Show More
@@ -1,345 +1,345 b'' | |||
|
1 | 1 | # narrowbundle2.py - bundle2 extensions for narrow repository support |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2017 Google, Inc. |
|
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 struct |
|
10 | 10 | |
|
11 | 11 | from mercurial.i18n import _ |
|
12 | 12 | from mercurial import ( |
|
13 | 13 | bundle2, |
|
14 | 14 | changegroup, |
|
15 | 15 | error, |
|
16 | 16 | exchange, |
|
17 | 17 | localrepo, |
|
18 | 18 | narrowspec, |
|
19 | 19 | repair, |
|
20 | 20 | requirements, |
|
21 | 21 | scmutil, |
|
22 | 22 | transaction, |
|
23 | 23 | util, |
|
24 | 24 | wireprototypes, |
|
25 | 25 | ) |
|
26 | 26 | |
|
27 | 27 | _NARROWACL_SECTION = b'narrowacl' |
|
28 | 28 | _CHANGESPECPART = b'narrow:changespec' |
|
29 | 29 | _RESSPECS = b'narrow:responsespec' |
|
30 | 30 | _SPECPART = b'narrow:spec' |
|
31 | 31 | _SPECPART_INCLUDE = b'include' |
|
32 | 32 | _SPECPART_EXCLUDE = b'exclude' |
|
33 | 33 | _KILLNODESIGNAL = b'KILL' |
|
34 | 34 | _DONESIGNAL = b'DONE' |
|
35 | 35 | _ELIDEDCSHEADER = b'>20s20s20sl' # cset id, p1, p2, len(text) |
|
36 | 36 | _ELIDEDMFHEADER = b'>20s20s20s20sl' # manifest id, p1, p2, link id, len(text) |
|
37 | 37 | _CSHEADERSIZE = struct.calcsize(_ELIDEDCSHEADER) |
|
38 | 38 | _MFHEADERSIZE = struct.calcsize(_ELIDEDMFHEADER) |
|
39 | 39 | |
|
40 | 40 | # Serve a changegroup for a client with a narrow clone. |
|
41 | 41 | def getbundlechangegrouppart_narrow( |
|
42 | 42 | bundler, |
|
43 | 43 | repo, |
|
44 | 44 | source, |
|
45 | 45 | bundlecaps=None, |
|
46 | 46 | b2caps=None, |
|
47 | 47 | heads=None, |
|
48 | 48 | common=None, |
|
49 | 49 | **kwargs |
|
50 | 50 | ): |
|
51 | 51 | assert repo.ui.configbool(b'experimental', b'narrowservebrokenellipses') |
|
52 | 52 | |
|
53 | 53 | cgversions = b2caps.get(b'changegroup') |
|
54 | 54 | cgversions = [ |
|
55 | 55 | v |
|
56 | 56 | for v in cgversions |
|
57 | 57 | if v in changegroup.supportedoutgoingversions(repo) |
|
58 | 58 | ] |
|
59 | 59 | if not cgversions: |
|
60 | 60 | raise ValueError(_(b'no common changegroup version')) |
|
61 | 61 | version = max(cgversions) |
|
62 | 62 | |
|
63 | 63 | include = sorted(filter(bool, kwargs.get('includepats', []))) |
|
64 | 64 | exclude = sorted(filter(bool, kwargs.get('excludepats', []))) |
|
65 | 65 | generateellipsesbundle2( |
|
66 | 66 | bundler, |
|
67 | 67 | repo, |
|
68 | 68 | include, |
|
69 | 69 | exclude, |
|
70 | 70 | version, |
|
71 | 71 | common, |
|
72 | 72 | heads, |
|
73 | 73 | kwargs.get('depth', None), |
|
74 | 74 | ) |
|
75 | 75 | |
|
76 | 76 | |
|
77 | 77 | def generateellipsesbundle2( |
|
78 | 78 | bundler, |
|
79 | 79 | repo, |
|
80 | 80 | include, |
|
81 | 81 | exclude, |
|
82 | 82 | version, |
|
83 | 83 | common, |
|
84 | 84 | heads, |
|
85 | 85 | depth, |
|
86 | 86 | ): |
|
87 | 87 | match = narrowspec.match(repo.root, include=include, exclude=exclude) |
|
88 | 88 | if depth is not None: |
|
89 | 89 | depth = int(depth) |
|
90 | 90 | if depth < 1: |
|
91 | 91 | raise error.Abort(_(b'depth must be positive, got %d') % depth) |
|
92 | 92 | |
|
93 | 93 | heads = set(heads or repo.heads()) |
|
94 | 94 | common = set(common or [repo.nullid]) |
|
95 | 95 | |
|
96 | 96 | visitnodes, relevant_nodes, ellipsisroots = exchange._computeellipsis( |
|
97 | 97 | repo, common, heads, set(), match, depth=depth |
|
98 | 98 | ) |
|
99 | 99 | |
|
100 | 100 | repo.ui.debug(b'Found %d relevant revs\n' % len(relevant_nodes)) |
|
101 | 101 | if visitnodes: |
|
102 | 102 | packer = changegroup.getbundler( |
|
103 | 103 | version, |
|
104 | 104 | repo, |
|
105 | 105 | matcher=match, |
|
106 | 106 | ellipses=True, |
|
107 | 107 | shallow=depth is not None, |
|
108 | 108 | ellipsisroots=ellipsisroots, |
|
109 | 109 | fullnodes=relevant_nodes, |
|
110 | 110 | ) |
|
111 | 111 | cgdata = packer.generate(common, visitnodes, False, b'narrow_widen') |
|
112 | 112 | |
|
113 | 113 | part = bundler.newpart(b'changegroup', data=cgdata) |
|
114 | 114 | part.addparam(b'version', version) |
|
115 | 115 | if scmutil.istreemanifest(repo): |
|
116 | 116 | part.addparam(b'treemanifest', b'1') |
|
117 | 117 | |
|
118 | 118 | |
|
119 | 119 | def generate_ellipses_bundle2_for_widening( |
|
120 | 120 | bundler, |
|
121 | 121 | repo, |
|
122 | 122 | oldmatch, |
|
123 | 123 | newmatch, |
|
124 | 124 | version, |
|
125 | 125 | common, |
|
126 | 126 | known, |
|
127 | 127 | ): |
|
128 | 128 | common = set(common or [repo.nullid]) |
|
129 | 129 | # Steps: |
|
130 | 130 | # 1. Send kill for "$known & ::common" |
|
131 | 131 | # |
|
132 | 132 | # 2. Send changegroup for ::common |
|
133 | 133 | # |
|
134 | 134 | # 3. Proceed. |
|
135 | 135 | # |
|
136 | 136 | # In the future, we can send kills for only the specific |
|
137 | 137 | # nodes we know should go away or change shape, and then |
|
138 | 138 | # send a data stream that tells the client something like this: |
|
139 | 139 | # |
|
140 | 140 | # a) apply this changegroup |
|
141 | 141 | # b) apply nodes XXX, YYY, ZZZ that you already have |
|
142 | 142 | # c) goto a |
|
143 | 143 | # |
|
144 | 144 | # until they've built up the full new state. |
|
145 | 145 | knownrevs = {repo.changelog.rev(n) for n in known} |
|
146 | 146 | # TODO: we could send only roots() of this set, and the |
|
147 | 147 | # list of nodes in common, and the client could work out |
|
148 | 148 | # what to strip, instead of us explicitly sending every |
|
149 | 149 | # single node. |
|
150 | 150 | deadrevs = knownrevs |
|
151 | 151 | |
|
152 | 152 | def genkills(): |
|
153 | 153 | for r in deadrevs: |
|
154 | 154 | yield _KILLNODESIGNAL |
|
155 | 155 | yield repo.changelog.node(r) |
|
156 | 156 | yield _DONESIGNAL |
|
157 | 157 | |
|
158 | 158 | bundler.newpart(_CHANGESPECPART, data=genkills()) |
|
159 | 159 | newvisit, newfull, newellipsis = exchange._computeellipsis( |
|
160 | 160 | repo, set(), common, knownrevs, newmatch |
|
161 | 161 | ) |
|
162 | 162 | if newvisit: |
|
163 | 163 | packer = changegroup.getbundler( |
|
164 | 164 | version, |
|
165 | 165 | repo, |
|
166 | 166 | matcher=newmatch, |
|
167 | 167 | ellipses=True, |
|
168 | 168 | shallow=False, |
|
169 | 169 | ellipsisroots=newellipsis, |
|
170 | 170 | fullnodes=newfull, |
|
171 | 171 | ) |
|
172 | 172 | cgdata = packer.generate(common, newvisit, False, b'narrow_widen') |
|
173 | 173 | |
|
174 | 174 | part = bundler.newpart(b'changegroup', data=cgdata) |
|
175 | 175 | part.addparam(b'version', version) |
|
176 | 176 | if scmutil.istreemanifest(repo): |
|
177 | 177 | part.addparam(b'treemanifest', b'1') |
|
178 | 178 | |
|
179 | 179 | |
|
180 | 180 | @bundle2.parthandler(_SPECPART, (_SPECPART_INCLUDE, _SPECPART_EXCLUDE)) |
|
181 | 181 | def _handlechangespec_2(op, inpart): |
|
182 | 182 | # XXX: This bundle2 handling is buggy and should be removed after hg5.2 is |
|
183 | 183 | # released. New servers will send a mandatory bundle2 part named |
|
184 | 184 | # 'Narrowspec' and will send specs as data instead of params. |
|
185 | 185 | # Refer to issue5952 and 6019 |
|
186 | 186 | includepats = set(inpart.params.get(_SPECPART_INCLUDE, b'').splitlines()) |
|
187 | 187 | excludepats = set(inpart.params.get(_SPECPART_EXCLUDE, b'').splitlines()) |
|
188 | 188 | narrowspec.validatepatterns(includepats) |
|
189 | 189 | narrowspec.validatepatterns(excludepats) |
|
190 | 190 | |
|
191 | 191 | if not requirements.NARROW_REQUIREMENT in op.repo.requirements: |
|
192 | 192 | op.repo.requirements.add(requirements.NARROW_REQUIREMENT) |
|
193 | 193 | scmutil.writereporequirements(op.repo) |
|
194 | 194 | op.repo.setnarrowpats(includepats, excludepats) |
|
195 | 195 | narrowspec.copytoworkingcopy(op.repo) |
|
196 | 196 | |
|
197 | 197 | |
|
198 | 198 | @bundle2.parthandler(_RESSPECS) |
|
199 | 199 | def _handlenarrowspecs(op, inpart): |
|
200 | 200 | data = inpart.read() |
|
201 | 201 | inc, exc = data.split(b'\0') |
|
202 | 202 | includepats = set(inc.splitlines()) |
|
203 | 203 | excludepats = set(exc.splitlines()) |
|
204 | 204 | narrowspec.validatepatterns(includepats) |
|
205 | 205 | narrowspec.validatepatterns(excludepats) |
|
206 | 206 | |
|
207 | 207 | if requirements.NARROW_REQUIREMENT not in op.repo.requirements: |
|
208 | 208 | op.repo.requirements.add(requirements.NARROW_REQUIREMENT) |
|
209 | 209 | scmutil.writereporequirements(op.repo) |
|
210 | 210 | op.repo.setnarrowpats(includepats, excludepats) |
|
211 | 211 | narrowspec.copytoworkingcopy(op.repo) |
|
212 | 212 | |
|
213 | 213 | |
|
214 | 214 | @bundle2.parthandler(_CHANGESPECPART) |
|
215 | 215 | def _handlechangespec(op, inpart): |
|
216 | 216 | repo = op.repo |
|
217 | 217 | cl = repo.changelog |
|
218 | 218 | |
|
219 | 219 | # changesets which need to be stripped entirely. either they're no longer |
|
220 | 220 | # needed in the new narrow spec, or the server is sending a replacement |
|
221 | 221 | # in the changegroup part. |
|
222 | 222 | clkills = set() |
|
223 | 223 | |
|
224 | 224 | # A changespec part contains all the updates to ellipsis nodes |
|
225 | 225 | # that will happen as a result of widening or narrowing a |
|
226 | 226 | # repo. All the changes that this block encounters are ellipsis |
|
227 | 227 | # nodes or flags to kill an existing ellipsis. |
|
228 | 228 | chunksignal = changegroup.readexactly(inpart, 4) |
|
229 | 229 | while chunksignal != _DONESIGNAL: |
|
230 | 230 | if chunksignal == _KILLNODESIGNAL: |
|
231 | 231 | # a node used to be an ellipsis but isn't anymore |
|
232 | 232 | ck = changegroup.readexactly(inpart, 20) |
|
233 | 233 | if cl.hasnode(ck): |
|
234 | 234 | clkills.add(ck) |
|
235 | 235 | else: |
|
236 | 236 | raise error.Abort( |
|
237 | 237 | _(b'unexpected changespec node chunk type: %s') % chunksignal |
|
238 | 238 | ) |
|
239 | 239 | chunksignal = changegroup.readexactly(inpart, 4) |
|
240 | 240 | |
|
241 | 241 | if clkills: |
|
242 | 242 | # preserve bookmarks that repair.strip() would otherwise strip |
|
243 | 243 | op._bookmarksbackup = repo._bookmarks |
|
244 | 244 | |
|
245 | 245 | class dummybmstore(dict): |
|
246 | 246 | def applychanges(self, repo, tr, changes): |
|
247 | 247 | pass |
|
248 | 248 | |
|
249 | 249 | localrepo.localrepository._bookmarks.set(repo, dummybmstore()) |
|
250 | 250 | chgrpfile = repair.strip( |
|
251 | 251 | op.ui, repo, list(clkills), backup=True, topic=b'widen' |
|
252 | 252 | ) |
|
253 | 253 | if chgrpfile: |
|
254 | 254 | op._widen_uninterr = repo.ui.uninterruptible() |
|
255 | 255 | op._widen_uninterr.__enter__() |
|
256 | 256 | # presence of _widen_bundle attribute activates widen handler later |
|
257 | 257 | op._widen_bundle = chgrpfile |
|
258 | 258 | # Set the new narrowspec if we're widening. The setnewnarrowpats() method |
|
259 | 259 | # will currently always be there when using the core+narrowhg server, but |
|
260 | 260 | # other servers may include a changespec part even when not widening (e.g. |
|
261 | 261 | # because we're deepening a shallow repo). |
|
262 | 262 | if util.safehasattr(repo, 'setnewnarrowpats'): |
|
263 | 263 | op.gettransaction() |
|
264 | 264 | repo.setnewnarrowpats() |
|
265 | 265 | |
|
266 | 266 | |
|
267 | 267 | def handlechangegroup_widen(op, inpart): |
|
268 | 268 | """Changegroup exchange handler which restores temporarily-stripped nodes""" |
|
269 | 269 | # We saved a bundle with stripped node data we must now restore. |
|
270 | 270 | # This approach is based on mercurial/repair.py@6ee26a53c111. |
|
271 | 271 | repo = op.repo |
|
272 | 272 | ui = op.ui |
|
273 | 273 | |
|
274 | 274 | chgrpfile = op._widen_bundle |
|
275 | 275 | del op._widen_bundle |
|
276 | 276 | vfs = repo.vfs |
|
277 | 277 | |
|
278 | 278 | ui.note(_(b"adding branch\n")) |
|
279 | 279 | f = vfs.open(chgrpfile, b"rb") |
|
280 | 280 | try: |
|
281 | 281 | gen = exchange.readbundle(ui, f, chgrpfile, vfs) |
|
282 | 282 | # silence internal shuffling chatter |
|
283 | 283 | maybe_silent = ( |
|
284 | 284 | ui.silent() if not ui.verbose else util.nullcontextmanager() |
|
285 | 285 | ) |
|
286 | 286 | with maybe_silent: |
|
287 | 287 | if isinstance(gen, bundle2.unbundle20): |
|
288 | 288 | with repo.transaction(b'strip') as tr: |
|
289 | 289 | bundle2.processbundle(repo, gen, lambda: tr) |
|
290 | 290 | else: |
|
291 | 291 | gen.apply( |
|
292 | 292 | repo, b'strip', b'bundle:' + vfs.join(chgrpfile), True |
|
293 | 293 | ) |
|
294 | 294 | finally: |
|
295 | 295 | f.close() |
|
296 | 296 | |
|
297 | transaction.cleanup_undo_files(repo) | |
|
297 | transaction.cleanup_undo_files(repo.ui.warn, repo.vfs_map) | |
|
298 | 298 | |
|
299 | 299 | # Remove partial backup only if there were no exceptions |
|
300 | 300 | op._widen_uninterr.__exit__(None, None, None) |
|
301 | 301 | vfs.unlink(chgrpfile) |
|
302 | 302 | |
|
303 | 303 | |
|
304 | 304 | def setup(): |
|
305 | 305 | """Enable narrow repo support in bundle2-related extension points.""" |
|
306 | 306 | getbundleargs = wireprototypes.GETBUNDLE_ARGUMENTS |
|
307 | 307 | |
|
308 | 308 | getbundleargs[b'narrow'] = b'boolean' |
|
309 | 309 | getbundleargs[b'depth'] = b'plain' |
|
310 | 310 | getbundleargs[b'oldincludepats'] = b'csv' |
|
311 | 311 | getbundleargs[b'oldexcludepats'] = b'csv' |
|
312 | 312 | getbundleargs[b'known'] = b'csv' |
|
313 | 313 | |
|
314 | 314 | # Extend changegroup serving to handle requests from narrow clients. |
|
315 | 315 | origcgfn = exchange.getbundle2partsmapping[b'changegroup'] |
|
316 | 316 | |
|
317 | 317 | def wrappedcgfn(*args, **kwargs): |
|
318 | 318 | repo = args[1] |
|
319 | 319 | if repo.ui.has_section(_NARROWACL_SECTION): |
|
320 | 320 | kwargs = exchange.applynarrowacl(repo, kwargs) |
|
321 | 321 | |
|
322 | 322 | if kwargs.get('narrow', False) and repo.ui.configbool( |
|
323 | 323 | b'experimental', b'narrowservebrokenellipses' |
|
324 | 324 | ): |
|
325 | 325 | getbundlechangegrouppart_narrow(*args, **kwargs) |
|
326 | 326 | else: |
|
327 | 327 | origcgfn(*args, **kwargs) |
|
328 | 328 | |
|
329 | 329 | exchange.getbundle2partsmapping[b'changegroup'] = wrappedcgfn |
|
330 | 330 | |
|
331 | 331 | # Extend changegroup receiver so client can fixup after widen requests. |
|
332 | 332 | origcghandler = bundle2.parthandlermapping[b'changegroup'] |
|
333 | 333 | |
|
334 | 334 | def wrappedcghandler(op, inpart): |
|
335 | 335 | origcghandler(op, inpart) |
|
336 | 336 | if util.safehasattr(op, '_widen_bundle'): |
|
337 | 337 | handlechangegroup_widen(op, inpart) |
|
338 | 338 | if util.safehasattr(op, '_bookmarksbackup'): |
|
339 | 339 | localrepo.localrepository._bookmarks.set( |
|
340 | 340 | op.repo, op._bookmarksbackup |
|
341 | 341 | ) |
|
342 | 342 | del op._bookmarksbackup |
|
343 | 343 | |
|
344 | 344 | wrappedcghandler.params = origcghandler.params |
|
345 | 345 | bundle2.parthandlermapping[b'changegroup'] = wrappedcghandler |
@@ -1,547 +1,547 b'' | |||
|
1 | 1 | # repair.py - functions for repository repair for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005, 2006 Chris Mason <mason@suse.com> |
|
4 | 4 | # Copyright 2007 Olivia Mackall |
|
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 | |
|
10 | 10 | from .i18n import _ |
|
11 | 11 | from .node import ( |
|
12 | 12 | hex, |
|
13 | 13 | short, |
|
14 | 14 | ) |
|
15 | 15 | from . import ( |
|
16 | 16 | bundle2, |
|
17 | 17 | changegroup, |
|
18 | 18 | discovery, |
|
19 | 19 | error, |
|
20 | 20 | exchange, |
|
21 | 21 | obsolete, |
|
22 | 22 | obsutil, |
|
23 | 23 | pathutil, |
|
24 | 24 | phases, |
|
25 | 25 | requirements, |
|
26 | 26 | scmutil, |
|
27 | 27 | transaction, |
|
28 | 28 | util, |
|
29 | 29 | ) |
|
30 | 30 | from .utils import ( |
|
31 | 31 | hashutil, |
|
32 | 32 | urlutil, |
|
33 | 33 | ) |
|
34 | 34 | |
|
35 | 35 | |
|
36 | 36 | def backupbundle( |
|
37 | 37 | repo, bases, heads, node, suffix, compress=True, obsolescence=True |
|
38 | 38 | ): |
|
39 | 39 | """create a bundle with the specified revisions as a backup""" |
|
40 | 40 | |
|
41 | 41 | backupdir = b"strip-backup" |
|
42 | 42 | vfs = repo.vfs |
|
43 | 43 | if not vfs.isdir(backupdir): |
|
44 | 44 | vfs.mkdir(backupdir) |
|
45 | 45 | |
|
46 | 46 | # Include a hash of all the nodes in the filename for uniqueness |
|
47 | 47 | allcommits = repo.set(b'%ln::%ln', bases, heads) |
|
48 | 48 | allhashes = sorted(c.hex() for c in allcommits) |
|
49 | 49 | totalhash = hashutil.sha1(b''.join(allhashes)).digest() |
|
50 | 50 | name = b"%s/%s-%s-%s.hg" % ( |
|
51 | 51 | backupdir, |
|
52 | 52 | short(node), |
|
53 | 53 | hex(totalhash[:4]), |
|
54 | 54 | suffix, |
|
55 | 55 | ) |
|
56 | 56 | |
|
57 | 57 | cgversion = changegroup.localversion(repo) |
|
58 | 58 | comp = None |
|
59 | 59 | if cgversion != b'01': |
|
60 | 60 | bundletype = b"HG20" |
|
61 | 61 | if compress: |
|
62 | 62 | comp = b'BZ' |
|
63 | 63 | elif compress: |
|
64 | 64 | bundletype = b"HG10BZ" |
|
65 | 65 | else: |
|
66 | 66 | bundletype = b"HG10UN" |
|
67 | 67 | |
|
68 | 68 | outgoing = discovery.outgoing(repo, missingroots=bases, ancestorsof=heads) |
|
69 | 69 | contentopts = { |
|
70 | 70 | b'cg.version': cgversion, |
|
71 | 71 | b'obsolescence': obsolescence, |
|
72 | 72 | b'phases': True, |
|
73 | 73 | } |
|
74 | 74 | return bundle2.writenewbundle( |
|
75 | 75 | repo.ui, |
|
76 | 76 | repo, |
|
77 | 77 | b'strip', |
|
78 | 78 | name, |
|
79 | 79 | bundletype, |
|
80 | 80 | outgoing, |
|
81 | 81 | contentopts, |
|
82 | 82 | vfs, |
|
83 | 83 | compression=comp, |
|
84 | 84 | ) |
|
85 | 85 | |
|
86 | 86 | |
|
87 | 87 | def _collectfiles(repo, striprev): |
|
88 | 88 | """find out the filelogs affected by the strip""" |
|
89 | 89 | files = set() |
|
90 | 90 | |
|
91 | 91 | for x in range(striprev, len(repo)): |
|
92 | 92 | files.update(repo[x].files()) |
|
93 | 93 | |
|
94 | 94 | return sorted(files) |
|
95 | 95 | |
|
96 | 96 | |
|
97 | 97 | def _collectrevlog(revlog, striprev): |
|
98 | 98 | _, brokenset = revlog.getstrippoint(striprev) |
|
99 | 99 | return [revlog.linkrev(r) for r in brokenset] |
|
100 | 100 | |
|
101 | 101 | |
|
102 | 102 | def _collectbrokencsets(repo, files, striprev): |
|
103 | 103 | """return the changesets which will be broken by the truncation""" |
|
104 | 104 | s = set() |
|
105 | 105 | |
|
106 | 106 | for revlog in manifestrevlogs(repo): |
|
107 | 107 | s.update(_collectrevlog(revlog, striprev)) |
|
108 | 108 | for fname in files: |
|
109 | 109 | s.update(_collectrevlog(repo.file(fname), striprev)) |
|
110 | 110 | |
|
111 | 111 | return s |
|
112 | 112 | |
|
113 | 113 | |
|
114 | 114 | def strip(ui, repo, nodelist, backup=True, topic=b'backup'): |
|
115 | 115 | # This function requires the caller to lock the repo, but it operates |
|
116 | 116 | # within a transaction of its own, and thus requires there to be no current |
|
117 | 117 | # transaction when it is called. |
|
118 | 118 | if repo.currenttransaction() is not None: |
|
119 | 119 | raise error.ProgrammingError(b'cannot strip from inside a transaction') |
|
120 | 120 | |
|
121 | 121 | # Simple way to maintain backwards compatibility for this |
|
122 | 122 | # argument. |
|
123 | 123 | if backup in [b'none', b'strip']: |
|
124 | 124 | backup = False |
|
125 | 125 | |
|
126 | 126 | repo = repo.unfiltered() |
|
127 | 127 | repo.destroying() |
|
128 | 128 | vfs = repo.vfs |
|
129 | 129 | # load bookmark before changelog to avoid side effect from outdated |
|
130 | 130 | # changelog (see repo._refreshchangelog) |
|
131 | 131 | repo._bookmarks |
|
132 | 132 | cl = repo.changelog |
|
133 | 133 | |
|
134 | 134 | # TODO handle undo of merge sets |
|
135 | 135 | if isinstance(nodelist, bytes): |
|
136 | 136 | nodelist = [nodelist] |
|
137 | 137 | striplist = [cl.rev(node) for node in nodelist] |
|
138 | 138 | striprev = min(striplist) |
|
139 | 139 | |
|
140 | 140 | files = _collectfiles(repo, striprev) |
|
141 | 141 | saverevs = _collectbrokencsets(repo, files, striprev) |
|
142 | 142 | |
|
143 | 143 | # Some revisions with rev > striprev may not be descendants of striprev. |
|
144 | 144 | # We have to find these revisions and put them in a bundle, so that |
|
145 | 145 | # we can restore them after the truncations. |
|
146 | 146 | # To create the bundle we use repo.changegroupsubset which requires |
|
147 | 147 | # the list of heads and bases of the set of interesting revisions. |
|
148 | 148 | # (head = revision in the set that has no descendant in the set; |
|
149 | 149 | # base = revision in the set that has no ancestor in the set) |
|
150 | 150 | tostrip = set(striplist) |
|
151 | 151 | saveheads = set(saverevs) |
|
152 | 152 | for r in cl.revs(start=striprev + 1): |
|
153 | 153 | if any(p in tostrip for p in cl.parentrevs(r)): |
|
154 | 154 | tostrip.add(r) |
|
155 | 155 | |
|
156 | 156 | if r not in tostrip: |
|
157 | 157 | saverevs.add(r) |
|
158 | 158 | saveheads.difference_update(cl.parentrevs(r)) |
|
159 | 159 | saveheads.add(r) |
|
160 | 160 | saveheads = [cl.node(r) for r in saveheads] |
|
161 | 161 | |
|
162 | 162 | # compute base nodes |
|
163 | 163 | if saverevs: |
|
164 | 164 | descendants = set(cl.descendants(saverevs)) |
|
165 | 165 | saverevs.difference_update(descendants) |
|
166 | 166 | savebases = [cl.node(r) for r in saverevs] |
|
167 | 167 | stripbases = [cl.node(r) for r in tostrip] |
|
168 | 168 | |
|
169 | 169 | stripobsidx = obsmarkers = () |
|
170 | 170 | if repo.ui.configbool(b'devel', b'strip-obsmarkers'): |
|
171 | 171 | obsmarkers = obsutil.exclusivemarkers(repo, stripbases) |
|
172 | 172 | if obsmarkers: |
|
173 | 173 | stripobsidx = [ |
|
174 | 174 | i for i, m in enumerate(repo.obsstore) if m in obsmarkers |
|
175 | 175 | ] |
|
176 | 176 | |
|
177 | 177 | newbmtarget, updatebm = _bookmarkmovements(repo, tostrip) |
|
178 | 178 | |
|
179 | 179 | backupfile = None |
|
180 | 180 | node = nodelist[-1] |
|
181 | 181 | if backup: |
|
182 | 182 | backupfile = _createstripbackup(repo, stripbases, node, topic) |
|
183 | 183 | # create a changegroup for all the branches we need to keep |
|
184 | 184 | tmpbundlefile = None |
|
185 | 185 | if saveheads: |
|
186 | 186 | # do not compress temporary bundle if we remove it from disk later |
|
187 | 187 | # |
|
188 | 188 | # We do not include obsolescence, it might re-introduce prune markers |
|
189 | 189 | # we are trying to strip. This is harmless since the stripped markers |
|
190 | 190 | # are already backed up and we did not touched the markers for the |
|
191 | 191 | # saved changesets. |
|
192 | 192 | tmpbundlefile = backupbundle( |
|
193 | 193 | repo, |
|
194 | 194 | savebases, |
|
195 | 195 | saveheads, |
|
196 | 196 | node, |
|
197 | 197 | b'temp', |
|
198 | 198 | compress=False, |
|
199 | 199 | obsolescence=False, |
|
200 | 200 | ) |
|
201 | 201 | |
|
202 | 202 | with ui.uninterruptible(): |
|
203 | 203 | try: |
|
204 | 204 | with repo.transaction(b"strip") as tr: |
|
205 | 205 | # TODO this code violates the interface abstraction of the |
|
206 | 206 | # transaction and makes assumptions that file storage is |
|
207 | 207 | # using append-only files. We'll need some kind of storage |
|
208 | 208 | # API to handle stripping for us. |
|
209 | 209 | oldfiles = set(tr._offsetmap.keys()) |
|
210 | 210 | oldfiles.update(tr._newfiles) |
|
211 | 211 | |
|
212 | 212 | tr.startgroup() |
|
213 | 213 | cl.strip(striprev, tr) |
|
214 | 214 | stripmanifest(repo, striprev, tr, files) |
|
215 | 215 | |
|
216 | 216 | for fn in files: |
|
217 | 217 | repo.file(fn).strip(striprev, tr) |
|
218 | 218 | tr.endgroup() |
|
219 | 219 | |
|
220 | 220 | entries = tr.readjournal() |
|
221 | 221 | |
|
222 | 222 | for file, troffset in entries: |
|
223 | 223 | if file in oldfiles: |
|
224 | 224 | continue |
|
225 | 225 | with repo.svfs(file, b'a', checkambig=True) as fp: |
|
226 | 226 | fp.truncate(troffset) |
|
227 | 227 | if troffset == 0: |
|
228 | 228 | repo.store.markremoved(file) |
|
229 | 229 | |
|
230 | 230 | deleteobsmarkers(repo.obsstore, stripobsidx) |
|
231 | 231 | del repo.obsstore |
|
232 | 232 | repo.invalidatevolatilesets() |
|
233 | 233 | repo._phasecache.filterunknown(repo) |
|
234 | 234 | |
|
235 | 235 | if tmpbundlefile: |
|
236 | 236 | ui.note(_(b"adding branch\n")) |
|
237 | 237 | f = vfs.open(tmpbundlefile, b"rb") |
|
238 | 238 | gen = exchange.readbundle(ui, f, tmpbundlefile, vfs) |
|
239 | 239 | # silence internal shuffling chatter |
|
240 | 240 | maybe_silent = ( |
|
241 | 241 | repo.ui.silent() |
|
242 | 242 | if not repo.ui.verbose |
|
243 | 243 | else util.nullcontextmanager() |
|
244 | 244 | ) |
|
245 | 245 | with maybe_silent: |
|
246 | 246 | tmpbundleurl = b'bundle:' + vfs.join(tmpbundlefile) |
|
247 | 247 | txnname = b'strip' |
|
248 | 248 | if not isinstance(gen, bundle2.unbundle20): |
|
249 | 249 | txnname = b"strip\n%s" % urlutil.hidepassword( |
|
250 | 250 | tmpbundleurl |
|
251 | 251 | ) |
|
252 | 252 | with repo.transaction(txnname) as tr: |
|
253 | 253 | bundle2.applybundle( |
|
254 | 254 | repo, gen, tr, source=b'strip', url=tmpbundleurl |
|
255 | 255 | ) |
|
256 | 256 | f.close() |
|
257 | 257 | |
|
258 | 258 | with repo.transaction(b'repair') as tr: |
|
259 | 259 | bmchanges = [(m, repo[newbmtarget].node()) for m in updatebm] |
|
260 | 260 | repo._bookmarks.applychanges(repo, tr, bmchanges) |
|
261 | 261 | |
|
262 | transaction.cleanup_undo_files(repo) | |
|
262 | transaction.cleanup_undo_files(repo.ui.warn, repo.vfs_map) | |
|
263 | 263 | |
|
264 | 264 | except: # re-raises |
|
265 | 265 | if backupfile: |
|
266 | 266 | ui.warn( |
|
267 | 267 | _(b"strip failed, backup bundle stored in '%s'\n") |
|
268 | 268 | % vfs.join(backupfile) |
|
269 | 269 | ) |
|
270 | 270 | if tmpbundlefile: |
|
271 | 271 | ui.warn( |
|
272 | 272 | _(b"strip failed, unrecovered changes stored in '%s'\n") |
|
273 | 273 | % vfs.join(tmpbundlefile) |
|
274 | 274 | ) |
|
275 | 275 | ui.warn( |
|
276 | 276 | _( |
|
277 | 277 | b"(fix the problem, then recover the changesets with " |
|
278 | 278 | b"\"hg unbundle '%s'\")\n" |
|
279 | 279 | ) |
|
280 | 280 | % vfs.join(tmpbundlefile) |
|
281 | 281 | ) |
|
282 | 282 | raise |
|
283 | 283 | else: |
|
284 | 284 | if tmpbundlefile: |
|
285 | 285 | # Remove temporary bundle only if there were no exceptions |
|
286 | 286 | vfs.unlink(tmpbundlefile) |
|
287 | 287 | |
|
288 | 288 | repo.destroyed() |
|
289 | 289 | # return the backup file path (or None if 'backup' was False) so |
|
290 | 290 | # extensions can use it |
|
291 | 291 | return backupfile |
|
292 | 292 | |
|
293 | 293 | |
|
294 | 294 | def softstrip(ui, repo, nodelist, backup=True, topic=b'backup'): |
|
295 | 295 | """perform a "soft" strip using the archived phase""" |
|
296 | 296 | tostrip = [c.node() for c in repo.set(b'sort(%ln::)', nodelist)] |
|
297 | 297 | if not tostrip: |
|
298 | 298 | return None |
|
299 | 299 | |
|
300 | 300 | backupfile = None |
|
301 | 301 | if backup: |
|
302 | 302 | node = tostrip[0] |
|
303 | 303 | backupfile = _createstripbackup(repo, tostrip, node, topic) |
|
304 | 304 | |
|
305 | 305 | newbmtarget, updatebm = _bookmarkmovements(repo, tostrip) |
|
306 | 306 | with repo.transaction(b'strip') as tr: |
|
307 | 307 | phases.retractboundary(repo, tr, phases.archived, tostrip) |
|
308 | 308 | bmchanges = [(m, repo[newbmtarget].node()) for m in updatebm] |
|
309 | 309 | repo._bookmarks.applychanges(repo, tr, bmchanges) |
|
310 | 310 | return backupfile |
|
311 | 311 | |
|
312 | 312 | |
|
313 | 313 | def _bookmarkmovements(repo, tostrip): |
|
314 | 314 | # compute necessary bookmark movement |
|
315 | 315 | bm = repo._bookmarks |
|
316 | 316 | updatebm = [] |
|
317 | 317 | for m in bm: |
|
318 | 318 | rev = repo[bm[m]].rev() |
|
319 | 319 | if rev in tostrip: |
|
320 | 320 | updatebm.append(m) |
|
321 | 321 | newbmtarget = None |
|
322 | 322 | # If we need to move bookmarks, compute bookmark |
|
323 | 323 | # targets. Otherwise we can skip doing this logic. |
|
324 | 324 | if updatebm: |
|
325 | 325 | # For a set s, max(parents(s) - s) is the same as max(heads(::s - s)), |
|
326 | 326 | # but is much faster |
|
327 | 327 | newbmtarget = repo.revs(b'max(parents(%ld) - (%ld))', tostrip, tostrip) |
|
328 | 328 | if newbmtarget: |
|
329 | 329 | newbmtarget = repo[newbmtarget.first()].node() |
|
330 | 330 | else: |
|
331 | 331 | newbmtarget = b'.' |
|
332 | 332 | return newbmtarget, updatebm |
|
333 | 333 | |
|
334 | 334 | |
|
335 | 335 | def _createstripbackup(repo, stripbases, node, topic): |
|
336 | 336 | # backup the changeset we are about to strip |
|
337 | 337 | vfs = repo.vfs |
|
338 | 338 | cl = repo.changelog |
|
339 | 339 | backupfile = backupbundle(repo, stripbases, cl.heads(), node, topic) |
|
340 | 340 | repo.ui.status(_(b"saved backup bundle to %s\n") % vfs.join(backupfile)) |
|
341 | 341 | repo.ui.log( |
|
342 | 342 | b"backupbundle", b"saved backup bundle to %s\n", vfs.join(backupfile) |
|
343 | 343 | ) |
|
344 | 344 | return backupfile |
|
345 | 345 | |
|
346 | 346 | |
|
347 | 347 | def safestriproots(ui, repo, nodes): |
|
348 | 348 | """return list of roots of nodes where descendants are covered by nodes""" |
|
349 | 349 | torev = repo.unfiltered().changelog.rev |
|
350 | 350 | revs = {torev(n) for n in nodes} |
|
351 | 351 | # tostrip = wanted - unsafe = wanted - ancestors(orphaned) |
|
352 | 352 | # orphaned = affected - wanted |
|
353 | 353 | # affected = descendants(roots(wanted)) |
|
354 | 354 | # wanted = revs |
|
355 | 355 | revset = b'%ld - ( ::( (roots(%ld):: and not _phase(%s)) -%ld) )' |
|
356 | 356 | tostrip = set(repo.revs(revset, revs, revs, phases.internal, revs)) |
|
357 | 357 | notstrip = revs - tostrip |
|
358 | 358 | if notstrip: |
|
359 | 359 | nodestr = b', '.join(sorted(short(repo[n].node()) for n in notstrip)) |
|
360 | 360 | ui.warn( |
|
361 | 361 | _(b'warning: orphaned descendants detected, not stripping %s\n') |
|
362 | 362 | % nodestr |
|
363 | 363 | ) |
|
364 | 364 | return [c.node() for c in repo.set(b'roots(%ld)', tostrip)] |
|
365 | 365 | |
|
366 | 366 | |
|
367 | 367 | class stripcallback: |
|
368 | 368 | """used as a transaction postclose callback""" |
|
369 | 369 | |
|
370 | 370 | def __init__(self, ui, repo, backup, topic): |
|
371 | 371 | self.ui = ui |
|
372 | 372 | self.repo = repo |
|
373 | 373 | self.backup = backup |
|
374 | 374 | self.topic = topic or b'backup' |
|
375 | 375 | self.nodelist = [] |
|
376 | 376 | |
|
377 | 377 | def addnodes(self, nodes): |
|
378 | 378 | self.nodelist.extend(nodes) |
|
379 | 379 | |
|
380 | 380 | def __call__(self, tr): |
|
381 | 381 | roots = safestriproots(self.ui, self.repo, self.nodelist) |
|
382 | 382 | if roots: |
|
383 | 383 | strip(self.ui, self.repo, roots, self.backup, self.topic) |
|
384 | 384 | |
|
385 | 385 | |
|
386 | 386 | def delayedstrip(ui, repo, nodelist, topic=None, backup=True): |
|
387 | 387 | """like strip, but works inside transaction and won't strip irreverent revs |
|
388 | 388 | |
|
389 | 389 | nodelist must explicitly contain all descendants. Otherwise a warning will |
|
390 | 390 | be printed that some nodes are not stripped. |
|
391 | 391 | |
|
392 | 392 | Will do a backup if `backup` is True. The last non-None "topic" will be |
|
393 | 393 | used as the backup topic name. The default backup topic name is "backup". |
|
394 | 394 | """ |
|
395 | 395 | tr = repo.currenttransaction() |
|
396 | 396 | if not tr: |
|
397 | 397 | nodes = safestriproots(ui, repo, nodelist) |
|
398 | 398 | return strip(ui, repo, nodes, backup=backup, topic=topic) |
|
399 | 399 | # transaction postclose callbacks are called in alphabet order. |
|
400 | 400 | # use '\xff' as prefix so we are likely to be called last. |
|
401 | 401 | callback = tr.getpostclose(b'\xffstrip') |
|
402 | 402 | if callback is None: |
|
403 | 403 | callback = stripcallback(ui, repo, backup=backup, topic=topic) |
|
404 | 404 | tr.addpostclose(b'\xffstrip', callback) |
|
405 | 405 | if topic: |
|
406 | 406 | callback.topic = topic |
|
407 | 407 | callback.addnodes(nodelist) |
|
408 | 408 | |
|
409 | 409 | |
|
410 | 410 | def stripmanifest(repo, striprev, tr, files): |
|
411 | 411 | for revlog in manifestrevlogs(repo): |
|
412 | 412 | revlog.strip(striprev, tr) |
|
413 | 413 | |
|
414 | 414 | |
|
415 | 415 | def manifestrevlogs(repo): |
|
416 | 416 | yield repo.manifestlog.getstorage(b'') |
|
417 | 417 | if scmutil.istreemanifest(repo): |
|
418 | 418 | # This logic is safe if treemanifest isn't enabled, but also |
|
419 | 419 | # pointless, so we skip it if treemanifest isn't enabled. |
|
420 | 420 | for t, unencoded, size in repo.store.datafiles(): |
|
421 | 421 | if unencoded.startswith(b'meta/') and unencoded.endswith( |
|
422 | 422 | b'00manifest.i' |
|
423 | 423 | ): |
|
424 | 424 | dir = unencoded[5:-12] |
|
425 | 425 | yield repo.manifestlog.getstorage(dir) |
|
426 | 426 | |
|
427 | 427 | |
|
428 | 428 | def rebuildfncache(ui, repo, only_data=False): |
|
429 | 429 | """Rebuilds the fncache file from repo history. |
|
430 | 430 | |
|
431 | 431 | Missing entries will be added. Extra entries will be removed. |
|
432 | 432 | """ |
|
433 | 433 | repo = repo.unfiltered() |
|
434 | 434 | |
|
435 | 435 | if requirements.FNCACHE_REQUIREMENT not in repo.requirements: |
|
436 | 436 | ui.warn( |
|
437 | 437 | _( |
|
438 | 438 | b'(not rebuilding fncache because repository does not ' |
|
439 | 439 | b'support fncache)\n' |
|
440 | 440 | ) |
|
441 | 441 | ) |
|
442 | 442 | return |
|
443 | 443 | |
|
444 | 444 | with repo.lock(): |
|
445 | 445 | fnc = repo.store.fncache |
|
446 | 446 | fnc.ensureloaded(warn=ui.warn) |
|
447 | 447 | |
|
448 | 448 | oldentries = set(fnc.entries) |
|
449 | 449 | newentries = set() |
|
450 | 450 | seenfiles = set() |
|
451 | 451 | |
|
452 | 452 | if only_data: |
|
453 | 453 | # Trust the listing of .i from the fncache, but not the .d. This is |
|
454 | 454 | # much faster, because we only need to stat every possible .d files, |
|
455 | 455 | # instead of reading the full changelog |
|
456 | 456 | for f in fnc: |
|
457 | 457 | if f[:5] == b'data/' and f[-2:] == b'.i': |
|
458 | 458 | seenfiles.add(f[5:-2]) |
|
459 | 459 | newentries.add(f) |
|
460 | 460 | dataf = f[:-2] + b'.d' |
|
461 | 461 | if repo.store._exists(dataf): |
|
462 | 462 | newentries.add(dataf) |
|
463 | 463 | else: |
|
464 | 464 | progress = ui.makeprogress( |
|
465 | 465 | _(b'rebuilding'), unit=_(b'changesets'), total=len(repo) |
|
466 | 466 | ) |
|
467 | 467 | for rev in repo: |
|
468 | 468 | progress.update(rev) |
|
469 | 469 | |
|
470 | 470 | ctx = repo[rev] |
|
471 | 471 | for f in ctx.files(): |
|
472 | 472 | # This is to minimize I/O. |
|
473 | 473 | if f in seenfiles: |
|
474 | 474 | continue |
|
475 | 475 | seenfiles.add(f) |
|
476 | 476 | |
|
477 | 477 | i = b'data/%s.i' % f |
|
478 | 478 | d = b'data/%s.d' % f |
|
479 | 479 | |
|
480 | 480 | if repo.store._exists(i): |
|
481 | 481 | newentries.add(i) |
|
482 | 482 | if repo.store._exists(d): |
|
483 | 483 | newentries.add(d) |
|
484 | 484 | |
|
485 | 485 | progress.complete() |
|
486 | 486 | |
|
487 | 487 | if requirements.TREEMANIFEST_REQUIREMENT in repo.requirements: |
|
488 | 488 | # This logic is safe if treemanifest isn't enabled, but also |
|
489 | 489 | # pointless, so we skip it if treemanifest isn't enabled. |
|
490 | 490 | for dir in pathutil.dirs(seenfiles): |
|
491 | 491 | i = b'meta/%s/00manifest.i' % dir |
|
492 | 492 | d = b'meta/%s/00manifest.d' % dir |
|
493 | 493 | |
|
494 | 494 | if repo.store._exists(i): |
|
495 | 495 | newentries.add(i) |
|
496 | 496 | if repo.store._exists(d): |
|
497 | 497 | newentries.add(d) |
|
498 | 498 | |
|
499 | 499 | addcount = len(newentries - oldentries) |
|
500 | 500 | removecount = len(oldentries - newentries) |
|
501 | 501 | for p in sorted(oldentries - newentries): |
|
502 | 502 | ui.write(_(b'removing %s\n') % p) |
|
503 | 503 | for p in sorted(newentries - oldentries): |
|
504 | 504 | ui.write(_(b'adding %s\n') % p) |
|
505 | 505 | |
|
506 | 506 | if addcount or removecount: |
|
507 | 507 | ui.write( |
|
508 | 508 | _(b'%d items added, %d removed from fncache\n') |
|
509 | 509 | % (addcount, removecount) |
|
510 | 510 | ) |
|
511 | 511 | fnc.entries = newentries |
|
512 | 512 | fnc._dirty = True |
|
513 | 513 | |
|
514 | 514 | with repo.transaction(b'fncache') as tr: |
|
515 | 515 | fnc.write(tr) |
|
516 | 516 | else: |
|
517 | 517 | ui.write(_(b'fncache already up to date\n')) |
|
518 | 518 | |
|
519 | 519 | |
|
520 | 520 | def deleteobsmarkers(obsstore, indices): |
|
521 | 521 | """Delete some obsmarkers from obsstore and return how many were deleted |
|
522 | 522 | |
|
523 | 523 | 'indices' is a list of ints which are the indices |
|
524 | 524 | of the markers to be deleted. |
|
525 | 525 | |
|
526 | 526 | Every invocation of this function completely rewrites the obsstore file, |
|
527 | 527 | skipping the markers we want to be removed. The new temporary file is |
|
528 | 528 | created, remaining markers are written there and on .close() this file |
|
529 | 529 | gets atomically renamed to obsstore, thus guaranteeing consistency.""" |
|
530 | 530 | if not indices: |
|
531 | 531 | # we don't want to rewrite the obsstore with the same content |
|
532 | 532 | return |
|
533 | 533 | |
|
534 | 534 | left = [] |
|
535 | 535 | current = obsstore._all |
|
536 | 536 | n = 0 |
|
537 | 537 | for i, m in enumerate(current): |
|
538 | 538 | if i in indices: |
|
539 | 539 | n += 1 |
|
540 | 540 | continue |
|
541 | 541 | left.append(m) |
|
542 | 542 | |
|
543 | 543 | newobsstorefile = obsstore.svfs(b'obsstore', b'w', atomictemp=True) |
|
544 | 544 | for bytes in obsolete.encodemarkers(left, True, obsstore._version): |
|
545 | 545 | newobsstorefile.write(bytes) |
|
546 | 546 | newobsstorefile.close() |
|
547 | 547 | return n |
@@ -1,935 +1,935 b'' | |||
|
1 | 1 | # streamclone.py - producing and consuming streaming repository data |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2015 Gregory Szorc <gregory.szorc@gmail.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | |
|
9 | 9 | import contextlib |
|
10 | 10 | import os |
|
11 | 11 | import struct |
|
12 | 12 | |
|
13 | 13 | from .i18n import _ |
|
14 | 14 | from .pycompat import open |
|
15 | 15 | from .interfaces import repository |
|
16 | 16 | from . import ( |
|
17 | 17 | bookmarks, |
|
18 | 18 | cacheutil, |
|
19 | 19 | error, |
|
20 | 20 | narrowspec, |
|
21 | 21 | phases, |
|
22 | 22 | pycompat, |
|
23 | 23 | requirements as requirementsmod, |
|
24 | 24 | scmutil, |
|
25 | 25 | store, |
|
26 | 26 | transaction, |
|
27 | 27 | util, |
|
28 | 28 | ) |
|
29 | 29 | from .revlogutils import ( |
|
30 | 30 | nodemap, |
|
31 | 31 | ) |
|
32 | 32 | |
|
33 | 33 | |
|
34 | 34 | def new_stream_clone_requirements(default_requirements, streamed_requirements): |
|
35 | 35 | """determine the final set of requirement for a new stream clone |
|
36 | 36 | |
|
37 | 37 | this method combine the "default" requirements that a new repository would |
|
38 | 38 | use with the constaint we get from the stream clone content. We keep local |
|
39 | 39 | configuration choice when possible. |
|
40 | 40 | """ |
|
41 | 41 | requirements = set(default_requirements) |
|
42 | 42 | requirements -= requirementsmod.STREAM_FIXED_REQUIREMENTS |
|
43 | 43 | requirements.update(streamed_requirements) |
|
44 | 44 | return requirements |
|
45 | 45 | |
|
46 | 46 | |
|
47 | 47 | def streamed_requirements(repo): |
|
48 | 48 | """the set of requirement the new clone will have to support |
|
49 | 49 | |
|
50 | 50 | This is used for advertising the stream options and to generate the actual |
|
51 | 51 | stream content.""" |
|
52 | 52 | requiredformats = ( |
|
53 | 53 | repo.requirements & requirementsmod.STREAM_FIXED_REQUIREMENTS |
|
54 | 54 | ) |
|
55 | 55 | return requiredformats |
|
56 | 56 | |
|
57 | 57 | |
|
58 | 58 | def canperformstreamclone(pullop, bundle2=False): |
|
59 | 59 | """Whether it is possible to perform a streaming clone as part of pull. |
|
60 | 60 | |
|
61 | 61 | ``bundle2`` will cause the function to consider stream clone through |
|
62 | 62 | bundle2 and only through bundle2. |
|
63 | 63 | |
|
64 | 64 | Returns a tuple of (supported, requirements). ``supported`` is True if |
|
65 | 65 | streaming clone is supported and False otherwise. ``requirements`` is |
|
66 | 66 | a set of repo requirements from the remote, or ``None`` if stream clone |
|
67 | 67 | isn't supported. |
|
68 | 68 | """ |
|
69 | 69 | repo = pullop.repo |
|
70 | 70 | remote = pullop.remote |
|
71 | 71 | |
|
72 | 72 | bundle2supported = False |
|
73 | 73 | if pullop.canusebundle2: |
|
74 | 74 | if b'v2' in pullop.remotebundle2caps.get(b'stream', []): |
|
75 | 75 | bundle2supported = True |
|
76 | 76 | # else |
|
77 | 77 | # Server doesn't support bundle2 stream clone or doesn't support |
|
78 | 78 | # the versions we support. Fall back and possibly allow legacy. |
|
79 | 79 | |
|
80 | 80 | # Ensures legacy code path uses available bundle2. |
|
81 | 81 | if bundle2supported and not bundle2: |
|
82 | 82 | return False, None |
|
83 | 83 | # Ensures bundle2 doesn't try to do a stream clone if it isn't supported. |
|
84 | 84 | elif bundle2 and not bundle2supported: |
|
85 | 85 | return False, None |
|
86 | 86 | |
|
87 | 87 | # Streaming clone only works on empty repositories. |
|
88 | 88 | if len(repo): |
|
89 | 89 | return False, None |
|
90 | 90 | |
|
91 | 91 | # Streaming clone only works if all data is being requested. |
|
92 | 92 | if pullop.heads: |
|
93 | 93 | return False, None |
|
94 | 94 | |
|
95 | 95 | streamrequested = pullop.streamclonerequested |
|
96 | 96 | |
|
97 | 97 | # If we don't have a preference, let the server decide for us. This |
|
98 | 98 | # likely only comes into play in LANs. |
|
99 | 99 | if streamrequested is None: |
|
100 | 100 | # The server can advertise whether to prefer streaming clone. |
|
101 | 101 | streamrequested = remote.capable(b'stream-preferred') |
|
102 | 102 | |
|
103 | 103 | if not streamrequested: |
|
104 | 104 | return False, None |
|
105 | 105 | |
|
106 | 106 | # In order for stream clone to work, the client has to support all the |
|
107 | 107 | # requirements advertised by the server. |
|
108 | 108 | # |
|
109 | 109 | # The server advertises its requirements via the "stream" and "streamreqs" |
|
110 | 110 | # capability. "stream" (a value-less capability) is advertised if and only |
|
111 | 111 | # if the only requirement is "revlogv1." Else, the "streamreqs" capability |
|
112 | 112 | # is advertised and contains a comma-delimited list of requirements. |
|
113 | 113 | requirements = set() |
|
114 | 114 | if remote.capable(b'stream'): |
|
115 | 115 | requirements.add(requirementsmod.REVLOGV1_REQUIREMENT) |
|
116 | 116 | else: |
|
117 | 117 | streamreqs = remote.capable(b'streamreqs') |
|
118 | 118 | # This is weird and shouldn't happen with modern servers. |
|
119 | 119 | if not streamreqs: |
|
120 | 120 | pullop.repo.ui.warn( |
|
121 | 121 | _( |
|
122 | 122 | b'warning: stream clone requested but server has them ' |
|
123 | 123 | b'disabled\n' |
|
124 | 124 | ) |
|
125 | 125 | ) |
|
126 | 126 | return False, None |
|
127 | 127 | |
|
128 | 128 | streamreqs = set(streamreqs.split(b',')) |
|
129 | 129 | # Server requires something we don't support. Bail. |
|
130 | 130 | missingreqs = streamreqs - repo.supported |
|
131 | 131 | if missingreqs: |
|
132 | 132 | pullop.repo.ui.warn( |
|
133 | 133 | _( |
|
134 | 134 | b'warning: stream clone requested but client is missing ' |
|
135 | 135 | b'requirements: %s\n' |
|
136 | 136 | ) |
|
137 | 137 | % b', '.join(sorted(missingreqs)) |
|
138 | 138 | ) |
|
139 | 139 | pullop.repo.ui.warn( |
|
140 | 140 | _( |
|
141 | 141 | b'(see https://www.mercurial-scm.org/wiki/MissingRequirement ' |
|
142 | 142 | b'for more information)\n' |
|
143 | 143 | ) |
|
144 | 144 | ) |
|
145 | 145 | return False, None |
|
146 | 146 | requirements = streamreqs |
|
147 | 147 | |
|
148 | 148 | return True, requirements |
|
149 | 149 | |
|
150 | 150 | |
|
151 | 151 | def maybeperformlegacystreamclone(pullop): |
|
152 | 152 | """Possibly perform a legacy stream clone operation. |
|
153 | 153 | |
|
154 | 154 | Legacy stream clones are performed as part of pull but before all other |
|
155 | 155 | operations. |
|
156 | 156 | |
|
157 | 157 | A legacy stream clone will not be performed if a bundle2 stream clone is |
|
158 | 158 | supported. |
|
159 | 159 | """ |
|
160 | 160 | from . import localrepo |
|
161 | 161 | |
|
162 | 162 | supported, requirements = canperformstreamclone(pullop) |
|
163 | 163 | |
|
164 | 164 | if not supported: |
|
165 | 165 | return |
|
166 | 166 | |
|
167 | 167 | repo = pullop.repo |
|
168 | 168 | remote = pullop.remote |
|
169 | 169 | |
|
170 | 170 | # Save remote branchmap. We will use it later to speed up branchcache |
|
171 | 171 | # creation. |
|
172 | 172 | rbranchmap = None |
|
173 | 173 | if remote.capable(b'branchmap'): |
|
174 | 174 | with remote.commandexecutor() as e: |
|
175 | 175 | rbranchmap = e.callcommand(b'branchmap', {}).result() |
|
176 | 176 | |
|
177 | 177 | repo.ui.status(_(b'streaming all changes\n')) |
|
178 | 178 | |
|
179 | 179 | with remote.commandexecutor() as e: |
|
180 | 180 | fp = e.callcommand(b'stream_out', {}).result() |
|
181 | 181 | |
|
182 | 182 | # TODO strictly speaking, this code should all be inside the context |
|
183 | 183 | # manager because the context manager is supposed to ensure all wire state |
|
184 | 184 | # is flushed when exiting. But the legacy peers don't do this, so it |
|
185 | 185 | # doesn't matter. |
|
186 | 186 | l = fp.readline() |
|
187 | 187 | try: |
|
188 | 188 | resp = int(l) |
|
189 | 189 | except ValueError: |
|
190 | 190 | raise error.ResponseError( |
|
191 | 191 | _(b'unexpected response from remote server:'), l |
|
192 | 192 | ) |
|
193 | 193 | if resp == 1: |
|
194 | 194 | raise error.Abort(_(b'operation forbidden by server')) |
|
195 | 195 | elif resp == 2: |
|
196 | 196 | raise error.Abort(_(b'locking the remote repository failed')) |
|
197 | 197 | elif resp != 0: |
|
198 | 198 | raise error.Abort(_(b'the server sent an unknown error code')) |
|
199 | 199 | |
|
200 | 200 | l = fp.readline() |
|
201 | 201 | try: |
|
202 | 202 | filecount, bytecount = map(int, l.split(b' ', 1)) |
|
203 | 203 | except (ValueError, TypeError): |
|
204 | 204 | raise error.ResponseError( |
|
205 | 205 | _(b'unexpected response from remote server:'), l |
|
206 | 206 | ) |
|
207 | 207 | |
|
208 | 208 | with repo.lock(): |
|
209 | 209 | consumev1(repo, fp, filecount, bytecount) |
|
210 | 210 | repo.requirements = new_stream_clone_requirements( |
|
211 | 211 | repo.requirements, |
|
212 | 212 | requirements, |
|
213 | 213 | ) |
|
214 | 214 | repo.svfs.options = localrepo.resolvestorevfsoptions( |
|
215 | 215 | repo.ui, repo.requirements, repo.features |
|
216 | 216 | ) |
|
217 | 217 | scmutil.writereporequirements(repo) |
|
218 | 218 | nodemap.post_stream_cleanup(repo) |
|
219 | 219 | |
|
220 | 220 | if rbranchmap: |
|
221 | 221 | repo._branchcaches.replace(repo, rbranchmap) |
|
222 | 222 | |
|
223 | 223 | repo.invalidate() |
|
224 | 224 | |
|
225 | 225 | |
|
226 | 226 | def allowservergeneration(repo): |
|
227 | 227 | """Whether streaming clones are allowed from the server.""" |
|
228 | 228 | if repository.REPO_FEATURE_STREAM_CLONE not in repo.features: |
|
229 | 229 | return False |
|
230 | 230 | |
|
231 | 231 | if not repo.ui.configbool(b'server', b'uncompressed', untrusted=True): |
|
232 | 232 | return False |
|
233 | 233 | |
|
234 | 234 | # The way stream clone works makes it impossible to hide secret changesets. |
|
235 | 235 | # So don't allow this by default. |
|
236 | 236 | secret = phases.hassecret(repo) |
|
237 | 237 | if secret: |
|
238 | 238 | return repo.ui.configbool(b'server', b'uncompressedallowsecret') |
|
239 | 239 | |
|
240 | 240 | return True |
|
241 | 241 | |
|
242 | 242 | |
|
243 | 243 | # This is it's own function so extensions can override it. |
|
244 | 244 | def _walkstreamfiles(repo, matcher=None): |
|
245 | 245 | return repo.store.walk(matcher) |
|
246 | 246 | |
|
247 | 247 | |
|
248 | 248 | def generatev1(repo): |
|
249 | 249 | """Emit content for version 1 of a streaming clone. |
|
250 | 250 | |
|
251 | 251 | This returns a 3-tuple of (file count, byte size, data iterator). |
|
252 | 252 | |
|
253 | 253 | The data iterator consists of N entries for each file being transferred. |
|
254 | 254 | Each file entry starts as a line with the file name and integer size |
|
255 | 255 | delimited by a null byte. |
|
256 | 256 | |
|
257 | 257 | The raw file data follows. Following the raw file data is the next file |
|
258 | 258 | entry, or EOF. |
|
259 | 259 | |
|
260 | 260 | When used on the wire protocol, an additional line indicating protocol |
|
261 | 261 | success will be prepended to the stream. This function is not responsible |
|
262 | 262 | for adding it. |
|
263 | 263 | |
|
264 | 264 | This function will obtain a repository lock to ensure a consistent view of |
|
265 | 265 | the store is captured. It therefore may raise LockError. |
|
266 | 266 | """ |
|
267 | 267 | entries = [] |
|
268 | 268 | total_bytes = 0 |
|
269 | 269 | # Get consistent snapshot of repo, lock during scan. |
|
270 | 270 | with repo.lock(): |
|
271 | 271 | repo.ui.debug(b'scanning\n') |
|
272 | 272 | for file_type, name, size in _walkstreamfiles(repo): |
|
273 | 273 | if size: |
|
274 | 274 | entries.append((name, size)) |
|
275 | 275 | total_bytes += size |
|
276 | 276 | _test_sync_point_walk_1(repo) |
|
277 | 277 | _test_sync_point_walk_2(repo) |
|
278 | 278 | |
|
279 | 279 | repo.ui.debug( |
|
280 | 280 | b'%d files, %d bytes to transfer\n' % (len(entries), total_bytes) |
|
281 | 281 | ) |
|
282 | 282 | |
|
283 | 283 | svfs = repo.svfs |
|
284 | 284 | debugflag = repo.ui.debugflag |
|
285 | 285 | |
|
286 | 286 | def emitrevlogdata(): |
|
287 | 287 | for name, size in entries: |
|
288 | 288 | if debugflag: |
|
289 | 289 | repo.ui.debug(b'sending %s (%d bytes)\n' % (name, size)) |
|
290 | 290 | # partially encode name over the wire for backwards compat |
|
291 | 291 | yield b'%s\0%d\n' % (store.encodedir(name), size) |
|
292 | 292 | # auditing at this stage is both pointless (paths are already |
|
293 | 293 | # trusted by the local repo) and expensive |
|
294 | 294 | with svfs(name, b'rb', auditpath=False) as fp: |
|
295 | 295 | if size <= 65536: |
|
296 | 296 | yield fp.read(size) |
|
297 | 297 | else: |
|
298 | 298 | for chunk in util.filechunkiter(fp, limit=size): |
|
299 | 299 | yield chunk |
|
300 | 300 | |
|
301 | 301 | return len(entries), total_bytes, emitrevlogdata() |
|
302 | 302 | |
|
303 | 303 | |
|
304 | 304 | def generatev1wireproto(repo): |
|
305 | 305 | """Emit content for version 1 of streaming clone suitable for the wire. |
|
306 | 306 | |
|
307 | 307 | This is the data output from ``generatev1()`` with 2 header lines. The |
|
308 | 308 | first line indicates overall success. The 2nd contains the file count and |
|
309 | 309 | byte size of payload. |
|
310 | 310 | |
|
311 | 311 | The success line contains "0" for success, "1" for stream generation not |
|
312 | 312 | allowed, and "2" for error locking the repository (possibly indicating |
|
313 | 313 | a permissions error for the server process). |
|
314 | 314 | """ |
|
315 | 315 | if not allowservergeneration(repo): |
|
316 | 316 | yield b'1\n' |
|
317 | 317 | return |
|
318 | 318 | |
|
319 | 319 | try: |
|
320 | 320 | filecount, bytecount, it = generatev1(repo) |
|
321 | 321 | except error.LockError: |
|
322 | 322 | yield b'2\n' |
|
323 | 323 | return |
|
324 | 324 | |
|
325 | 325 | # Indicates successful response. |
|
326 | 326 | yield b'0\n' |
|
327 | 327 | yield b'%d %d\n' % (filecount, bytecount) |
|
328 | 328 | for chunk in it: |
|
329 | 329 | yield chunk |
|
330 | 330 | |
|
331 | 331 | |
|
332 | 332 | def generatebundlev1(repo, compression=b'UN'): |
|
333 | 333 | """Emit content for version 1 of a stream clone bundle. |
|
334 | 334 | |
|
335 | 335 | The first 4 bytes of the output ("HGS1") denote this as stream clone |
|
336 | 336 | bundle version 1. |
|
337 | 337 | |
|
338 | 338 | The next 2 bytes indicate the compression type. Only "UN" is currently |
|
339 | 339 | supported. |
|
340 | 340 | |
|
341 | 341 | The next 16 bytes are two 64-bit big endian unsigned integers indicating |
|
342 | 342 | file count and byte count, respectively. |
|
343 | 343 | |
|
344 | 344 | The next 2 bytes is a 16-bit big endian unsigned short declaring the length |
|
345 | 345 | of the requirements string, including a trailing \0. The following N bytes |
|
346 | 346 | are the requirements string, which is ASCII containing a comma-delimited |
|
347 | 347 | list of repo requirements that are needed to support the data. |
|
348 | 348 | |
|
349 | 349 | The remaining content is the output of ``generatev1()`` (which may be |
|
350 | 350 | compressed in the future). |
|
351 | 351 | |
|
352 | 352 | Returns a tuple of (requirements, data generator). |
|
353 | 353 | """ |
|
354 | 354 | if compression != b'UN': |
|
355 | 355 | raise ValueError(b'we do not support the compression argument yet') |
|
356 | 356 | |
|
357 | 357 | requirements = streamed_requirements(repo) |
|
358 | 358 | requires = b','.join(sorted(requirements)) |
|
359 | 359 | |
|
360 | 360 | def gen(): |
|
361 | 361 | yield b'HGS1' |
|
362 | 362 | yield compression |
|
363 | 363 | |
|
364 | 364 | filecount, bytecount, it = generatev1(repo) |
|
365 | 365 | repo.ui.status( |
|
366 | 366 | _(b'writing %d bytes for %d files\n') % (bytecount, filecount) |
|
367 | 367 | ) |
|
368 | 368 | |
|
369 | 369 | yield struct.pack(b'>QQ', filecount, bytecount) |
|
370 | 370 | yield struct.pack(b'>H', len(requires) + 1) |
|
371 | 371 | yield requires + b'\0' |
|
372 | 372 | |
|
373 | 373 | # This is where we'll add compression in the future. |
|
374 | 374 | assert compression == b'UN' |
|
375 | 375 | |
|
376 | 376 | progress = repo.ui.makeprogress( |
|
377 | 377 | _(b'bundle'), total=bytecount, unit=_(b'bytes') |
|
378 | 378 | ) |
|
379 | 379 | progress.update(0) |
|
380 | 380 | |
|
381 | 381 | for chunk in it: |
|
382 | 382 | progress.increment(step=len(chunk)) |
|
383 | 383 | yield chunk |
|
384 | 384 | |
|
385 | 385 | progress.complete() |
|
386 | 386 | |
|
387 | 387 | return requirements, gen() |
|
388 | 388 | |
|
389 | 389 | |
|
390 | 390 | def consumev1(repo, fp, filecount, bytecount): |
|
391 | 391 | """Apply the contents from version 1 of a streaming clone file handle. |
|
392 | 392 | |
|
393 | 393 | This takes the output from "stream_out" and applies it to the specified |
|
394 | 394 | repository. |
|
395 | 395 | |
|
396 | 396 | Like "stream_out," the status line added by the wire protocol is not |
|
397 | 397 | handled by this function. |
|
398 | 398 | """ |
|
399 | 399 | with repo.lock(): |
|
400 | 400 | repo.ui.status( |
|
401 | 401 | _(b'%d files to transfer, %s of data\n') |
|
402 | 402 | % (filecount, util.bytecount(bytecount)) |
|
403 | 403 | ) |
|
404 | 404 | progress = repo.ui.makeprogress( |
|
405 | 405 | _(b'clone'), total=bytecount, unit=_(b'bytes') |
|
406 | 406 | ) |
|
407 | 407 | progress.update(0) |
|
408 | 408 | start = util.timer() |
|
409 | 409 | |
|
410 | 410 | # TODO: get rid of (potential) inconsistency |
|
411 | 411 | # |
|
412 | 412 | # If transaction is started and any @filecache property is |
|
413 | 413 | # changed at this point, it causes inconsistency between |
|
414 | 414 | # in-memory cached property and streamclone-ed file on the |
|
415 | 415 | # disk. Nested transaction prevents transaction scope "clone" |
|
416 | 416 | # below from writing in-memory changes out at the end of it, |
|
417 | 417 | # even though in-memory changes are discarded at the end of it |
|
418 | 418 | # regardless of transaction nesting. |
|
419 | 419 | # |
|
420 | 420 | # But transaction nesting can't be simply prohibited, because |
|
421 | 421 | # nesting occurs also in ordinary case (e.g. enabling |
|
422 | 422 | # clonebundles). |
|
423 | 423 | |
|
424 | 424 | with repo.transaction(b'clone'): |
|
425 | 425 | with repo.svfs.backgroundclosing(repo.ui, expectedcount=filecount): |
|
426 | 426 | for i in range(filecount): |
|
427 | 427 | # XXX doesn't support '\n' or '\r' in filenames |
|
428 | 428 | l = fp.readline() |
|
429 | 429 | try: |
|
430 | 430 | name, size = l.split(b'\0', 1) |
|
431 | 431 | size = int(size) |
|
432 | 432 | except (ValueError, TypeError): |
|
433 | 433 | raise error.ResponseError( |
|
434 | 434 | _(b'unexpected response from remote server:'), l |
|
435 | 435 | ) |
|
436 | 436 | if repo.ui.debugflag: |
|
437 | 437 | repo.ui.debug( |
|
438 | 438 | b'adding %s (%s)\n' % (name, util.bytecount(size)) |
|
439 | 439 | ) |
|
440 | 440 | # for backwards compat, name was partially encoded |
|
441 | 441 | path = store.decodedir(name) |
|
442 | 442 | with repo.svfs(path, b'w', backgroundclose=True) as ofp: |
|
443 | 443 | for chunk in util.filechunkiter(fp, limit=size): |
|
444 | 444 | progress.increment(step=len(chunk)) |
|
445 | 445 | ofp.write(chunk) |
|
446 | 446 | |
|
447 | 447 | # force @filecache properties to be reloaded from |
|
448 | 448 | # streamclone-ed file at next access |
|
449 | 449 | repo.invalidate(clearfilecache=True) |
|
450 | 450 | |
|
451 | 451 | elapsed = util.timer() - start |
|
452 | 452 | if elapsed <= 0: |
|
453 | 453 | elapsed = 0.001 |
|
454 | 454 | progress.complete() |
|
455 | 455 | repo.ui.status( |
|
456 | 456 | _(b'transferred %s in %.1f seconds (%s/sec)\n') |
|
457 | 457 | % ( |
|
458 | 458 | util.bytecount(bytecount), |
|
459 | 459 | elapsed, |
|
460 | 460 | util.bytecount(bytecount / elapsed), |
|
461 | 461 | ) |
|
462 | 462 | ) |
|
463 | 463 | |
|
464 | 464 | |
|
465 | 465 | def readbundle1header(fp): |
|
466 | 466 | compression = fp.read(2) |
|
467 | 467 | if compression != b'UN': |
|
468 | 468 | raise error.Abort( |
|
469 | 469 | _( |
|
470 | 470 | b'only uncompressed stream clone bundles are ' |
|
471 | 471 | b'supported; got %s' |
|
472 | 472 | ) |
|
473 | 473 | % compression |
|
474 | 474 | ) |
|
475 | 475 | |
|
476 | 476 | filecount, bytecount = struct.unpack(b'>QQ', fp.read(16)) |
|
477 | 477 | requireslen = struct.unpack(b'>H', fp.read(2))[0] |
|
478 | 478 | requires = fp.read(requireslen) |
|
479 | 479 | |
|
480 | 480 | if not requires.endswith(b'\0'): |
|
481 | 481 | raise error.Abort( |
|
482 | 482 | _( |
|
483 | 483 | b'malformed stream clone bundle: ' |
|
484 | 484 | b'requirements not properly encoded' |
|
485 | 485 | ) |
|
486 | 486 | ) |
|
487 | 487 | |
|
488 | 488 | requirements = set(requires.rstrip(b'\0').split(b',')) |
|
489 | 489 | |
|
490 | 490 | return filecount, bytecount, requirements |
|
491 | 491 | |
|
492 | 492 | |
|
493 | 493 | def applybundlev1(repo, fp): |
|
494 | 494 | """Apply the content from a stream clone bundle version 1. |
|
495 | 495 | |
|
496 | 496 | We assume the 4 byte header has been read and validated and the file handle |
|
497 | 497 | is at the 2 byte compression identifier. |
|
498 | 498 | """ |
|
499 | 499 | if len(repo): |
|
500 | 500 | raise error.Abort( |
|
501 | 501 | _(b'cannot apply stream clone bundle on non-empty repo') |
|
502 | 502 | ) |
|
503 | 503 | |
|
504 | 504 | filecount, bytecount, requirements = readbundle1header(fp) |
|
505 | 505 | missingreqs = requirements - repo.supported |
|
506 | 506 | if missingreqs: |
|
507 | 507 | raise error.Abort( |
|
508 | 508 | _(b'unable to apply stream clone: unsupported format: %s') |
|
509 | 509 | % b', '.join(sorted(missingreqs)) |
|
510 | 510 | ) |
|
511 | 511 | |
|
512 | 512 | consumev1(repo, fp, filecount, bytecount) |
|
513 | 513 | nodemap.post_stream_cleanup(repo) |
|
514 | 514 | |
|
515 | 515 | |
|
516 | 516 | class streamcloneapplier: |
|
517 | 517 | """Class to manage applying streaming clone bundles. |
|
518 | 518 | |
|
519 | 519 | We need to wrap ``applybundlev1()`` in a dedicated type to enable bundle |
|
520 | 520 | readers to perform bundle type-specific functionality. |
|
521 | 521 | """ |
|
522 | 522 | |
|
523 | 523 | def __init__(self, fh): |
|
524 | 524 | self._fh = fh |
|
525 | 525 | |
|
526 | 526 | def apply(self, repo): |
|
527 | 527 | return applybundlev1(repo, self._fh) |
|
528 | 528 | |
|
529 | 529 | |
|
530 | 530 | # type of file to stream |
|
531 | 531 | _fileappend = 0 # append only file |
|
532 | 532 | _filefull = 1 # full snapshot file |
|
533 | 533 | |
|
534 | 534 | # Source of the file |
|
535 | 535 | _srcstore = b's' # store (svfs) |
|
536 | 536 | _srccache = b'c' # cache (cache) |
|
537 | 537 | |
|
538 | 538 | # This is it's own function so extensions can override it. |
|
539 | 539 | def _walkstreamfullstorefiles(repo): |
|
540 | 540 | """list snapshot file from the store""" |
|
541 | 541 | fnames = [] |
|
542 | 542 | if not repo.publishing(): |
|
543 | 543 | fnames.append(b'phaseroots') |
|
544 | 544 | return fnames |
|
545 | 545 | |
|
546 | 546 | |
|
547 | 547 | def _filterfull(entry, copy, vfsmap): |
|
548 | 548 | """actually copy the snapshot files""" |
|
549 | 549 | src, name, ftype, data = entry |
|
550 | 550 | if ftype != _filefull: |
|
551 | 551 | return entry |
|
552 | 552 | return (src, name, ftype, copy(vfsmap[src].join(name))) |
|
553 | 553 | |
|
554 | 554 | |
|
555 | 555 | @contextlib.contextmanager |
|
556 | 556 | def maketempcopies(): |
|
557 | 557 | """return a function to temporary copy file""" |
|
558 | 558 | |
|
559 | 559 | files = [] |
|
560 | 560 | dst_dir = pycompat.mkdtemp(prefix=b'hg-clone-') |
|
561 | 561 | try: |
|
562 | 562 | |
|
563 | 563 | def copy(src): |
|
564 | 564 | fd, dst = pycompat.mkstemp( |
|
565 | 565 | prefix=os.path.basename(src), dir=dst_dir |
|
566 | 566 | ) |
|
567 | 567 | os.close(fd) |
|
568 | 568 | files.append(dst) |
|
569 | 569 | util.copyfiles(src, dst, hardlink=True) |
|
570 | 570 | return dst |
|
571 | 571 | |
|
572 | 572 | yield copy |
|
573 | 573 | finally: |
|
574 | 574 | for tmp in files: |
|
575 | 575 | util.tryunlink(tmp) |
|
576 | 576 | util.tryrmdir(dst_dir) |
|
577 | 577 | |
|
578 | 578 | |
|
579 | 579 | def _makemap(repo): |
|
580 | 580 | """make a (src -> vfs) map for the repo""" |
|
581 | 581 | vfsmap = { |
|
582 | 582 | _srcstore: repo.svfs, |
|
583 | 583 | _srccache: repo.cachevfs, |
|
584 | 584 | } |
|
585 | 585 | # we keep repo.vfs out of the on purpose, ther are too many danger there |
|
586 | 586 | # (eg: .hg/hgrc) |
|
587 | 587 | assert repo.vfs not in vfsmap.values() |
|
588 | 588 | |
|
589 | 589 | return vfsmap |
|
590 | 590 | |
|
591 | 591 | |
|
592 | 592 | def _emit2(repo, entries, totalfilesize): |
|
593 | 593 | """actually emit the stream bundle""" |
|
594 | 594 | vfsmap = _makemap(repo) |
|
595 | 595 | # we keep repo.vfs out of the on purpose, ther are too many danger there |
|
596 | 596 | # (eg: .hg/hgrc), |
|
597 | 597 | # |
|
598 | 598 | # this assert is duplicated (from _makemap) as author might think this is |
|
599 | 599 | # fine, while this is really not fine. |
|
600 | 600 | if repo.vfs in vfsmap.values(): |
|
601 | 601 | raise error.ProgrammingError( |
|
602 | 602 | b'repo.vfs must not be added to vfsmap for security reasons' |
|
603 | 603 | ) |
|
604 | 604 | |
|
605 | 605 | progress = repo.ui.makeprogress( |
|
606 | 606 | _(b'bundle'), total=totalfilesize, unit=_(b'bytes') |
|
607 | 607 | ) |
|
608 | 608 | progress.update(0) |
|
609 | 609 | with maketempcopies() as copy, progress: |
|
610 | 610 | # copy is delayed until we are in the try |
|
611 | 611 | entries = [_filterfull(e, copy, vfsmap) for e in entries] |
|
612 | 612 | yield None # this release the lock on the repository |
|
613 | 613 | totalbytecount = 0 |
|
614 | 614 | |
|
615 | 615 | for src, name, ftype, data in entries: |
|
616 | 616 | vfs = vfsmap[src] |
|
617 | 617 | yield src |
|
618 | 618 | yield util.uvarintencode(len(name)) |
|
619 | 619 | if ftype == _fileappend: |
|
620 | 620 | fp = vfs(name) |
|
621 | 621 | size = data |
|
622 | 622 | elif ftype == _filefull: |
|
623 | 623 | fp = open(data, b'rb') |
|
624 | 624 | size = util.fstat(fp).st_size |
|
625 | 625 | bytecount = 0 |
|
626 | 626 | try: |
|
627 | 627 | yield util.uvarintencode(size) |
|
628 | 628 | yield name |
|
629 | 629 | if size <= 65536: |
|
630 | 630 | chunks = (fp.read(size),) |
|
631 | 631 | else: |
|
632 | 632 | chunks = util.filechunkiter(fp, limit=size) |
|
633 | 633 | for chunk in chunks: |
|
634 | 634 | bytecount += len(chunk) |
|
635 | 635 | totalbytecount += len(chunk) |
|
636 | 636 | progress.update(totalbytecount) |
|
637 | 637 | yield chunk |
|
638 | 638 | if bytecount != size: |
|
639 | 639 | # Would most likely be caused by a race due to `hg strip` or |
|
640 | 640 | # a revlog split |
|
641 | 641 | raise error.Abort( |
|
642 | 642 | _( |
|
643 | 643 | b'clone could only read %d bytes from %s, but ' |
|
644 | 644 | b'expected %d bytes' |
|
645 | 645 | ) |
|
646 | 646 | % (bytecount, name, size) |
|
647 | 647 | ) |
|
648 | 648 | finally: |
|
649 | 649 | fp.close() |
|
650 | 650 | |
|
651 | 651 | |
|
652 | 652 | def _test_sync_point_walk_1(repo): |
|
653 | 653 | """a function for synchronisation during tests""" |
|
654 | 654 | |
|
655 | 655 | |
|
656 | 656 | def _test_sync_point_walk_2(repo): |
|
657 | 657 | """a function for synchronisation during tests""" |
|
658 | 658 | |
|
659 | 659 | |
|
660 | 660 | def _v2_walk(repo, includes, excludes, includeobsmarkers): |
|
661 | 661 | """emit a seris of files information useful to clone a repo |
|
662 | 662 | |
|
663 | 663 | return (entries, totalfilesize) |
|
664 | 664 | |
|
665 | 665 | entries is a list of tuple (vfs-key, file-path, file-type, size) |
|
666 | 666 | |
|
667 | 667 | - `vfs-key`: is a key to the right vfs to write the file (see _makemap) |
|
668 | 668 | - `name`: file path of the file to copy (to be feed to the vfss) |
|
669 | 669 | - `file-type`: do this file need to be copied with the source lock ? |
|
670 | 670 | - `size`: the size of the file (or None) |
|
671 | 671 | """ |
|
672 | 672 | assert repo._currentlock(repo._lockref) is not None |
|
673 | 673 | entries = [] |
|
674 | 674 | totalfilesize = 0 |
|
675 | 675 | |
|
676 | 676 | matcher = None |
|
677 | 677 | if includes or excludes: |
|
678 | 678 | matcher = narrowspec.match(repo.root, includes, excludes) |
|
679 | 679 | |
|
680 | 680 | for rl_type, name, size in _walkstreamfiles(repo, matcher): |
|
681 | 681 | if size: |
|
682 | 682 | ft = _fileappend |
|
683 | 683 | if rl_type & store.FILEFLAGS_VOLATILE: |
|
684 | 684 | ft = _filefull |
|
685 | 685 | entries.append((_srcstore, name, ft, size)) |
|
686 | 686 | totalfilesize += size |
|
687 | 687 | for name in _walkstreamfullstorefiles(repo): |
|
688 | 688 | if repo.svfs.exists(name): |
|
689 | 689 | totalfilesize += repo.svfs.lstat(name).st_size |
|
690 | 690 | entries.append((_srcstore, name, _filefull, None)) |
|
691 | 691 | if includeobsmarkers and repo.svfs.exists(b'obsstore'): |
|
692 | 692 | totalfilesize += repo.svfs.lstat(b'obsstore').st_size |
|
693 | 693 | entries.append((_srcstore, b'obsstore', _filefull, None)) |
|
694 | 694 | for name in cacheutil.cachetocopy(repo): |
|
695 | 695 | if repo.cachevfs.exists(name): |
|
696 | 696 | totalfilesize += repo.cachevfs.lstat(name).st_size |
|
697 | 697 | entries.append((_srccache, name, _filefull, None)) |
|
698 | 698 | return entries, totalfilesize |
|
699 | 699 | |
|
700 | 700 | |
|
701 | 701 | def generatev2(repo, includes, excludes, includeobsmarkers): |
|
702 | 702 | """Emit content for version 2 of a streaming clone. |
|
703 | 703 | |
|
704 | 704 | the data stream consists the following entries: |
|
705 | 705 | 1) A char representing the file destination (eg: store or cache) |
|
706 | 706 | 2) A varint containing the length of the filename |
|
707 | 707 | 3) A varint containing the length of file data |
|
708 | 708 | 4) N bytes containing the filename (the internal, store-agnostic form) |
|
709 | 709 | 5) N bytes containing the file data |
|
710 | 710 | |
|
711 | 711 | Returns a 3-tuple of (file count, file size, data iterator). |
|
712 | 712 | """ |
|
713 | 713 | |
|
714 | 714 | with repo.lock(): |
|
715 | 715 | |
|
716 | 716 | repo.ui.debug(b'scanning\n') |
|
717 | 717 | |
|
718 | 718 | entries, totalfilesize = _v2_walk( |
|
719 | 719 | repo, |
|
720 | 720 | includes=includes, |
|
721 | 721 | excludes=excludes, |
|
722 | 722 | includeobsmarkers=includeobsmarkers, |
|
723 | 723 | ) |
|
724 | 724 | |
|
725 | 725 | chunks = _emit2(repo, entries, totalfilesize) |
|
726 | 726 | first = next(chunks) |
|
727 | 727 | assert first is None |
|
728 | 728 | _test_sync_point_walk_1(repo) |
|
729 | 729 | _test_sync_point_walk_2(repo) |
|
730 | 730 | |
|
731 | 731 | return len(entries), totalfilesize, chunks |
|
732 | 732 | |
|
733 | 733 | |
|
734 | 734 | @contextlib.contextmanager |
|
735 | 735 | def nested(*ctxs): |
|
736 | 736 | this = ctxs[0] |
|
737 | 737 | rest = ctxs[1:] |
|
738 | 738 | with this: |
|
739 | 739 | if rest: |
|
740 | 740 | with nested(*rest): |
|
741 | 741 | yield |
|
742 | 742 | else: |
|
743 | 743 | yield |
|
744 | 744 | |
|
745 | 745 | |
|
746 | 746 | def consumev2(repo, fp, filecount, filesize): |
|
747 | 747 | """Apply the contents from a version 2 streaming clone. |
|
748 | 748 | |
|
749 | 749 | Data is read from an object that only needs to provide a ``read(size)`` |
|
750 | 750 | method. |
|
751 | 751 | """ |
|
752 | 752 | with repo.lock(): |
|
753 | 753 | repo.ui.status( |
|
754 | 754 | _(b'%d files to transfer, %s of data\n') |
|
755 | 755 | % (filecount, util.bytecount(filesize)) |
|
756 | 756 | ) |
|
757 | 757 | |
|
758 | 758 | start = util.timer() |
|
759 | 759 | progress = repo.ui.makeprogress( |
|
760 | 760 | _(b'clone'), total=filesize, unit=_(b'bytes') |
|
761 | 761 | ) |
|
762 | 762 | progress.update(0) |
|
763 | 763 | |
|
764 | 764 | vfsmap = _makemap(repo) |
|
765 | 765 | # we keep repo.vfs out of the on purpose, ther are too many danger |
|
766 | 766 | # there (eg: .hg/hgrc), |
|
767 | 767 | # |
|
768 | 768 | # this assert is duplicated (from _makemap) as author might think this |
|
769 | 769 | # is fine, while this is really not fine. |
|
770 | 770 | if repo.vfs in vfsmap.values(): |
|
771 | 771 | raise error.ProgrammingError( |
|
772 | 772 | b'repo.vfs must not be added to vfsmap for security reasons' |
|
773 | 773 | ) |
|
774 | 774 | |
|
775 | 775 | with repo.transaction(b'clone'): |
|
776 | 776 | ctxs = (vfs.backgroundclosing(repo.ui) for vfs in vfsmap.values()) |
|
777 | 777 | with nested(*ctxs): |
|
778 | 778 | for i in range(filecount): |
|
779 | 779 | src = util.readexactly(fp, 1) |
|
780 | 780 | vfs = vfsmap[src] |
|
781 | 781 | namelen = util.uvarintdecodestream(fp) |
|
782 | 782 | datalen = util.uvarintdecodestream(fp) |
|
783 | 783 | |
|
784 | 784 | name = util.readexactly(fp, namelen) |
|
785 | 785 | |
|
786 | 786 | if repo.ui.debugflag: |
|
787 | 787 | repo.ui.debug( |
|
788 | 788 | b'adding [%s] %s (%s)\n' |
|
789 | 789 | % (src, name, util.bytecount(datalen)) |
|
790 | 790 | ) |
|
791 | 791 | |
|
792 | 792 | with vfs(name, b'w') as ofp: |
|
793 | 793 | for chunk in util.filechunkiter(fp, limit=datalen): |
|
794 | 794 | progress.increment(step=len(chunk)) |
|
795 | 795 | ofp.write(chunk) |
|
796 | 796 | |
|
797 | 797 | # force @filecache properties to be reloaded from |
|
798 | 798 | # streamclone-ed file at next access |
|
799 | 799 | repo.invalidate(clearfilecache=True) |
|
800 | 800 | |
|
801 | 801 | elapsed = util.timer() - start |
|
802 | 802 | if elapsed <= 0: |
|
803 | 803 | elapsed = 0.001 |
|
804 | 804 | repo.ui.status( |
|
805 | 805 | _(b'transferred %s in %.1f seconds (%s/sec)\n') |
|
806 | 806 | % ( |
|
807 | 807 | util.bytecount(progress.pos), |
|
808 | 808 | elapsed, |
|
809 | 809 | util.bytecount(progress.pos / elapsed), |
|
810 | 810 | ) |
|
811 | 811 | ) |
|
812 | 812 | progress.complete() |
|
813 | 813 | |
|
814 | 814 | |
|
815 | 815 | def applybundlev2(repo, fp, filecount, filesize, requirements): |
|
816 | 816 | from . import localrepo |
|
817 | 817 | |
|
818 | 818 | missingreqs = [r for r in requirements if r not in repo.supported] |
|
819 | 819 | if missingreqs: |
|
820 | 820 | raise error.Abort( |
|
821 | 821 | _(b'unable to apply stream clone: unsupported format: %s') |
|
822 | 822 | % b', '.join(sorted(missingreqs)) |
|
823 | 823 | ) |
|
824 | 824 | |
|
825 | 825 | consumev2(repo, fp, filecount, filesize) |
|
826 | 826 | |
|
827 | 827 | repo.requirements = new_stream_clone_requirements( |
|
828 | 828 | repo.requirements, |
|
829 | 829 | requirements, |
|
830 | 830 | ) |
|
831 | 831 | repo.svfs.options = localrepo.resolvestorevfsoptions( |
|
832 | 832 | repo.ui, repo.requirements, repo.features |
|
833 | 833 | ) |
|
834 | 834 | scmutil.writereporequirements(repo) |
|
835 | 835 | nodemap.post_stream_cleanup(repo) |
|
836 | 836 | |
|
837 | 837 | |
|
838 | 838 | def _copy_files(src_vfs_map, dst_vfs_map, entries, progress): |
|
839 | 839 | hardlink = [True] |
|
840 | 840 | |
|
841 | 841 | def copy_used(): |
|
842 | 842 | hardlink[0] = False |
|
843 | 843 | progress.topic = _(b'copying') |
|
844 | 844 | |
|
845 | 845 | for k, path, size in entries: |
|
846 | 846 | src_vfs = src_vfs_map[k] |
|
847 | 847 | dst_vfs = dst_vfs_map[k] |
|
848 | 848 | src_path = src_vfs.join(path) |
|
849 | 849 | dst_path = dst_vfs.join(path) |
|
850 | 850 | # We cannot use dirname and makedirs of dst_vfs here because the store |
|
851 | 851 | # encoding confuses them. See issue 6581 for details. |
|
852 | 852 | dirname = os.path.dirname(dst_path) |
|
853 | 853 | if not os.path.exists(dirname): |
|
854 | 854 | util.makedirs(dirname) |
|
855 | 855 | dst_vfs.register_file(path) |
|
856 | 856 | # XXX we could use the #nb_bytes argument. |
|
857 | 857 | util.copyfile( |
|
858 | 858 | src_path, |
|
859 | 859 | dst_path, |
|
860 | 860 | hardlink=hardlink[0], |
|
861 | 861 | no_hardlink_cb=copy_used, |
|
862 | 862 | check_fs_hardlink=False, |
|
863 | 863 | ) |
|
864 | 864 | progress.increment() |
|
865 | 865 | return hardlink[0] |
|
866 | 866 | |
|
867 | 867 | |
|
868 | 868 | def local_copy(src_repo, dest_repo): |
|
869 | 869 | """copy all content from one local repository to another |
|
870 | 870 | |
|
871 | 871 | This is useful for local clone""" |
|
872 | 872 | src_store_requirements = { |
|
873 | 873 | r |
|
874 | 874 | for r in src_repo.requirements |
|
875 | 875 | if r not in requirementsmod.WORKING_DIR_REQUIREMENTS |
|
876 | 876 | } |
|
877 | 877 | dest_store_requirements = { |
|
878 | 878 | r |
|
879 | 879 | for r in dest_repo.requirements |
|
880 | 880 | if r not in requirementsmod.WORKING_DIR_REQUIREMENTS |
|
881 | 881 | } |
|
882 | 882 | assert src_store_requirements == dest_store_requirements |
|
883 | 883 | |
|
884 | 884 | with dest_repo.lock(): |
|
885 | 885 | with src_repo.lock(): |
|
886 | 886 | |
|
887 | 887 | # bookmark is not integrated to the streaming as it might use the |
|
888 | 888 | # `repo.vfs` and they are too many sentitive data accessible |
|
889 | 889 | # through `repo.vfs` to expose it to streaming clone. |
|
890 | 890 | src_book_vfs = bookmarks.bookmarksvfs(src_repo) |
|
891 | 891 | srcbookmarks = src_book_vfs.join(b'bookmarks') |
|
892 | 892 | bm_count = 0 |
|
893 | 893 | if os.path.exists(srcbookmarks): |
|
894 | 894 | bm_count = 1 |
|
895 | 895 | |
|
896 | 896 | entries, totalfilesize = _v2_walk( |
|
897 | 897 | src_repo, |
|
898 | 898 | includes=None, |
|
899 | 899 | excludes=None, |
|
900 | 900 | includeobsmarkers=True, |
|
901 | 901 | ) |
|
902 | 902 | src_vfs_map = _makemap(src_repo) |
|
903 | 903 | dest_vfs_map = _makemap(dest_repo) |
|
904 | 904 | progress = src_repo.ui.makeprogress( |
|
905 | 905 | topic=_(b'linking'), |
|
906 | 906 | total=len(entries) + bm_count, |
|
907 | 907 | unit=_(b'files'), |
|
908 | 908 | ) |
|
909 | 909 | # copy files |
|
910 | 910 | # |
|
911 | 911 | # We could copy the full file while the source repository is locked |
|
912 | 912 | # and the other one without the lock. However, in the linking case, |
|
913 | 913 | # this would also requires checks that nobody is appending any data |
|
914 | 914 | # to the files while we do the clone, so this is not done yet. We |
|
915 | 915 | # could do this blindly when copying files. |
|
916 | 916 | files = ((k, path, size) for k, path, ftype, size in entries) |
|
917 | 917 | hardlink = _copy_files(src_vfs_map, dest_vfs_map, files, progress) |
|
918 | 918 | |
|
919 | 919 | # copy bookmarks over |
|
920 | 920 | if bm_count: |
|
921 | 921 | dst_book_vfs = bookmarks.bookmarksvfs(dest_repo) |
|
922 | 922 | dstbookmarks = dst_book_vfs.join(b'bookmarks') |
|
923 | 923 | util.copyfile(srcbookmarks, dstbookmarks) |
|
924 | 924 | progress.complete() |
|
925 | 925 | if hardlink: |
|
926 | 926 | msg = b'linked %d files\n' |
|
927 | 927 | else: |
|
928 | 928 | msg = b'copied %d files\n' |
|
929 | 929 | src_repo.ui.debug(msg % (len(entries) + bm_count)) |
|
930 | 930 | |
|
931 | 931 | with dest_repo.transaction(b"localclone") as tr: |
|
932 | 932 | dest_repo.store.write(tr) |
|
933 | 933 | |
|
934 | 934 | # clean up transaction file as they do not make sense |
|
935 | transaction.cleanup_undo_files(dest_repo) | |
|
935 | transaction.cleanup_undo_files(dest_repo.ui.warn, dest_repo.vfs_map) |
@@ -1,884 +1,884 b'' | |||
|
1 | 1 | # transaction.py - simple journaling scheme for mercurial |
|
2 | 2 | # |
|
3 | 3 | # This transaction scheme is intended to gracefully handle program |
|
4 | 4 | # errors and interruptions. More serious failures like system crashes |
|
5 | 5 | # can be recovered with an fsck-like tool. As the whole repository is |
|
6 | 6 | # effectively log-structured, this should amount to simply truncating |
|
7 | 7 | # anything that isn't referenced in the changelog. |
|
8 | 8 | # |
|
9 | 9 | # Copyright 2005, 2006 Olivia Mackall <olivia@selenic.com> |
|
10 | 10 | # |
|
11 | 11 | # This software may be used and distributed according to the terms of the |
|
12 | 12 | # GNU General Public License version 2 or any later version. |
|
13 | 13 | |
|
14 | 14 | import errno |
|
15 | 15 | import os |
|
16 | 16 | |
|
17 | 17 | from .i18n import _ |
|
18 | 18 | from . import ( |
|
19 | 19 | error, |
|
20 | 20 | pycompat, |
|
21 | 21 | util, |
|
22 | 22 | ) |
|
23 | 23 | from .utils import stringutil |
|
24 | 24 | |
|
25 | 25 | version = 2 |
|
26 | 26 | |
|
27 | 27 | GEN_GROUP_ALL = b'all' |
|
28 | 28 | GEN_GROUP_PRE_FINALIZE = b'prefinalize' |
|
29 | 29 | GEN_GROUP_POST_FINALIZE = b'postfinalize' |
|
30 | 30 | |
|
31 | 31 | |
|
32 | 32 | def active(func): |
|
33 | 33 | def _active(self, *args, **kwds): |
|
34 | 34 | if self._count == 0: |
|
35 | 35 | raise error.ProgrammingError( |
|
36 | 36 | b'cannot use transaction when it is already committed/aborted' |
|
37 | 37 | ) |
|
38 | 38 | return func(self, *args, **kwds) |
|
39 | 39 | |
|
40 | 40 | return _active |
|
41 | 41 | |
|
42 | 42 | |
|
43 | 43 | UNDO_BACKUP = b'undo.backupfiles' |
|
44 | 44 | |
|
45 | 45 | UNDO_FILES_MAY_NEED_CLEANUP = [ |
|
46 | 46 | (b'plain', b'undo.desc'), |
|
47 | 47 | # Always delete undo last to make sure we detect that a clean up is needed if |
|
48 | 48 | # the process is interrupted. |
|
49 | 49 | (b'store', b'undo'), |
|
50 | 50 | ] |
|
51 | 51 | |
|
52 | 52 | |
|
53 | def cleanup_undo_files(repo): | |
|
53 | def cleanup_undo_files(report, vfsmap): | |
|
54 | 54 | """remove "undo" files used by the rollback logic |
|
55 | 55 | |
|
56 | 56 | This is useful to prevent rollback running in situation were it does not |
|
57 | 57 | make sense. For example after a strip. |
|
58 | 58 | """ |
|
59 | 59 | backup_entries = [] |
|
60 | 60 | undo_files = [] |
|
61 | vfsmap = repo.vfs_map | |
|
61 | svfs = vfsmap[b'store'] | |
|
62 | 62 | try: |
|
63 |
with |
|
|
64 |
backup_entries = read_backup_files(repo |
|
|
63 | with svfs(UNDO_BACKUP) as f: | |
|
64 | backup_entries = read_backup_files(report, f) | |
|
65 | 65 | except OSError as e: |
|
66 | 66 | if e.errno != errno.ENOENT: |
|
67 | 67 | msg = _(b'could not read %s: %s\n') |
|
68 |
msg %= ( |
|
|
69 |
repo |
|
|
68 | msg %= (svfs.join(UNDO_BACKUP), stringutil.forcebytestr(e)) | |
|
69 | report(msg) | |
|
70 | 70 | |
|
71 | 71 | for location, f, backup_path, c in backup_entries: |
|
72 | 72 | if location in vfsmap and backup_path: |
|
73 | 73 | undo_files.append((vfsmap[location], backup_path)) |
|
74 | 74 | |
|
75 |
undo_files.append(( |
|
|
75 | undo_files.append((svfs, UNDO_BACKUP)) | |
|
76 | 76 | for location, undo_path in UNDO_FILES_MAY_NEED_CLEANUP: |
|
77 | 77 | undo_files.append((vfsmap[location], undo_path)) |
|
78 | 78 | for undovfs, undofile in undo_files: |
|
79 | 79 | try: |
|
80 | 80 | undovfs.unlink(undofile) |
|
81 | 81 | except OSError as e: |
|
82 | 82 | if e.errno != errno.ENOENT: |
|
83 | 83 | msg = _(b'error removing %s: %s\n') |
|
84 | 84 | msg %= (undovfs.join(undofile), stringutil.forcebytestr(e)) |
|
85 |
repo |
|
|
85 | report(msg) | |
|
86 | 86 | |
|
87 | 87 | |
|
88 | 88 | def _playback( |
|
89 | 89 | journal, |
|
90 | 90 | report, |
|
91 | 91 | opener, |
|
92 | 92 | vfsmap, |
|
93 | 93 | entries, |
|
94 | 94 | backupentries, |
|
95 | 95 | unlink=True, |
|
96 | 96 | checkambigfiles=None, |
|
97 | 97 | ): |
|
98 | 98 | for f, o in sorted(dict(entries).items()): |
|
99 | 99 | if o or not unlink: |
|
100 | 100 | checkambig = checkambigfiles and (f, b'') in checkambigfiles |
|
101 | 101 | try: |
|
102 | 102 | fp = opener(f, b'a', checkambig=checkambig) |
|
103 | 103 | if fp.tell() < o: |
|
104 | 104 | raise error.Abort( |
|
105 | 105 | _( |
|
106 | 106 | b"attempted to truncate %s to %d bytes, but it was " |
|
107 | 107 | b"already %d bytes\n" |
|
108 | 108 | ) |
|
109 | 109 | % (f, o, fp.tell()) |
|
110 | 110 | ) |
|
111 | 111 | fp.truncate(o) |
|
112 | 112 | fp.close() |
|
113 | 113 | except IOError: |
|
114 | 114 | report(_(b"failed to truncate %s\n") % f) |
|
115 | 115 | raise |
|
116 | 116 | else: |
|
117 | 117 | try: |
|
118 | 118 | opener.unlink(f) |
|
119 | 119 | except FileNotFoundError: |
|
120 | 120 | pass |
|
121 | 121 | |
|
122 | 122 | backupfiles = [] |
|
123 | 123 | for l, f, b, c in backupentries: |
|
124 | 124 | if l not in vfsmap and c: |
|
125 | 125 | report(b"couldn't handle %s: unknown cache location %s\n" % (b, l)) |
|
126 | 126 | vfs = vfsmap[l] |
|
127 | 127 | try: |
|
128 | 128 | if f and b: |
|
129 | 129 | filepath = vfs.join(f) |
|
130 | 130 | backuppath = vfs.join(b) |
|
131 | 131 | checkambig = checkambigfiles and (f, l) in checkambigfiles |
|
132 | 132 | try: |
|
133 | 133 | util.copyfile(backuppath, filepath, checkambig=checkambig) |
|
134 | 134 | backupfiles.append(b) |
|
135 | 135 | except IOError as exc: |
|
136 | 136 | e_msg = stringutil.forcebytestr(exc) |
|
137 | 137 | report(_(b"failed to recover %s (%s)\n") % (f, e_msg)) |
|
138 | 138 | else: |
|
139 | 139 | target = f or b |
|
140 | 140 | try: |
|
141 | 141 | vfs.unlink(target) |
|
142 | 142 | except FileNotFoundError: |
|
143 | 143 | pass |
|
144 | 144 | except (IOError, OSError, error.Abort): |
|
145 | 145 | if not c: |
|
146 | 146 | raise |
|
147 | 147 | |
|
148 | 148 | backuppath = b"%s.backupfiles" % journal |
|
149 | 149 | if opener.exists(backuppath): |
|
150 | 150 | opener.unlink(backuppath) |
|
151 | 151 | opener.unlink(journal) |
|
152 | 152 | try: |
|
153 | 153 | for f in backupfiles: |
|
154 | 154 | if opener.exists(f): |
|
155 | 155 | opener.unlink(f) |
|
156 | 156 | except (IOError, OSError, error.Abort): |
|
157 | 157 | # only pure backup file remains, it is sage to ignore any error |
|
158 | 158 | pass |
|
159 | 159 | |
|
160 | 160 | |
|
161 | 161 | class transaction(util.transactional): |
|
162 | 162 | def __init__( |
|
163 | 163 | self, |
|
164 | 164 | report, |
|
165 | 165 | opener, |
|
166 | 166 | vfsmap, |
|
167 | 167 | journalname, |
|
168 | 168 | undoname=None, |
|
169 | 169 | after=None, |
|
170 | 170 | createmode=None, |
|
171 | 171 | validator=None, |
|
172 | 172 | releasefn=None, |
|
173 | 173 | checkambigfiles=None, |
|
174 | 174 | name='<unnamed>', |
|
175 | 175 | ): |
|
176 | 176 | """Begin a new transaction |
|
177 | 177 | |
|
178 | 178 | Begins a new transaction that allows rolling back writes in the event of |
|
179 | 179 | an exception. |
|
180 | 180 | |
|
181 | 181 | * `after`: called after the transaction has been committed |
|
182 | 182 | * `createmode`: the mode of the journal file that will be created |
|
183 | 183 | * `releasefn`: called after releasing (with transaction and result) |
|
184 | 184 | |
|
185 | 185 | `checkambigfiles` is a set of (path, vfs-location) tuples, |
|
186 | 186 | which determine whether file stat ambiguity should be avoided |
|
187 | 187 | for corresponded files. |
|
188 | 188 | """ |
|
189 | 189 | self._count = 1 |
|
190 | 190 | self._usages = 1 |
|
191 | 191 | self._report = report |
|
192 | 192 | # a vfs to the store content |
|
193 | 193 | self._opener = opener |
|
194 | 194 | # a map to access file in various {location -> vfs} |
|
195 | 195 | vfsmap = vfsmap.copy() |
|
196 | 196 | vfsmap[b''] = opener # set default value |
|
197 | 197 | self._vfsmap = vfsmap |
|
198 | 198 | self._after = after |
|
199 | 199 | self._offsetmap = {} |
|
200 | 200 | self._newfiles = set() |
|
201 | 201 | self._journal = journalname |
|
202 | 202 | self._journal_files = [] |
|
203 | 203 | self._undoname = undoname |
|
204 | 204 | self._queue = [] |
|
205 | 205 | # A callback to do something just after releasing transaction. |
|
206 | 206 | if releasefn is None: |
|
207 | 207 | releasefn = lambda tr, success: None |
|
208 | 208 | self._releasefn = releasefn |
|
209 | 209 | |
|
210 | 210 | self._checkambigfiles = set() |
|
211 | 211 | if checkambigfiles: |
|
212 | 212 | self._checkambigfiles.update(checkambigfiles) |
|
213 | 213 | |
|
214 | 214 | self._names = [name] |
|
215 | 215 | |
|
216 | 216 | # A dict dedicated to precisely tracking the changes introduced in the |
|
217 | 217 | # transaction. |
|
218 | 218 | self.changes = {} |
|
219 | 219 | |
|
220 | 220 | # a dict of arguments to be passed to hooks |
|
221 | 221 | self.hookargs = {} |
|
222 | 222 | self._file = opener.open(self._journal, b"w+") |
|
223 | 223 | |
|
224 | 224 | # a list of ('location', 'path', 'backuppath', cache) entries. |
|
225 | 225 | # - if 'backuppath' is empty, no file existed at backup time |
|
226 | 226 | # - if 'path' is empty, this is a temporary transaction file |
|
227 | 227 | # - if 'location' is not empty, the path is outside main opener reach. |
|
228 | 228 | # use 'location' value as a key in a vfsmap to find the right 'vfs' |
|
229 | 229 | # (cache is currently unused) |
|
230 | 230 | self._backupentries = [] |
|
231 | 231 | self._backupmap = {} |
|
232 | 232 | self._backupjournal = b"%s.backupfiles" % self._journal |
|
233 | 233 | self._backupsfile = opener.open(self._backupjournal, b'w') |
|
234 | 234 | self._backupsfile.write(b'%d\n' % version) |
|
235 | 235 | |
|
236 | 236 | if createmode is not None: |
|
237 | 237 | opener.chmod(self._journal, createmode & 0o666) |
|
238 | 238 | opener.chmod(self._backupjournal, createmode & 0o666) |
|
239 | 239 | |
|
240 | 240 | # hold file generations to be performed on commit |
|
241 | 241 | self._filegenerators = {} |
|
242 | 242 | # hold callback to write pending data for hooks |
|
243 | 243 | self._pendingcallback = {} |
|
244 | 244 | # True is any pending data have been written ever |
|
245 | 245 | self._anypending = False |
|
246 | 246 | # holds callback to call when writing the transaction |
|
247 | 247 | self._finalizecallback = {} |
|
248 | 248 | # holds callback to call when validating the transaction |
|
249 | 249 | # should raise exception if anything is wrong |
|
250 | 250 | self._validatecallback = {} |
|
251 | 251 | if validator is not None: |
|
252 | 252 | self._validatecallback[b'001-userhooks'] = validator |
|
253 | 253 | # hold callback for post transaction close |
|
254 | 254 | self._postclosecallback = {} |
|
255 | 255 | # holds callbacks to call during abort |
|
256 | 256 | self._abortcallback = {} |
|
257 | 257 | |
|
258 | 258 | def __repr__(self): |
|
259 | 259 | name = '/'.join(self._names) |
|
260 | 260 | return '<transaction name=%s, count=%d, usages=%d>' % ( |
|
261 | 261 | name, |
|
262 | 262 | self._count, |
|
263 | 263 | self._usages, |
|
264 | 264 | ) |
|
265 | 265 | |
|
266 | 266 | def __del__(self): |
|
267 | 267 | if self._journal: |
|
268 | 268 | self._abort() |
|
269 | 269 | |
|
270 | 270 | @property |
|
271 | 271 | def finalized(self): |
|
272 | 272 | return self._finalizecallback is None |
|
273 | 273 | |
|
274 | 274 | @active |
|
275 | 275 | def startgroup(self): |
|
276 | 276 | """delay registration of file entry |
|
277 | 277 | |
|
278 | 278 | This is used by strip to delay vision of strip offset. The transaction |
|
279 | 279 | sees either none or all of the strip actions to be done.""" |
|
280 | 280 | self._queue.append([]) |
|
281 | 281 | |
|
282 | 282 | @active |
|
283 | 283 | def endgroup(self): |
|
284 | 284 | """apply delayed registration of file entry. |
|
285 | 285 | |
|
286 | 286 | This is used by strip to delay vision of strip offset. The transaction |
|
287 | 287 | sees either none or all of the strip actions to be done.""" |
|
288 | 288 | q = self._queue.pop() |
|
289 | 289 | for f, o in q: |
|
290 | 290 | self._addentry(f, o) |
|
291 | 291 | |
|
292 | 292 | @active |
|
293 | 293 | def add(self, file, offset): |
|
294 | 294 | """record the state of an append-only file before update""" |
|
295 | 295 | if ( |
|
296 | 296 | file in self._newfiles |
|
297 | 297 | or file in self._offsetmap |
|
298 | 298 | or file in self._backupmap |
|
299 | 299 | ): |
|
300 | 300 | return |
|
301 | 301 | if self._queue: |
|
302 | 302 | self._queue[-1].append((file, offset)) |
|
303 | 303 | return |
|
304 | 304 | |
|
305 | 305 | self._addentry(file, offset) |
|
306 | 306 | |
|
307 | 307 | def _addentry(self, file, offset): |
|
308 | 308 | """add a append-only entry to memory and on-disk state""" |
|
309 | 309 | if ( |
|
310 | 310 | file in self._newfiles |
|
311 | 311 | or file in self._offsetmap |
|
312 | 312 | or file in self._backupmap |
|
313 | 313 | ): |
|
314 | 314 | return |
|
315 | 315 | if offset: |
|
316 | 316 | self._offsetmap[file] = offset |
|
317 | 317 | else: |
|
318 | 318 | self._newfiles.add(file) |
|
319 | 319 | # add enough data to the journal to do the truncate |
|
320 | 320 | self._file.write(b"%s\0%d\n" % (file, offset)) |
|
321 | 321 | self._file.flush() |
|
322 | 322 | |
|
323 | 323 | @active |
|
324 | 324 | def addbackup(self, file, hardlink=True, location=b''): |
|
325 | 325 | """Adds a backup of the file to the transaction |
|
326 | 326 | |
|
327 | 327 | Calling addbackup() creates a hardlink backup of the specified file |
|
328 | 328 | that is used to recover the file in the event of the transaction |
|
329 | 329 | aborting. |
|
330 | 330 | |
|
331 | 331 | * `file`: the file path, relative to .hg/store |
|
332 | 332 | * `hardlink`: use a hardlink to quickly create the backup |
|
333 | 333 | """ |
|
334 | 334 | if self._queue: |
|
335 | 335 | msg = b'cannot use transaction.addbackup inside "group"' |
|
336 | 336 | raise error.ProgrammingError(msg) |
|
337 | 337 | |
|
338 | 338 | if ( |
|
339 | 339 | file in self._newfiles |
|
340 | 340 | or file in self._offsetmap |
|
341 | 341 | or file in self._backupmap |
|
342 | 342 | ): |
|
343 | 343 | return |
|
344 | 344 | vfs = self._vfsmap[location] |
|
345 | 345 | dirname, filename = vfs.split(file) |
|
346 | 346 | backupfilename = b"%s.backup.%s" % (self._journal, filename) |
|
347 | 347 | backupfile = vfs.reljoin(dirname, backupfilename) |
|
348 | 348 | if vfs.exists(file): |
|
349 | 349 | filepath = vfs.join(file) |
|
350 | 350 | backuppath = vfs.join(backupfile) |
|
351 | 351 | util.copyfile(filepath, backuppath, hardlink=hardlink) |
|
352 | 352 | else: |
|
353 | 353 | backupfile = b'' |
|
354 | 354 | |
|
355 | 355 | self._addbackupentry((location, file, backupfile, False)) |
|
356 | 356 | |
|
357 | 357 | def _addbackupentry(self, entry): |
|
358 | 358 | """register a new backup entry and write it to disk""" |
|
359 | 359 | self._backupentries.append(entry) |
|
360 | 360 | self._backupmap[entry[1]] = len(self._backupentries) - 1 |
|
361 | 361 | self._backupsfile.write(b"%s\0%s\0%s\0%d\n" % entry) |
|
362 | 362 | self._backupsfile.flush() |
|
363 | 363 | |
|
364 | 364 | @active |
|
365 | 365 | def registertmp(self, tmpfile, location=b''): |
|
366 | 366 | """register a temporary transaction file |
|
367 | 367 | |
|
368 | 368 | Such files will be deleted when the transaction exits (on both |
|
369 | 369 | failure and success). |
|
370 | 370 | """ |
|
371 | 371 | self._addbackupentry((location, b'', tmpfile, False)) |
|
372 | 372 | |
|
373 | 373 | @active |
|
374 | 374 | def addfilegenerator( |
|
375 | 375 | self, |
|
376 | 376 | genid, |
|
377 | 377 | filenames, |
|
378 | 378 | genfunc, |
|
379 | 379 | order=0, |
|
380 | 380 | location=b'', |
|
381 | 381 | post_finalize=False, |
|
382 | 382 | ): |
|
383 | 383 | """add a function to generates some files at transaction commit |
|
384 | 384 | |
|
385 | 385 | The `genfunc` argument is a function capable of generating proper |
|
386 | 386 | content of each entry in the `filename` tuple. |
|
387 | 387 | |
|
388 | 388 | At transaction close time, `genfunc` will be called with one file |
|
389 | 389 | object argument per entries in `filenames`. |
|
390 | 390 | |
|
391 | 391 | The transaction itself is responsible for the backup, creation and |
|
392 | 392 | final write of such file. |
|
393 | 393 | |
|
394 | 394 | The `genid` argument is used to ensure the same set of file is only |
|
395 | 395 | generated once. Call to `addfilegenerator` for a `genid` already |
|
396 | 396 | present will overwrite the old entry. |
|
397 | 397 | |
|
398 | 398 | The `order` argument may be used to control the order in which multiple |
|
399 | 399 | generator will be executed. |
|
400 | 400 | |
|
401 | 401 | The `location` arguments may be used to indicate the files are located |
|
402 | 402 | outside of the the standard directory for transaction. It should match |
|
403 | 403 | one of the key of the `transaction.vfsmap` dictionary. |
|
404 | 404 | |
|
405 | 405 | The `post_finalize` argument can be set to `True` for file generation |
|
406 | 406 | that must be run after the transaction has been finalized. |
|
407 | 407 | """ |
|
408 | 408 | # For now, we are unable to do proper backup and restore of custom vfs |
|
409 | 409 | # but for bookmarks that are handled outside this mechanism. |
|
410 | 410 | entry = (order, filenames, genfunc, location, post_finalize) |
|
411 | 411 | self._filegenerators[genid] = entry |
|
412 | 412 | |
|
413 | 413 | @active |
|
414 | 414 | def removefilegenerator(self, genid): |
|
415 | 415 | """reverse of addfilegenerator, remove a file generator function""" |
|
416 | 416 | if genid in self._filegenerators: |
|
417 | 417 | del self._filegenerators[genid] |
|
418 | 418 | |
|
419 | 419 | def _generatefiles(self, suffix=b'', group=GEN_GROUP_ALL): |
|
420 | 420 | # write files registered for generation |
|
421 | 421 | any = False |
|
422 | 422 | |
|
423 | 423 | if group == GEN_GROUP_ALL: |
|
424 | 424 | skip_post = skip_pre = False |
|
425 | 425 | else: |
|
426 | 426 | skip_pre = group == GEN_GROUP_POST_FINALIZE |
|
427 | 427 | skip_post = group == GEN_GROUP_PRE_FINALIZE |
|
428 | 428 | |
|
429 | 429 | for id, entry in sorted(self._filegenerators.items()): |
|
430 | 430 | any = True |
|
431 | 431 | order, filenames, genfunc, location, post_finalize = entry |
|
432 | 432 | |
|
433 | 433 | # for generation at closing, check if it's before or after finalize |
|
434 | 434 | if skip_post and post_finalize: |
|
435 | 435 | continue |
|
436 | 436 | elif skip_pre and not post_finalize: |
|
437 | 437 | continue |
|
438 | 438 | |
|
439 | 439 | vfs = self._vfsmap[location] |
|
440 | 440 | files = [] |
|
441 | 441 | try: |
|
442 | 442 | for name in filenames: |
|
443 | 443 | name += suffix |
|
444 | 444 | if suffix: |
|
445 | 445 | self.registertmp(name, location=location) |
|
446 | 446 | checkambig = False |
|
447 | 447 | else: |
|
448 | 448 | self.addbackup(name, location=location) |
|
449 | 449 | checkambig = (name, location) in self._checkambigfiles |
|
450 | 450 | files.append( |
|
451 | 451 | vfs(name, b'w', atomictemp=True, checkambig=checkambig) |
|
452 | 452 | ) |
|
453 | 453 | genfunc(*files) |
|
454 | 454 | for f in files: |
|
455 | 455 | f.close() |
|
456 | 456 | # skip discard() loop since we're sure no open file remains |
|
457 | 457 | del files[:] |
|
458 | 458 | finally: |
|
459 | 459 | for f in files: |
|
460 | 460 | f.discard() |
|
461 | 461 | return any |
|
462 | 462 | |
|
463 | 463 | @active |
|
464 | 464 | def findoffset(self, file): |
|
465 | 465 | if file in self._newfiles: |
|
466 | 466 | return 0 |
|
467 | 467 | return self._offsetmap.get(file) |
|
468 | 468 | |
|
469 | 469 | @active |
|
470 | 470 | def readjournal(self): |
|
471 | 471 | self._file.seek(0) |
|
472 | 472 | entries = [] |
|
473 | 473 | for l in self._file.readlines(): |
|
474 | 474 | file, troffset = l.split(b'\0') |
|
475 | 475 | entries.append((file, int(troffset))) |
|
476 | 476 | return entries |
|
477 | 477 | |
|
478 | 478 | @active |
|
479 | 479 | def replace(self, file, offset): |
|
480 | 480 | """ |
|
481 | 481 | replace can only replace already committed entries |
|
482 | 482 | that are not pending in the queue |
|
483 | 483 | """ |
|
484 | 484 | if file in self._newfiles: |
|
485 | 485 | if not offset: |
|
486 | 486 | return |
|
487 | 487 | self._newfiles.remove(file) |
|
488 | 488 | self._offsetmap[file] = offset |
|
489 | 489 | elif file in self._offsetmap: |
|
490 | 490 | if not offset: |
|
491 | 491 | del self._offsetmap[file] |
|
492 | 492 | self._newfiles.add(file) |
|
493 | 493 | else: |
|
494 | 494 | self._offsetmap[file] = offset |
|
495 | 495 | else: |
|
496 | 496 | raise KeyError(file) |
|
497 | 497 | self._file.write(b"%s\0%d\n" % (file, offset)) |
|
498 | 498 | self._file.flush() |
|
499 | 499 | |
|
500 | 500 | @active |
|
501 | 501 | def nest(self, name='<unnamed>'): |
|
502 | 502 | self._count += 1 |
|
503 | 503 | self._usages += 1 |
|
504 | 504 | self._names.append(name) |
|
505 | 505 | return self |
|
506 | 506 | |
|
507 | 507 | def release(self): |
|
508 | 508 | if self._count > 0: |
|
509 | 509 | self._usages -= 1 |
|
510 | 510 | if self._names: |
|
511 | 511 | self._names.pop() |
|
512 | 512 | # if the transaction scopes are left without being closed, fail |
|
513 | 513 | if self._count > 0 and self._usages == 0: |
|
514 | 514 | self._abort() |
|
515 | 515 | |
|
516 | 516 | def running(self): |
|
517 | 517 | return self._count > 0 |
|
518 | 518 | |
|
519 | 519 | def addpending(self, category, callback): |
|
520 | 520 | """add a callback to be called when the transaction is pending |
|
521 | 521 | |
|
522 | 522 | The transaction will be given as callback's first argument. |
|
523 | 523 | |
|
524 | 524 | Category is a unique identifier to allow overwriting an old callback |
|
525 | 525 | with a newer callback. |
|
526 | 526 | """ |
|
527 | 527 | self._pendingcallback[category] = callback |
|
528 | 528 | |
|
529 | 529 | @active |
|
530 | 530 | def writepending(self): |
|
531 | 531 | """write pending file to temporary version |
|
532 | 532 | |
|
533 | 533 | This is used to allow hooks to view a transaction before commit""" |
|
534 | 534 | categories = sorted(self._pendingcallback) |
|
535 | 535 | for cat in categories: |
|
536 | 536 | # remove callback since the data will have been flushed |
|
537 | 537 | any = self._pendingcallback.pop(cat)(self) |
|
538 | 538 | self._anypending = self._anypending or any |
|
539 | 539 | self._anypending |= self._generatefiles(suffix=b'.pending') |
|
540 | 540 | return self._anypending |
|
541 | 541 | |
|
542 | 542 | @active |
|
543 | 543 | def hasfinalize(self, category): |
|
544 | 544 | """check is a callback already exist for a category""" |
|
545 | 545 | return category in self._finalizecallback |
|
546 | 546 | |
|
547 | 547 | @active |
|
548 | 548 | def addfinalize(self, category, callback): |
|
549 | 549 | """add a callback to be called when the transaction is closed |
|
550 | 550 | |
|
551 | 551 | The transaction will be given as callback's first argument. |
|
552 | 552 | |
|
553 | 553 | Category is a unique identifier to allow overwriting old callbacks with |
|
554 | 554 | newer callbacks. |
|
555 | 555 | """ |
|
556 | 556 | self._finalizecallback[category] = callback |
|
557 | 557 | |
|
558 | 558 | @active |
|
559 | 559 | def addpostclose(self, category, callback): |
|
560 | 560 | """add or replace a callback to be called after the transaction closed |
|
561 | 561 | |
|
562 | 562 | The transaction will be given as callback's first argument. |
|
563 | 563 | |
|
564 | 564 | Category is a unique identifier to allow overwriting an old callback |
|
565 | 565 | with a newer callback. |
|
566 | 566 | """ |
|
567 | 567 | self._postclosecallback[category] = callback |
|
568 | 568 | |
|
569 | 569 | @active |
|
570 | 570 | def getpostclose(self, category): |
|
571 | 571 | """return a postclose callback added before, or None""" |
|
572 | 572 | return self._postclosecallback.get(category, None) |
|
573 | 573 | |
|
574 | 574 | @active |
|
575 | 575 | def addabort(self, category, callback): |
|
576 | 576 | """add a callback to be called when the transaction is aborted. |
|
577 | 577 | |
|
578 | 578 | The transaction will be given as the first argument to the callback. |
|
579 | 579 | |
|
580 | 580 | Category is a unique identifier to allow overwriting an old callback |
|
581 | 581 | with a newer callback. |
|
582 | 582 | """ |
|
583 | 583 | self._abortcallback[category] = callback |
|
584 | 584 | |
|
585 | 585 | @active |
|
586 | 586 | def addvalidator(self, category, callback): |
|
587 | 587 | """adds a callback to be called when validating the transaction. |
|
588 | 588 | |
|
589 | 589 | The transaction will be given as the first argument to the callback. |
|
590 | 590 | |
|
591 | 591 | callback should raise exception if to abort transaction""" |
|
592 | 592 | self._validatecallback[category] = callback |
|
593 | 593 | |
|
594 | 594 | @active |
|
595 | 595 | def close(self): |
|
596 | 596 | '''commit the transaction''' |
|
597 | 597 | if self._count == 1: |
|
598 | 598 | for category in sorted(self._validatecallback): |
|
599 | 599 | self._validatecallback[category](self) |
|
600 | 600 | self._validatecallback = None # Help prevent cycles. |
|
601 | 601 | self._generatefiles(group=GEN_GROUP_PRE_FINALIZE) |
|
602 | 602 | while self._finalizecallback: |
|
603 | 603 | callbacks = self._finalizecallback |
|
604 | 604 | self._finalizecallback = {} |
|
605 | 605 | categories = sorted(callbacks) |
|
606 | 606 | for cat in categories: |
|
607 | 607 | callbacks[cat](self) |
|
608 | 608 | # Prevent double usage and help clear cycles. |
|
609 | 609 | self._finalizecallback = None |
|
610 | 610 | self._generatefiles(group=GEN_GROUP_POST_FINALIZE) |
|
611 | 611 | |
|
612 | 612 | self._count -= 1 |
|
613 | 613 | if self._count != 0: |
|
614 | 614 | return |
|
615 | 615 | self._file.close() |
|
616 | 616 | self._backupsfile.close() |
|
617 | 617 | # cleanup temporary files |
|
618 | 618 | for l, f, b, c in self._backupentries: |
|
619 | 619 | if l not in self._vfsmap and c: |
|
620 | 620 | self._report( |
|
621 | 621 | b"couldn't remove %s: unknown cache location %s\n" % (b, l) |
|
622 | 622 | ) |
|
623 | 623 | continue |
|
624 | 624 | vfs = self._vfsmap[l] |
|
625 | 625 | if not f and b and vfs.exists(b): |
|
626 | 626 | try: |
|
627 | 627 | vfs.unlink(b) |
|
628 | 628 | except (IOError, OSError, error.Abort) as inst: |
|
629 | 629 | if not c: |
|
630 | 630 | raise |
|
631 | 631 | # Abort may be raise by read only opener |
|
632 | 632 | self._report( |
|
633 | 633 | b"couldn't remove %s: %s\n" % (vfs.join(b), inst) |
|
634 | 634 | ) |
|
635 | 635 | self._offsetmap = {} |
|
636 | 636 | self._newfiles = set() |
|
637 | 637 | self._writeundo() |
|
638 | 638 | if self._after: |
|
639 | 639 | self._after() |
|
640 | 640 | self._after = None # Help prevent cycles. |
|
641 | 641 | if self._opener.isfile(self._backupjournal): |
|
642 | 642 | self._opener.unlink(self._backupjournal) |
|
643 | 643 | if self._opener.isfile(self._journal): |
|
644 | 644 | self._opener.unlink(self._journal) |
|
645 | 645 | for l, _f, b, c in self._backupentries: |
|
646 | 646 | if l not in self._vfsmap and c: |
|
647 | 647 | self._report( |
|
648 | 648 | b"couldn't remove %s: unknown cache location" |
|
649 | 649 | b"%s\n" % (b, l) |
|
650 | 650 | ) |
|
651 | 651 | continue |
|
652 | 652 | vfs = self._vfsmap[l] |
|
653 | 653 | if b and vfs.exists(b): |
|
654 | 654 | try: |
|
655 | 655 | vfs.unlink(b) |
|
656 | 656 | except (IOError, OSError, error.Abort) as inst: |
|
657 | 657 | if not c: |
|
658 | 658 | raise |
|
659 | 659 | # Abort may be raise by read only opener |
|
660 | 660 | self._report( |
|
661 | 661 | b"couldn't remove %s: %s\n" % (vfs.join(b), inst) |
|
662 | 662 | ) |
|
663 | 663 | self._backupentries = [] |
|
664 | 664 | self._journal = None |
|
665 | 665 | |
|
666 | 666 | self._releasefn(self, True) # notify success of closing transaction |
|
667 | 667 | self._releasefn = None # Help prevent cycles. |
|
668 | 668 | |
|
669 | 669 | # run post close action |
|
670 | 670 | categories = sorted(self._postclosecallback) |
|
671 | 671 | for cat in categories: |
|
672 | 672 | self._postclosecallback[cat](self) |
|
673 | 673 | # Prevent double usage and help clear cycles. |
|
674 | 674 | self._postclosecallback = None |
|
675 | 675 | |
|
676 | 676 | @active |
|
677 | 677 | def abort(self): |
|
678 | 678 | """abort the transaction (generally called on error, or when the |
|
679 | 679 | transaction is not explicitly committed before going out of |
|
680 | 680 | scope)""" |
|
681 | 681 | self._abort() |
|
682 | 682 | |
|
683 | 683 | @active |
|
684 | 684 | def add_journal(self, vfs_id, path): |
|
685 | 685 | self._journal_files.append((vfs_id, path)) |
|
686 | 686 | |
|
687 | 687 | def _writeundo(self): |
|
688 | 688 | """write transaction data for possible future undo call""" |
|
689 | 689 | if self._undoname is None: |
|
690 | 690 | return |
|
691 | 691 | |
|
692 | 692 | def undoname(fn: bytes) -> bytes: |
|
693 | 693 | base, name = os.path.split(fn) |
|
694 | 694 | assert name.startswith(self._journal) |
|
695 | 695 | new_name = name.replace(self._journal, self._undoname, 1) |
|
696 | 696 | return os.path.join(base, new_name) |
|
697 | 697 | |
|
698 | 698 | undo_backup_path = b"%s.backupfiles" % self._undoname |
|
699 | 699 | undobackupfile = self._opener.open(undo_backup_path, b'w') |
|
700 | 700 | undobackupfile.write(b'%d\n' % version) |
|
701 | 701 | for l, f, b, c in self._backupentries: |
|
702 | 702 | if not f: # temporary file |
|
703 | 703 | continue |
|
704 | 704 | if not b: |
|
705 | 705 | u = b'' |
|
706 | 706 | else: |
|
707 | 707 | if l not in self._vfsmap and c: |
|
708 | 708 | self._report( |
|
709 | 709 | b"couldn't remove %s: unknown cache location" |
|
710 | 710 | b"%s\n" % (b, l) |
|
711 | 711 | ) |
|
712 | 712 | continue |
|
713 | 713 | vfs = self._vfsmap[l] |
|
714 | 714 | u = undoname(b) |
|
715 | 715 | util.copyfile(vfs.join(b), vfs.join(u), hardlink=True) |
|
716 | 716 | undobackupfile.write(b"%s\0%s\0%s\0%d\n" % (l, f, u, c)) |
|
717 | 717 | undobackupfile.close() |
|
718 | 718 | for vfs, src in self._journal_files: |
|
719 | 719 | dest = undoname(src) |
|
720 | 720 | # if src and dest refer to a same file, vfs.rename is a no-op, |
|
721 | 721 | # leaving both src and dest on disk. delete dest to make sure |
|
722 | 722 | # the rename couldn't be such a no-op. |
|
723 | 723 | vfs.tryunlink(dest) |
|
724 | 724 | try: |
|
725 | 725 | vfs.rename(src, dest) |
|
726 | 726 | except FileNotFoundError: # journal file does not yet exist |
|
727 | 727 | pass |
|
728 | 728 | |
|
729 | 729 | def _abort(self): |
|
730 | 730 | entries = self.readjournal() |
|
731 | 731 | self._count = 0 |
|
732 | 732 | self._usages = 0 |
|
733 | 733 | self._file.close() |
|
734 | 734 | self._backupsfile.close() |
|
735 | 735 | |
|
736 | 736 | quick = self._can_quick_abort(entries) |
|
737 | 737 | try: |
|
738 | 738 | if not quick: |
|
739 | 739 | self._report(_(b"transaction abort!\n")) |
|
740 | 740 | for cat in sorted(self._abortcallback): |
|
741 | 741 | self._abortcallback[cat](self) |
|
742 | 742 | # Prevent double usage and help clear cycles. |
|
743 | 743 | self._abortcallback = None |
|
744 | 744 | if quick: |
|
745 | 745 | self._do_quick_abort(entries) |
|
746 | 746 | else: |
|
747 | 747 | self._do_full_abort(entries) |
|
748 | 748 | finally: |
|
749 | 749 | self._journal = None |
|
750 | 750 | self._releasefn(self, False) # notify failure of transaction |
|
751 | 751 | self._releasefn = None # Help prevent cycles. |
|
752 | 752 | |
|
753 | 753 | def _can_quick_abort(self, entries): |
|
754 | 754 | """False if any semantic content have been written on disk |
|
755 | 755 | |
|
756 | 756 | True if nothing, except temporary files has been writen on disk.""" |
|
757 | 757 | if entries: |
|
758 | 758 | return False |
|
759 | 759 | for e in self._backupentries: |
|
760 | 760 | if e[1]: |
|
761 | 761 | return False |
|
762 | 762 | return True |
|
763 | 763 | |
|
764 | 764 | def _do_quick_abort(self, entries): |
|
765 | 765 | """(Silently) do a quick cleanup (see _can_quick_abort)""" |
|
766 | 766 | assert self._can_quick_abort(entries) |
|
767 | 767 | tmp_files = [e for e in self._backupentries if not e[1]] |
|
768 | 768 | for vfs_id, old_path, tmp_path, xxx in tmp_files: |
|
769 | 769 | vfs = self._vfsmap[vfs_id] |
|
770 | 770 | try: |
|
771 | 771 | vfs.unlink(tmp_path) |
|
772 | 772 | except FileNotFoundError: |
|
773 | 773 | pass |
|
774 | 774 | if self._backupjournal: |
|
775 | 775 | self._opener.unlink(self._backupjournal) |
|
776 | 776 | if self._journal: |
|
777 | 777 | self._opener.unlink(self._journal) |
|
778 | 778 | |
|
779 | 779 | def _do_full_abort(self, entries): |
|
780 | 780 | """(Noisily) rollback all the change introduced by the transaction""" |
|
781 | 781 | try: |
|
782 | 782 | _playback( |
|
783 | 783 | self._journal, |
|
784 | 784 | self._report, |
|
785 | 785 | self._opener, |
|
786 | 786 | self._vfsmap, |
|
787 | 787 | entries, |
|
788 | 788 | self._backupentries, |
|
789 | 789 | False, |
|
790 | 790 | checkambigfiles=self._checkambigfiles, |
|
791 | 791 | ) |
|
792 | 792 | self._report(_(b"rollback completed\n")) |
|
793 | 793 | except BaseException as exc: |
|
794 | 794 | self._report(_(b"rollback failed - please run hg recover\n")) |
|
795 | 795 | self._report( |
|
796 | 796 | _(b"(failure reason: %s)\n") % stringutil.forcebytestr(exc) |
|
797 | 797 | ) |
|
798 | 798 | |
|
799 | 799 | |
|
800 | 800 | BAD_VERSION_MSG = _( |
|
801 | 801 | b"journal was created by a different version of Mercurial\n" |
|
802 | 802 | ) |
|
803 | 803 | |
|
804 | 804 | |
|
805 | 805 | def read_backup_files(report, fp): |
|
806 | 806 | """parse an (already open) backup file an return contained backup entries |
|
807 | 807 | |
|
808 | 808 | entries are in the form: (location, file, backupfile, xxx) |
|
809 | 809 | |
|
810 | 810 | :location: the vfs identifier (vfsmap's key) |
|
811 | 811 | :file: original file path (in the vfs) |
|
812 | 812 | :backupfile: path of the backup (in the vfs) |
|
813 | 813 | :cache: a boolean currently always set to False |
|
814 | 814 | """ |
|
815 | 815 | lines = fp.readlines() |
|
816 | 816 | backupentries = [] |
|
817 | 817 | if lines: |
|
818 | 818 | ver = lines[0][:-1] |
|
819 | 819 | if ver != (b'%d' % version): |
|
820 | 820 | report(BAD_VERSION_MSG) |
|
821 | 821 | else: |
|
822 | 822 | for line in lines[1:]: |
|
823 | 823 | if line: |
|
824 | 824 | # Shave off the trailing newline |
|
825 | 825 | line = line[:-1] |
|
826 | 826 | l, f, b, c = line.split(b'\0') |
|
827 | 827 | backupentries.append((l, f, b, bool(c))) |
|
828 | 828 | return backupentries |
|
829 | 829 | |
|
830 | 830 | |
|
831 | 831 | def rollback( |
|
832 | 832 | opener, |
|
833 | 833 | vfsmap, |
|
834 | 834 | file, |
|
835 | 835 | report, |
|
836 | 836 | checkambigfiles=None, |
|
837 | 837 | skip_journal_pattern=None, |
|
838 | 838 | ): |
|
839 | 839 | """Rolls back the transaction contained in the given file |
|
840 | 840 | |
|
841 | 841 | Reads the entries in the specified file, and the corresponding |
|
842 | 842 | '*.backupfiles' file, to recover from an incomplete transaction. |
|
843 | 843 | |
|
844 | 844 | * `file`: a file containing a list of entries, specifying where |
|
845 | 845 | to truncate each file. The file should contain a list of |
|
846 | 846 | file\0offset pairs, delimited by newlines. The corresponding |
|
847 | 847 | '*.backupfiles' file should contain a list of file\0backupfile |
|
848 | 848 | pairs, delimited by \0. |
|
849 | 849 | |
|
850 | 850 | `checkambigfiles` is a set of (path, vfs-location) tuples, |
|
851 | 851 | which determine whether file stat ambiguity should be avoided at |
|
852 | 852 | restoring corresponded files. |
|
853 | 853 | """ |
|
854 | 854 | entries = [] |
|
855 | 855 | backupentries = [] |
|
856 | 856 | |
|
857 | 857 | with opener.open(file) as fp: |
|
858 | 858 | lines = fp.readlines() |
|
859 | 859 | for l in lines: |
|
860 | 860 | try: |
|
861 | 861 | f, o = l.split(b'\0') |
|
862 | 862 | entries.append((f, int(o))) |
|
863 | 863 | except ValueError: |
|
864 | 864 | report( |
|
865 | 865 | _(b"couldn't read journal entry %r!\n") % pycompat.bytestr(l) |
|
866 | 866 | ) |
|
867 | 867 | |
|
868 | 868 | backupjournal = b"%s.backupfiles" % file |
|
869 | 869 | if opener.exists(backupjournal): |
|
870 | 870 | with opener.open(backupjournal) as fp: |
|
871 | 871 | backupentries = read_backup_files(report, fp) |
|
872 | 872 | if skip_journal_pattern is not None: |
|
873 | 873 | keep = lambda x: not skip_journal_pattern.match(x[1]) |
|
874 | 874 | backupentries = [x for x in backupentries if keep(x)] |
|
875 | 875 | |
|
876 | 876 | _playback( |
|
877 | 877 | file, |
|
878 | 878 | report, |
|
879 | 879 | opener, |
|
880 | 880 | vfsmap, |
|
881 | 881 | entries, |
|
882 | 882 | backupentries, |
|
883 | 883 | checkambigfiles=checkambigfiles, |
|
884 | 884 | ) |
General Comments 0
You need to be logged in to leave comments.
Login now