Mercurial > public > mercurial-scm > hg
annotate hgext/largefiles/lfcommands.py @ 15170:c1a4a3220711
largefiles: fix over-long lines
| author | Matt Mackall <mpm@selenic.com> |
|---|---|
| date | Thu, 29 Sep 2011 17:04:57 -0500 |
| parents | cfccd3bee7b3 |
| children | 547da6115d1d |
| rev | line source |
|---|---|
| 15168 | 1 # Copyright 2009-2010 Gregory P. Ward |
| 2 # Copyright 2009-2010 Intelerad Medical Systems Incorporated | |
| 3 # Copyright 2010-2011 Fog Creek Software | |
| 4 # Copyright 2010-2011 Unity Technologies | |
| 5 # | |
| 6 # This software may be used and distributed according to the terms of the | |
| 7 # GNU General Public License version 2 or any later version. | |
| 8 | |
| 9 '''High-level command functions: lfadd() et. al, plus the cmdtable.''' | |
| 10 | |
| 11 import os | |
| 12 import shutil | |
| 13 | |
| 14 from mercurial import util, match as match_, hg, node, context, error | |
| 15 from mercurial.i18n import _ | |
| 16 | |
| 17 import lfutil | |
| 18 import basestore | |
| 19 | |
| 20 # -- Commands ---------------------------------------------------------- | |
| 21 | |
| 22 def lfconvert(ui, src, dest, *pats, **opts): | |
| 23 '''Convert a normal repository to a largefiles repository | |
| 24 | |
| 25 Convert source repository creating an identical repository, except that all | |
| 26 files that match the patterns given, or are over the given size will be | |
| 27 added as largefiles. The size used to determine whether or not to track a | |
| 28 file as a largefile is the size of the first version of the file. After | |
| 29 running this command you will need to make sure that largefiles is enabled | |
| 30 anywhere you intend to push the new repository.''' | |
| 31 | |
| 32 if opts['tonormal']: | |
| 33 tolfile = False | |
| 34 else: | |
| 35 tolfile = True | |
| 36 size = opts['size'] | |
| 37 if not size: | |
| 38 size = ui.config(lfutil.longname, 'size', default=None) | |
| 39 try: | |
| 40 size = int(size) | |
| 41 except ValueError: | |
|
15170
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
42 raise util.Abort(_('largefiles.size must be integer, was %s\n') |
|
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
43 % size) |
| 15168 | 44 except TypeError: |
| 45 raise util.Abort(_('size must be specified')) | |
| 46 | |
| 47 try: | |
| 48 rsrc = hg.repository(ui, src) | |
| 49 if not rsrc.local(): | |
| 50 raise util.Abort(_('%s is not a local Mercurial repo') % src) | |
| 51 except error.RepoError, err: | |
| 52 ui.traceback() | |
| 53 raise util.Abort(err.args[0]) | |
| 54 if os.path.exists(dest): | |
| 55 if not os.path.isdir(dest): | |
| 56 raise util.Abort(_('destination %s already exists') % dest) | |
| 57 elif os.listdir(dest): | |
| 58 raise util.Abort(_('destination %s is not empty') % dest) | |
| 59 try: | |
| 60 ui.status(_('initializing destination %s\n') % dest) | |
| 61 rdst = hg.repository(ui, dest, create=True) | |
| 62 if not rdst.local(): | |
| 63 raise util.Abort(_('%s is not a local Mercurial repo') % dest) | |
| 64 except error.RepoError: | |
| 65 ui.traceback() | |
| 66 raise util.Abort(_('%s is not a repo') % dest) | |
| 67 | |
| 68 try: | |
| 69 # Lock destination to prevent modification while it is converted to. | |
| 70 # Don't need to lock src because we are just reading from its history | |
| 71 # which can't change. | |
| 72 dst_lock = rdst.lock() | |
| 73 | |
| 74 # Get a list of all changesets in the source. The easy way to do this | |
| 75 # is to simply walk the changelog, using changelog.nodesbewteen(). | |
| 76 # Take a look at mercurial/revlog.py:639 for more details. | |
| 77 # Use a generator instead of a list to decrease memory usage | |
| 78 ctxs = (rsrc[ctx] for ctx in rsrc.changelog.nodesbetween(None, | |
| 79 rsrc.heads())[0]) | |
| 80 revmap = {node.nullid: node.nullid} | |
| 81 if tolfile: | |
| 82 lfiles = set() | |
| 83 normalfiles = set() | |
| 84 if not pats: | |
| 85 pats = ui.config(lfutil.longname, 'patterns', default=()) | |
| 86 if pats: | |
| 87 pats = pats.split(' ') | |
| 88 if pats: | |
| 89 matcher = match_.match(rsrc.root, '', list(pats)) | |
| 90 else: | |
| 91 matcher = None | |
| 92 | |
| 93 lfiletohash = {} | |
| 94 for ctx in ctxs: | |
| 95 ui.progress(_('converting revisions'), ctx.rev(), | |
| 96 unit=_('revision'), total=rsrc['tip'].rev()) | |
| 97 _lfconvert_addchangeset(rsrc, rdst, ctx, revmap, | |
| 98 lfiles, normalfiles, matcher, size, lfiletohash) | |
| 99 ui.progress(_('converting revisions'), None) | |
| 100 | |
| 101 if os.path.exists(rdst.wjoin(lfutil.shortname)): | |
| 102 shutil.rmtree(rdst.wjoin(lfutil.shortname)) | |
| 103 | |
| 104 for f in lfiletohash.keys(): | |
| 105 if os.path.isfile(rdst.wjoin(f)): | |
| 106 os.unlink(rdst.wjoin(f)) | |
| 107 try: | |
| 108 os.removedirs(os.path.dirname(rdst.wjoin(f))) | |
| 109 except: | |
| 110 pass | |
| 111 | |
| 112 else: | |
| 113 for ctx in ctxs: | |
| 114 ui.progress(_('converting revisions'), ctx.rev(), | |
| 115 unit=_('revision'), total=rsrc['tip'].rev()) | |
| 116 _addchangeset(ui, rsrc, rdst, ctx, revmap) | |
| 117 | |
| 118 ui.progress(_('converting revisions'), None) | |
| 119 except: | |
| 120 # we failed, remove the new directory | |
| 121 shutil.rmtree(rdst.root) | |
| 122 raise | |
| 123 finally: | |
| 124 dst_lock.release() | |
| 125 | |
| 126 def _addchangeset(ui, rsrc, rdst, ctx, revmap): | |
| 127 # Convert src parents to dst parents | |
| 128 parents = [] | |
| 129 for p in ctx.parents(): | |
| 130 parents.append(revmap[p.node()]) | |
| 131 while len(parents) < 2: | |
| 132 parents.append(node.nullid) | |
| 133 | |
| 134 # Generate list of changed files | |
| 135 files = set(ctx.files()) | |
| 136 if node.nullid not in parents: | |
| 137 mc = ctx.manifest() | |
| 138 mp1 = ctx.parents()[0].manifest() | |
| 139 mp2 = ctx.parents()[1].manifest() | |
| 140 files |= (set(mp1) | set(mp2)) - set(mc) | |
| 141 for f in mc: | |
| 142 if mc[f] != mp1.get(f, None) or mc[f] != mp2.get(f, None): | |
| 143 files.add(f) | |
| 144 | |
| 145 def getfilectx(repo, memctx, f): | |
| 146 if lfutil.standin(f) in files: | |
| 147 # if the file isn't in the manifest then it was removed | |
| 148 # or renamed, raise IOError to indicate this | |
| 149 try: | |
| 150 fctx = ctx.filectx(lfutil.standin(f)) | |
| 151 except error.LookupError: | |
| 152 raise IOError() | |
| 153 renamed = fctx.renamed() | |
| 154 if renamed: | |
| 155 renamed = lfutil.splitstandin(renamed[0]) | |
| 156 | |
| 157 hash = fctx.data().strip() | |
| 158 path = lfutil.findfile(rsrc, hash) | |
| 159 ### TODO: What if the file is not cached? | |
| 160 data = '' | |
| 161 fd = None | |
| 162 try: | |
| 163 fd = open(path, 'rb') | |
| 164 data = fd.read() | |
| 165 finally: | |
| 166 if fd: fd.close() | |
| 167 return context.memfilectx(f, data, 'l' in fctx.flags(), | |
| 168 'x' in fctx.flags(), renamed) | |
| 169 else: | |
| 170 try: | |
| 171 fctx = ctx.filectx(f) | |
| 172 except error.LookupError: | |
| 173 raise IOError() | |
| 174 renamed = fctx.renamed() | |
| 175 if renamed: | |
| 176 renamed = renamed[0] | |
| 177 data = fctx.data() | |
| 178 if f == '.hgtags': | |
| 179 newdata = [] | |
| 180 for line in data.splitlines(): | |
| 181 id, name = line.split(' ', 1) | |
| 182 newdata.append('%s %s\n' % (node.hex(revmap[node.bin(id)]), | |
| 183 name)) | |
| 184 data = ''.join(newdata) | |
| 185 return context.memfilectx(f, data, 'l' in fctx.flags(), | |
| 186 'x' in fctx.flags(), renamed) | |
| 187 | |
| 188 dstfiles = [] | |
| 189 for file in files: | |
| 190 if lfutil.isstandin(file): | |
| 191 dstfiles.append(lfutil.splitstandin(file)) | |
| 192 else: | |
| 193 dstfiles.append(file) | |
| 194 # Commit | |
| 195 mctx = context.memctx(rdst, parents, ctx.description(), dstfiles, | |
| 196 getfilectx, ctx.user(), ctx.date(), ctx.extra()) | |
| 197 ret = rdst.commitctx(mctx) | |
| 198 rdst.dirstate.setparents(ret) | |
| 199 revmap[ctx.node()] = rdst.changelog.tip() | |
| 200 | |
| 201 def _lfconvert_addchangeset(rsrc, rdst, ctx, revmap, lfiles, normalfiles, | |
| 202 matcher, size, lfiletohash): | |
| 203 # Convert src parents to dst parents | |
| 204 parents = [] | |
| 205 for p in ctx.parents(): | |
| 206 parents.append(revmap[p.node()]) | |
| 207 while len(parents) < 2: | |
| 208 parents.append(node.nullid) | |
| 209 | |
| 210 # Generate list of changed files | |
| 211 files = set(ctx.files()) | |
| 212 if node.nullid not in parents: | |
| 213 mc = ctx.manifest() | |
| 214 mp1 = ctx.parents()[0].manifest() | |
| 215 mp2 = ctx.parents()[1].manifest() | |
| 216 files |= (set(mp1) | set(mp2)) - set(mc) | |
| 217 for f in mc: | |
| 218 if mc[f] != mp1.get(f, None) or mc[f] != mp2.get(f, None): | |
| 219 files.add(f) | |
| 220 | |
| 221 dstfiles = [] | |
| 222 for f in files: | |
| 223 if f not in lfiles and f not in normalfiles: | |
| 224 islfile = _islfile(f, ctx, matcher, size) | |
| 225 # If this file was renamed or copied then copy | |
| 226 # the lfileness of its predecessor | |
| 227 if f in ctx.manifest(): | |
| 228 fctx = ctx.filectx(f) | |
| 229 renamed = fctx.renamed() | |
| 230 renamedlfile = renamed and renamed[0] in lfiles | |
| 231 islfile |= renamedlfile | |
| 232 if 'l' in fctx.flags(): | |
| 233 if renamedlfile: | |
| 234 raise util.Abort( | |
|
15170
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
235 _('Renamed/copied largefile %s becomes symlink') |
|
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
236 % f) |
| 15168 | 237 islfile = False |
| 238 if islfile: | |
| 239 lfiles.add(f) | |
| 240 else: | |
| 241 normalfiles.add(f) | |
| 242 | |
| 243 if f in lfiles: | |
| 244 dstfiles.append(lfutil.standin(f)) | |
| 245 # lfile in manifest if it has not been removed/renamed | |
| 246 if f in ctx.manifest(): | |
| 247 if 'l' in ctx.filectx(f).flags(): | |
| 248 if renamed and renamed[0] in lfiles: | |
| 249 raise util.Abort(_('largefile %s becomes symlink') % f) | |
| 250 | |
| 251 # lfile was modified, update standins | |
| 252 fullpath = rdst.wjoin(f) | |
| 253 lfutil.createdir(os.path.dirname(fullpath)) | |
| 254 m = util.sha1('') | |
| 255 m.update(ctx[f].data()) | |
| 256 hash = m.hexdigest() | |
| 257 if f not in lfiletohash or lfiletohash[f] != hash: | |
| 258 try: | |
| 259 fd = open(fullpath, 'wb') | |
| 260 fd.write(ctx[f].data()) | |
| 261 finally: | |
| 262 if fd: | |
| 263 fd.close() | |
| 264 executable = 'x' in ctx[f].flags() | |
| 265 os.chmod(fullpath, lfutil.getmode(executable)) | |
| 266 lfutil.writestandin(rdst, lfutil.standin(f), hash, | |
| 267 executable) | |
| 268 lfiletohash[f] = hash | |
| 269 else: | |
| 270 # normal file | |
| 271 dstfiles.append(f) | |
| 272 | |
| 273 def getfilectx(repo, memctx, f): | |
| 274 if lfutil.isstandin(f): | |
| 275 # if the file isn't in the manifest then it was removed | |
| 276 # or renamed, raise IOError to indicate this | |
| 277 srcfname = lfutil.splitstandin(f) | |
| 278 try: | |
| 279 fctx = ctx.filectx(srcfname) | |
| 280 except error.LookupError: | |
| 281 raise IOError() | |
| 282 renamed = fctx.renamed() | |
| 283 if renamed: | |
| 284 # standin is always a lfile because lfileness | |
| 285 # doesn't change after rename or copy | |
| 286 renamed = lfutil.standin(renamed[0]) | |
| 287 | |
| 288 return context.memfilectx(f, lfiletohash[srcfname], 'l' in | |
| 289 fctx.flags(), 'x' in fctx.flags(), renamed) | |
| 290 else: | |
| 291 try: | |
| 292 fctx = ctx.filectx(f) | |
| 293 except error.LookupError: | |
| 294 raise IOError() | |
| 295 renamed = fctx.renamed() | |
| 296 if renamed: | |
| 297 renamed = renamed[0] | |
| 298 | |
| 299 data = fctx.data() | |
| 300 if f == '.hgtags': | |
| 301 newdata = [] | |
| 302 for line in data.splitlines(): | |
| 303 id, name = line.split(' ', 1) | |
| 304 newdata.append('%s %s\n' % (node.hex(revmap[node.bin(id)]), | |
| 305 name)) | |
| 306 data = ''.join(newdata) | |
| 307 return context.memfilectx(f, data, 'l' in fctx.flags(), | |
| 308 'x' in fctx.flags(), renamed) | |
| 309 | |
| 310 # Commit | |
| 311 mctx = context.memctx(rdst, parents, ctx.description(), dstfiles, | |
| 312 getfilectx, ctx.user(), ctx.date(), ctx.extra()) | |
| 313 ret = rdst.commitctx(mctx) | |
| 314 rdst.dirstate.setparents(ret) | |
| 315 revmap[ctx.node()] = rdst.changelog.tip() | |
| 316 | |
| 317 def _islfile(file, ctx, matcher, size): | |
| 318 ''' | |
| 319 A file is a lfile if it matches a pattern or is over | |
| 320 the given size. | |
| 321 ''' | |
| 322 # Never store hgtags or hgignore as lfiles | |
| 323 if file == '.hgtags' or file == '.hgignore' or file == '.hgsigs': | |
| 324 return False | |
| 325 if matcher and matcher(file): | |
| 326 return True | |
| 327 try: | |
| 328 return ctx.filectx(file).size() >= size * 1024 * 1024 | |
| 329 except error.LookupError: | |
| 330 return False | |
| 331 | |
| 332 def uploadlfiles(ui, rsrc, rdst, files): | |
| 333 '''upload largefiles to the central store''' | |
| 334 | |
| 335 # Don't upload locally. All largefiles are in the system wide cache | |
| 336 # so the other repo can just get them from there. | |
| 337 if not files or rdst.local(): | |
| 338 return | |
| 339 | |
| 340 store = basestore._openstore(rsrc, rdst, put=True) | |
| 341 | |
| 342 at = 0 | |
| 343 files = filter(lambda h: not store.exists(h), files) | |
| 344 for hash in files: | |
|
15170
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
345 ui.progress(_('uploading largefiles'), at, unit='largefile', |
|
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
346 total=len(files)) |
| 15168 | 347 source = lfutil.findfile(rsrc, hash) |
| 348 if not source: | |
|
15170
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
349 raise util.Abort(_('Missing largefile %s needs to be uploaded') |
|
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
350 % hash) |
| 15168 | 351 # XXX check for errors here |
| 352 store.put(source, hash) | |
| 353 at += 1 | |
| 354 ui.progress('uploading largefiles', None) | |
| 355 | |
| 356 def verifylfiles(ui, repo, all=False, contents=False): | |
| 357 '''Verify that every big file revision in the current changeset | |
| 358 exists in the central store. With --contents, also verify that | |
| 359 the contents of each big file revision are correct (SHA-1 hash | |
| 360 matches the revision ID). With --all, check every changeset in | |
| 361 this repository.''' | |
| 362 if all: | |
| 363 # Pass a list to the function rather than an iterator because we know a | |
| 364 # list will work. | |
| 365 revs = range(len(repo)) | |
| 366 else: | |
| 367 revs = ['.'] | |
| 368 | |
| 369 store = basestore._openstore(repo) | |
| 370 return store.verify(revs, contents=contents) | |
| 371 | |
| 372 def cachelfiles(ui, repo, node): | |
| 373 '''cachelfiles ensures that all largefiles needed by the specified revision | |
| 374 are present in the repository's largefile cache. | |
| 375 | |
| 376 returns a tuple (cached, missing). cached is the list of files downloaded | |
| 377 by this operation; missing is the list of files that were needed but could | |
| 378 not be found.''' | |
| 379 lfiles = lfutil.listlfiles(repo, node) | |
| 380 toget = [] | |
| 381 | |
| 382 for lfile in lfiles: | |
| 383 expectedhash = repo[node][lfutil.standin(lfile)].data().strip() | |
| 384 # if it exists and its hash matches, it might have been locally | |
| 385 # modified before updating and the user chose 'local'. in this case, | |
| 386 # it will not be in any store, so don't look for it. | |
| 387 if (not os.path.exists(repo.wjoin(lfile)) \ | |
| 388 or expectedhash != lfutil.hashfile(repo.wjoin(lfile))) and \ | |
| 389 not lfutil.findfile(repo, expectedhash): | |
| 390 toget.append((lfile, expectedhash)) | |
| 391 | |
| 392 if toget: | |
| 393 store = basestore._openstore(repo) | |
| 394 ret = store.get(toget) | |
| 395 return ret | |
| 396 | |
| 397 return ([], []) | |
| 398 | |
| 399 def updatelfiles(ui, repo, filelist=None, printmessage=True): | |
| 400 wlock = repo.wlock() | |
| 401 try: | |
| 402 lfdirstate = lfutil.openlfdirstate(ui, repo) | |
| 403 lfiles = set(lfutil.listlfiles(repo)) | set(lfdirstate) | |
| 404 | |
| 405 if filelist is not None: | |
| 406 lfiles = [f for f in lfiles if f in filelist] | |
| 407 | |
| 408 printed = False | |
| 409 if printmessage and lfiles: | |
| 410 ui.status(_('getting changed largefiles\n')) | |
| 411 printed = True | |
| 412 cachelfiles(ui, repo, '.') | |
| 413 | |
| 414 updated, removed = 0, 0 | |
| 415 for i in map(lambda f: _updatelfile(repo, lfdirstate, f), lfiles): | |
| 416 # increment the appropriate counter according to _updatelfile's | |
| 417 # return value | |
| 418 updated += i > 0 and i or 0 | |
| 419 removed -= i < 0 and i or 0 | |
| 420 if printmessage and (removed or updated) and not printed: | |
| 421 ui.status(_('getting changed largefiles\n')) | |
| 422 printed = True | |
| 423 | |
| 424 lfdirstate.write() | |
| 425 if printed and printmessage: | |
| 426 ui.status(_('%d largefiles updated, %d removed\n') % (updated, | |
| 427 removed)) | |
| 428 finally: | |
| 429 wlock.release() | |
| 430 | |
| 431 def _updatelfile(repo, lfdirstate, lfile): | |
| 432 '''updates a single largefile and copies the state of its standin from | |
| 433 the repository's dirstate to its state in the lfdirstate. | |
| 434 | |
| 435 returns 1 if the file was modified, -1 if the file was removed, 0 if the | |
| 436 file was unchanged, and None if the needed largefile was missing from the | |
| 437 cache.''' | |
| 438 ret = 0 | |
| 439 abslfile = repo.wjoin(lfile) | |
| 440 absstandin = repo.wjoin(lfutil.standin(lfile)) | |
| 441 if os.path.exists(absstandin): | |
| 442 if os.path.exists(absstandin+'.orig'): | |
| 443 shutil.copyfile(abslfile, abslfile+'.orig') | |
| 444 expecthash = lfutil.readstandin(repo, lfile) | |
| 445 if expecthash != '' and \ | |
| 446 (not os.path.exists(abslfile) or \ | |
| 447 expecthash != lfutil.hashfile(abslfile)): | |
| 448 if not lfutil.copyfromcache(repo, expecthash, lfile): | |
| 449 return None # don't try to set the mode or update the dirstate | |
| 450 ret = 1 | |
| 451 mode = os.stat(absstandin).st_mode | |
| 452 if mode != os.stat(abslfile).st_mode: | |
| 453 os.chmod(abslfile, mode) | |
| 454 ret = 1 | |
| 455 else: | |
| 456 if os.path.exists(abslfile): | |
| 457 os.unlink(abslfile) | |
| 458 ret = -1 | |
| 459 state = repo.dirstate[lfutil.standin(lfile)] | |
| 460 if state == 'n': | |
| 461 lfdirstate.normal(lfile) | |
| 462 elif state == 'r': | |
| 463 lfdirstate.remove(lfile) | |
| 464 elif state == 'a': | |
| 465 lfdirstate.add(lfile) | |
| 466 elif state == '?': | |
| 467 try: | |
| 468 # Mercurial >= 1.9 | |
| 469 lfdirstate.drop(lfile) | |
| 470 except AttributeError: | |
| 471 # Mercurial <= 1.8 | |
| 472 lfdirstate.forget(lfile) | |
| 473 return ret | |
| 474 | |
| 475 # -- hg commands declarations ------------------------------------------------ | |
| 476 | |
| 477 | |
| 478 cmdtable = { | |
| 479 'lfconvert': (lfconvert, | |
| 480 [('s', 'size', 0, 'All files over this size (in megabytes) ' | |
|
15170
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
481 'will be considered largefiles. This can also be specified ' |
|
c1a4a3220711
largefiles: fix over-long lines
Matt Mackall <mpm@selenic.com>
parents:
15168
diff
changeset
|
482 'in your hgrc as [largefiles].size.'), |
| 15168 | 483 ('','tonormal',False, |
| 484 'Convert from a largefiles repo to a normal repo')], | |
| 485 _('hg lfconvert SOURCE DEST [FILE ...]')), | |
| 486 } |