inmemory'): False}, b"largefiles" |
|
1179 | 1185 | ): |
|
1180 | 1186 | return orig(ui, repo, **opts) |
|
1181 | 1187 | finally: |
|
1182 | 1188 | repo._lfstatuswriters.pop() |
|
1183 | 1189 | repo._lfcommithooks.pop() |
|
1184 | 1190 | |
|
1185 | 1191 | |
|
1186 | 1192 | @eh.extsetup |
|
1187 | 1193 | def overriderebase(ui): |
|
1188 | 1194 | try: |
|
1189 | 1195 | rebase = extensions.find(b'rebase') |
|
1190 | 1196 | except KeyError: |
|
1191 | 1197 | pass |
|
1192 | 1198 | else: |
|
1193 | 1199 | |
|
1194 | 1200 | def _dorebase(orig, *args, **kwargs): |
|
1195 | 1201 | kwargs['inmemory'] = False |
|
1196 | 1202 | return orig(*args, **kwargs) |
|
1197 | 1203 | |
|
1198 | 1204 | extensions.wrapfunction(rebase, '_dorebase', _dorebase) |
|
1199 | 1205 | |
|
1200 | 1206 | |
|
1201 | 1207 | @eh.wrapcommand(b'archive') |
|
1202 | 1208 | def overridearchivecmd(orig, ui, repo, dest, **opts): |
|
1203 | 1209 | with lfstatus(repo.unfiltered()): |
|
1204 | 1210 | return orig(ui, repo.unfiltered(), dest, **opts) |
|
1205 | 1211 | |
|
1206 | 1212 | |
|
1207 | 1213 | @eh.wrapfunction(webcommands, 'archive') |
|
1208 | 1214 | def hgwebarchive(orig, web): |
|
1209 | 1215 | with lfstatus(web.repo): |
|
1210 | 1216 | return orig(web) |
|
1211 | 1217 | |
|
1212 | 1218 | |
|
1213 | 1219 | @eh.wrapfunction(archival, 'archive') |
|
1214 | 1220 | def overridearchive( |
|
1215 | 1221 | orig, |
|
1216 | 1222 | repo, |
|
1217 | 1223 | dest, |
|
1218 | 1224 | node, |
|
1219 | 1225 | kind, |
|
1220 | 1226 | decode=True, |
|
1221 | 1227 | match=None, |
|
1222 | 1228 | prefix=b'', |
|
1223 | 1229 | mtime=None, |
|
1224 | 1230 | subrepos=None, |
|
1225 | 1231 | ): |
|
1226 | 1232 | # For some reason setting repo.lfstatus in hgwebarchive only changes the |
|
1227 | 1233 | # unfiltered repo's attr, so check that as well. |
|
1228 | 1234 | if not repo.lfstatus and not repo.unfiltered().lfstatus: |
|
1229 | 1235 | return orig( |
|
1230 | 1236 | repo, dest, node, kind, decode, match, prefix, mtime, subrepos |
|
1231 | 1237 | ) |
|
1232 | 1238 | |
|
1233 | 1239 | # No need to lock because we are only reading history and |
|
1234 | 1240 | # largefile caches, neither of which are modified. |
|
1235 | 1241 | if node is not None: |
|
1236 | 1242 | lfcommands.cachelfiles(repo.ui, repo, node) |
|
1237 | 1243 | |
|
1238 | 1244 | if kind not in archival.archivers: |
|
1239 | 1245 | raise error.Abort(_(b"unknown archive type '%s'") % kind) |
|
1240 | 1246 | |
|
1241 | 1247 | ctx = repo[node] |
|
1242 | 1248 | |
|
1243 | 1249 | if kind == b'files': |
|
1244 | 1250 | if prefix: |
|
1245 | 1251 | raise error.Abort(_(b'cannot give prefix when archiving to files')) |
|
1246 | 1252 | else: |
|
1247 | 1253 | prefix = archival.tidyprefix(dest, kind, prefix) |
|
1248 | 1254 | |
|
1249 | 1255 | def write(name, mode, islink, getdata): |
|
1250 | 1256 | if match and not match(name): |
|
1251 | 1257 | return |
|
1252 | 1258 | data = getdata() |
|
1253 | 1259 | if decode: |
|
1254 | 1260 | data = repo.wwritedata(name, data) |
|
1255 | 1261 | archiver.addfile(prefix + name, mode, islink, data) |
|
1256 | 1262 | |
|
1257 | 1263 | archiver = archival.archivers[kind](dest, mtime or ctx.date()[0]) |
|
1258 | 1264 | |
|
1259 | 1265 | if repo.ui.configbool(b"ui", b"archivemeta"): |
|
1260 | 1266 | write( |
|
1261 | 1267 | b'.hg_archival.txt', |
|
1262 | 1268 | 0o644, |
|
1263 | 1269 | False, |
|
1264 | 1270 | lambda: archival.buildmetadata(ctx), |
|
1265 | 1271 | ) |
|
1266 | 1272 | |
|
1267 | 1273 | for f in ctx: |
|
1268 | 1274 | ff = ctx.flags(f) |
|
1269 | 1275 | getdata = ctx[f].data |
|
1270 | 1276 | lfile = lfutil.splitstandin(f) |
|
1271 | 1277 | if lfile is not None: |
|
1272 | 1278 | if node is not None: |
|
1273 | 1279 | path = lfutil.findfile(repo, getdata().strip()) |
|
1274 | 1280 | |
|
1275 | 1281 | if path is None: |
|
1276 | 1282 | raise error.Abort( |
|
1277 | 1283 | _( |
|
1278 | 1284 | b'largefile %s not found in repo store or system cache' |
|
1279 | 1285 | ) |
|
1280 | 1286 | % lfile |
|
1281 | 1287 | ) |
|
1282 | 1288 | else: |
|
1283 | 1289 | path = lfile |
|
1284 | 1290 | |
|
1285 | 1291 | f = lfile |
|
1286 | 1292 | |
|
1287 | 1293 | getdata = lambda: util.readfile(path) |
|
1288 | 1294 | write(f, b'x' in ff and 0o755 or 0o644, b'l' in ff, getdata) |
|
1289 | 1295 | |
|
1290 | 1296 | if subrepos: |
|
1291 | 1297 | for subpath in sorted(ctx.substate): |
|
1292 | 1298 | sub = ctx.workingsub(subpath) |
|
1293 | 1299 | submatch = matchmod.subdirmatcher(subpath, match) |
|
1294 | 1300 | subprefix = prefix + subpath + b'/' |
|
1295 | 1301 | |
|
1296 | 1302 | # TODO: Only hgsubrepo instances have `_repo`, so figure out how to |
|
1297 | 1303 | # infer and possibly set lfstatus in hgsubrepoarchive. That would |
|
1298 | 1304 | # allow only hgsubrepos to set this, instead of the current scheme |
|
1299 | 1305 | # where the parent sets this for the child. |
|
1300 | 1306 | with ( |
|
1301 | 1307 | hasattr(sub, '_repo') |
|
1302 | 1308 | and lfstatus(sub._repo) |
|
1303 | 1309 | or util.nullcontextmanager() |
|
1304 | 1310 | ): |
|
1305 | 1311 | sub.archive(archiver, subprefix, submatch) |
|
1306 | 1312 | |
|
1307 | 1313 | archiver.done() |
|
1308 | 1314 | |
|
1309 | 1315 | |
|
1310 | 1316 | @eh.wrapfunction(subrepo.hgsubrepo, 'archive') |
|
1311 | 1317 | def hgsubrepoarchive(orig, repo, archiver, prefix, match=None, decode=True): |
|
1312 | 1318 | lfenabled = hasattr(repo._repo, '_largefilesenabled') |
|
1313 | 1319 | if not lfenabled or not repo._repo.lfstatus: |
|
1314 | 1320 | return orig(repo, archiver, prefix, match, decode) |
|
1315 | 1321 | |
|
1316 | 1322 | repo._get(repo._state + (b'hg',)) |
|
1317 | 1323 | rev = repo._state[1] |
|
1318 | 1324 | ctx = repo._repo[rev] |
|
1319 | 1325 | |
|
1320 | 1326 | if ctx.node() is not None: |
|
1321 | 1327 | lfcommands.cachelfiles(repo.ui, repo._repo, ctx.node()) |
|
1322 | 1328 | |
|
1323 | 1329 | def write(name, mode, islink, getdata): |
|
1324 | 1330 | # At this point, the standin has been replaced with the largefile name, |
|
1325 | 1331 | # so the normal matcher works here without the lfutil variants. |
|
1326 | 1332 | if match and not match(f): |
|
1327 | 1333 | return |
|
1328 | 1334 | data = getdata() |
|
1329 | 1335 | if decode: |
|
1330 | 1336 | data = repo._repo.wwritedata(name, data) |
|
1331 | 1337 | |
|
1332 | 1338 | archiver.addfile(prefix + name, mode, islink, data) |
|
1333 | 1339 | |
|
1334 | 1340 | for f in ctx: |
|
1335 | 1341 | ff = ctx.flags(f) |
|
1336 | 1342 | getdata = ctx[f].data |
|
1337 | 1343 | lfile = lfutil.splitstandin(f) |
|
1338 | 1344 | if lfile is not None: |
|
1339 | 1345 | if ctx.node() is not None: |
|
1340 | 1346 | path = lfutil.findfile(repo._repo, getdata().strip()) |
|
1341 | 1347 | |
|
1342 | 1348 | if path is None: |
|
1343 | 1349 | raise error.Abort( |
|
1344 | 1350 | _( |
|
1345 | 1351 | b'largefile %s not found in repo store or system cache' |
|
1346 | 1352 | ) |
|
1347 | 1353 | % lfile |
|
1348 | 1354 | ) |
|
1349 | 1355 | else: |
|
1350 | 1356 | path = lfile |
|
1351 | 1357 | |
|
1352 | 1358 | f = lfile |
|
1353 | 1359 | |
|
1354 | 1360 | getdata = lambda: util.readfile(os.path.join(prefix, path)) |
|
1355 | 1361 | |
|
1356 | 1362 | write(f, b'x' in ff and 0o755 or 0o644, b'l' in ff, getdata) |
|
1357 | 1363 | |
|
1358 | 1364 | for subpath in sorted(ctx.substate): |
|
1359 | 1365 | sub = ctx.workingsub(subpath) |
|
1360 | 1366 | submatch = matchmod.subdirmatcher(subpath, match) |
|
1361 | 1367 | subprefix = prefix + subpath + b'/' |
|
1362 | 1368 | # TODO: Only hgsubrepo instances have `_repo`, so figure out how to |
|
1363 | 1369 | # infer and possibly set lfstatus at the top of this function. That |
|
1364 | 1370 | # would allow only hgsubrepos to set this, instead of the current scheme |
|
1365 | 1371 | # where the parent sets this for the child. |
|
1366 | 1372 | with ( |
|
1367 | 1373 | hasattr(sub, '_repo') |
|
1368 | 1374 | and lfstatus(sub._repo) |
|
1369 | 1375 | or util.nullcontextmanager() |
|
1370 | 1376 | ): |
|
1371 | 1377 | sub.archive(archiver, subprefix, submatch, decode) |
|
1372 | 1378 | |
|
1373 | 1379 | |
|
1374 | 1380 | # If a largefile is modified, the change is not reflected in its |
|
1375 | 1381 | # standin until a commit. cmdutil.bailifchanged() raises an exception |
|
1376 | 1382 | # if the repo has uncommitted changes. Wrap it to also check if |
|
1377 | 1383 | # largefiles were changed. This is used by bisect, backout and fetch. |
|
1378 | 1384 | @eh.wrapfunction(cmdutil, 'bailifchanged') |
|
1379 | 1385 | def overridebailifchanged(orig, repo, *args, **kwargs): |
|
1380 | 1386 | orig(repo, *args, **kwargs) |
|
1381 | 1387 | with lfstatus(repo): |
|
1382 | 1388 | s = repo.status() |
|
1383 | 1389 | if s.modified or s.added or s.removed or s.deleted: |
|
1384 | 1390 | raise error.Abort(_(b'uncommitted changes')) |
|
1385 | 1391 | |
|
1386 | 1392 | |
|
1387 | 1393 | @eh.wrapfunction(cmdutil, 'postcommitstatus') |
|
1388 | 1394 | def postcommitstatus(orig, repo, *args, **kwargs): |
|
1389 | 1395 | with lfstatus(repo): |
|
1390 | 1396 | return orig(repo, *args, **kwargs) |
|
1391 | 1397 | |
|
1392 | 1398 | |
|
1393 | 1399 | @eh.wrapfunction(cmdutil, 'forget') |
|
1394 | 1400 | def cmdutilforget( |
|
1395 | 1401 | orig, ui, repo, match, prefix, uipathfn, explicitonly, dryrun, interactive |
|
1396 | 1402 | ): |
|
1397 | 1403 | normalmatcher = composenormalfilematcher(match, repo[None].manifest()) |
|
1398 | 1404 | bad, forgot = orig( |
|
1399 | 1405 | ui, |
|
1400 | 1406 | repo, |
|
1401 | 1407 | normalmatcher, |
|
1402 | 1408 | prefix, |
|
1403 | 1409 | uipathfn, |
|
1404 | 1410 | explicitonly, |
|
1405 | 1411 | dryrun, |
|
1406 | 1412 | interactive, |
|
1407 | 1413 | ) |
|
1408 | 1414 | m = composelargefilematcher(match, repo[None].manifest()) |
|
1409 | 1415 | |
|
1410 | 1416 | with lfstatus(repo): |
|
1411 | 1417 | s = repo.status(match=m, clean=True) |
|
1412 | 1418 | manifest = repo[None].manifest() |
|
1413 | 1419 | forget = sorted(s.modified + s.added + s.deleted + s.clean) |
|
1414 | 1420 | forget = [f for f in forget if lfutil.standin(f) in manifest] |
|
1415 | 1421 | |
|
1416 | 1422 | for f in forget: |
|
1417 | 1423 | fstandin = lfutil.standin(f) |
|
1418 | 1424 | if fstandin not in repo.dirstate and not repo.wvfs.isdir(fstandin): |
|
1419 | 1425 | ui.warn( |
|
1420 | 1426 | _(b'not removing %s: file is already untracked\n') % uipathfn(f) |
|
1421 | 1427 | ) |
|
1422 | 1428 | bad.append(f) |
|
1423 | 1429 | |
|
1424 | 1430 | for f in forget: |
|
1425 | 1431 | if ui.verbose or not m.exact(f): |
|
1426 | 1432 | ui.status(_(b'removing %s\n') % uipathfn(f)) |
|
1427 | 1433 | |
|
1428 | 1434 | # Need to lock because standin files are deleted then removed from the |
|
1429 | 1435 | # repository and we could race in-between. |
|
1430 | 1436 | with repo.wlock(): |
|
1431 | 1437 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
1432 | 1438 | for f in forget: |
|
1433 | 1439 | lfdirstate.set_untracked(f) |
|
1434 | 1440 | lfdirstate.write(repo.currenttransaction()) |
|
1435 | 1441 | standins = [lfutil.standin(f) for f in forget] |
|
1436 | 1442 | for f in standins: |
|
1437 | 1443 | repo.wvfs.unlinkpath(f, ignoremissing=True) |
|
1438 | 1444 | rejected = repo[None].forget(standins) |
|
1439 | 1445 | |
|
1440 | 1446 | bad.extend(f for f in rejected if f in m.files()) |
|
1441 | 1447 | forgot.extend(f for f in forget if f not in rejected) |
|
1442 | 1448 | return bad, forgot |
|
1443 | 1449 | |
|
1444 | 1450 | |
|
1445 | 1451 | def _getoutgoings(repo, other, missing, addfunc): |
|
1446 | 1452 | """get pairs of filename and largefile hash in outgoing revisions |
|
1447 | 1453 | in 'missing'. |
|
1448 | 1454 | |
|
1449 | 1455 | largefiles already existing on 'other' repository are ignored. |
|
1450 | 1456 | |
|
1451 | 1457 | 'addfunc' is invoked with each unique pairs of filename and |
|
1452 | 1458 | largefile hash value. |
|
1453 | 1459 | """ |
|
1454 | 1460 | knowns = set() |
|
1455 | 1461 | lfhashes = set() |
|
1456 | 1462 | |
|
1457 | 1463 | def dedup(fn, lfhash): |
|
1458 | 1464 | k = (fn, lfhash) |
|
1459 | 1465 | if k not in knowns: |
|
1460 | 1466 | knowns.add(k) |
|
1461 | 1467 | lfhashes.add(lfhash) |
|
1462 | 1468 | |
|
1463 | 1469 | lfutil.getlfilestoupload(repo, missing, dedup) |
|
1464 | 1470 | if lfhashes: |
|
1465 | 1471 | lfexists = storefactory.openstore(repo, other).exists(lfhashes) |
|
1466 | 1472 | for fn, lfhash in knowns: |
|
1467 | 1473 | if not lfexists[lfhash]: # lfhash doesn't exist on "other" |
|
1468 | 1474 | addfunc(fn, lfhash) |
|
1469 | 1475 | |
|
1470 | 1476 | |
|
1471 | 1477 | def outgoinghook(ui, repo, other, opts, missing): |
|
1472 | 1478 | if opts.pop(b'large', None): |
|
1473 | 1479 | lfhashes = set() |
|
1474 | 1480 | if ui.debugflag: |
|
1475 | 1481 | toupload = {} |
|
1476 | 1482 | |
|
1477 | 1483 | def addfunc(fn, lfhash): |
|
1478 | 1484 | if fn not in toupload: |
|
1479 | 1485 | toupload[fn] = [] # pytype: disable=unsupported-operands |
|
1480 | 1486 | toupload[fn].append(lfhash) |
|
1481 | 1487 | lfhashes.add(lfhash) |
|
1482 | 1488 | |
|
1483 | 1489 | def showhashes(fn): |
|
1484 | 1490 | for lfhash in sorted(toupload[fn]): |
|
1485 | 1491 | ui.debug(b' %s\n' % lfhash) |
|
1486 | 1492 | |
|
1487 | 1493 | else: |
|
1488 | 1494 | toupload = set() |
|
1489 | 1495 | |
|
1490 | 1496 | def addfunc(fn, lfhash): |
|
1491 | 1497 | toupload.add(fn) |
|
1492 | 1498 | lfhashes.add(lfhash) |
|
1493 | 1499 | |
|
1494 | 1500 | def showhashes(fn): |
|
1495 | 1501 | pass |
|
1496 | 1502 | |
|
1497 | 1503 | _getoutgoings(repo, other, missing, addfunc) |
|
1498 | 1504 | |
|
1499 | 1505 | if not toupload: |
|
1500 | 1506 | ui.status(_(b'largefiles: no files to upload\n')) |
|
1501 | 1507 | else: |
|
1502 | 1508 | ui.status( |
|
1503 | 1509 | _(b'largefiles to upload (%d entities):\n') % (len(lfhashes)) |
|
1504 | 1510 | ) |
|
1505 | 1511 | for file in sorted(toupload): |
|
1506 | 1512 | ui.status(lfutil.splitstandin(file) + b'\n') |
|
1507 | 1513 | showhashes(file) |
|
1508 | 1514 | ui.status(b'\n') |
|
1509 | 1515 | |
|
1510 | 1516 | |
|
1511 | 1517 | @eh.wrapcommand( |
|
1512 | 1518 | b'outgoing', opts=[(b'', b'large', None, _(b'display outgoing largefiles'))] |
|
1513 | 1519 | ) |
|
1514 | 1520 | def _outgoingcmd(orig, *args, **kwargs): |
|
1515 | 1521 | # Nothing to do here other than add the extra help option- the hook above |
|
1516 | 1522 | # processes it. |
|
1517 | 1523 | return orig(*args, **kwargs) |
|
1518 | 1524 | |
|
1519 | 1525 | |
|
1520 | 1526 | def summaryremotehook(ui, repo, opts, changes): |
|
1521 | 1527 | largeopt = opts.get(b'large', False) |
|
1522 | 1528 | if changes is None: |
|
1523 | 1529 | if largeopt: |
|
1524 | 1530 | return (False, True) # only outgoing check is needed |
|
1525 | 1531 | else: |
|
1526 | 1532 | return (False, False) |
|
1527 | 1533 | elif largeopt: |
|
1528 | 1534 | url, branch, peer, outgoing = changes[1] |
|
1529 | 1535 | if peer is None: |
|
1530 | 1536 | # i18n: column positioning for "hg summary" |
|
1531 | 1537 | ui.status(_(b'largefiles: (no remote repo)\n')) |
|
1532 | 1538 | return |
|
1533 | 1539 | |
|
1534 | 1540 | toupload = set() |
|
1535 | 1541 | lfhashes = set() |
|
1536 | 1542 | |
|
1537 | 1543 | def addfunc(fn, lfhash): |
|
1538 | 1544 | toupload.add(fn) |
|
1539 | 1545 | lfhashes.add(lfhash) |
|
1540 | 1546 | |
|
1541 | 1547 | _getoutgoings(repo, peer, outgoing.missing, addfunc) |
|
1542 | 1548 | |
|
1543 | 1549 | if not toupload: |
|
1544 | 1550 | # i18n: column positioning for "hg summary" |
|
1545 | 1551 | ui.status(_(b'largefiles: (no files to upload)\n')) |
|
1546 | 1552 | else: |
|
1547 | 1553 | # i18n: column positioning for "hg summary" |
|
1548 | 1554 | ui.status( |
|
1549 | 1555 | _(b'largefiles: %d entities for %d files to upload\n') |
|
1550 | 1556 | % (len(lfhashes), len(toupload)) |
|
1551 | 1557 | ) |
|
1552 | 1558 | |
|
1553 | 1559 | |
|
1554 | 1560 | @eh.wrapcommand( |
|
1555 | 1561 | b'summary', opts=[(b'', b'large', None, _(b'display outgoing largefiles'))] |
|
1556 | 1562 | ) |
|
1557 | 1563 | def overridesummary(orig, ui, repo, *pats, **opts): |
|
1558 | 1564 | with lfstatus(repo): |
|
1559 | 1565 | orig(ui, repo, *pats, **opts) |
|
1560 | 1566 | |
|
1561 | 1567 | |
|
1562 | 1568 | @eh.wrapfunction(scmutil, 'addremove') |
|
1563 | 1569 | def scmutiladdremove( |
|
1564 | 1570 | orig, |
|
1565 | 1571 | repo, |
|
1566 | 1572 | matcher, |
|
1567 | 1573 | prefix, |
|
1568 | 1574 | uipathfn, |
|
1569 | 1575 | opts=None, |
|
1570 | 1576 | open_tr=None, |
|
1571 | 1577 | ): |
|
1572 | 1578 | if opts is None: |
|
1573 | 1579 | opts = {} |
|
1574 | 1580 | if not lfutil.islfilesrepo(repo): |
|
1575 | 1581 | return orig(repo, matcher, prefix, uipathfn, opts, open_tr=open_tr) |
|
1576 | 1582 | |
|
1577 | 1583 | # open the transaction and changing_files context |
|
1578 | 1584 | if open_tr is not None: |
|
1579 | 1585 | open_tr() |
|
1580 | 1586 | |
|
1581 | 1587 | # Get the list of missing largefiles so we can remove them |
|
1582 | 1588 | with repo.dirstate.running_status(repo): |
|
1583 | 1589 | lfdirstate = lfutil.openlfdirstate(repo.ui, repo) |
|
1584 | 1590 | unsure, s, mtime_boundary = lfdirstate.status( |
|
1585 | 1591 | matchmod.always(), |
|
1586 | 1592 | subrepos=[], |
|
1587 | 1593 | ignored=False, |
|
1588 | 1594 | clean=False, |
|
1589 | 1595 | unknown=False, |
|
1590 | 1596 | ) |
|
1591 | 1597 | |
|
1592 | 1598 | # Call into the normal remove code, but the removing of the standin, we want |
|
1593 | 1599 | # to have handled by original addremove. Monkey patching here makes sure |
|
1594 | 1600 | # we don't remove the standin in the largefiles code, preventing a very |
|
1595 | 1601 | # confused state later. |
|
1596 | 1602 | if s.deleted: |
|
1597 | 1603 | m = copy.copy(matcher) |
|
1604 | m._was_tampered_with = True | |
|
1598 | 1605 | |
|
1599 | 1606 | # The m._files and m._map attributes are not changed to the deleted list |
|
1600 | 1607 | # because that affects the m.exact() test, which in turn governs whether |
|
1601 | 1608 | # or not the file name is printed, and how. Simply limit the original |
|
1602 | 1609 | # matches to those in the deleted status list. |
|
1603 | 1610 | matchfn = m.matchfn |
|
1604 | 1611 | m.matchfn = lambda f: f in s.deleted and matchfn(f) |
|
1605 | 1612 | |
|
1606 | 1613 | removelargefiles( |
|
1607 | 1614 | repo.ui, |
|
1608 | 1615 | repo, |
|
1609 | 1616 | True, |
|
1610 | 1617 | m, |
|
1611 | 1618 | uipathfn, |
|
1612 | 1619 | opts.get(b'dry_run'), |
|
1613 | 1620 | **pycompat.strkwargs(opts) |
|
1614 | 1621 | ) |
|
1615 | 1622 | # Call into the normal add code, and any files that *should* be added as |
|
1616 | 1623 | # largefiles will be |
|
1617 | 1624 | added, bad = addlargefiles( |
|
1618 | 1625 | repo.ui, repo, True, matcher, uipathfn, **pycompat.strkwargs(opts) |
|
1619 | 1626 | ) |
|
1620 | 1627 | # Now that we've handled largefiles, hand off to the original addremove |
|
1621 | 1628 | # function to take care of the rest. Make sure it doesn't do anything with |
|
1622 | 1629 | # largefiles by passing a matcher that will ignore them. |
|
1623 | 1630 | matcher = composenormalfilematcher(matcher, repo[None].manifest(), added) |
|
1624 | 1631 | |
|
1625 | 1632 | return orig(repo, matcher, prefix, uipathfn, opts, open_tr=open_tr) |
|
1626 | 1633 | |
|
1627 | 1634 | |
|
1628 | 1635 | # Calling purge with --all will cause the largefiles to be deleted. |
|
1629 | 1636 | # Override repo.status to prevent this from happening. |
|
1630 | 1637 | @eh.wrapcommand(b'purge') |
|
1631 | 1638 | def overridepurge(orig, ui, repo, *dirs, **opts): |
|
1632 | 1639 | # XXX Monkey patching a repoview will not work. The assigned attribute will |
|
1633 | 1640 | # be set on the unfiltered repo, but we will only lookup attributes in the |
|
1634 | 1641 | # unfiltered repo if the lookup in the repoview object itself fails. As the |
|
1635 | 1642 | # monkey patched method exists on the repoview class the lookup will not |
|
1636 | 1643 | # fail. As a result, the original version will shadow the monkey patched |
|
1637 | 1644 | # one, defeating the monkey patch. |
|
1638 | 1645 | # |
|
1639 | 1646 | # As a work around we use an unfiltered repo here. We should do something |
|
1640 | 1647 | # cleaner instead. |
|
1641 | 1648 | repo = repo.unfiltered() |
|
1642 | 1649 | oldstatus = repo.status |
|
1643 | 1650 | |
|
1644 | 1651 | def overridestatus( |
|
1645 | 1652 | node1=b'.', |
|
1646 | 1653 | node2=None, |
|
1647 | 1654 | match=None, |
|
1648 | 1655 | ignored=False, |
|
1649 | 1656 | clean=False, |
|
1650 | 1657 | unknown=False, |
|
1651 | 1658 | listsubrepos=False, |
|
1652 | 1659 | ): |
|
1653 | 1660 | r = oldstatus( |
|
1654 | 1661 | node1, node2, match, ignored, clean, unknown, listsubrepos |
|
1655 | 1662 | ) |
|
1656 | 1663 | lfdirstate = lfutil.openlfdirstate(ui, repo) |
|
1657 | 1664 | unknown = [ |
|
1658 | 1665 | f for f in r.unknown if not lfdirstate.get_entry(f).any_tracked |
|
1659 | 1666 | ] |
|
1660 | 1667 | ignored = [ |
|
1661 | 1668 | f for f in r.ignored if not lfdirstate.get_entry(f).any_tracked |
|
1662 | 1669 | ] |
|
1663 | 1670 | return scmutil.status( |
|
1664 | 1671 | r.modified, r.added, r.removed, r.deleted, unknown, ignored, r.clean |
|
1665 | 1672 | ) |
|
1666 | 1673 | |
|
1667 | 1674 | repo.status = overridestatus |
|
1668 | 1675 | orig(ui, repo, *dirs, **opts) |
|
1669 | 1676 | repo.status = oldstatus |
|
1670 | 1677 | |
|
1671 | 1678 | |
|
1672 | 1679 | @eh.wrapcommand(b'rollback') |
|
1673 | 1680 | def overriderollback(orig, ui, repo, **opts): |
|
1674 | 1681 | with repo.wlock(): |
|
1675 | 1682 | before = repo.dirstate.parents() |
|
1676 | 1683 | orphans = { |
|
1677 | 1684 | f |
|
1678 | 1685 | for f in repo.dirstate |
|
1679 | 1686 | if lfutil.isstandin(f) and not repo.dirstate.get_entry(f).removed |
|
1680 | 1687 | } |
|
1681 | 1688 | result = orig(ui, repo, **opts) |
|
1682 | 1689 | after = repo.dirstate.parents() |
|
1683 | 1690 | if before == after: |
|
1684 | 1691 | return result # no need to restore standins |
|
1685 | 1692 | |
|
1686 | 1693 | pctx = repo[b'.'] |
|
1687 | 1694 | for f in repo.dirstate: |
|
1688 | 1695 | if lfutil.isstandin(f): |
|
1689 | 1696 | orphans.discard(f) |
|
1690 | 1697 | if repo.dirstate.get_entry(f).removed: |
|
1691 | 1698 | repo.wvfs.unlinkpath(f, ignoremissing=True) |
|
1692 | 1699 | elif f in pctx: |
|
1693 | 1700 | fctx = pctx[f] |
|
1694 | 1701 | repo.wwrite(f, fctx.data(), fctx.flags()) |
|
1695 | 1702 | else: |
|
1696 | 1703 | # content of standin is not so important in 'a', |
|
1697 | 1704 | # 'm' or 'n' (coming from the 2nd parent) cases |
|
1698 | 1705 | lfutil.writestandin(repo, f, b'', False) |
|
1699 | 1706 | for standin in orphans: |
|
1700 | 1707 | repo.wvfs.unlinkpath(standin, ignoremissing=True) |
|
1701 | 1708 | |
|
1702 | 1709 | return result |
|
1703 | 1710 | |
|
1704 | 1711 | |
|
1705 | 1712 | @eh.wrapcommand(b'transplant', extension=b'transplant') |
|
1706 | 1713 | def overridetransplant(orig, ui, repo, *revs, **opts): |
|
1707 | 1714 | resuming = opts.get('continue') |
|
1708 | 1715 | repo._lfcommithooks.append(lfutil.automatedcommithook(resuming)) |
|
1709 | 1716 | repo._lfstatuswriters.append(lambda *msg, **opts: None) |
|
1710 | 1717 | try: |
|
1711 | 1718 | result = orig(ui, repo, *revs, **opts) |
|
1712 | 1719 | finally: |
|
1713 | 1720 | repo._lfstatuswriters.pop() |
|
1714 | 1721 | repo._lfcommithooks.pop() |
|
1715 | 1722 | return result |
|
1716 | 1723 | |
|
1717 | 1724 | |
|
1718 | 1725 | @eh.wrapcommand(b'cat') |
|
1719 | 1726 | def overridecat(orig, ui, repo, file1, *pats, **opts): |
|
1720 | 1727 | ctx = logcmdutil.revsingle(repo, opts.get('rev')) |
|
1721 | 1728 | err = 1 |
|
1722 | 1729 | notbad = set() |
|
1723 | 1730 | m = scmutil.match(ctx, (file1,) + pats, pycompat.byteskwargs(opts)) |
|
1731 | m._was_tampered_with = True | |
|
1724 | 1732 | origmatchfn = m.matchfn |
|
1725 | 1733 | |
|
1726 | 1734 | def lfmatchfn(f): |
|
1727 | 1735 | if origmatchfn(f): |
|
1728 | 1736 | return True |
|
1729 | 1737 | lf = lfutil.splitstandin(f) |
|
1730 | 1738 | if lf is None: |
|
1731 | 1739 | return False |
|
1732 | 1740 | notbad.add(lf) |
|
1733 | 1741 | return origmatchfn(lf) |
|
1734 | 1742 | |
|
1735 | 1743 | m.matchfn = lfmatchfn |
|
1736 | 1744 | origbadfn = m.bad |
|
1737 | 1745 | |
|
1738 | 1746 | def lfbadfn(f, msg): |
|
1739 | 1747 | if not f in notbad: |
|
1740 | 1748 | origbadfn(f, msg) |
|
1741 | 1749 | |
|
1742 | 1750 | m.bad = lfbadfn |
|
1743 | 1751 | |
|
1744 | 1752 | origvisitdirfn = m.visitdir |
|
1745 | 1753 | |
|
1746 | 1754 | def lfvisitdirfn(dir): |
|
1747 | 1755 | if dir == lfutil.shortname: |
|
1748 | 1756 | return True |
|
1749 | 1757 | ret = origvisitdirfn(dir) |
|
1750 | 1758 | if ret: |
|
1751 | 1759 | return ret |
|
1752 | 1760 | lf = lfutil.splitstandin(dir) |
|
1753 | 1761 | if lf is None: |
|
1754 | 1762 | return False |
|
1755 | 1763 | return origvisitdirfn(lf) |
|
1756 | 1764 | |
|
1757 | 1765 | m.visitdir = lfvisitdirfn |
|
1758 | 1766 | |
|
1759 | 1767 | for f in ctx.walk(m): |
|
1760 | 1768 | with cmdutil.makefileobj(ctx, opts.get('output'), pathname=f) as fp: |
|
1761 | 1769 | lf = lfutil.splitstandin(f) |
|
1762 | 1770 | if lf is None or origmatchfn(f): |
|
1763 | 1771 | # duplicating unreachable code from commands.cat |
|
1764 | 1772 | data = ctx[f].data() |
|
1765 | 1773 | if opts.get('decode'): |
|
1766 | 1774 | data = repo.wwritedata(f, data) |
|
1767 | 1775 | fp.write(data) |
|
1768 | 1776 | else: |
|
1769 | 1777 | hash = lfutil.readasstandin(ctx[f]) |
|
1770 | 1778 | if not lfutil.inusercache(repo.ui, hash): |
|
1771 | 1779 | store = storefactory.openstore(repo) |
|
1772 | 1780 | success, missing = store.get([(lf, hash)]) |
|
1773 | 1781 | if len(success) != 1: |
|
1774 | 1782 | raise error.Abort( |
|
1775 | 1783 | _( |
|
1776 | 1784 | b'largefile %s is not in cache and could not be ' |
|
1777 | 1785 | b'downloaded' |
|
1778 | 1786 | ) |
|
1779 | 1787 | % lf |
|
1780 | 1788 | ) |
|
1781 | 1789 | path = lfutil.usercachepath(repo.ui, hash) |
|
1782 | 1790 | with open(path, b"rb") as fpin: |
|
1783 | 1791 | for chunk in util.filechunkiter(fpin): |
|
1784 | 1792 | fp.write(chunk) |
|
1785 | 1793 | err = 0 |
|
1786 | 1794 | return err |
|
1787 | 1795 | |
|
1788 | 1796 | |
|
1789 | 1797 | @eh.wrapfunction(merge, '_update') |
|
1790 | 1798 | def mergeupdate(orig, repo, node, branchmerge, force, *args, **kwargs): |
|
1791 | 1799 | matcher = kwargs.get('matcher', None) |
|
1792 | 1800 | # note if this is a partial update |
|
1793 | 1801 | partial = matcher and not matcher.always() |
|
1794 | 1802 | with repo.wlock(), repo.dirstate.changing_parents(repo): |
|
1795 | 1803 | # branch | | | |
|
1796 | 1804 | # merge | force | partial | action |
|
1797 | 1805 | # -------+-------+---------+-------------- |
|
1798 | 1806 | # x | x | x | linear-merge |
|
1799 | 1807 | # o | x | x | branch-merge |
|
1800 | 1808 | # x | o | x | overwrite (as clean update) |
|
1801 | 1809 | # o | o | x | force-branch-merge (*1) |
|
1802 | 1810 | # x | x | o | (*) |
|
1803 | 1811 | # o | x | o | (*) |
|
1804 | 1812 | # x | o | o | overwrite (as revert) |
|
1805 | 1813 | # o | o | o | (*) |
|
1806 | 1814 | # |
|
1807 | 1815 | # (*) don't care |
|
1808 | 1816 | # (*1) deprecated, but used internally (e.g: "rebase --collapse") |
|
1809 | 1817 | with repo.dirstate.running_status(repo): |
|
1810 | 1818 | lfdirstate = lfutil.openlfdirstate(repo.ui, repo) |
|
1811 | 1819 | unsure, s, mtime_boundary = lfdirstate.status( |
|
1812 | 1820 | matchmod.always(), |
|
1813 | 1821 | subrepos=[], |
|
1814 | 1822 | ignored=False, |
|
1815 | 1823 | clean=True, |
|
1816 | 1824 | unknown=False, |
|
1817 | 1825 | ) |
|
1818 | 1826 | oldclean = set(s.clean) |
|
1819 | 1827 | pctx = repo[b'.'] |
|
1820 | 1828 | dctx = repo[node] |
|
1821 | 1829 | for lfile in unsure + s.modified: |
|
1822 | 1830 | lfileabs = repo.wvfs.join(lfile) |
|
1823 | 1831 | if not repo.wvfs.exists(lfileabs): |
|
1824 | 1832 | continue |
|
1825 | 1833 | lfhash = lfutil.hashfile(lfileabs) |
|
1826 | 1834 | standin = lfutil.standin(lfile) |
|
1827 | 1835 | lfutil.writestandin( |
|
1828 | 1836 | repo, standin, lfhash, lfutil.getexecutable(lfileabs) |
|
1829 | 1837 | ) |
|
1830 | 1838 | if standin in pctx and lfhash == lfutil.readasstandin( |
|
1831 | 1839 | pctx[standin] |
|
1832 | 1840 | ): |
|
1833 | 1841 | oldclean.add(lfile) |
|
1834 | 1842 | for lfile in s.added: |
|
1835 | 1843 | fstandin = lfutil.standin(lfile) |
|
1836 | 1844 | if fstandin not in dctx: |
|
1837 | 1845 | # in this case, content of standin file is meaningless |
|
1838 | 1846 | # (in dctx, lfile is unknown, or normal file) |
|
1839 | 1847 | continue |
|
1840 | 1848 | lfutil.updatestandin(repo, lfile, fstandin) |
|
1841 | 1849 | # mark all clean largefiles as dirty, just in case the update gets |
|
1842 | 1850 | # interrupted before largefiles and lfdirstate are synchronized |
|
1843 | 1851 | for lfile in oldclean: |
|
1844 | 1852 | entry = lfdirstate.get_entry(lfile) |
|
1845 | 1853 | lfdirstate.hacky_extension_update_file( |
|
1846 | 1854 | lfile, |
|
1847 | 1855 | wc_tracked=entry.tracked, |
|
1848 | 1856 | p1_tracked=entry.p1_tracked, |
|
1849 | 1857 | p2_info=entry.p2_info, |
|
1850 | 1858 | possibly_dirty=True, |
|
1851 | 1859 | ) |
|
1852 | 1860 | lfdirstate.write(repo.currenttransaction()) |
|
1853 | 1861 | |
|
1854 | 1862 | oldstandins = lfutil.getstandinsstate(repo) |
|
1855 | 1863 | wc = kwargs.get('wc') |
|
1856 | 1864 | if wc and wc.isinmemory(): |
|
1857 | 1865 | # largefiles is not a good candidate for in-memory merge (large |
|
1858 | 1866 | # files, custom dirstate, matcher usage). |
|
1859 | 1867 | raise error.ProgrammingError( |
|
1860 | 1868 | b'largefiles is not compatible with in-memory merge' |
|
1861 | 1869 | ) |
|
1862 | 1870 | result = orig(repo, node, branchmerge, force, *args, **kwargs) |
|
1863 | 1871 | |
|
1864 | 1872 | newstandins = lfutil.getstandinsstate(repo) |
|
1865 | 1873 | filelist = lfutil.getlfilestoupdate(oldstandins, newstandins) |
|
1866 | 1874 | |
|
1867 | 1875 | # to avoid leaving all largefiles as dirty and thus rehash them, mark |
|
1868 | 1876 | # all the ones that didn't change as clean |
|
1869 | 1877 | for lfile in oldclean.difference(filelist): |
|
1870 | 1878 | lfdirstate.update_file(lfile, p1_tracked=True, wc_tracked=True) |
|
1871 | 1879 | |
|
1872 | 1880 | if branchmerge or force or partial: |
|
1873 | 1881 | filelist.extend(s.deleted + s.removed) |
|
1874 | 1882 | |
|
1875 | 1883 | lfcommands.updatelfiles( |
|
1876 | 1884 | repo.ui, repo, filelist=filelist, normallookup=partial |
|
1877 | 1885 | ) |
|
1878 | 1886 | |
|
1879 | 1887 | return result |
|
1880 | 1888 | |
|
1881 | 1889 | |
|
1882 | 1890 | @eh.wrapfunction(scmutil, 'marktouched') |
|
1883 | 1891 | def scmutilmarktouched(orig, repo, files, *args, **kwargs): |
|
1884 | 1892 | result = orig(repo, files, *args, **kwargs) |
|
1885 | 1893 | |
|
1886 | 1894 | filelist = [] |
|
1887 | 1895 | for f in files: |
|
1888 | 1896 | lf = lfutil.splitstandin(f) |
|
1889 | 1897 | if lf is not None: |
|
1890 | 1898 | filelist.append(lf) |
|
1891 | 1899 | if filelist: |
|
1892 | 1900 | lfcommands.updatelfiles( |
|
1893 | 1901 | repo.ui, |
|
1894 | 1902 | repo, |
|
1895 | 1903 | filelist=filelist, |
|
1896 | 1904 | printmessage=False, |
|
1897 | 1905 | normallookup=True, |
|
1898 | 1906 | ) |
|
1899 | 1907 | |
|
1900 | 1908 | return result |
|
1901 | 1909 | |
|
1902 | 1910 | |
|
1903 | 1911 | @eh.wrapfunction(upgrade_actions, 'preservedrequirements') |
|
1904 | 1912 | @eh.wrapfunction(upgrade_actions, 'supporteddestrequirements') |
|
1905 | 1913 | def upgraderequirements(orig, repo): |
|
1906 | 1914 | reqs = orig(repo) |
|
1907 | 1915 | if b'largefiles' in repo.requirements: |
|
1908 | 1916 | reqs.add(b'largefiles') |
|
1909 | 1917 | return reqs |
|
1910 | 1918 | |
|
1911 | 1919 | |
|
1912 | 1920 | _lfscheme = b'largefile://' |
|
1913 | 1921 | |
|
1914 | 1922 | |
|
1915 | 1923 | @eh.wrapfunction(urlmod, 'open') |
|
1916 | 1924 | def openlargefile(orig, ui, url_, data=None, **kwargs): |
|
1917 | 1925 | if url_.startswith(_lfscheme): |
|
1918 | 1926 | if data: |
|
1919 | 1927 | msg = b"cannot use data on a 'largefile://' url" |
|
1920 | 1928 | raise error.ProgrammingError(msg) |
|
1921 | 1929 | lfid = url_[len(_lfscheme) :] |
|
1922 | 1930 | return storefactory.getlfile(ui, lfid) |
|
1923 | 1931 | else: |
|
1924 | 1932 | return orig(ui, url_, data=data, **kwargs) |
@@ -1,1706 +1,1712 b'' | |||
|
1 | 1 | # match.py - filename matching |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2008, 2009 Olivia Mackall <olivia@selenic.com> and others |
|
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 | |
|
9 | 9 | import bisect |
|
10 | 10 | import copy |
|
11 | 11 | import itertools |
|
12 | 12 | import os |
|
13 | 13 | import re |
|
14 | 14 | |
|
15 | 15 | from .i18n import _ |
|
16 | 16 | from .pycompat import open |
|
17 | 17 | from . import ( |
|
18 | 18 | encoding, |
|
19 | 19 | error, |
|
20 | 20 | pathutil, |
|
21 | 21 | policy, |
|
22 | 22 | pycompat, |
|
23 | 23 | util, |
|
24 | 24 | ) |
|
25 | 25 | from .utils import stringutil |
|
26 | 26 | |
|
27 | 27 | rustmod = policy.importrust('dirstate') |
|
28 | 28 | |
|
29 | 29 | allpatternkinds = ( |
|
30 | 30 | b're', |
|
31 | 31 | b'glob', |
|
32 | 32 | b'path', |
|
33 | 33 | b'filepath', |
|
34 | 34 | b'relglob', |
|
35 | 35 | b'relpath', |
|
36 | 36 | b'relre', |
|
37 | 37 | b'rootglob', |
|
38 | 38 | b'listfile', |
|
39 | 39 | b'listfile0', |
|
40 | 40 | b'set', |
|
41 | 41 | b'include', |
|
42 | 42 | b'subinclude', |
|
43 | 43 | b'rootfilesin', |
|
44 | 44 | ) |
|
45 | 45 | cwdrelativepatternkinds = (b'relpath', b'glob') |
|
46 | 46 | |
|
47 | 47 | propertycache = util.propertycache |
|
48 | 48 | |
|
49 | 49 | |
|
50 | 50 | def _rematcher(regex): |
|
51 | 51 | """compile the regexp with the best available regexp engine and return a |
|
52 | 52 | matcher function""" |
|
53 | 53 | m = util.re.compile(regex) |
|
54 | 54 | try: |
|
55 | 55 | # slightly faster, provided by facebook's re2 bindings |
|
56 | 56 | return m.test_match |
|
57 | 57 | except AttributeError: |
|
58 | 58 | return m.match |
|
59 | 59 | |
|
60 | 60 | |
|
61 | 61 | def _expandsets(cwd, kindpats, ctx=None, listsubrepos=False, badfn=None): |
|
62 | 62 | '''Returns the kindpats list with the 'set' patterns expanded to matchers''' |
|
63 | 63 | matchers = [] |
|
64 | 64 | other = [] |
|
65 | 65 | |
|
66 | 66 | for kind, pat, source in kindpats: |
|
67 | 67 | if kind == b'set': |
|
68 | 68 | if ctx is None: |
|
69 | 69 | raise error.ProgrammingError( |
|
70 | 70 | b"fileset expression with no context" |
|
71 | 71 | ) |
|
72 | 72 | matchers.append(ctx.matchfileset(cwd, pat, badfn=badfn)) |
|
73 | 73 | |
|
74 | 74 | if listsubrepos: |
|
75 | 75 | for subpath in ctx.substate: |
|
76 | 76 | sm = ctx.sub(subpath).matchfileset(cwd, pat, badfn=badfn) |
|
77 | 77 | pm = prefixdirmatcher(subpath, sm, badfn=badfn) |
|
78 | 78 | matchers.append(pm) |
|
79 | 79 | |
|
80 | 80 | continue |
|
81 | 81 | other.append((kind, pat, source)) |
|
82 | 82 | return matchers, other |
|
83 | 83 | |
|
84 | 84 | |
|
85 | 85 | def _expandsubinclude(kindpats, root): |
|
86 | 86 | """Returns the list of subinclude matcher args and the kindpats without the |
|
87 | 87 | subincludes in it.""" |
|
88 | 88 | relmatchers = [] |
|
89 | 89 | other = [] |
|
90 | 90 | |
|
91 | 91 | for kind, pat, source in kindpats: |
|
92 | 92 | if kind == b'subinclude': |
|
93 | 93 | sourceroot = pathutil.dirname(util.normpath(source)) |
|
94 | 94 | pat = util.pconvert(pat) |
|
95 | 95 | path = pathutil.join(sourceroot, pat) |
|
96 | 96 | |
|
97 | 97 | newroot = pathutil.dirname(path) |
|
98 | 98 | matcherargs = (newroot, b'', [], [b'include:%s' % path]) |
|
99 | 99 | |
|
100 | 100 | prefix = pathutil.canonpath(root, root, newroot) |
|
101 | 101 | if prefix: |
|
102 | 102 | prefix += b'/' |
|
103 | 103 | relmatchers.append((prefix, matcherargs)) |
|
104 | 104 | else: |
|
105 | 105 | other.append((kind, pat, source)) |
|
106 | 106 | |
|
107 | 107 | return relmatchers, other |
|
108 | 108 | |
|
109 | 109 | |
|
110 | 110 | def _kindpatsalwaysmatch(kindpats): |
|
111 | 111 | """Checks whether the kindspats match everything, as e.g. |
|
112 | 112 | 'relpath:.' does. |
|
113 | 113 | """ |
|
114 | 114 | for kind, pat, source in kindpats: |
|
115 | 115 | if pat != b'' or kind not in [b'relpath', b'glob']: |
|
116 | 116 | return False |
|
117 | 117 | return True |
|
118 | 118 | |
|
119 | 119 | |
|
120 | 120 | def _buildkindpatsmatcher( |
|
121 | 121 | matchercls, |
|
122 | 122 | root, |
|
123 | 123 | cwd, |
|
124 | 124 | kindpats, |
|
125 | 125 | ctx=None, |
|
126 | 126 | listsubrepos=False, |
|
127 | 127 | badfn=None, |
|
128 | 128 | ): |
|
129 | 129 | matchers = [] |
|
130 | 130 | fms, kindpats = _expandsets( |
|
131 | 131 | cwd, |
|
132 | 132 | kindpats, |
|
133 | 133 | ctx=ctx, |
|
134 | 134 | listsubrepos=listsubrepos, |
|
135 | 135 | badfn=badfn, |
|
136 | 136 | ) |
|
137 | 137 | if kindpats: |
|
138 | 138 | m = matchercls(root, kindpats, badfn=badfn) |
|
139 | 139 | matchers.append(m) |
|
140 | 140 | if fms: |
|
141 | 141 | matchers.extend(fms) |
|
142 | 142 | if not matchers: |
|
143 | 143 | return nevermatcher(badfn=badfn) |
|
144 | 144 | if len(matchers) == 1: |
|
145 | 145 | return matchers[0] |
|
146 | 146 | return unionmatcher(matchers) |
|
147 | 147 | |
|
148 | 148 | |
|
149 | 149 | def match( |
|
150 | 150 | root, |
|
151 | 151 | cwd, |
|
152 | 152 | patterns=None, |
|
153 | 153 | include=None, |
|
154 | 154 | exclude=None, |
|
155 | 155 | default=b'glob', |
|
156 | 156 | auditor=None, |
|
157 | 157 | ctx=None, |
|
158 | 158 | listsubrepos=False, |
|
159 | 159 | warn=None, |
|
160 | 160 | badfn=None, |
|
161 | 161 | icasefs=False, |
|
162 | 162 | ): |
|
163 | 163 | r"""build an object to match a set of file patterns |
|
164 | 164 | |
|
165 | 165 | arguments: |
|
166 | 166 | root - the canonical root of the tree you're matching against |
|
167 | 167 | cwd - the current working directory, if relevant |
|
168 | 168 | patterns - patterns to find |
|
169 | 169 | include - patterns to include (unless they are excluded) |
|
170 | 170 | exclude - patterns to exclude (even if they are included) |
|
171 | 171 | default - if a pattern in patterns has no explicit type, assume this one |
|
172 | 172 | auditor - optional path auditor |
|
173 | 173 | ctx - optional changecontext |
|
174 | 174 | listsubrepos - if True, recurse into subrepositories |
|
175 | 175 | warn - optional function used for printing warnings |
|
176 | 176 | badfn - optional bad() callback for this matcher instead of the default |
|
177 | 177 | icasefs - make a matcher for wdir on case insensitive filesystems, which |
|
178 | 178 | normalizes the given patterns to the case in the filesystem |
|
179 | 179 | |
|
180 | 180 | a pattern is one of: |
|
181 | 181 | 'glob:<glob>' - a glob relative to cwd |
|
182 | 182 | 're:<regexp>' - a regular expression |
|
183 | 183 | 'path:<path>' - a path relative to repository root, which is matched |
|
184 | 184 | recursively |
|
185 | 185 | 'filepath:<path>' - an exact path to a single file, relative to the |
|
186 | 186 | repository root |
|
187 | 187 | 'rootfilesin:<path>' - a path relative to repository root, which is |
|
188 | 188 | matched non-recursively (will not match subdirectories) |
|
189 | 189 | 'relglob:<glob>' - an unrooted glob (*.c matches C files in all dirs) |
|
190 | 190 | 'relpath:<path>' - a path relative to cwd |
|
191 | 191 | 'relre:<regexp>' - a regexp that needn't match the start of a name |
|
192 | 192 | 'set:<fileset>' - a fileset expression |
|
193 | 193 | 'include:<path>' - a file of patterns to read and include |
|
194 | 194 | 'subinclude:<path>' - a file of patterns to match against files under |
|
195 | 195 | the same directory |
|
196 | 196 | '<something>' - a pattern of the specified default type |
|
197 | 197 | |
|
198 | 198 | >>> def _match(root, *args, **kwargs): |
|
199 | 199 | ... return match(util.localpath(root), *args, **kwargs) |
|
200 | 200 | |
|
201 | 201 | Usually a patternmatcher is returned: |
|
202 | 202 | >>> _match(b'/foo', b'.', [br're:.*\.c$', b'path:foo/a', b'*.py']) |
|
203 | 203 | <patternmatcher patterns='[^/]*\\.py$|foo/a(?:/|$)|.*\\.c$'> |
|
204 | 204 | |
|
205 | 205 | Combining 'patterns' with 'include' (resp. 'exclude') gives an |
|
206 | 206 | intersectionmatcher (resp. a differencematcher): |
|
207 | 207 | >>> type(_match(b'/foo', b'.', [br're:.*\.c$'], include=[b'path:lib'])) |
|
208 | 208 | <class 'mercurial.match.intersectionmatcher'> |
|
209 | 209 | >>> type(_match(b'/foo', b'.', [br're:.*\.c$'], exclude=[b'path:build'])) |
|
210 | 210 | <class 'mercurial.match.differencematcher'> |
|
211 | 211 | |
|
212 | 212 | Notice that, if 'patterns' is empty, an alwaysmatcher is returned: |
|
213 | 213 | >>> _match(b'/foo', b'.', []) |
|
214 | 214 | <alwaysmatcher> |
|
215 | 215 | |
|
216 | 216 | The 'default' argument determines which kind of pattern is assumed if a |
|
217 | 217 | pattern has no prefix: |
|
218 | 218 | >>> _match(b'/foo', b'.', [br'.*\.c$'], default=b're') |
|
219 | 219 | <patternmatcher patterns='.*\\.c$'> |
|
220 | 220 | >>> _match(b'/foo', b'.', [b'main.py'], default=b'relpath') |
|
221 | 221 | <patternmatcher patterns='main\\.py(?:/|$)'> |
|
222 | 222 | >>> _match(b'/foo', b'.', [b'main.py'], default=b're') |
|
223 | 223 | <patternmatcher patterns='main.py'> |
|
224 | 224 | |
|
225 | 225 | The primary use of matchers is to check whether a value (usually a file |
|
226 | 226 | name) matches againset one of the patterns given at initialization. There |
|
227 | 227 | are two ways of doing this check. |
|
228 | 228 | |
|
229 | 229 | >>> m = _match(b'/foo', b'', [br're:.*\.c$', b'relpath:a']) |
|
230 | 230 | |
|
231 | 231 | 1. Calling the matcher with a file name returns True if any pattern |
|
232 | 232 | matches that file name: |
|
233 | 233 | >>> m(b'a') |
|
234 | 234 | True |
|
235 | 235 | >>> m(b'main.c') |
|
236 | 236 | True |
|
237 | 237 | >>> m(b'test.py') |
|
238 | 238 | False |
|
239 | 239 | |
|
240 | 240 | 2. Using the exact() method only returns True if the file name matches one |
|
241 | 241 | of the exact patterns (i.e. not re: or glob: patterns): |
|
242 | 242 | >>> m.exact(b'a') |
|
243 | 243 | True |
|
244 | 244 | >>> m.exact(b'main.c') |
|
245 | 245 | False |
|
246 | 246 | """ |
|
247 | 247 | assert os.path.isabs(root) |
|
248 | 248 | cwd = os.path.join(root, util.localpath(cwd)) |
|
249 | 249 | normalize = _donormalize |
|
250 | 250 | if icasefs: |
|
251 | 251 | dirstate = ctx.repo().dirstate |
|
252 | 252 | dsnormalize = dirstate.normalize |
|
253 | 253 | |
|
254 | 254 | def normalize(patterns, default, root, cwd, auditor, warn): |
|
255 | 255 | kp = _donormalize(patterns, default, root, cwd, auditor, warn) |
|
256 | 256 | kindpats = [] |
|
257 | 257 | for kind, pats, source in kp: |
|
258 | 258 | if kind not in (b're', b'relre'): # regex can't be normalized |
|
259 | 259 | p = pats |
|
260 | 260 | pats = dsnormalize(pats) |
|
261 | 261 | |
|
262 | 262 | # Preserve the original to handle a case only rename. |
|
263 | 263 | if p != pats and p in dirstate: |
|
264 | 264 | kindpats.append((kind, p, source)) |
|
265 | 265 | |
|
266 | 266 | kindpats.append((kind, pats, source)) |
|
267 | 267 | return kindpats |
|
268 | 268 | |
|
269 | 269 | if patterns: |
|
270 | 270 | kindpats = normalize(patterns, default, root, cwd, auditor, warn) |
|
271 | 271 | if _kindpatsalwaysmatch(kindpats): |
|
272 | 272 | m = alwaysmatcher(badfn) |
|
273 | 273 | else: |
|
274 | 274 | m = _buildkindpatsmatcher( |
|
275 | 275 | patternmatcher, |
|
276 | 276 | root, |
|
277 | 277 | cwd, |
|
278 | 278 | kindpats, |
|
279 | 279 | ctx=ctx, |
|
280 | 280 | listsubrepos=listsubrepos, |
|
281 | 281 | badfn=badfn, |
|
282 | 282 | ) |
|
283 | 283 | else: |
|
284 | 284 | # It's a little strange that no patterns means to match everything. |
|
285 | 285 | # Consider changing this to match nothing (probably using nevermatcher). |
|
286 | 286 | m = alwaysmatcher(badfn) |
|
287 | 287 | |
|
288 | 288 | if include: |
|
289 | 289 | kindpats = normalize(include, b'glob', root, cwd, auditor, warn) |
|
290 | 290 | im = _buildkindpatsmatcher( |
|
291 | 291 | includematcher, |
|
292 | 292 | root, |
|
293 | 293 | cwd, |
|
294 | 294 | kindpats, |
|
295 | 295 | ctx=ctx, |
|
296 | 296 | listsubrepos=listsubrepos, |
|
297 | 297 | badfn=None, |
|
298 | 298 | ) |
|
299 | 299 | m = intersectmatchers(m, im) |
|
300 | 300 | if exclude: |
|
301 | 301 | kindpats = normalize(exclude, b'glob', root, cwd, auditor, warn) |
|
302 | 302 | em = _buildkindpatsmatcher( |
|
303 | 303 | includematcher, |
|
304 | 304 | root, |
|
305 | 305 | cwd, |
|
306 | 306 | kindpats, |
|
307 | 307 | ctx=ctx, |
|
308 | 308 | listsubrepos=listsubrepos, |
|
309 | 309 | badfn=None, |
|
310 | 310 | ) |
|
311 | 311 | m = differencematcher(m, em) |
|
312 | 312 | return m |
|
313 | 313 | |
|
314 | 314 | |
|
315 | 315 | def exact(files, badfn=None): |
|
316 | 316 | return exactmatcher(files, badfn=badfn) |
|
317 | 317 | |
|
318 | 318 | |
|
319 | 319 | def always(badfn=None): |
|
320 | 320 | return alwaysmatcher(badfn) |
|
321 | 321 | |
|
322 | 322 | |
|
323 | 323 | def never(badfn=None): |
|
324 | 324 | return nevermatcher(badfn) |
|
325 | 325 | |
|
326 | 326 | |
|
327 | 327 | def badmatch(match, badfn): |
|
328 | 328 | """Make a copy of the given matcher, replacing its bad method with the given |
|
329 | 329 | one. |
|
330 | 330 | """ |
|
331 | 331 | m = copy.copy(match) |
|
332 | 332 | m.bad = badfn |
|
333 | 333 | return m |
|
334 | 334 | |
|
335 | 335 | |
|
336 | 336 | def _donormalize(patterns, default, root, cwd, auditor=None, warn=None): |
|
337 | 337 | """Convert 'kind:pat' from the patterns list to tuples with kind and |
|
338 | 338 | normalized and rooted patterns and with listfiles expanded.""" |
|
339 | 339 | kindpats = [] |
|
340 | 340 | kinds_to_normalize = ( |
|
341 | 341 | b'relglob', |
|
342 | 342 | b'path', |
|
343 | 343 | b'filepath', |
|
344 | 344 | b'rootfilesin', |
|
345 | 345 | b'rootglob', |
|
346 | 346 | ) |
|
347 | 347 | |
|
348 | 348 | for kind, pat in [_patsplit(p, default) for p in patterns]: |
|
349 | 349 | if kind in cwdrelativepatternkinds: |
|
350 | 350 | pat = pathutil.canonpath(root, cwd, pat, auditor=auditor) |
|
351 | 351 | elif kind in kinds_to_normalize: |
|
352 | 352 | pat = util.normpath(pat) |
|
353 | 353 | elif kind in (b'listfile', b'listfile0'): |
|
354 | 354 | try: |
|
355 | 355 | files = util.readfile(pat) |
|
356 | 356 | if kind == b'listfile0': |
|
357 | 357 | files = files.split(b'\0') |
|
358 | 358 | else: |
|
359 | 359 | files = files.splitlines() |
|
360 | 360 | files = [f for f in files if f] |
|
361 | 361 | except EnvironmentError: |
|
362 | 362 | raise error.Abort(_(b"unable to read file list (%s)") % pat) |
|
363 | 363 | for k, p, source in _donormalize( |
|
364 | 364 | files, default, root, cwd, auditor, warn |
|
365 | 365 | ): |
|
366 | 366 | kindpats.append((k, p, pat)) |
|
367 | 367 | continue |
|
368 | 368 | elif kind == b'include': |
|
369 | 369 | try: |
|
370 | 370 | fullpath = os.path.join(root, util.localpath(pat)) |
|
371 | 371 | includepats = readpatternfile(fullpath, warn) |
|
372 | 372 | for k, p, source in _donormalize( |
|
373 | 373 | includepats, default, root, cwd, auditor, warn |
|
374 | 374 | ): |
|
375 | 375 | kindpats.append((k, p, source or pat)) |
|
376 | 376 | except error.Abort as inst: |
|
377 | 377 | raise error.Abort( |
|
378 | 378 | b'%s: %s' |
|
379 | 379 | % ( |
|
380 | 380 | pat, |
|
381 | 381 | inst.message, |
|
382 | 382 | ) |
|
383 | 383 | ) |
|
384 | 384 | except IOError as inst: |
|
385 | 385 | if warn: |
|
386 | 386 | warn( |
|
387 | 387 | _(b"skipping unreadable pattern file '%s': %s\n") |
|
388 | 388 | % (pat, stringutil.forcebytestr(inst.strerror)) |
|
389 | 389 | ) |
|
390 | 390 | continue |
|
391 | 391 | # else: re or relre - which cannot be normalized |
|
392 | 392 | kindpats.append((kind, pat, b'')) |
|
393 | 393 | return kindpats |
|
394 | 394 | |
|
395 | 395 | |
|
396 | 396 | class basematcher: |
|
397 | 397 | def __init__(self, badfn=None): |
|
398 | self._was_tampered_with = False | |
|
398 | 399 | if badfn is not None: |
|
399 | 400 | self.bad = badfn |
|
400 | 401 | |
|
402 | def was_tampered_with(self): | |
|
403 | # [_was_tampered_with] is used to track if when extensions changed the matcher | |
|
404 | # behavior (crazy stuff!), so we disable the rust fast path. | |
|
405 | return self._was_tampered_with | |
|
406 | ||
|
401 | 407 | def __call__(self, fn): |
|
402 | 408 | return self.matchfn(fn) |
|
403 | 409 | |
|
404 | 410 | # Callbacks related to how the matcher is used by dirstate.walk. |
|
405 | 411 | # Subscribers to these events must monkeypatch the matcher object. |
|
406 | 412 | def bad(self, f, msg): |
|
407 | 413 | """Callback from dirstate.walk for each explicit file that can't be |
|
408 | 414 | found/accessed, with an error message.""" |
|
409 | 415 | |
|
410 | 416 | # If an traversedir is set, it will be called when a directory discovered |
|
411 | 417 | # by recursive traversal is visited. |
|
412 | 418 | traversedir = None |
|
413 | 419 | |
|
414 | 420 | @propertycache |
|
415 | 421 | def _files(self): |
|
416 | 422 | return [] |
|
417 | 423 | |
|
418 | 424 | def files(self): |
|
419 | 425 | """Explicitly listed files or patterns or roots: |
|
420 | 426 | if no patterns or .always(): empty list, |
|
421 | 427 | if exact: list exact files, |
|
422 | 428 | if not .anypats(): list all files and dirs, |
|
423 | 429 | else: optimal roots""" |
|
424 | 430 | return self._files |
|
425 | 431 | |
|
426 | 432 | @propertycache |
|
427 | 433 | def _fileset(self): |
|
428 | 434 | return set(self._files) |
|
429 | 435 | |
|
430 | 436 | def exact(self, f): |
|
431 | 437 | '''Returns True if f is in .files().''' |
|
432 | 438 | return f in self._fileset |
|
433 | 439 | |
|
434 | 440 | def matchfn(self, f): |
|
435 | 441 | return False |
|
436 | 442 | |
|
437 | 443 | def visitdir(self, dir): |
|
438 | 444 | """Decides whether a directory should be visited based on whether it |
|
439 | 445 | has potential matches in it or one of its subdirectories. This is |
|
440 | 446 | based on the match's primary, included, and excluded patterns. |
|
441 | 447 | |
|
442 | 448 | Returns the string 'all' if the given directory and all subdirectories |
|
443 | 449 | should be visited. Otherwise returns True or False indicating whether |
|
444 | 450 | the given directory should be visited. |
|
445 | 451 | """ |
|
446 | 452 | return True |
|
447 | 453 | |
|
448 | 454 | def visitchildrenset(self, dir): |
|
449 | 455 | """Decides whether a directory should be visited based on whether it |
|
450 | 456 | has potential matches in it or one of its subdirectories, and |
|
451 | 457 | potentially lists which subdirectories of that directory should be |
|
452 | 458 | visited. This is based on the match's primary, included, and excluded |
|
453 | 459 | patterns. |
|
454 | 460 | |
|
455 | 461 | This function is very similar to 'visitdir', and the following mapping |
|
456 | 462 | can be applied: |
|
457 | 463 | |
|
458 | 464 | visitdir | visitchildrenlist |
|
459 | 465 | ----------+------------------- |
|
460 | 466 | False | set() |
|
461 | 467 | 'all' | 'all' |
|
462 | 468 | True | 'this' OR non-empty set of subdirs -or files- to visit |
|
463 | 469 | |
|
464 | 470 | Example: |
|
465 | 471 | Assume matchers ['path:foo/bar', 'rootfilesin:qux'], we would return |
|
466 | 472 | the following values (assuming the implementation of visitchildrenset |
|
467 | 473 | is capable of recognizing this; some implementations are not). |
|
468 | 474 | |
|
469 | 475 | '' -> {'foo', 'qux'} |
|
470 | 476 | 'baz' -> set() |
|
471 | 477 | 'foo' -> {'bar'} |
|
472 | 478 | # Ideally this would be 'all', but since the prefix nature of matchers |
|
473 | 479 | # is applied to the entire matcher, we have to downgrade this to |
|
474 | 480 | # 'this' due to the non-prefix 'rootfilesin'-kind matcher being mixed |
|
475 | 481 | # in. |
|
476 | 482 | 'foo/bar' -> 'this' |
|
477 | 483 | 'qux' -> 'this' |
|
478 | 484 | |
|
479 | 485 | Important: |
|
480 | 486 | Most matchers do not know if they're representing files or |
|
481 | 487 | directories. They see ['path:dir/f'] and don't know whether 'f' is a |
|
482 | 488 | file or a directory, so visitchildrenset('dir') for most matchers will |
|
483 | 489 | return {'f'}, but if the matcher knows it's a file (like exactmatcher |
|
484 | 490 | does), it may return 'this'. Do not rely on the return being a set |
|
485 | 491 | indicating that there are no files in this dir to investigate (or |
|
486 | 492 | equivalently that if there are files to investigate in 'dir' that it |
|
487 | 493 | will always return 'this'). |
|
488 | 494 | """ |
|
489 | 495 | return b'this' |
|
490 | 496 | |
|
491 | 497 | def always(self): |
|
492 | 498 | """Matcher will match everything and .files() will be empty -- |
|
493 | 499 | optimization might be possible.""" |
|
494 | 500 | return False |
|
495 | 501 | |
|
496 | 502 | def isexact(self): |
|
497 | 503 | """Matcher will match exactly the list of files in .files() -- |
|
498 | 504 | optimization might be possible.""" |
|
499 | 505 | return False |
|
500 | 506 | |
|
501 | 507 | def prefix(self): |
|
502 | 508 | """Matcher will match the paths in .files() recursively -- |
|
503 | 509 | optimization might be possible.""" |
|
504 | 510 | return False |
|
505 | 511 | |
|
506 | 512 | def anypats(self): |
|
507 | 513 | """None of .always(), .isexact(), and .prefix() is true -- |
|
508 | 514 | optimizations will be difficult.""" |
|
509 | 515 | return not self.always() and not self.isexact() and not self.prefix() |
|
510 | 516 | |
|
511 | 517 | |
|
512 | 518 | class alwaysmatcher(basematcher): |
|
513 | 519 | '''Matches everything.''' |
|
514 | 520 | |
|
515 | 521 | def __init__(self, badfn=None): |
|
516 | 522 | super(alwaysmatcher, self).__init__(badfn) |
|
517 | 523 | |
|
518 | 524 | def always(self): |
|
519 | 525 | return True |
|
520 | 526 | |
|
521 | 527 | def matchfn(self, f): |
|
522 | 528 | return True |
|
523 | 529 | |
|
524 | 530 | def visitdir(self, dir): |
|
525 | 531 | return b'all' |
|
526 | 532 | |
|
527 | 533 | def visitchildrenset(self, dir): |
|
528 | 534 | return b'all' |
|
529 | 535 | |
|
530 | 536 | def __repr__(self): |
|
531 | 537 | return r'<alwaysmatcher>' |
|
532 | 538 | |
|
533 | 539 | |
|
534 | 540 | class nevermatcher(basematcher): |
|
535 | 541 | '''Matches nothing.''' |
|
536 | 542 | |
|
537 | 543 | def __init__(self, badfn=None): |
|
538 | 544 | super(nevermatcher, self).__init__(badfn) |
|
539 | 545 | |
|
540 | 546 | # It's a little weird to say that the nevermatcher is an exact matcher |
|
541 | 547 | # or a prefix matcher, but it seems to make sense to let callers take |
|
542 | 548 | # fast paths based on either. There will be no exact matches, nor any |
|
543 | 549 | # prefixes (files() returns []), so fast paths iterating over them should |
|
544 | 550 | # be efficient (and correct). |
|
545 | 551 | def isexact(self): |
|
546 | 552 | return True |
|
547 | 553 | |
|
548 | 554 | def prefix(self): |
|
549 | 555 | return True |
|
550 | 556 | |
|
551 | 557 | def visitdir(self, dir): |
|
552 | 558 | return False |
|
553 | 559 | |
|
554 | 560 | def visitchildrenset(self, dir): |
|
555 | 561 | return set() |
|
556 | 562 | |
|
557 | 563 | def __repr__(self): |
|
558 | 564 | return r'<nevermatcher>' |
|
559 | 565 | |
|
560 | 566 | |
|
561 | 567 | class predicatematcher(basematcher): |
|
562 | 568 | """A matcher adapter for a simple boolean function""" |
|
563 | 569 | |
|
564 | 570 | def __init__(self, predfn, predrepr=None, badfn=None): |
|
565 | 571 | super(predicatematcher, self).__init__(badfn) |
|
566 | 572 | self.matchfn = predfn |
|
567 | 573 | self._predrepr = predrepr |
|
568 | 574 | |
|
569 | 575 | @encoding.strmethod |
|
570 | 576 | def __repr__(self): |
|
571 | 577 | s = stringutil.buildrepr(self._predrepr) or pycompat.byterepr( |
|
572 | 578 | self.matchfn |
|
573 | 579 | ) |
|
574 | 580 | return b'<predicatenmatcher pred=%s>' % s |
|
575 | 581 | |
|
576 | 582 | |
|
577 | 583 | def path_or_parents_in_set(path, prefix_set): |
|
578 | 584 | """Returns True if `path` (or any parent of `path`) is in `prefix_set`.""" |
|
579 | 585 | l = len(prefix_set) |
|
580 | 586 | if l == 0: |
|
581 | 587 | return False |
|
582 | 588 | if path in prefix_set: |
|
583 | 589 | return True |
|
584 | 590 | # If there's more than 5 paths in prefix_set, it's *probably* quicker to |
|
585 | 591 | # "walk up" the directory hierarchy instead, with the assumption that most |
|
586 | 592 | # directory hierarchies are relatively shallow and hash lookup is cheap. |
|
587 | 593 | if l > 5: |
|
588 | 594 | return any( |
|
589 | 595 | parentdir in prefix_set for parentdir in pathutil.finddirs(path) |
|
590 | 596 | ) |
|
591 | 597 | |
|
592 | 598 | # FIXME: Ideally we'd never get to this point if this is the case - we'd |
|
593 | 599 | # recognize ourselves as an 'always' matcher and skip this. |
|
594 | 600 | if b'' in prefix_set: |
|
595 | 601 | return True |
|
596 | 602 | |
|
597 | 603 | sl = ord(b'/') |
|
598 | 604 | |
|
599 | 605 | # We already checked that path isn't in prefix_set exactly, so |
|
600 | 606 | # `path[len(pf)] should never raise IndexError. |
|
601 | 607 | return any(path.startswith(pf) and path[len(pf)] == sl for pf in prefix_set) |
|
602 | 608 | |
|
603 | 609 | |
|
604 | 610 | class patternmatcher(basematcher): |
|
605 | 611 | r"""Matches a set of (kind, pat, source) against a 'root' directory. |
|
606 | 612 | |
|
607 | 613 | >>> kindpats = [ |
|
608 | 614 | ... (b're', br'.*\.c$', b''), |
|
609 | 615 | ... (b'path', b'foo/a', b''), |
|
610 | 616 | ... (b'relpath', b'b', b''), |
|
611 | 617 | ... (b'glob', b'*.h', b''), |
|
612 | 618 | ... ] |
|
613 | 619 | >>> m = patternmatcher(b'foo', kindpats) |
|
614 | 620 | >>> m(b'main.c') # matches re:.*\.c$ |
|
615 | 621 | True |
|
616 | 622 | >>> m(b'b.txt') |
|
617 | 623 | False |
|
618 | 624 | >>> m(b'foo/a') # matches path:foo/a |
|
619 | 625 | True |
|
620 | 626 | >>> m(b'a') # does not match path:b, since 'root' is 'foo' |
|
621 | 627 | False |
|
622 | 628 | >>> m(b'b') # matches relpath:b, since 'root' is 'foo' |
|
623 | 629 | True |
|
624 | 630 | >>> m(b'lib.h') # matches glob:*.h |
|
625 | 631 | True |
|
626 | 632 | |
|
627 | 633 | >>> m.files() |
|
628 | 634 | [b'', b'foo/a', b'', b'b'] |
|
629 | 635 | >>> m.exact(b'foo/a') |
|
630 | 636 | True |
|
631 | 637 | >>> m.exact(b'b') |
|
632 | 638 | True |
|
633 | 639 | >>> m.exact(b'lib.h') # exact matches are for (rel)path kinds |
|
634 | 640 | False |
|
635 | 641 | """ |
|
636 | 642 | |
|
637 | 643 | def __init__(self, root, kindpats, badfn=None): |
|
638 | 644 | super(patternmatcher, self).__init__(badfn) |
|
639 | 645 | kindpats.sort() |
|
640 | 646 | |
|
641 | 647 | roots, dirs, parents = _rootsdirsandparents(kindpats) |
|
642 | 648 | self._files = _explicitfiles(kindpats) |
|
643 | 649 | self._dirs_explicit = set(dirs) |
|
644 | 650 | self._dirs = parents |
|
645 | 651 | self._prefix = _prefix(kindpats) |
|
646 | 652 | self._pats, self._matchfn = _buildmatch(kindpats, b'$', root) |
|
647 | 653 | |
|
648 | 654 | def matchfn(self, fn): |
|
649 | 655 | if fn in self._fileset: |
|
650 | 656 | return True |
|
651 | 657 | return self._matchfn(fn) |
|
652 | 658 | |
|
653 | 659 | def visitdir(self, dir): |
|
654 | 660 | if self._prefix and dir in self._fileset: |
|
655 | 661 | return b'all' |
|
656 | 662 | return ( |
|
657 | 663 | dir in self._dirs |
|
658 | 664 | or path_or_parents_in_set(dir, self._fileset) |
|
659 | 665 | or path_or_parents_in_set(dir, self._dirs_explicit) |
|
660 | 666 | ) |
|
661 | 667 | |
|
662 | 668 | def visitchildrenset(self, dir): |
|
663 | 669 | ret = self.visitdir(dir) |
|
664 | 670 | if ret is True: |
|
665 | 671 | return b'this' |
|
666 | 672 | elif not ret: |
|
667 | 673 | return set() |
|
668 | 674 | assert ret == b'all' |
|
669 | 675 | return b'all' |
|
670 | 676 | |
|
671 | 677 | def prefix(self): |
|
672 | 678 | return self._prefix |
|
673 | 679 | |
|
674 | 680 | @encoding.strmethod |
|
675 | 681 | def __repr__(self): |
|
676 | 682 | return b'<patternmatcher patterns=%r>' % pycompat.bytestr(self._pats) |
|
677 | 683 | |
|
678 | 684 | |
|
679 | 685 | # This is basically a reimplementation of pathutil.dirs that stores the |
|
680 | 686 | # children instead of just a count of them, plus a small optional optimization |
|
681 | 687 | # to avoid some directories we don't need. |
|
682 | 688 | class _dirchildren: |
|
683 | 689 | def __init__(self, paths, onlyinclude=None): |
|
684 | 690 | self._dirs = {} |
|
685 | 691 | self._onlyinclude = onlyinclude or [] |
|
686 | 692 | addpath = self.addpath |
|
687 | 693 | for f in paths: |
|
688 | 694 | addpath(f) |
|
689 | 695 | |
|
690 | 696 | def addpath(self, path): |
|
691 | 697 | if path == b'': |
|
692 | 698 | return |
|
693 | 699 | dirs = self._dirs |
|
694 | 700 | findsplitdirs = _dirchildren._findsplitdirs |
|
695 | 701 | for d, b in findsplitdirs(path): |
|
696 | 702 | if d not in self._onlyinclude: |
|
697 | 703 | continue |
|
698 | 704 | dirs.setdefault(d, set()).add(b) |
|
699 | 705 | |
|
700 | 706 | @staticmethod |
|
701 | 707 | def _findsplitdirs(path): |
|
702 | 708 | # yields (dirname, basename) tuples, walking back to the root. This is |
|
703 | 709 | # very similar to pathutil.finddirs, except: |
|
704 | 710 | # - produces a (dirname, basename) tuple, not just 'dirname' |
|
705 | 711 | # Unlike manifest._splittopdir, this does not suffix `dirname` with a |
|
706 | 712 | # slash. |
|
707 | 713 | oldpos = len(path) |
|
708 | 714 | pos = path.rfind(b'/') |
|
709 | 715 | while pos != -1: |
|
710 | 716 | yield path[:pos], path[pos + 1 : oldpos] |
|
711 | 717 | oldpos = pos |
|
712 | 718 | pos = path.rfind(b'/', 0, pos) |
|
713 | 719 | yield b'', path[:oldpos] |
|
714 | 720 | |
|
715 | 721 | def get(self, path): |
|
716 | 722 | return self._dirs.get(path, set()) |
|
717 | 723 | |
|
718 | 724 | |
|
719 | 725 | class includematcher(basematcher): |
|
720 | 726 | def __init__(self, root, kindpats, badfn=None): |
|
721 | 727 | super(includematcher, self).__init__(badfn) |
|
722 | 728 | if rustmod is not None: |
|
723 | 729 | # We need to pass the patterns to Rust because they can contain |
|
724 | 730 | # patterns from the user interface |
|
725 | 731 | self._kindpats = kindpats |
|
726 | 732 | self._pats, self.matchfn = _buildmatch(kindpats, b'(?:/|$)', root) |
|
727 | 733 | self._prefix = _prefix(kindpats) |
|
728 | 734 | roots, dirs, parents = _rootsdirsandparents(kindpats) |
|
729 | 735 | # roots are directories which are recursively included. |
|
730 | 736 | self._roots = set(roots) |
|
731 | 737 | # dirs are directories which are non-recursively included. |
|
732 | 738 | self._dirs = set(dirs) |
|
733 | 739 | # parents are directories which are non-recursively included because |
|
734 | 740 | # they are needed to get to items in _dirs or _roots. |
|
735 | 741 | self._parents = parents |
|
736 | 742 | |
|
737 | 743 | def visitdir(self, dir): |
|
738 | 744 | if self._prefix and dir in self._roots: |
|
739 | 745 | return b'all' |
|
740 | 746 | return ( |
|
741 | 747 | dir in self._dirs |
|
742 | 748 | or dir in self._parents |
|
743 | 749 | or path_or_parents_in_set(dir, self._roots) |
|
744 | 750 | ) |
|
745 | 751 | |
|
746 | 752 | @propertycache |
|
747 | 753 | def _allparentschildren(self): |
|
748 | 754 | # It may seem odd that we add dirs, roots, and parents, and then |
|
749 | 755 | # restrict to only parents. This is to catch the case of: |
|
750 | 756 | # dirs = ['foo/bar'] |
|
751 | 757 | # parents = ['foo'] |
|
752 | 758 | # if we asked for the children of 'foo', but had only added |
|
753 | 759 | # self._parents, we wouldn't be able to respond ['bar']. |
|
754 | 760 | return _dirchildren( |
|
755 | 761 | itertools.chain(self._dirs, self._roots, self._parents), |
|
756 | 762 | onlyinclude=self._parents, |
|
757 | 763 | ) |
|
758 | 764 | |
|
759 | 765 | def visitchildrenset(self, dir): |
|
760 | 766 | if self._prefix and dir in self._roots: |
|
761 | 767 | return b'all' |
|
762 | 768 | # Note: this does *not* include the 'dir in self._parents' case from |
|
763 | 769 | # visitdir, that's handled below. |
|
764 | 770 | if ( |
|
765 | 771 | b'' in self._roots |
|
766 | 772 | or dir in self._dirs |
|
767 | 773 | or path_or_parents_in_set(dir, self._roots) |
|
768 | 774 | ): |
|
769 | 775 | return b'this' |
|
770 | 776 | |
|
771 | 777 | if dir in self._parents: |
|
772 | 778 | return self._allparentschildren.get(dir) or set() |
|
773 | 779 | return set() |
|
774 | 780 | |
|
775 | 781 | @encoding.strmethod |
|
776 | 782 | def __repr__(self): |
|
777 | 783 | return b'<includematcher includes=%r>' % pycompat.bytestr(self._pats) |
|
778 | 784 | |
|
779 | 785 | |
|
780 | 786 | class exactmatcher(basematcher): |
|
781 | 787 | r"""Matches the input files exactly. They are interpreted as paths, not |
|
782 | 788 | patterns (so no kind-prefixes). |
|
783 | 789 | |
|
784 | 790 | >>> m = exactmatcher([b'a.txt', br're:.*\.c$']) |
|
785 | 791 | >>> m(b'a.txt') |
|
786 | 792 | True |
|
787 | 793 | >>> m(b'b.txt') |
|
788 | 794 | False |
|
789 | 795 | |
|
790 | 796 | Input files that would be matched are exactly those returned by .files() |
|
791 | 797 | >>> m.files() |
|
792 | 798 | ['a.txt', 're:.*\\.c$'] |
|
793 | 799 | |
|
794 | 800 | So pattern 're:.*\.c$' is not considered as a regex, but as a file name |
|
795 | 801 | >>> m(b'main.c') |
|
796 | 802 | False |
|
797 | 803 | >>> m(br're:.*\.c$') |
|
798 | 804 | True |
|
799 | 805 | """ |
|
800 | 806 | |
|
801 | 807 | def __init__(self, files, badfn=None): |
|
802 | 808 | super(exactmatcher, self).__init__(badfn) |
|
803 | 809 | |
|
804 | 810 | if isinstance(files, list): |
|
805 | 811 | self._files = files |
|
806 | 812 | else: |
|
807 | 813 | self._files = list(files) |
|
808 | 814 | |
|
809 | 815 | matchfn = basematcher.exact |
|
810 | 816 | |
|
811 | 817 | @propertycache |
|
812 | 818 | def _dirs(self): |
|
813 | 819 | return set(pathutil.dirs(self._fileset)) |
|
814 | 820 | |
|
815 | 821 | def visitdir(self, dir): |
|
816 | 822 | return dir in self._dirs |
|
817 | 823 | |
|
818 | 824 | @propertycache |
|
819 | 825 | def _visitchildrenset_candidates(self): |
|
820 | 826 | """A memoized set of candidates for visitchildrenset.""" |
|
821 | 827 | return self._fileset | self._dirs - {b''} |
|
822 | 828 | |
|
823 | 829 | @propertycache |
|
824 | 830 | def _sorted_visitchildrenset_candidates(self): |
|
825 | 831 | """A memoized sorted list of candidates for visitchildrenset.""" |
|
826 | 832 | return sorted(self._visitchildrenset_candidates) |
|
827 | 833 | |
|
828 | 834 | def visitchildrenset(self, dir): |
|
829 | 835 | if not self._fileset or dir not in self._dirs: |
|
830 | 836 | return set() |
|
831 | 837 | |
|
832 | 838 | if dir == b'': |
|
833 | 839 | candidates = self._visitchildrenset_candidates |
|
834 | 840 | else: |
|
835 | 841 | candidates = self._sorted_visitchildrenset_candidates |
|
836 | 842 | d = dir + b'/' |
|
837 | 843 | # Use bisect to find the first element potentially starting with d |
|
838 | 844 | # (i.e. >= d). This should always find at least one element (we'll |
|
839 | 845 | # assert later if this is not the case). |
|
840 | 846 | first = bisect.bisect_left(candidates, d) |
|
841 | 847 | # We need a representation of the first element that is > d that |
|
842 | 848 | # does not start with d, so since we added a `/` on the end of dir, |
|
843 | 849 | # we'll add whatever comes after slash (we could probably assume |
|
844 | 850 | # that `0` is after `/`, but let's not) to the end of dir instead. |
|
845 | 851 | dnext = dir + encoding.strtolocal(chr(ord(b'/') + 1)) |
|
846 | 852 | # Use bisect to find the first element >= d_next |
|
847 | 853 | last = bisect.bisect_left(candidates, dnext, lo=first) |
|
848 | 854 | dlen = len(d) |
|
849 | 855 | candidates = {c[dlen:] for c in candidates[first:last]} |
|
850 | 856 | # self._dirs includes all of the directories, recursively, so if |
|
851 | 857 | # we're attempting to match foo/bar/baz.txt, it'll have '', 'foo', |
|
852 | 858 | # 'foo/bar' in it. Thus we can safely ignore a candidate that has a |
|
853 | 859 | # '/' in it, indicating a it's for a subdir-of-a-subdir; the |
|
854 | 860 | # immediate subdir will be in there without a slash. |
|
855 | 861 | ret = {c for c in candidates if b'/' not in c} |
|
856 | 862 | # We really do not expect ret to be empty, since that would imply that |
|
857 | 863 | # there's something in _dirs that didn't have a file in _fileset. |
|
858 | 864 | assert ret |
|
859 | 865 | return ret |
|
860 | 866 | |
|
861 | 867 | def isexact(self): |
|
862 | 868 | return True |
|
863 | 869 | |
|
864 | 870 | @encoding.strmethod |
|
865 | 871 | def __repr__(self): |
|
866 | 872 | return b'<exactmatcher files=%r>' % self._files |
|
867 | 873 | |
|
868 | 874 | |
|
869 | 875 | class differencematcher(basematcher): |
|
870 | 876 | """Composes two matchers by matching if the first matches and the second |
|
871 | 877 | does not. |
|
872 | 878 | |
|
873 | 879 | The second matcher's non-matching-attributes (bad, traversedir) are ignored. |
|
874 | 880 | """ |
|
875 | 881 | |
|
876 | 882 | def __init__(self, m1, m2): |
|
877 | 883 | super(differencematcher, self).__init__() |
|
878 | 884 | self._m1 = m1 |
|
879 | 885 | self._m2 = m2 |
|
880 | 886 | self.bad = m1.bad |
|
881 | 887 | self.traversedir = m1.traversedir |
|
882 | 888 | |
|
883 | 889 | def matchfn(self, f): |
|
884 | 890 | return self._m1(f) and not self._m2(f) |
|
885 | 891 | |
|
886 | 892 | @propertycache |
|
887 | 893 | def _files(self): |
|
888 | 894 | if self.isexact(): |
|
889 | 895 | return [f for f in self._m1.files() if self(f)] |
|
890 | 896 | # If m1 is not an exact matcher, we can't easily figure out the set of |
|
891 | 897 | # files, because its files() are not always files. For example, if |
|
892 | 898 | # m1 is "path:dir" and m2 is "rootfileins:.", we don't |
|
893 | 899 | # want to remove "dir" from the set even though it would match m2, |
|
894 | 900 | # because the "dir" in m1 may not be a file. |
|
895 | 901 | return self._m1.files() |
|
896 | 902 | |
|
897 | 903 | def visitdir(self, dir): |
|
898 | 904 | if self._m2.visitdir(dir) == b'all': |
|
899 | 905 | return False |
|
900 | 906 | elif not self._m2.visitdir(dir): |
|
901 | 907 | # m2 does not match dir, we can return 'all' here if possible |
|
902 | 908 | return self._m1.visitdir(dir) |
|
903 | 909 | return bool(self._m1.visitdir(dir)) |
|
904 | 910 | |
|
905 | 911 | def visitchildrenset(self, dir): |
|
906 | 912 | m2_set = self._m2.visitchildrenset(dir) |
|
907 | 913 | if m2_set == b'all': |
|
908 | 914 | return set() |
|
909 | 915 | m1_set = self._m1.visitchildrenset(dir) |
|
910 | 916 | # Possible values for m1: 'all', 'this', set(...), set() |
|
911 | 917 | # Possible values for m2: 'this', set(...), set() |
|
912 | 918 | # If m2 has nothing under here that we care about, return m1, even if |
|
913 | 919 | # it's 'all'. This is a change in behavior from visitdir, which would |
|
914 | 920 | # return True, not 'all', for some reason. |
|
915 | 921 | if not m2_set: |
|
916 | 922 | return m1_set |
|
917 | 923 | if m1_set in [b'all', b'this']: |
|
918 | 924 | # Never return 'all' here if m2_set is any kind of non-empty (either |
|
919 | 925 | # 'this' or set(foo)), since m2 might return set() for a |
|
920 | 926 | # subdirectory. |
|
921 | 927 | return b'this' |
|
922 | 928 | # Possible values for m1: set(...), set() |
|
923 | 929 | # Possible values for m2: 'this', set(...) |
|
924 | 930 | # We ignore m2's set results. They're possibly incorrect: |
|
925 | 931 | # m1 = path:dir/subdir, m2=rootfilesin:dir, visitchildrenset(''): |
|
926 | 932 | # m1 returns {'dir'}, m2 returns {'dir'}, if we subtracted we'd |
|
927 | 933 | # return set(), which is *not* correct, we still need to visit 'dir'! |
|
928 | 934 | return m1_set |
|
929 | 935 | |
|
930 | 936 | def isexact(self): |
|
931 | 937 | return self._m1.isexact() |
|
932 | 938 | |
|
933 | 939 | @encoding.strmethod |
|
934 | 940 | def __repr__(self): |
|
935 | 941 | return b'<differencematcher m1=%r, m2=%r>' % (self._m1, self._m2) |
|
936 | 942 | |
|
937 | 943 | |
|
938 | 944 | def intersectmatchers(m1, m2): |
|
939 | 945 | """Composes two matchers by matching if both of them match. |
|
940 | 946 | |
|
941 | 947 | The second matcher's non-matching-attributes (bad, traversedir) are ignored. |
|
942 | 948 | """ |
|
943 | 949 | if m1 is None or m2 is None: |
|
944 | 950 | return m1 or m2 |
|
945 | 951 | if m1.always(): |
|
946 | 952 | m = copy.copy(m2) |
|
947 | 953 | # TODO: Consider encapsulating these things in a class so there's only |
|
948 | 954 | # one thing to copy from m1. |
|
949 | 955 | m.bad = m1.bad |
|
950 | 956 | m.traversedir = m1.traversedir |
|
951 | 957 | return m |
|
952 | 958 | if m2.always(): |
|
953 | 959 | m = copy.copy(m1) |
|
954 | 960 | return m |
|
955 | 961 | return intersectionmatcher(m1, m2) |
|
956 | 962 | |
|
957 | 963 | |
|
958 | 964 | class intersectionmatcher(basematcher): |
|
959 | 965 | def __init__(self, m1, m2): |
|
960 | 966 | super(intersectionmatcher, self).__init__() |
|
961 | 967 | self._m1 = m1 |
|
962 | 968 | self._m2 = m2 |
|
963 | 969 | self.bad = m1.bad |
|
964 | 970 | self.traversedir = m1.traversedir |
|
965 | 971 | |
|
966 | 972 | @propertycache |
|
967 | 973 | def _files(self): |
|
968 | 974 | if self.isexact(): |
|
969 | 975 | m1, m2 = self._m1, self._m2 |
|
970 | 976 | if not m1.isexact(): |
|
971 | 977 | m1, m2 = m2, m1 |
|
972 | 978 | return [f for f in m1.files() if m2(f)] |
|
973 | 979 | # It neither m1 nor m2 is an exact matcher, we can't easily intersect |
|
974 | 980 | # the set of files, because their files() are not always files. For |
|
975 | 981 | # example, if intersecting a matcher "-I glob:foo.txt" with matcher of |
|
976 | 982 | # "path:dir2", we don't want to remove "dir2" from the set. |
|
977 | 983 | return self._m1.files() + self._m2.files() |
|
978 | 984 | |
|
979 | 985 | def matchfn(self, f): |
|
980 | 986 | return self._m1(f) and self._m2(f) |
|
981 | 987 | |
|
982 | 988 | def visitdir(self, dir): |
|
983 | 989 | visit1 = self._m1.visitdir(dir) |
|
984 | 990 | if visit1 == b'all': |
|
985 | 991 | return self._m2.visitdir(dir) |
|
986 | 992 | # bool() because visit1=True + visit2='all' should not be 'all' |
|
987 | 993 | return bool(visit1 and self._m2.visitdir(dir)) |
|
988 | 994 | |
|
989 | 995 | def visitchildrenset(self, dir): |
|
990 | 996 | m1_set = self._m1.visitchildrenset(dir) |
|
991 | 997 | if not m1_set: |
|
992 | 998 | return set() |
|
993 | 999 | m2_set = self._m2.visitchildrenset(dir) |
|
994 | 1000 | if not m2_set: |
|
995 | 1001 | return set() |
|
996 | 1002 | |
|
997 | 1003 | if m1_set == b'all': |
|
998 | 1004 | return m2_set |
|
999 | 1005 | elif m2_set == b'all': |
|
1000 | 1006 | return m1_set |
|
1001 | 1007 | |
|
1002 | 1008 | if m1_set == b'this' or m2_set == b'this': |
|
1003 | 1009 | return b'this' |
|
1004 | 1010 | |
|
1005 | 1011 | assert isinstance(m1_set, set) and isinstance(m2_set, set) |
|
1006 | 1012 | return m1_set.intersection(m2_set) |
|
1007 | 1013 | |
|
1008 | 1014 | def always(self): |
|
1009 | 1015 | return self._m1.always() and self._m2.always() |
|
1010 | 1016 | |
|
1011 | 1017 | def isexact(self): |
|
1012 | 1018 | return self._m1.isexact() or self._m2.isexact() |
|
1013 | 1019 | |
|
1014 | 1020 | @encoding.strmethod |
|
1015 | 1021 | def __repr__(self): |
|
1016 | 1022 | return b'<intersectionmatcher m1=%r, m2=%r>' % (self._m1, self._m2) |
|
1017 | 1023 | |
|
1018 | 1024 | |
|
1019 | 1025 | class subdirmatcher(basematcher): |
|
1020 | 1026 | """Adapt a matcher to work on a subdirectory only. |
|
1021 | 1027 | |
|
1022 | 1028 | The paths are remapped to remove/insert the path as needed: |
|
1023 | 1029 | |
|
1024 | 1030 | >>> from . import pycompat |
|
1025 | 1031 | >>> m1 = match(util.localpath(b'/root'), b'', [b'a.txt', b'sub/b.txt'], auditor=lambda name: None) |
|
1026 | 1032 | >>> m2 = subdirmatcher(b'sub', m1) |
|
1027 | 1033 | >>> m2(b'a.txt') |
|
1028 | 1034 | False |
|
1029 | 1035 | >>> m2(b'b.txt') |
|
1030 | 1036 | True |
|
1031 | 1037 | >>> m2.matchfn(b'a.txt') |
|
1032 | 1038 | False |
|
1033 | 1039 | >>> m2.matchfn(b'b.txt') |
|
1034 | 1040 | True |
|
1035 | 1041 | >>> m2.files() |
|
1036 | 1042 | ['b.txt'] |
|
1037 | 1043 | >>> m2.exact(b'b.txt') |
|
1038 | 1044 | True |
|
1039 | 1045 | >>> def bad(f, msg): |
|
1040 | 1046 | ... print(pycompat.sysstr(b"%s: %s" % (f, msg))) |
|
1041 | 1047 | >>> m1.bad = bad |
|
1042 | 1048 | >>> m2.bad(b'x.txt', b'No such file') |
|
1043 | 1049 | sub/x.txt: No such file |
|
1044 | 1050 | """ |
|
1045 | 1051 | |
|
1046 | 1052 | def __init__(self, path, matcher): |
|
1047 | 1053 | super(subdirmatcher, self).__init__() |
|
1048 | 1054 | self._path = path |
|
1049 | 1055 | self._matcher = matcher |
|
1050 | 1056 | self._always = matcher.always() |
|
1051 | 1057 | |
|
1052 | 1058 | self._files = [ |
|
1053 | 1059 | f[len(path) + 1 :] |
|
1054 | 1060 | for f in matcher._files |
|
1055 | 1061 | if f.startswith(path + b"/") |
|
1056 | 1062 | ] |
|
1057 | 1063 | |
|
1058 | 1064 | # If the parent repo had a path to this subrepo and the matcher is |
|
1059 | 1065 | # a prefix matcher, this submatcher always matches. |
|
1060 | 1066 | if matcher.prefix(): |
|
1061 | 1067 | self._always = any(f == path for f in matcher._files) |
|
1062 | 1068 | |
|
1063 | 1069 | def bad(self, f, msg): |
|
1064 | 1070 | self._matcher.bad(self._path + b"/" + f, msg) |
|
1065 | 1071 | |
|
1066 | 1072 | def matchfn(self, f): |
|
1067 | 1073 | # Some information is lost in the superclass's constructor, so we |
|
1068 | 1074 | # can not accurately create the matching function for the subdirectory |
|
1069 | 1075 | # from the inputs. Instead, we override matchfn() and visitdir() to |
|
1070 | 1076 | # call the original matcher with the subdirectory path prepended. |
|
1071 | 1077 | return self._matcher.matchfn(self._path + b"/" + f) |
|
1072 | 1078 | |
|
1073 | 1079 | def visitdir(self, dir): |
|
1074 | 1080 | if dir == b'': |
|
1075 | 1081 | dir = self._path |
|
1076 | 1082 | else: |
|
1077 | 1083 | dir = self._path + b"/" + dir |
|
1078 | 1084 | return self._matcher.visitdir(dir) |
|
1079 | 1085 | |
|
1080 | 1086 | def visitchildrenset(self, dir): |
|
1081 | 1087 | if dir == b'': |
|
1082 | 1088 | dir = self._path |
|
1083 | 1089 | else: |
|
1084 | 1090 | dir = self._path + b"/" + dir |
|
1085 | 1091 | return self._matcher.visitchildrenset(dir) |
|
1086 | 1092 | |
|
1087 | 1093 | def always(self): |
|
1088 | 1094 | return self._always |
|
1089 | 1095 | |
|
1090 | 1096 | def prefix(self): |
|
1091 | 1097 | return self._matcher.prefix() and not self._always |
|
1092 | 1098 | |
|
1093 | 1099 | @encoding.strmethod |
|
1094 | 1100 | def __repr__(self): |
|
1095 | 1101 | return b'<subdirmatcher path=%r, matcher=%r>' % ( |
|
1096 | 1102 | self._path, |
|
1097 | 1103 | self._matcher, |
|
1098 | 1104 | ) |
|
1099 | 1105 | |
|
1100 | 1106 | |
|
1101 | 1107 | class prefixdirmatcher(basematcher): |
|
1102 | 1108 | """Adapt a matcher to work on a parent directory. |
|
1103 | 1109 | |
|
1104 | 1110 | The matcher's non-matching-attributes (bad, traversedir) are ignored. |
|
1105 | 1111 | |
|
1106 | 1112 | The prefix path should usually be the relative path from the root of |
|
1107 | 1113 | this matcher to the root of the wrapped matcher. |
|
1108 | 1114 | |
|
1109 | 1115 | >>> m1 = match(util.localpath(b'/root/d/e'), b'f', [b'../a.txt', b'b.txt'], auditor=lambda name: None) |
|
1110 | 1116 | >>> m2 = prefixdirmatcher(b'd/e', m1) |
|
1111 | 1117 | >>> m2(b'a.txt') |
|
1112 | 1118 | False |
|
1113 | 1119 | >>> m2(b'd/e/a.txt') |
|
1114 | 1120 | True |
|
1115 | 1121 | >>> m2(b'd/e/b.txt') |
|
1116 | 1122 | False |
|
1117 | 1123 | >>> m2.files() |
|
1118 | 1124 | ['d/e/a.txt', 'd/e/f/b.txt'] |
|
1119 | 1125 | >>> m2.exact(b'd/e/a.txt') |
|
1120 | 1126 | True |
|
1121 | 1127 | >>> m2.visitdir(b'd') |
|
1122 | 1128 | True |
|
1123 | 1129 | >>> m2.visitdir(b'd/e') |
|
1124 | 1130 | True |
|
1125 | 1131 | >>> m2.visitdir(b'd/e/f') |
|
1126 | 1132 | True |
|
1127 | 1133 | >>> m2.visitdir(b'd/e/g') |
|
1128 | 1134 | False |
|
1129 | 1135 | >>> m2.visitdir(b'd/ef') |
|
1130 | 1136 | False |
|
1131 | 1137 | """ |
|
1132 | 1138 | |
|
1133 | 1139 | def __init__(self, path, matcher, badfn=None): |
|
1134 | 1140 | super(prefixdirmatcher, self).__init__(badfn) |
|
1135 | 1141 | if not path: |
|
1136 | 1142 | raise error.ProgrammingError(b'prefix path must not be empty') |
|
1137 | 1143 | self._path = path |
|
1138 | 1144 | self._pathprefix = path + b'/' |
|
1139 | 1145 | self._matcher = matcher |
|
1140 | 1146 | |
|
1141 | 1147 | @propertycache |
|
1142 | 1148 | def _files(self): |
|
1143 | 1149 | return [self._pathprefix + f for f in self._matcher._files] |
|
1144 | 1150 | |
|
1145 | 1151 | def matchfn(self, f): |
|
1146 | 1152 | if not f.startswith(self._pathprefix): |
|
1147 | 1153 | return False |
|
1148 | 1154 | return self._matcher.matchfn(f[len(self._pathprefix) :]) |
|
1149 | 1155 | |
|
1150 | 1156 | @propertycache |
|
1151 | 1157 | def _pathdirs(self): |
|
1152 | 1158 | return set(pathutil.finddirs(self._path)) |
|
1153 | 1159 | |
|
1154 | 1160 | def visitdir(self, dir): |
|
1155 | 1161 | if dir == self._path: |
|
1156 | 1162 | return self._matcher.visitdir(b'') |
|
1157 | 1163 | if dir.startswith(self._pathprefix): |
|
1158 | 1164 | return self._matcher.visitdir(dir[len(self._pathprefix) :]) |
|
1159 | 1165 | return dir in self._pathdirs |
|
1160 | 1166 | |
|
1161 | 1167 | def visitchildrenset(self, dir): |
|
1162 | 1168 | if dir == self._path: |
|
1163 | 1169 | return self._matcher.visitchildrenset(b'') |
|
1164 | 1170 | if dir.startswith(self._pathprefix): |
|
1165 | 1171 | return self._matcher.visitchildrenset(dir[len(self._pathprefix) :]) |
|
1166 | 1172 | if dir in self._pathdirs: |
|
1167 | 1173 | return b'this' |
|
1168 | 1174 | return set() |
|
1169 | 1175 | |
|
1170 | 1176 | def isexact(self): |
|
1171 | 1177 | return self._matcher.isexact() |
|
1172 | 1178 | |
|
1173 | 1179 | def prefix(self): |
|
1174 | 1180 | return self._matcher.prefix() |
|
1175 | 1181 | |
|
1176 | 1182 | @encoding.strmethod |
|
1177 | 1183 | def __repr__(self): |
|
1178 | 1184 | return b'<prefixdirmatcher path=%r, matcher=%r>' % ( |
|
1179 | 1185 | pycompat.bytestr(self._path), |
|
1180 | 1186 | self._matcher, |
|
1181 | 1187 | ) |
|
1182 | 1188 | |
|
1183 | 1189 | |
|
1184 | 1190 | class unionmatcher(basematcher): |
|
1185 | 1191 | """A matcher that is the union of several matchers. |
|
1186 | 1192 | |
|
1187 | 1193 | The non-matching-attributes (bad, traversedir) are taken from the first |
|
1188 | 1194 | matcher. |
|
1189 | 1195 | """ |
|
1190 | 1196 | |
|
1191 | 1197 | def __init__(self, matchers): |
|
1192 | 1198 | m1 = matchers[0] |
|
1193 | 1199 | super(unionmatcher, self).__init__() |
|
1194 | 1200 | self.traversedir = m1.traversedir |
|
1195 | 1201 | self._matchers = matchers |
|
1196 | 1202 | |
|
1197 | 1203 | def matchfn(self, f): |
|
1198 | 1204 | for match in self._matchers: |
|
1199 | 1205 | if match(f): |
|
1200 | 1206 | return True |
|
1201 | 1207 | return False |
|
1202 | 1208 | |
|
1203 | 1209 | def visitdir(self, dir): |
|
1204 | 1210 | r = False |
|
1205 | 1211 | for m in self._matchers: |
|
1206 | 1212 | v = m.visitdir(dir) |
|
1207 | 1213 | if v == b'all': |
|
1208 | 1214 | return v |
|
1209 | 1215 | r |= v |
|
1210 | 1216 | return r |
|
1211 | 1217 | |
|
1212 | 1218 | def visitchildrenset(self, dir): |
|
1213 | 1219 | r = set() |
|
1214 | 1220 | this = False |
|
1215 | 1221 | for m in self._matchers: |
|
1216 | 1222 | v = m.visitchildrenset(dir) |
|
1217 | 1223 | if not v: |
|
1218 | 1224 | continue |
|
1219 | 1225 | if v == b'all': |
|
1220 | 1226 | return v |
|
1221 | 1227 | if this or v == b'this': |
|
1222 | 1228 | this = True |
|
1223 | 1229 | # don't break, we might have an 'all' in here. |
|
1224 | 1230 | continue |
|
1225 | 1231 | assert isinstance(v, set) |
|
1226 | 1232 | r = r.union(v) |
|
1227 | 1233 | if this: |
|
1228 | 1234 | return b'this' |
|
1229 | 1235 | return r |
|
1230 | 1236 | |
|
1231 | 1237 | @encoding.strmethod |
|
1232 | 1238 | def __repr__(self): |
|
1233 | 1239 | return b'<unionmatcher matchers=%r>' % self._matchers |
|
1234 | 1240 | |
|
1235 | 1241 | |
|
1236 | 1242 | def patkind(pattern, default=None): |
|
1237 | 1243 | r"""If pattern is 'kind:pat' with a known kind, return kind. |
|
1238 | 1244 | |
|
1239 | 1245 | >>> patkind(br're:.*\.c$') |
|
1240 | 1246 | 're' |
|
1241 | 1247 | >>> patkind(b'glob:*.c') |
|
1242 | 1248 | 'glob' |
|
1243 | 1249 | >>> patkind(b'relpath:test.py') |
|
1244 | 1250 | 'relpath' |
|
1245 | 1251 | >>> patkind(b'main.py') |
|
1246 | 1252 | >>> patkind(b'main.py', default=b're') |
|
1247 | 1253 | 're' |
|
1248 | 1254 | """ |
|
1249 | 1255 | return _patsplit(pattern, default)[0] |
|
1250 | 1256 | |
|
1251 | 1257 | |
|
1252 | 1258 | def _patsplit(pattern, default): |
|
1253 | 1259 | """Split a string into the optional pattern kind prefix and the actual |
|
1254 | 1260 | pattern.""" |
|
1255 | 1261 | if b':' in pattern: |
|
1256 | 1262 | kind, pat = pattern.split(b':', 1) |
|
1257 | 1263 | if kind in allpatternkinds: |
|
1258 | 1264 | return kind, pat |
|
1259 | 1265 | return default, pattern |
|
1260 | 1266 | |
|
1261 | 1267 | |
|
1262 | 1268 | def _globre(pat): |
|
1263 | 1269 | r"""Convert an extended glob string to a regexp string. |
|
1264 | 1270 | |
|
1265 | 1271 | >>> from . import pycompat |
|
1266 | 1272 | >>> def bprint(s): |
|
1267 | 1273 | ... print(pycompat.sysstr(s)) |
|
1268 | 1274 | >>> bprint(_globre(br'?')) |
|
1269 | 1275 | . |
|
1270 | 1276 | >>> bprint(_globre(br'*')) |
|
1271 | 1277 | [^/]* |
|
1272 | 1278 | >>> bprint(_globre(br'**')) |
|
1273 | 1279 | .* |
|
1274 | 1280 | >>> bprint(_globre(br'**/a')) |
|
1275 | 1281 | (?:.*/)?a |
|
1276 | 1282 | >>> bprint(_globre(br'a/**/b')) |
|
1277 | 1283 | a/(?:.*/)?b |
|
1278 | 1284 | >>> bprint(_globre(br'[a*?!^][^b][!c]')) |
|
1279 | 1285 | [a*?!^][\^b][^c] |
|
1280 | 1286 | >>> bprint(_globre(br'{a,b}')) |
|
1281 | 1287 | (?:a|b) |
|
1282 | 1288 | >>> bprint(_globre(br'.\*\?')) |
|
1283 | 1289 | \.\*\? |
|
1284 | 1290 | """ |
|
1285 | 1291 | i, n = 0, len(pat) |
|
1286 | 1292 | res = b'' |
|
1287 | 1293 | group = 0 |
|
1288 | 1294 | escape = util.stringutil.regexbytesescapemap.get |
|
1289 | 1295 | |
|
1290 | 1296 | def peek(): |
|
1291 | 1297 | return i < n and pat[i : i + 1] |
|
1292 | 1298 | |
|
1293 | 1299 | while i < n: |
|
1294 | 1300 | c = pat[i : i + 1] |
|
1295 | 1301 | i += 1 |
|
1296 | 1302 | if c not in b'*?[{},\\': |
|
1297 | 1303 | res += escape(c, c) |
|
1298 | 1304 | elif c == b'*': |
|
1299 | 1305 | if peek() == b'*': |
|
1300 | 1306 | i += 1 |
|
1301 | 1307 | if peek() == b'/': |
|
1302 | 1308 | i += 1 |
|
1303 | 1309 | res += b'(?:.*/)?' |
|
1304 | 1310 | else: |
|
1305 | 1311 | res += b'.*' |
|
1306 | 1312 | else: |
|
1307 | 1313 | res += b'[^/]*' |
|
1308 | 1314 | elif c == b'?': |
|
1309 | 1315 | res += b'.' |
|
1310 | 1316 | elif c == b'[': |
|
1311 | 1317 | j = i |
|
1312 | 1318 | if j < n and pat[j : j + 1] in b'!]': |
|
1313 | 1319 | j += 1 |
|
1314 | 1320 | while j < n and pat[j : j + 1] != b']': |
|
1315 | 1321 | j += 1 |
|
1316 | 1322 | if j >= n: |
|
1317 | 1323 | res += b'\\[' |
|
1318 | 1324 | else: |
|
1319 | 1325 | stuff = pat[i:j].replace(b'\\', b'\\\\') |
|
1320 | 1326 | i = j + 1 |
|
1321 | 1327 | if stuff[0:1] == b'!': |
|
1322 | 1328 | stuff = b'^' + stuff[1:] |
|
1323 | 1329 | elif stuff[0:1] == b'^': |
|
1324 | 1330 | stuff = b'\\' + stuff |
|
1325 | 1331 | res = b'%s[%s]' % (res, stuff) |
|
1326 | 1332 | elif c == b'{': |
|
1327 | 1333 | group += 1 |
|
1328 | 1334 | res += b'(?:' |
|
1329 | 1335 | elif c == b'}' and group: |
|
1330 | 1336 | res += b')' |
|
1331 | 1337 | group -= 1 |
|
1332 | 1338 | elif c == b',' and group: |
|
1333 | 1339 | res += b'|' |
|
1334 | 1340 | elif c == b'\\': |
|
1335 | 1341 | p = peek() |
|
1336 | 1342 | if p: |
|
1337 | 1343 | i += 1 |
|
1338 | 1344 | res += escape(p, p) |
|
1339 | 1345 | else: |
|
1340 | 1346 | res += escape(c, c) |
|
1341 | 1347 | else: |
|
1342 | 1348 | res += escape(c, c) |
|
1343 | 1349 | return res |
|
1344 | 1350 | |
|
1345 | 1351 | |
|
1346 | 1352 | FLAG_RE = util.re.compile(br'^\(\?([aiLmsux]+)\)(.*)') |
|
1347 | 1353 | |
|
1348 | 1354 | |
|
1349 | 1355 | def _regex(kind, pat, globsuffix): |
|
1350 | 1356 | """Convert a (normalized) pattern of any kind into a |
|
1351 | 1357 | regular expression. |
|
1352 | 1358 | globsuffix is appended to the regexp of globs.""" |
|
1353 | 1359 | if not pat and kind in (b'glob', b'relpath'): |
|
1354 | 1360 | return b'' |
|
1355 | 1361 | if kind == b're': |
|
1356 | 1362 | return pat |
|
1357 | 1363 | if kind == b'filepath': |
|
1358 | 1364 | raise error.ProgrammingError( |
|
1359 | 1365 | "'filepath:' patterns should not be converted to a regex" |
|
1360 | 1366 | ) |
|
1361 | 1367 | if kind in (b'path', b'relpath'): |
|
1362 | 1368 | if pat == b'.': |
|
1363 | 1369 | return b'' |
|
1364 | 1370 | return util.stringutil.reescape(pat) + b'(?:/|$)' |
|
1365 | 1371 | if kind == b'rootfilesin': |
|
1366 | 1372 | if pat == b'.': |
|
1367 | 1373 | escaped = b'' |
|
1368 | 1374 | else: |
|
1369 | 1375 | # Pattern is a directory name. |
|
1370 | 1376 | escaped = util.stringutil.reescape(pat) + b'/' |
|
1371 | 1377 | # Anything after the pattern must be a non-directory. |
|
1372 | 1378 | return escaped + b'[^/]+$' |
|
1373 | 1379 | if kind == b'relglob': |
|
1374 | 1380 | globre = _globre(pat) |
|
1375 | 1381 | if globre.startswith(b'[^/]*'): |
|
1376 | 1382 | # When pat has the form *XYZ (common), make the returned regex more |
|
1377 | 1383 | # legible by returning the regex for **XYZ instead of **/*XYZ. |
|
1378 | 1384 | return b'.*' + globre[len(b'[^/]*') :] + globsuffix |
|
1379 | 1385 | return b'(?:|.*/)' + globre + globsuffix |
|
1380 | 1386 | if kind == b'relre': |
|
1381 | 1387 | flag = None |
|
1382 | 1388 | m = FLAG_RE.match(pat) |
|
1383 | 1389 | if m: |
|
1384 | 1390 | flag, pat = m.groups() |
|
1385 | 1391 | if not pat.startswith(b'^'): |
|
1386 | 1392 | pat = b'.*' + pat |
|
1387 | 1393 | if flag is not None: |
|
1388 | 1394 | pat = br'(?%s:%s)' % (flag, pat) |
|
1389 | 1395 | return pat |
|
1390 | 1396 | if kind in (b'glob', b'rootglob'): |
|
1391 | 1397 | return _globre(pat) + globsuffix |
|
1392 | 1398 | raise error.ProgrammingError(b'not a regex pattern: %s:%s' % (kind, pat)) |
|
1393 | 1399 | |
|
1394 | 1400 | |
|
1395 | 1401 | def _buildmatch(kindpats, globsuffix, root): |
|
1396 | 1402 | """Return regexp string and a matcher function for kindpats. |
|
1397 | 1403 | globsuffix is appended to the regexp of globs.""" |
|
1398 | 1404 | matchfuncs = [] |
|
1399 | 1405 | |
|
1400 | 1406 | subincludes, kindpats = _expandsubinclude(kindpats, root) |
|
1401 | 1407 | if subincludes: |
|
1402 | 1408 | submatchers = {} |
|
1403 | 1409 | |
|
1404 | 1410 | def matchsubinclude(f): |
|
1405 | 1411 | for prefix, matcherargs in subincludes: |
|
1406 | 1412 | if f.startswith(prefix): |
|
1407 | 1413 | mf = submatchers.get(prefix) |
|
1408 | 1414 | if mf is None: |
|
1409 | 1415 | mf = match(*matcherargs) |
|
1410 | 1416 | submatchers[prefix] = mf |
|
1411 | 1417 | |
|
1412 | 1418 | if mf(f[len(prefix) :]): |
|
1413 | 1419 | return True |
|
1414 | 1420 | return False |
|
1415 | 1421 | |
|
1416 | 1422 | matchfuncs.append(matchsubinclude) |
|
1417 | 1423 | |
|
1418 | 1424 | regex = b'' |
|
1419 | 1425 | if kindpats: |
|
1420 | 1426 | if all(k == b'rootfilesin' for k, p, s in kindpats): |
|
1421 | 1427 | dirs = {p for k, p, s in kindpats} |
|
1422 | 1428 | |
|
1423 | 1429 | def mf(f): |
|
1424 | 1430 | i = f.rfind(b'/') |
|
1425 | 1431 | if i >= 0: |
|
1426 | 1432 | dir = f[:i] |
|
1427 | 1433 | else: |
|
1428 | 1434 | dir = b'.' |
|
1429 | 1435 | return dir in dirs |
|
1430 | 1436 | |
|
1431 | 1437 | regex = b'rootfilesin: %s' % stringutil.pprint(list(sorted(dirs))) |
|
1432 | 1438 | matchfuncs.append(mf) |
|
1433 | 1439 | else: |
|
1434 | 1440 | regex, mf = _buildregexmatch(kindpats, globsuffix) |
|
1435 | 1441 | matchfuncs.append(mf) |
|
1436 | 1442 | |
|
1437 | 1443 | if len(matchfuncs) == 1: |
|
1438 | 1444 | return regex, matchfuncs[0] |
|
1439 | 1445 | else: |
|
1440 | 1446 | return regex, lambda f: any(mf(f) for mf in matchfuncs) |
|
1441 | 1447 | |
|
1442 | 1448 | |
|
1443 | 1449 | MAX_RE_SIZE = 20000 |
|
1444 | 1450 | |
|
1445 | 1451 | |
|
1446 | 1452 | def _joinregexes(regexps): |
|
1447 | 1453 | """gather multiple regular expressions into a single one""" |
|
1448 | 1454 | return b'|'.join(regexps) |
|
1449 | 1455 | |
|
1450 | 1456 | |
|
1451 | 1457 | def _buildregexmatch(kindpats, globsuffix): |
|
1452 | 1458 | """Build a match function from a list of kinds and kindpats, |
|
1453 | 1459 | return regexp string and a matcher function. |
|
1454 | 1460 | |
|
1455 | 1461 | Test too large input |
|
1456 | 1462 | >>> _buildregexmatch([ |
|
1457 | 1463 | ... (b'relglob', b'?' * MAX_RE_SIZE, b'') |
|
1458 | 1464 | ... ], b'$') |
|
1459 | 1465 | Traceback (most recent call last): |
|
1460 | 1466 | ... |
|
1461 | 1467 | Abort: matcher pattern is too long (20009 bytes) |
|
1462 | 1468 | """ |
|
1463 | 1469 | try: |
|
1464 | 1470 | allgroups = [] |
|
1465 | 1471 | regexps = [] |
|
1466 | 1472 | exact = set() |
|
1467 | 1473 | for kind, pattern, _source in kindpats: |
|
1468 | 1474 | if kind == b'filepath': |
|
1469 | 1475 | exact.add(pattern) |
|
1470 | 1476 | continue |
|
1471 | 1477 | regexps.append(_regex(kind, pattern, globsuffix)) |
|
1472 | 1478 | |
|
1473 | 1479 | fullregexp = _joinregexes(regexps) |
|
1474 | 1480 | |
|
1475 | 1481 | startidx = 0 |
|
1476 | 1482 | groupsize = 0 |
|
1477 | 1483 | for idx, r in enumerate(regexps): |
|
1478 | 1484 | piecesize = len(r) |
|
1479 | 1485 | if piecesize > MAX_RE_SIZE: |
|
1480 | 1486 | msg = _(b"matcher pattern is too long (%d bytes)") % piecesize |
|
1481 | 1487 | raise error.Abort(msg) |
|
1482 | 1488 | elif (groupsize + piecesize) > MAX_RE_SIZE: |
|
1483 | 1489 | group = regexps[startidx:idx] |
|
1484 | 1490 | allgroups.append(_joinregexes(group)) |
|
1485 | 1491 | startidx = idx |
|
1486 | 1492 | groupsize = 0 |
|
1487 | 1493 | groupsize += piecesize + 1 |
|
1488 | 1494 | |
|
1489 | 1495 | if startidx == 0: |
|
1490 | 1496 | matcher = _rematcher(fullregexp) |
|
1491 | 1497 | func = lambda s: bool(matcher(s)) |
|
1492 | 1498 | else: |
|
1493 | 1499 | group = regexps[startidx:] |
|
1494 | 1500 | allgroups.append(_joinregexes(group)) |
|
1495 | 1501 | allmatchers = [_rematcher(g) for g in allgroups] |
|
1496 | 1502 | func = lambda s: any(m(s) for m in allmatchers) |
|
1497 | 1503 | |
|
1498 | 1504 | actualfunc = func |
|
1499 | 1505 | if exact: |
|
1500 | 1506 | # An empty regex will always match, so only call the regex if |
|
1501 | 1507 | # there were any actual patterns to match. |
|
1502 | 1508 | if not regexps: |
|
1503 | 1509 | actualfunc = lambda s: s in exact |
|
1504 | 1510 | else: |
|
1505 | 1511 | actualfunc = lambda s: s in exact or func(s) |
|
1506 | 1512 | return fullregexp, actualfunc |
|
1507 | 1513 | except re.error: |
|
1508 | 1514 | for k, p, s in kindpats: |
|
1509 | 1515 | if k == b'filepath': |
|
1510 | 1516 | continue |
|
1511 | 1517 | try: |
|
1512 | 1518 | _rematcher(_regex(k, p, globsuffix)) |
|
1513 | 1519 | except re.error: |
|
1514 | 1520 | if s: |
|
1515 | 1521 | raise error.Abort( |
|
1516 | 1522 | _(b"%s: invalid pattern (%s): %s") % (s, k, p) |
|
1517 | 1523 | ) |
|
1518 | 1524 | else: |
|
1519 | 1525 | raise error.Abort(_(b"invalid pattern (%s): %s") % (k, p)) |
|
1520 | 1526 | raise error.Abort(_(b"invalid pattern")) |
|
1521 | 1527 | |
|
1522 | 1528 | |
|
1523 | 1529 | def _patternrootsanddirs(kindpats): |
|
1524 | 1530 | """Returns roots and directories corresponding to each pattern. |
|
1525 | 1531 | |
|
1526 | 1532 | This calculates the roots and directories exactly matching the patterns and |
|
1527 | 1533 | returns a tuple of (roots, dirs) for each. It does not return other |
|
1528 | 1534 | directories which may also need to be considered, like the parent |
|
1529 | 1535 | directories. |
|
1530 | 1536 | """ |
|
1531 | 1537 | r = [] |
|
1532 | 1538 | d = [] |
|
1533 | 1539 | for kind, pat, source in kindpats: |
|
1534 | 1540 | if kind in (b'glob', b'rootglob'): # find the non-glob prefix |
|
1535 | 1541 | root = [] |
|
1536 | 1542 | for p in pat.split(b'/'): |
|
1537 | 1543 | if b'[' in p or b'{' in p or b'*' in p or b'?' in p: |
|
1538 | 1544 | break |
|
1539 | 1545 | root.append(p) |
|
1540 | 1546 | r.append(b'/'.join(root)) |
|
1541 | 1547 | elif kind in (b'relpath', b'path', b'filepath'): |
|
1542 | 1548 | if pat == b'.': |
|
1543 | 1549 | pat = b'' |
|
1544 | 1550 | r.append(pat) |
|
1545 | 1551 | elif kind in (b'rootfilesin',): |
|
1546 | 1552 | if pat == b'.': |
|
1547 | 1553 | pat = b'' |
|
1548 | 1554 | d.append(pat) |
|
1549 | 1555 | else: # relglob, re, relre |
|
1550 | 1556 | r.append(b'') |
|
1551 | 1557 | return r, d |
|
1552 | 1558 | |
|
1553 | 1559 | |
|
1554 | 1560 | def _roots(kindpats): |
|
1555 | 1561 | '''Returns root directories to match recursively from the given patterns.''' |
|
1556 | 1562 | roots, dirs = _patternrootsanddirs(kindpats) |
|
1557 | 1563 | return roots |
|
1558 | 1564 | |
|
1559 | 1565 | |
|
1560 | 1566 | def _rootsdirsandparents(kindpats): |
|
1561 | 1567 | """Returns roots and exact directories from patterns. |
|
1562 | 1568 | |
|
1563 | 1569 | `roots` are directories to match recursively, `dirs` should |
|
1564 | 1570 | be matched non-recursively, and `parents` are the implicitly required |
|
1565 | 1571 | directories to walk to items in either roots or dirs. |
|
1566 | 1572 | |
|
1567 | 1573 | Returns a tuple of (roots, dirs, parents). |
|
1568 | 1574 | |
|
1569 | 1575 | >>> r = _rootsdirsandparents( |
|
1570 | 1576 | ... [(b'glob', b'g/h/*', b''), (b'glob', b'g/h', b''), |
|
1571 | 1577 | ... (b'glob', b'g*', b'')]) |
|
1572 | 1578 | >>> print(r[0:2], sorted(r[2])) # the set has an unstable output |
|
1573 | 1579 | (['g/h', 'g/h', ''], []) ['', 'g'] |
|
1574 | 1580 | >>> r = _rootsdirsandparents( |
|
1575 | 1581 | ... [(b'rootfilesin', b'g/h', b''), (b'rootfilesin', b'', b'')]) |
|
1576 | 1582 | >>> print(r[0:2], sorted(r[2])) # the set has an unstable output |
|
1577 | 1583 | ([], ['g/h', '']) ['', 'g'] |
|
1578 | 1584 | >>> r = _rootsdirsandparents( |
|
1579 | 1585 | ... [(b'relpath', b'r', b''), (b'path', b'p/p', b''), |
|
1580 | 1586 | ... (b'path', b'', b'')]) |
|
1581 | 1587 | >>> print(r[0:2], sorted(r[2])) # the set has an unstable output |
|
1582 | 1588 | (['r', 'p/p', ''], []) ['', 'p'] |
|
1583 | 1589 | >>> r = _rootsdirsandparents( |
|
1584 | 1590 | ... [(b'relglob', b'rg*', b''), (b're', b're/', b''), |
|
1585 | 1591 | ... (b'relre', b'rr', b'')]) |
|
1586 | 1592 | >>> print(r[0:2], sorted(r[2])) # the set has an unstable output |
|
1587 | 1593 | (['', '', ''], []) [''] |
|
1588 | 1594 | """ |
|
1589 | 1595 | r, d = _patternrootsanddirs(kindpats) |
|
1590 | 1596 | |
|
1591 | 1597 | p = set() |
|
1592 | 1598 | # Add the parents as non-recursive/exact directories, since they must be |
|
1593 | 1599 | # scanned to get to either the roots or the other exact directories. |
|
1594 | 1600 | p.update(pathutil.dirs(d)) |
|
1595 | 1601 | p.update(pathutil.dirs(r)) |
|
1596 | 1602 | |
|
1597 | 1603 | # FIXME: all uses of this function convert these to sets, do so before |
|
1598 | 1604 | # returning. |
|
1599 | 1605 | # FIXME: all uses of this function do not need anything in 'roots' and |
|
1600 | 1606 | # 'dirs' to also be in 'parents', consider removing them before returning. |
|
1601 | 1607 | return r, d, p |
|
1602 | 1608 | |
|
1603 | 1609 | |
|
1604 | 1610 | def _explicitfiles(kindpats): |
|
1605 | 1611 | """Returns the potential explicit filenames from the patterns. |
|
1606 | 1612 | |
|
1607 | 1613 | >>> _explicitfiles([(b'path', b'foo/bar', b'')]) |
|
1608 | 1614 | ['foo/bar'] |
|
1609 | 1615 | >>> _explicitfiles([(b'rootfilesin', b'foo/bar', b'')]) |
|
1610 | 1616 | [] |
|
1611 | 1617 | """ |
|
1612 | 1618 | # Keep only the pattern kinds where one can specify filenames (vs only |
|
1613 | 1619 | # directory names). |
|
1614 | 1620 | filable = [kp for kp in kindpats if kp[0] not in (b'rootfilesin',)] |
|
1615 | 1621 | return _roots(filable) |
|
1616 | 1622 | |
|
1617 | 1623 | |
|
1618 | 1624 | def _prefix(kindpats): |
|
1619 | 1625 | '''Whether all the patterns match a prefix (i.e. recursively)''' |
|
1620 | 1626 | for kind, pat, source in kindpats: |
|
1621 | 1627 | if kind not in (b'path', b'relpath'): |
|
1622 | 1628 | return False |
|
1623 | 1629 | return True |
|
1624 | 1630 | |
|
1625 | 1631 | |
|
1626 | 1632 | _commentre = None |
|
1627 | 1633 | |
|
1628 | 1634 | |
|
1629 | 1635 | def readpatternfile(filepath, warn, sourceinfo=False): |
|
1630 | 1636 | """parse a pattern file, returning a list of |
|
1631 | 1637 | patterns. These patterns should be given to compile() |
|
1632 | 1638 | to be validated and converted into a match function. |
|
1633 | 1639 | |
|
1634 | 1640 | trailing white space is dropped. |
|
1635 | 1641 | the escape character is backslash. |
|
1636 | 1642 | comments start with #. |
|
1637 | 1643 | empty lines are skipped. |
|
1638 | 1644 | |
|
1639 | 1645 | lines can be of the following formats: |
|
1640 | 1646 | |
|
1641 | 1647 | syntax: regexp # defaults following lines to non-rooted regexps |
|
1642 | 1648 | syntax: glob # defaults following lines to non-rooted globs |
|
1643 | 1649 | re:pattern # non-rooted regular expression |
|
1644 | 1650 | glob:pattern # non-rooted glob |
|
1645 | 1651 | rootglob:pat # rooted glob (same root as ^ in regexps) |
|
1646 | 1652 | pattern # pattern of the current default type |
|
1647 | 1653 | |
|
1648 | 1654 | if sourceinfo is set, returns a list of tuples: |
|
1649 | 1655 | (pattern, lineno, originalline). |
|
1650 | 1656 | This is useful to debug ignore patterns. |
|
1651 | 1657 | """ |
|
1652 | 1658 | |
|
1653 | 1659 | syntaxes = { |
|
1654 | 1660 | b're': b'relre:', |
|
1655 | 1661 | b'regexp': b'relre:', |
|
1656 | 1662 | b'glob': b'relglob:', |
|
1657 | 1663 | b'rootglob': b'rootglob:', |
|
1658 | 1664 | b'include': b'include', |
|
1659 | 1665 | b'subinclude': b'subinclude', |
|
1660 | 1666 | } |
|
1661 | 1667 | syntax = b'relre:' |
|
1662 | 1668 | patterns = [] |
|
1663 | 1669 | |
|
1664 | 1670 | fp = open(filepath, b'rb') |
|
1665 | 1671 | for lineno, line in enumerate(fp, start=1): |
|
1666 | 1672 | if b"#" in line: |
|
1667 | 1673 | global _commentre |
|
1668 | 1674 | if not _commentre: |
|
1669 | 1675 | _commentre = util.re.compile(br'((?:^|[^\\])(?:\\\\)*)#.*') |
|
1670 | 1676 | # remove comments prefixed by an even number of escapes |
|
1671 | 1677 | m = _commentre.search(line) |
|
1672 | 1678 | if m: |
|
1673 | 1679 | line = line[: m.end(1)] |
|
1674 | 1680 | # fixup properly escaped comments that survived the above |
|
1675 | 1681 | line = line.replace(b"\\#", b"#") |
|
1676 | 1682 | line = line.rstrip() |
|
1677 | 1683 | if not line: |
|
1678 | 1684 | continue |
|
1679 | 1685 | |
|
1680 | 1686 | if line.startswith(b'syntax:'): |
|
1681 | 1687 | s = line[7:].strip() |
|
1682 | 1688 | try: |
|
1683 | 1689 | syntax = syntaxes[s] |
|
1684 | 1690 | except KeyError: |
|
1685 | 1691 | if warn: |
|
1686 | 1692 | warn( |
|
1687 | 1693 | _(b"%s: ignoring invalid syntax '%s'\n") % (filepath, s) |
|
1688 | 1694 | ) |
|
1689 | 1695 | continue |
|
1690 | 1696 | |
|
1691 | 1697 | linesyntax = syntax |
|
1692 | 1698 | for s, rels in syntaxes.items(): |
|
1693 | 1699 | if line.startswith(rels): |
|
1694 | 1700 | linesyntax = rels |
|
1695 | 1701 | line = line[len(rels) :] |
|
1696 | 1702 | break |
|
1697 | 1703 | elif line.startswith(s + b':'): |
|
1698 | 1704 | linesyntax = rels |
|
1699 | 1705 | line = line[len(s) + 1 :] |
|
1700 | 1706 | break |
|
1701 | 1707 | if sourceinfo: |
|
1702 | 1708 | patterns.append((linesyntax + line, lineno, line)) |
|
1703 | 1709 | else: |
|
1704 | 1710 | patterns.append(linesyntax + line) |
|
1705 | 1711 | fp.close() |
|
1706 | 1712 | return patterns |
General Comments 0
You need to be logged in to leave comments.
Login now