largefiles: access to specific fields only if largefiles enabled (issue4547)

Even if largefiles extension is enabled in a repository, "repo"
object, which isn't "largefiles.reposetup()"-ed, is passed to
overridden functions in the cases below unexpectedly, because
extensions are enabled for each repositories strictly.

  (1) clone without -U:
  (2) pull with -U:
  (3) pull with --rebase:

    combination of "enabled@src", "disabled@dst" and
    "not-required@src" cause this situation.

       largefiles     requirement
    @src     @dst     @src            result
    -------- -------- --------------- --------------------
    enabled  disabled not-required    aborted unexpectedly
                      required        requirement error (intentional)
    -------- -------- --------------- --------------------
    enabled  enabled  *               success
    -------- -------- --------------- --------------------
    disabled enabled  *               success (only for "pull")
    -------- -------- --------------- --------------------
    disabled disabled not-required    success
                      required        requirement error (intentional)
    -------- -------- --------------- --------------------

  (4) update/revert with a subrepo disabling largefiles

In these cases, overridden functions cause accessing to largefiles
specific fields of not "largefiles.reposetup()"-ed "repo" object, and
execution is aborted.

  - (1), (2), (4) cause accessing to "_lfstatuswriters" in
    "getstatuswriter()" invoked via "updatelfiles()"

  - (3) causes accessing to "_lfcommithooks" in "overriderebase()"

For safe accessing to these fields, this patch examines whether passed
"repo" object is "largefiles.reposetup()"-ed or not before accessing
to them.

This patch chooses examining existence of newly introduced
"_largefilesenabled" instead of "_lfcommithooks" and
"_lfstatuswriters" directly, because the former is better name for the
generic "largefiles is enabled in this repo" mark than the latter.

In the future, all other overridden functions should avoid largefiles
specific processing for efficiency, and "_largefilesenabled" is better
also for such purpose.

BTW, "lfstatus" can't be used for such purpose, because some code
paths set it forcibly regardless of existence of it in specified
"repo" object.

hgext/largefiles/lfutil.py

@@ -1,586 +1,586 b''
 # Copyright 2009-2010 Gregory P. Ward
 # Copyright 2009-2010 Intelerad Medical Systems Incorporated
 # Copyright 2010-2011 Fog Creek Software
 # Copyright 2010-2011 Unity Technologies
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 '''largefiles utility code: must not import other modules in this package.'''
 
 import os
 import platform
 import shutil
 import stat
 import copy
 
 from mercurial import dirstate, httpconnection, match as match_, util, scmutil
 from mercurial.i18n import _
 from mercurial import node
 
 shortname = '.hglf'
 shortnameslash = shortname + '/'
 longname = 'largefiles'
 
 
 # -- Private worker functions ------------------------------------------
 
 def getminsize(ui, assumelfiles, opt, default=10):
     lfsize = opt
     if not lfsize and assumelfiles:
         lfsize = ui.config(longname, 'minsize', default=default)
     if lfsize:
         try:
             lfsize = float(lfsize)
         except ValueError:
             raise util.Abort(_('largefiles: size must be number (not %s)\n')
                              % lfsize)
     if lfsize is None:
         raise util.Abort(_('minimum size for largefiles must be specified'))
     return lfsize
 
 def link(src, dest):
     util.makedirs(os.path.dirname(dest))
     try:
         util.oslink(src, dest)
     except OSError:
         # if hardlinks fail, fallback on atomic copy
         dst = util.atomictempfile(dest)
         for chunk in util.filechunkiter(open(src, 'rb')):
             dst.write(chunk)
         dst.close()
         os.chmod(dest, os.stat(src).st_mode)
 
 def usercachepath(ui, hash):
     path = ui.configpath(longname, 'usercache', None)
     if path:
         path = os.path.join(path, hash)
     else:
         if os.name == 'nt':
             appdata = os.getenv('LOCALAPPDATA', os.getenv('APPDATA'))
             if appdata:
                 path = os.path.join(appdata, longname, hash)
         elif platform.system() == 'Darwin':
             home = os.getenv('HOME')
             if home:
                 path = os.path.join(home, 'Library', 'Caches',
                                     longname, hash)
         elif os.name == 'posix':
             path = os.getenv('XDG_CACHE_HOME')
             if path:
                 path = os.path.join(path, longname, hash)
             else:
                 home = os.getenv('HOME')
                 if home:
                     path = os.path.join(home, '.cache', longname, hash)
         else:
             raise util.Abort(_('unknown operating system: %s\n') % os.name)
     return path
 
 def inusercache(ui, hash):
     path = usercachepath(ui, hash)
     return path and os.path.exists(path)
 
 def findfile(repo, hash):
     if instore(repo, hash):
         repo.ui.note(_('found %s in store\n') % hash)
         return storepath(repo, hash)
     elif inusercache(repo.ui, hash):
         repo.ui.note(_('found %s in system cache\n') % hash)
         path = storepath(repo, hash)
         link(usercachepath(repo.ui, hash), path)
         return path
     return None
 
 class largefilesdirstate(dirstate.dirstate):
     def __getitem__(self, key):
         return super(largefilesdirstate, self).__getitem__(unixpath(key))
     def normal(self, f):
         return super(largefilesdirstate, self).normal(unixpath(f))
     def remove(self, f):
         return super(largefilesdirstate, self).remove(unixpath(f))
     def add(self, f):
         return super(largefilesdirstate, self).add(unixpath(f))
     def drop(self, f):
         return super(largefilesdirstate, self).drop(unixpath(f))
     def forget(self, f):
         return super(largefilesdirstate, self).forget(unixpath(f))
     def normallookup(self, f):
         return super(largefilesdirstate, self).normallookup(unixpath(f))
     def _ignore(self, f):
         return False
 
 def openlfdirstate(ui, repo, create=True):
     '''
     Return a dirstate object that tracks largefiles: i.e. its root is
     the repo root, but it is saved in .hg/largefiles/dirstate.
     '''
     lfstoredir = repo.join(longname)
     opener = scmutil.opener(lfstoredir)
     lfdirstate = largefilesdirstate(opener, ui, repo.root,
                                      repo.dirstate._validate)
 
     # If the largefiles dirstate does not exist, populate and create
     # it. This ensures that we create it on the first meaningful
     # largefiles operation in a new clone.
     if create and not os.path.exists(os.path.join(lfstoredir, 'dirstate')):
         matcher = getstandinmatcher(repo)
         standins = repo.dirstate.walk(matcher, [], False, False)
 
         if len(standins) > 0:
             util.makedirs(lfstoredir)
 
         for standin in standins:
             lfile = splitstandin(standin)
             lfdirstate.normallookup(lfile)
     return lfdirstate
 
 def lfdirstatestatus(lfdirstate, repo):
     wctx = repo['.']
     match = match_.always(repo.root, repo.getcwd())
     unsure, s = lfdirstate.status(match, [], False, False, False)
     modified, clean = s.modified, s.clean
     for lfile in unsure:
         try:
             fctx = wctx[standin(lfile)]
         except LookupError:
             fctx = None
         if not fctx or fctx.data().strip() != hashfile(repo.wjoin(lfile)):
             modified.append(lfile)
         else:
             clean.append(lfile)
             lfdirstate.normal(lfile)
     return s
 
 def listlfiles(repo, rev=None, matcher=None):
     '''return a list of largefiles in the working copy or the
     specified changeset'''
 
     if matcher is None:
         matcher = getstandinmatcher(repo)
 
     # ignore unknown files in working directory
     return [splitstandin(f)
             for f in repo[rev].walk(matcher)
             if rev is not None or repo.dirstate[f] != '?']
 
 def instore(repo, hash):
     return os.path.exists(storepath(repo, hash))
 
 def storepath(repo, hash):
     return repo.join(os.path.join(longname, hash))
 
 def copyfromcache(repo, hash, filename):
     '''Copy the specified largefile from the repo or system cache to
     filename in the repository. Return true on success or false if the
     file was not found in either cache (which should not happened:
     this is meant to be called only after ensuring that the needed
     largefile exists in the cache).'''
     path = findfile(repo, hash)
     if path is None:
         return False
     util.makedirs(os.path.dirname(repo.wjoin(filename)))
     # The write may fail before the file is fully written, but we
     # don't use atomic writes in the working copy.
     shutil.copy(path, repo.wjoin(filename))
     return True
 
 def copytostore(repo, rev, file, uploaded=False):
     hash = readstandin(repo, file, rev)
     if instore(repo, hash):
         return
     copytostoreabsolute(repo, repo.wjoin(file), hash)
 
 def copyalltostore(repo, node):
     '''Copy all largefiles in a given revision to the store'''
 
     ctx = repo[node]
     for filename in ctx.files():
         if isstandin(filename) and filename in ctx.manifest():
             realfile = splitstandin(filename)
             copytostore(repo, ctx.node(), realfile)
 
 
 def copytostoreabsolute(repo, file, hash):
     if inusercache(repo.ui, hash):
         link(usercachepath(repo.ui, hash), storepath(repo, hash))
     else:
         util.makedirs(os.path.dirname(storepath(repo, hash)))
         dst = util.atomictempfile(storepath(repo, hash),
                                   createmode=repo.store.createmode)
         for chunk in util.filechunkiter(open(file, 'rb')):
             dst.write(chunk)
         dst.close()
         linktousercache(repo, hash)
 
 def linktousercache(repo, hash):
     path = usercachepath(repo.ui, hash)
     if path:
         link(storepath(repo, hash), path)
 
 def getstandinmatcher(repo, pats=[], opts={}):
     '''Return a match object that applies pats to the standin directory'''
     standindir = repo.wjoin(shortname)
     if pats:
         pats = [os.path.join(standindir, pat) for pat in pats]
     else:
         # no patterns: relative to repo root
         pats = [standindir]
     # no warnings about missing files or directories
     match = scmutil.match(repo[None], pats, opts)
     match.bad = lambda f, msg: None
     return match
 
 def composestandinmatcher(repo, rmatcher):
     '''Return a matcher that accepts standins corresponding to the
     files accepted by rmatcher. Pass the list of files in the matcher
     as the paths specified by the user.'''
     smatcher = getstandinmatcher(repo, rmatcher.files())
     isstandin = smatcher.matchfn
     def composedmatchfn(f):
         return isstandin(f) and rmatcher.matchfn(splitstandin(f))
     smatcher.matchfn = composedmatchfn
 
     return smatcher
 
 def standin(filename):
     '''Return the repo-relative path to the standin for the specified big
     file.'''
     # Notes:
     # 1) Some callers want an absolute path, but for instance addlargefiles
     #    needs it repo-relative so it can be passed to repo[None].add().  So
     #    leave it up to the caller to use repo.wjoin() to get an absolute path.
     # 2) Join with '/' because that's what dirstate always uses, even on
     #    Windows. Change existing separator to '/' first in case we are
     #    passed filenames from an external source (like the command line).
     return shortnameslash + util.pconvert(filename)
 
 def isstandin(filename):
     '''Return true if filename is a big file standin. filename must be
     in Mercurial's internal form (slash-separated).'''
     return filename.startswith(shortnameslash)
 
 def splitstandin(filename):
     # Split on / because that's what dirstate always uses, even on Windows.
     # Change local separator to / first just in case we are passed filenames
     # from an external source (like the command line).
     bits = util.pconvert(filename).split('/', 1)
     if len(bits) == 2 and bits[0] == shortname:
         return bits[1]
     else:
         return None
 
 def updatestandin(repo, standin):
     file = repo.wjoin(splitstandin(standin))
     if os.path.exists(file):
         hash = hashfile(file)
         executable = getexecutable(file)
         writestandin(repo, standin, hash, executable)
 
 def readstandin(repo, filename, node=None):
     '''read hex hash from standin for filename at given node, or working
     directory if no node is given'''
     return repo[node][standin(filename)].data().strip()
 
 def writestandin(repo, standin, hash, executable):
     '''write hash to <repo.root>/<standin>'''
     repo.wwrite(standin, hash + '\n', executable and 'x' or '')
 
 def copyandhash(instream, outfile):
     '''Read bytes from instream (iterable) and write them to outfile,
     computing the SHA-1 hash of the data along the way. Return the hash.'''
     hasher = util.sha1('')
     for data in instream:
         hasher.update(data)
         outfile.write(data)
     return hasher.hexdigest()
 
 def hashrepofile(repo, file):
     return hashfile(repo.wjoin(file))
 
 def hashfile(file):
     if not os.path.exists(file):
         return ''
     hasher = util.sha1('')
     fd = open(file, 'rb')
     for data in util.filechunkiter(fd, 128 * 1024):
         hasher.update(data)
     fd.close()
     return hasher.hexdigest()
 
 def getexecutable(filename):
     mode = os.stat(filename).st_mode
     return ((mode & stat.S_IXUSR) and
             (mode & stat.S_IXGRP) and
             (mode & stat.S_IXOTH))
 
 def urljoin(first, second, *arg):
     def join(left, right):
         if not left.endswith('/'):
             left += '/'
         if right.startswith('/'):
             right = right[1:]
         return left + right
 
     url = join(first, second)
     for a in arg:
         url = join(url, a)
     return url
 
 def hexsha1(data):
     """hexsha1 returns the hex-encoded sha1 sum of the data in the file-like
     object data"""
     h = util.sha1()
     for chunk in util.filechunkiter(data):
         h.update(chunk)
     return h.hexdigest()
 
 def httpsendfile(ui, filename):
     return httpconnection.httpsendfile(ui, filename, 'rb')
 
 def unixpath(path):
     '''Return a version of path normalized for use with the lfdirstate.'''
     return util.pconvert(os.path.normpath(path))
 
 def islfilesrepo(repo):
     if ('largefiles' in repo.requirements and
             util.any(shortnameslash in f[0] for f in repo.store.datafiles())):
         return True
 
     return util.any(openlfdirstate(repo.ui, repo, False))
 
 class storeprotonotcapable(Exception):
     def __init__(self, storetypes):
         self.storetypes = storetypes
 
 def getstandinsstate(repo):
     standins = []
     matcher = getstandinmatcher(repo)
     for standin in repo.dirstate.walk(matcher, [], False, False):
         lfile = splitstandin(standin)
         try:
             hash = readstandin(repo, lfile)
         except IOError:
             hash = None
         standins.append((lfile, hash))
     return standins
 
 def synclfdirstate(repo, lfdirstate, lfile, normallookup):
     lfstandin = standin(lfile)
     if lfstandin in repo.dirstate:
         stat = repo.dirstate._map[lfstandin]
         state, mtime = stat[0], stat[3]
     else:
         state, mtime = '?', -1
     if state == 'n':
         if normallookup or mtime < 0:
             # state 'n' doesn't ensure 'clean' in this case
             lfdirstate.normallookup(lfile)
         else:
             lfdirstate.normal(lfile)
     elif state == 'm':
         lfdirstate.normallookup(lfile)
     elif state == 'r':
         lfdirstate.remove(lfile)
     elif state == 'a':
         lfdirstate.add(lfile)
     elif state == '?':
         lfdirstate.drop(lfile)
 
 def markcommitted(orig, ctx, node):
     repo = ctx._repo
 
     orig(node)
 
     # ATTENTION: "ctx.files()" may differ from "repo[node].files()"
     # because files coming from the 2nd parent are omitted in the latter.
     #
     # The former should be used to get targets of "synclfdirstate",
     # because such files:
     # - are marked as "a" by "patch.patch()" (e.g. via transplant), and
     # - have to be marked as "n" after commit, but
     # - aren't listed in "repo[node].files()"
 
     lfdirstate = openlfdirstate(repo.ui, repo)
     for f in ctx.files():
         if isstandin(f):
             lfile = splitstandin(f)
             synclfdirstate(repo, lfdirstate, lfile, False)
     lfdirstate.write()
 
     # As part of committing, copy all of the largefiles into the cache.
     copyalltostore(repo, node)
 
 def getlfilestoupdate(oldstandins, newstandins):
     changedstandins = set(oldstandins).symmetric_difference(set(newstandins))
     filelist = []
     for f in changedstandins:
         if f[0] not in filelist:
             filelist.append(f[0])
     return filelist
 
 def getlfilestoupload(repo, missing, addfunc):
     for i, n in enumerate(missing):
         repo.ui.progress(_('finding outgoing largefiles'), i,
             unit=_('revision'), total=len(missing))
         parents = [p for p in repo.changelog.parents(n) if p != node.nullid]
 
         oldlfstatus = repo.lfstatus
         repo.lfstatus = False
         try:
             ctx = repo[n]
         finally:
             repo.lfstatus = oldlfstatus
 
         files = set(ctx.files())
         if len(parents) == 2:
             mc = ctx.manifest()
             mp1 = ctx.parents()[0].manifest()
             mp2 = ctx.parents()[1].manifest()
             for f in mp1:
                 if f not in mc:
                     files.add(f)
             for f in mp2:
                 if f not in mc:
                     files.add(f)
             for f in mc:
                 if mc[f] != mp1.get(f, None) or mc[f] != mp2.get(f, None):
                     files.add(f)
         for fn in files:
             if isstandin(fn) and fn in ctx:
                 addfunc(fn, ctx[fn].data().strip())
     repo.ui.progress(_('finding outgoing largefiles'), None)
 
 def updatestandinsbymatch(repo, match):
     '''Update standins in the working directory according to specified match
 
     This returns (possibly modified) ``match`` object to be used for
     subsequent commit process.
     '''
 
     ui = repo.ui
 
     # Case 1: user calls commit with no specific files or
     # include/exclude patterns: refresh and commit all files that
     # are "dirty".
     if match is None or match.always():
         # Spend a bit of time here to get a list of files we know
         # are modified so we can compare only against those.
         # It can cost a lot of time (several seconds)
         # otherwise to update all standins if the largefiles are
         # large.
         lfdirstate = openlfdirstate(ui, repo)
         dirtymatch = match_.always(repo.root, repo.getcwd())
         unsure, s = lfdirstate.status(dirtymatch, [], False, False,
                                       False)
         modifiedfiles = unsure + s.modified + s.added + s.removed
         lfiles = listlfiles(repo)
         # this only loops through largefiles that exist (not
         # removed/renamed)
         for lfile in lfiles:
             if lfile in modifiedfiles:
                 if os.path.exists(
                         repo.wjoin(standin(lfile))):
                     # this handles the case where a rebase is being
                     # performed and the working copy is not updated
                     # yet.
                     if os.path.exists(repo.wjoin(lfile)):
                         updatestandin(repo,
                             standin(lfile))
 
         return match
 
     lfiles = listlfiles(repo)
     match._files = repo._subdirlfs(match.files(), lfiles)
 
     # Case 2: user calls commit with specified patterns: refresh
     # any matching big files.
     smatcher = composestandinmatcher(repo, match)
     standins = repo.dirstate.walk(smatcher, [], False, False)
 
     # No matching big files: get out of the way and pass control to
     # the usual commit() method.
     if not standins:
         return match
 
     # Refresh all matching big files.  It's possible that the
     # commit will end up failing, in which case the big files will
     # stay refreshed.  No harm done: the user modified them and
     # asked to commit them, so sooner or later we're going to
     # refresh the standins.  Might as well leave them refreshed.
     lfdirstate = openlfdirstate(ui, repo)
     for fstandin in standins:
         lfile = splitstandin(fstandin)
         if lfdirstate[lfile] != 'r':
             updatestandin(repo, fstandin)
 
     # Cook up a new matcher that only matches regular files or
     # standins corresponding to the big files requested by the
     # user.  Have to modify _files to prevent commit() from
     # complaining "not tracked" for big files.
     match = copy.copy(match)
     origmatchfn = match.matchfn
 
     # Check both the list of largefiles and the list of
     # standins because if a largefile was removed, it
     # won't be in the list of largefiles at this point
     match._files += sorted(standins)
 
     actualfiles = []
     for f in match._files:
         fstandin = standin(f)
 
         # ignore known largefiles and standins
         if f in lfiles or fstandin in standins:
             continue
 
         actualfiles.append(f)
     match._files = actualfiles
 
     def matchfn(f):
         if origmatchfn(f):
             return f not in lfiles
         else:
             return f in standins
 
     match.matchfn = matchfn
 
     return match
 
 class automatedcommithook(object):
     '''Stateful hook to update standins at the 1st commit of resuming
 
     For efficiency, updating standins in the working directory should
     be avoided while automated committing (like rebase, transplant and
     so on), because they should be updated before committing.
 
     But the 1st commit of resuming automated committing (e.g. ``rebase
     --continue``) should update them, because largefiles may be
     modified manually.
     '''
     def __init__(self, resuming):
         self.resuming = resuming
 
     def __call__(self, repo, match):
         if self.resuming:
             self.resuming = False # avoids updating at subsequent commits
             return updatestandinsbymatch(repo, match)
         else:
             return match
 
 def getstatuswriter(ui, repo, forcibly=None):
     '''Return the function to write largefiles specific status out
 
     If ``forcibly`` is ``None``, this returns the last element of
     ``repo._lfstatuswriters`` as "default" writer function.
 
     Otherwise, this returns the function to always write out (or
     ignore if ``not forcibly``) status.
     '''
-    if forcibly is None:
+    if forcibly is None and util.safehasattr(repo, '_largefilesenabled'):
         return repo._lfstatuswriters[-1]
     else:
         if forcibly:
             return ui.status # forcibly WRITE OUT
         else:
             return lambda *msg, **opts: None # forcibly IGNORE

hgext/largefiles/overrides.py

@@ -1,1346 +1,1349 b''
 # Copyright 2009-2010 Gregory P. Ward
 # Copyright 2009-2010 Intelerad Medical Systems Incorporated
 # Copyright 2010-2011 Fog Creek Software
 # Copyright 2010-2011 Unity Technologies
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 '''Overridden Mercurial commands and functions for the largefiles extension'''
 
 import os
 import copy
 
 from mercurial import hg, util, cmdutil, scmutil, match as match_, \
         archival, pathutil, revset
 from mercurial.i18n import _
 from mercurial.node import hex
 
 import lfutil
 import lfcommands
 import basestore
 
 # -- Utility functions: commonly/repeatedly needed functionality ---------------
 
 def composelargefilematcher(match, manifest):
     '''create a matcher that matches only the largefiles in the original
     matcher'''
     m = copy.copy(match)
     lfile = lambda f: lfutil.standin(f) in manifest
     m._files = filter(lfile, m._files)
     m._fmap = set(m._files)
     m._always = False
     origmatchfn = m.matchfn
     m.matchfn = lambda f: lfile(f) and origmatchfn(f)
     return m
 
 def composenormalfilematcher(match, manifest, exclude=None):
     excluded = set()
     if exclude is not None:
         excluded.update(exclude)
 
     m = copy.copy(match)
     notlfile = lambda f: not (lfutil.isstandin(f) or lfutil.standin(f) in
             manifest or f in excluded)
     m._files = filter(notlfile, m._files)
     m._fmap = set(m._files)
     m._always = False
     origmatchfn = m.matchfn
     m.matchfn = lambda f: notlfile(f) and origmatchfn(f)
     return m
 
 def installnormalfilesmatchfn(manifest):
     '''installmatchfn with a matchfn that ignores all largefiles'''
     def overridematch(ctx, pats=[], opts={}, globbed=False,
             default='relpath'):
         match = oldmatch(ctx, pats, opts, globbed, default)
         return composenormalfilematcher(match, manifest)
     oldmatch = installmatchfn(overridematch)
 
 def installmatchfn(f):
     '''monkey patch the scmutil module with a custom match function.
     Warning: it is monkey patching the _module_ on runtime! Not thread safe!'''
     oldmatch = scmutil.match
     setattr(f, 'oldmatch', oldmatch)
     scmutil.match = f
     return oldmatch
 
 def restorematchfn():
     '''restores scmutil.match to what it was before installmatchfn
     was called.  no-op if scmutil.match is its original function.
 
     Note that n calls to installmatchfn will require n calls to
     restore the original matchfn.'''
     scmutil.match = getattr(scmutil.match, 'oldmatch')
 
 def installmatchandpatsfn(f):
     oldmatchandpats = scmutil.matchandpats
     setattr(f, 'oldmatchandpats', oldmatchandpats)
     scmutil.matchandpats = f
     return oldmatchandpats
 
 def restorematchandpatsfn():
     '''restores scmutil.matchandpats to what it was before
     installmatchandpatsfn was called. No-op if scmutil.matchandpats
     is its original function.
 
     Note that n calls to installmatchandpatsfn will require n calls
     to restore the original matchfn.'''
     scmutil.matchandpats = getattr(scmutil.matchandpats, 'oldmatchandpats',
             scmutil.matchandpats)
 
 def addlargefiles(ui, repo, isaddremove, matcher, **opts):
     large = opts.get('large')
     lfsize = lfutil.getminsize(
         ui, lfutil.islfilesrepo(repo), opts.get('lfsize'))
 
     lfmatcher = None
     if lfutil.islfilesrepo(repo):
         lfpats = ui.configlist(lfutil.longname, 'patterns', default=[])
         if lfpats:
             lfmatcher = match_.match(repo.root, '', list(lfpats))
 
     lfnames = []
     m = copy.copy(matcher)
     m.bad = lambda x, y: None
     wctx = repo[None]
     for f in repo.walk(m):
         exact = m.exact(f)
         lfile = lfutil.standin(f) in wctx
         nfile = f in wctx
         exists = lfile or nfile
 
         # addremove in core gets fancy with the name, add doesn't
         if isaddremove:
             name = m.uipath(f)
         else:
             name = m.rel(f)
 
         # Don't warn the user when they attempt to add a normal tracked file.
         # The normal add code will do that for us.
         if exact and exists:
             if lfile:
                 ui.warn(_('%s already a largefile\n') % name)
             continue
 
         if (exact or not exists) and not lfutil.isstandin(f):
             # In case the file was removed previously, but not committed
             # (issue3507)
             if not repo.wvfs.exists(f):
                 continue
 
             abovemin = (lfsize and
                         repo.wvfs.lstat(f).st_size >= lfsize * 1024 * 1024)
             if large or abovemin or (lfmatcher and lfmatcher(f)):
                 lfnames.append(f)
                 if ui.verbose or not exact:
                     ui.status(_('adding %s as a largefile\n') % name)
 
     bad = []
 
     # Need to lock, otherwise there could be a race condition between
     # when standins are created and added to the repo.
     wlock = repo.wlock()
     try:
         if not opts.get('dry_run'):
             standins = []
             lfdirstate = lfutil.openlfdirstate(ui, repo)
             for f in lfnames:
                 standinname = lfutil.standin(f)
                 lfutil.writestandin(repo, standinname, hash='',
                     executable=lfutil.getexecutable(repo.wjoin(f)))
                 standins.append(standinname)
                 if lfdirstate[f] == 'r':
                     lfdirstate.normallookup(f)
                 else:
                     lfdirstate.add(f)
             lfdirstate.write()
             bad += [lfutil.splitstandin(f)
                     for f in repo[None].add(standins)
                     if f in m.files()]
 
         added = [f for f in lfnames if f not in bad]
     finally:
         wlock.release()
     return added, bad
 
 def removelargefiles(ui, repo, isaddremove, matcher, **opts):
     after = opts.get('after')
     m = composelargefilematcher(matcher, repo[None].manifest())
     try:
         repo.lfstatus = True
         s = repo.status(match=m, clean=not isaddremove)
     finally:
         repo.lfstatus = False
     manifest = repo[None].manifest()
     modified, added, deleted, clean = [[f for f in list
                                         if lfutil.standin(f) in manifest]
                                        for list in (s.modified, s.added,
                                                     s.deleted, s.clean)]
 
     def warn(files, msg):
         for f in files:
             ui.warn(msg % m.rel(f))
         return int(len(files) > 0)
 
     result = 0
 
     if after:
         remove = deleted
         result = warn(modified + added + clean,
                       _('not removing %s: file still exists\n'))
     else:
         remove = deleted + clean
         result = warn(modified, _('not removing %s: file is modified (use -f'
                                   ' to force removal)\n'))
         result = warn(added, _('not removing %s: file has been marked for add'
                                ' (use forget to undo)\n')) or result
 
     # Need to lock because standin files are deleted then removed from the
     # repository and we could race in-between.
     wlock = repo.wlock()
     try:
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         for f in sorted(remove):
             if ui.verbose or not m.exact(f):
                 # addremove in core gets fancy with the name, remove doesn't
                 if isaddremove:
                     name = m.uipath(f)
                 else:
                     name = m.rel(f)
                 ui.status(_('removing %s\n') % name)
 
             if not opts.get('dry_run'):
                 if not after:
                     util.unlinkpath(repo.wjoin(f), ignoremissing=True)
 
         if opts.get('dry_run'):
             return result
 
         remove = [lfutil.standin(f) for f in remove]
         # If this is being called by addremove, let the original addremove
         # function handle this.
         if not isaddremove:
             for f in remove:
                 util.unlinkpath(repo.wjoin(f), ignoremissing=True)
         repo[None].forget(remove)
 
         for f in remove:
             lfutil.synclfdirstate(repo, lfdirstate, lfutil.splitstandin(f),
                                   False)
 
         lfdirstate.write()
     finally:
         wlock.release()
 
     return result
 
 # For overriding mercurial.hgweb.webcommands so that largefiles will
 # appear at their right place in the manifests.
 def decodepath(orig, path):
     return lfutil.splitstandin(path) or path
 
 # -- Wrappers: modify existing commands --------------------------------
 
 def overrideadd(orig, ui, repo, *pats, **opts):
     if opts.get('normal') and opts.get('large'):
         raise util.Abort(_('--normal cannot be used with --large'))
     return orig(ui, repo, *pats, **opts)
 
 def cmdutiladd(orig, ui, repo, matcher, prefix, explicitonly, **opts):
     # The --normal flag short circuits this override
     if opts.get('normal'):
         return orig(ui, repo, matcher, prefix, explicitonly, **opts)
 
     ladded, lbad = addlargefiles(ui, repo, False, matcher, **opts)
     normalmatcher = composenormalfilematcher(matcher, repo[None].manifest(),
                                              ladded)
     bad = orig(ui, repo, normalmatcher, prefix, explicitonly, **opts)
 
     bad.extend(f for f in lbad)
     return bad
 
 def cmdutilremove(orig, ui, repo, matcher, prefix, after, force, subrepos):
     normalmatcher = composenormalfilematcher(matcher, repo[None].manifest())
     result = orig(ui, repo, normalmatcher, prefix, after, force, subrepos)
     return removelargefiles(ui, repo, False, matcher, after=after,
                             force=force) or result
 
 def overridestatusfn(orig, repo, rev2, **opts):
     try:
         repo._repo.lfstatus = True
         return orig(repo, rev2, **opts)
     finally:
         repo._repo.lfstatus = False
 
 def overridestatus(orig, ui, repo, *pats, **opts):
     try:
         repo.lfstatus = True
         return orig(ui, repo, *pats, **opts)
     finally:
         repo.lfstatus = False
 
 def overridedirty(orig, repo, ignoreupdate=False):
     try:
         repo._repo.lfstatus = True
         return orig(repo, ignoreupdate)
     finally:
         repo._repo.lfstatus = False
 
 def overridelog(orig, ui, repo, *pats, **opts):
     def overridematchandpats(ctx, pats=[], opts={}, globbed=False,
             default='relpath'):
         """Matcher that merges root directory with .hglf, suitable for log.
         It is still possible to match .hglf directly.
         For any listed files run log on the standin too.
         matchfn tries both the given filename and with .hglf stripped.
         """
         matchandpats = oldmatchandpats(ctx, pats, opts, globbed, default)
         m, p = copy.copy(matchandpats)
 
         if m.always():
             # We want to match everything anyway, so there's no benefit trying
             # to add standins.
             return matchandpats
 
         pats = set(p)
         # TODO: handling of patterns in both cases below
         if m._cwd:
             if os.path.isabs(m._cwd):
                 # TODO: handle largefile magic when invoked from other cwd
                 return matchandpats
             back = (m._cwd.count('/') + 1) * '../'
             pats.update(back + lfutil.standin(m._cwd + '/' + f) for f in p)
         else:
             pats.update(lfutil.standin(f) for f in p)
 
         for i in range(0, len(m._files)):
             standin = lfutil.standin(m._files[i])
             # If the "standin" is a directory, append instead of replace to
             # support naming a directory on the command line with only
             # largefiles.  The original directory is kept to support normal
             # files.
             if standin in repo[ctx.node()]:
                 m._files[i] = standin
             elif m._files[i] not in repo[ctx.node()] \
                     and repo.wvfs.isdir(standin):
                 m._files.append(standin)
             pats.add(standin)
 
         m._fmap = set(m._files)
         m._always = False
         origmatchfn = m.matchfn
         def lfmatchfn(f):
             lf = lfutil.splitstandin(f)
             if lf is not None and origmatchfn(lf):
                 return True
             r = origmatchfn(f)
             return r
         m.matchfn = lfmatchfn
 
         return m, pats
 
     # For hg log --patch, the match object is used in two different senses:
     # (1) to determine what revisions should be printed out, and
     # (2) to determine what files to print out diffs for.
     # The magic matchandpats override should be used for case (1) but not for
     # case (2).
     def overridemakelogfilematcher(repo, pats, opts):
         pctx = repo[None]
         match, pats = oldmatchandpats(pctx, pats, opts)
         return lambda rev: match
 
     oldmatchandpats = installmatchandpatsfn(overridematchandpats)
     oldmakelogfilematcher = cmdutil._makenofollowlogfilematcher
     setattr(cmdutil, '_makenofollowlogfilematcher', overridemakelogfilematcher)
 
     try:
         return orig(ui, repo, *pats, **opts)
     finally:
         restorematchandpatsfn()
         setattr(cmdutil, '_makenofollowlogfilematcher', oldmakelogfilematcher)
 
 def overrideverify(orig, ui, repo, *pats, **opts):
     large = opts.pop('large', False)
     all = opts.pop('lfa', False)
     contents = opts.pop('lfc', False)
 
     result = orig(ui, repo, *pats, **opts)
     if large or all or contents:
         result = result or lfcommands.verifylfiles(ui, repo, all, contents)
     return result
 
 def overridedebugstate(orig, ui, repo, *pats, **opts):
     large = opts.pop('large', False)
     if large:
         class fakerepo(object):
             dirstate = lfutil.openlfdirstate(ui, repo)
         orig(ui, fakerepo, *pats, **opts)
     else:
         orig(ui, repo, *pats, **opts)
 
 # Override needs to refresh standins so that update's normal merge
 # will go through properly. Then the other update hook (overriding repo.update)
 # will get the new files. Filemerge is also overridden so that the merge
 # will merge standins correctly.
 def overrideupdate(orig, ui, repo, *pats, **opts):
     # Need to lock between the standins getting updated and their
     # largefiles getting updated
     wlock = repo.wlock()
     try:
         if opts['check']:
             lfdirstate = lfutil.openlfdirstate(ui, repo)
             unsure, s = lfdirstate.status(
                 match_.always(repo.root, repo.getcwd()),
                 [], False, False, False)
 
             mod = len(s.modified) > 0
             for lfile in unsure:
                 standin = lfutil.standin(lfile)
                 if repo['.'][standin].data().strip() != \
                         lfutil.hashfile(repo.wjoin(lfile)):
                     mod = True
                 else:
                     lfdirstate.normal(lfile)
             lfdirstate.write()
             if mod:
                 raise util.Abort(_('uncommitted changes'))
         return orig(ui, repo, *pats, **opts)
     finally:
         wlock.release()
 
 # Before starting the manifest merge, merge.updates will call
 # _checkunknownfile to check if there are any files in the merged-in
 # changeset that collide with unknown files in the working copy.
 #
 # The largefiles are seen as unknown, so this prevents us from merging
 # in a file 'foo' if we already have a largefile with the same name.
 #
 # The overridden function filters the unknown files by removing any
 # largefiles. This makes the merge proceed and we can then handle this
 # case further in the overridden calculateupdates function below.
 def overridecheckunknownfile(origfn, repo, wctx, mctx, f, f2=None):
     if lfutil.standin(repo.dirstate.normalize(f)) in wctx:
         return False
     return origfn(repo, wctx, mctx, f, f2)
 
 # The manifest merge handles conflicts on the manifest level. We want
 # to handle changes in largefile-ness of files at this level too.
 #
 # The strategy is to run the original calculateupdates and then process
 # the action list it outputs. There are two cases we need to deal with:
 #
 # 1. Normal file in p1, largefile in p2. Here the largefile is
 #    detected via its standin file, which will enter the working copy
 #    with a "get" action. It is not "merge" since the standin is all
 #    Mercurial is concerned with at this level -- the link to the
 #    existing normal file is not relevant here.
 #
 # 2. Largefile in p1, normal file in p2. Here we get a "merge" action
 #    since the largefile will be present in the working copy and
 #    different from the normal file in p2. Mercurial therefore
 #    triggers a merge action.
 #
 # In both cases, we prompt the user and emit new actions to either
 # remove the standin (if the normal file was kept) or to remove the
 # normal file and get the standin (if the largefile was kept). The
 # default prompt answer is to use the largefile version since it was
 # presumably changed on purpose.
 #
 # Finally, the merge.applyupdates function will then take care of
 # writing the files into the working copy and lfcommands.updatelfiles
 # will update the largefiles.
 def overridecalculateupdates(origfn, repo, p1, p2, pas, branchmerge, force,
                              partial, acceptremote, followcopies):
     overwrite = force and not branchmerge
     actions, diverge, renamedelete = origfn(
         repo, p1, p2, pas, branchmerge, force, partial, acceptremote,
         followcopies)
 
     if overwrite:
         return actions, diverge, renamedelete
 
     # Convert to dictionary with filename as key and action as value.
     lfiles = set()
     for f in actions:
         splitstandin = f and lfutil.splitstandin(f)
         if splitstandin in p1:
             lfiles.add(splitstandin)
         elif lfutil.standin(f) in p1:
             lfiles.add(f)
 
     for lfile in lfiles:
         standin = lfutil.standin(lfile)
         (lm, largs, lmsg) = actions.get(lfile, (None, None, None))
         (sm, sargs, smsg) = actions.get(standin, (None, None, None))
         if sm in ('g', 'dc') and lm != 'r':
             # Case 1: normal file in the working copy, largefile in
             # the second parent
             usermsg = _('remote turned local normal file %s into a largefile\n'
                         'use (l)argefile or keep (n)ormal file?'
                         '$$ &Largefile $$ &Normal file') % lfile
             if repo.ui.promptchoice(usermsg, 0) == 0: # pick remote largefile
                 actions[lfile] = ('r', None, 'replaced by standin')
                 actions[standin] = ('g', sargs, 'replaces standin')
             else: # keep local normal file
                 actions[lfile] = ('k', None, 'replaces standin')
                 if branchmerge:
                     actions[standin] = ('k', None, 'replaced by non-standin')
                 else:
                     actions[standin] = ('r', None, 'replaced by non-standin')
         elif lm in ('g', 'dc') and sm != 'r':
             # Case 2: largefile in the working copy, normal file in
             # the second parent
             usermsg = _('remote turned local largefile %s into a normal file\n'
                     'keep (l)argefile or use (n)ormal file?'
                     '$$ &Largefile $$ &Normal file') % lfile
             if repo.ui.promptchoice(usermsg, 0) == 0: # keep local largefile
                 if branchmerge:
                     # largefile can be restored from standin safely
                     actions[lfile] = ('k', None, 'replaced by standin')
                     actions[standin] = ('k', None, 'replaces standin')
                 else:
                     # "lfile" should be marked as "removed" without
                     # removal of itself
                     actions[lfile] = ('lfmr', None,
                                       'forget non-standin largefile')
 
                     # linear-merge should treat this largefile as 're-added'
                     actions[standin] = ('a', None, 'keep standin')
             else: # pick remote normal file
                 actions[lfile] = ('g', largs, 'replaces standin')
                 actions[standin] = ('r', None, 'replaced by non-standin')
 
     return actions, diverge, renamedelete
 
 def mergerecordupdates(orig, repo, actions, branchmerge):
     if 'lfmr' in actions:
         lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
         for lfile, args, msg in actions['lfmr']:
             # this should be executed before 'orig', to execute 'remove'
             # before all other actions
             repo.dirstate.remove(lfile)
             # make sure lfile doesn't get synclfdirstate'd as normal
             lfdirstate.add(lfile)
         lfdirstate.write()
 
     return orig(repo, actions, branchmerge)
 
 
 # Override filemerge to prompt the user about how they wish to merge
 # largefiles. This will handle identical edits without prompting the user.
 def overridefilemerge(origfn, repo, mynode, orig, fcd, fco, fca, labels=None):
     if not lfutil.isstandin(orig):
         return origfn(repo, mynode, orig, fcd, fco, fca, labels=labels)
 
     ahash = fca.data().strip().lower()
     dhash = fcd.data().strip().lower()
     ohash = fco.data().strip().lower()
     if (ohash != ahash and
         ohash != dhash and
         (dhash == ahash or
          repo.ui.promptchoice(
              _('largefile %s has a merge conflict\nancestor was %s\n'
                'keep (l)ocal %s or\ntake (o)ther %s?'
                '$$ &Local $$ &Other') %
                (lfutil.splitstandin(orig), ahash, dhash, ohash),
              0) == 1)):
         repo.wwrite(fcd.path(), fco.data(), fco.flags())
     return 0
 
 # Copy first changes the matchers to match standins instead of
 # largefiles.  Then it overrides util.copyfile in that function it
 # checks if the destination largefile already exists. It also keeps a
 # list of copied files so that the largefiles can be copied and the
 # dirstate updated.
 def overridecopy(orig, ui, repo, pats, opts, rename=False):
     # doesn't remove largefile on rename
     if len(pats) < 2:
         # this isn't legal, let the original function deal with it
         return orig(ui, repo, pats, opts, rename)
 
     def makestandin(relpath):
         path = pathutil.canonpath(repo.root, repo.getcwd(), relpath)
         return os.path.join(repo.wjoin(lfutil.standin(path)))
 
     fullpats = scmutil.expandpats(pats)
     dest = fullpats[-1]
 
     if os.path.isdir(dest):
         if not os.path.isdir(makestandin(dest)):
             os.makedirs(makestandin(dest))
     # This could copy both lfiles and normal files in one command,
     # but we don't want to do that. First replace their matcher to
     # only match normal files and run it, then replace it to just
     # match largefiles and run it again.
     nonormalfiles = False
     nolfiles = False
     installnormalfilesmatchfn(repo[None].manifest())
     try:
         try:
             result = orig(ui, repo, pats, opts, rename)
         except util.Abort, e:
             if str(e) != _('no files to copy'):
                 raise e
             else:
                 nonormalfiles = True
             result = 0
     finally:
         restorematchfn()
 
     # The first rename can cause our current working directory to be removed.
     # In that case there is nothing left to copy/rename so just quit.
     try:
         repo.getcwd()
     except OSError:
         return result
 
     try:
         try:
             # When we call orig below it creates the standins but we don't add
             # them to the dir state until later so lock during that time.
             wlock = repo.wlock()
 
             manifest = repo[None].manifest()
             def overridematch(ctx, pats=[], opts={}, globbed=False,
                     default='relpath'):
                 newpats = []
                 # The patterns were previously mangled to add the standin
                 # directory; we need to remove that now
                 for pat in pats:
                     if match_.patkind(pat) is None and lfutil.shortname in pat:
                         newpats.append(pat.replace(lfutil.shortname, ''))
                     else:
                         newpats.append(pat)
                 match = oldmatch(ctx, newpats, opts, globbed, default)
                 m = copy.copy(match)
                 lfile = lambda f: lfutil.standin(f) in manifest
                 m._files = [lfutil.standin(f) for f in m._files if lfile(f)]
                 m._fmap = set(m._files)
                 origmatchfn = m.matchfn
                 m.matchfn = lambda f: (lfutil.isstandin(f) and
                                     (f in manifest) and
                                     origmatchfn(lfutil.splitstandin(f)) or
                                     None)
                 return m
             oldmatch = installmatchfn(overridematch)
             listpats = []
             for pat in pats:
                 if match_.patkind(pat) is not None:
                     listpats.append(pat)
                 else:
                     listpats.append(makestandin(pat))
 
             try:
                 origcopyfile = util.copyfile
                 copiedfiles = []
                 def overridecopyfile(src, dest):
                     if (lfutil.shortname in src and
                         dest.startswith(repo.wjoin(lfutil.shortname))):
                         destlfile = dest.replace(lfutil.shortname, '')
                         if not opts['force'] and os.path.exists(destlfile):
                             raise IOError('',
                                 _('destination largefile already exists'))
                     copiedfiles.append((src, dest))
                     origcopyfile(src, dest)
 
                 util.copyfile = overridecopyfile
                 result += orig(ui, repo, listpats, opts, rename)
             finally:
                 util.copyfile = origcopyfile
 
             lfdirstate = lfutil.openlfdirstate(ui, repo)
             for (src, dest) in copiedfiles:
                 if (lfutil.shortname in src and
                     dest.startswith(repo.wjoin(lfutil.shortname))):
                     srclfile = src.replace(repo.wjoin(lfutil.standin('')), '')
                     destlfile = dest.replace(repo.wjoin(lfutil.standin('')), '')
                     destlfiledir = os.path.dirname(repo.wjoin(destlfile)) or '.'
                     if not os.path.isdir(destlfiledir):
                         os.makedirs(destlfiledir)
                     if rename:
                         os.rename(repo.wjoin(srclfile), repo.wjoin(destlfile))
 
                         # The file is gone, but this deletes any empty parent
                         # directories as a side-effect.
                         util.unlinkpath(repo.wjoin(srclfile), True)
                         lfdirstate.remove(srclfile)
                     else:
                         util.copyfile(repo.wjoin(srclfile),
                                       repo.wjoin(destlfile))
 
                     lfdirstate.add(destlfile)
             lfdirstate.write()
         except util.Abort, e:
             if str(e) != _('no files to copy'):
                 raise e
             else:
                 nolfiles = True
     finally:
         restorematchfn()
         wlock.release()
 
     if nolfiles and nonormalfiles:
         raise util.Abort(_('no files to copy'))
 
     return result
 
 # When the user calls revert, we have to be careful to not revert any
 # changes to other largefiles accidentally. This means we have to keep
 # track of the largefiles that are being reverted so we only pull down
 # the necessary largefiles.
 #
 # Standins are only updated (to match the hash of largefiles) before
 # commits. Update the standins then run the original revert, changing
 # the matcher to hit standins instead of largefiles. Based on the
 # resulting standins update the largefiles.
 def overriderevert(orig, ui, repo, *pats, **opts):
     # Because we put the standins in a bad state (by updating them)
     # and then return them to a correct state we need to lock to
     # prevent others from changing them in their incorrect state.
     wlock = repo.wlock()
     try:
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         s = lfutil.lfdirstatestatus(lfdirstate, repo)
         lfdirstate.write()
         for lfile in s.modified:
             lfutil.updatestandin(repo, lfutil.standin(lfile))
         for lfile in s.deleted:
             if (os.path.exists(repo.wjoin(lfutil.standin(lfile)))):
                 os.unlink(repo.wjoin(lfutil.standin(lfile)))
 
         oldstandins = lfutil.getstandinsstate(repo)
 
         def overridematch(ctx, pats=[], opts={}, globbed=False,
                 default='relpath'):
             match = oldmatch(ctx, pats, opts, globbed, default)
             m = copy.copy(match)
             def tostandin(f):
                 if lfutil.standin(f) in ctx:
                     return lfutil.standin(f)
                 elif lfutil.standin(f) in repo[None]:
                     return None
                 return f
             m._files = [tostandin(f) for f in m._files]
             m._files = [f for f in m._files if f is not None]
             m._fmap = set(m._files)
             origmatchfn = m.matchfn
             def matchfn(f):
                 if lfutil.isstandin(f):
                     return (origmatchfn(lfutil.splitstandin(f)) and
                             (f in repo[None] or f in ctx))
                 return origmatchfn(f)
             m.matchfn = matchfn
             return m
         oldmatch = installmatchfn(overridematch)
         try:
             orig(ui, repo, *pats, **opts)
         finally:
             restorematchfn()
 
         newstandins = lfutil.getstandinsstate(repo)
         filelist = lfutil.getlfilestoupdate(oldstandins, newstandins)
         # lfdirstate should be 'normallookup'-ed for updated files,
         # because reverting doesn't touch dirstate for 'normal' files
         # when target revision is explicitly specified: in such case,
         # 'n' and valid timestamp in dirstate doesn't ensure 'clean'
         # of target (standin) file.
         lfcommands.updatelfiles(ui, repo, filelist, printmessage=False,
                                 normallookup=True)
 
     finally:
         wlock.release()
 
 # after pulling changesets, we need to take some extra care to get
 # largefiles updated remotely
 def overridepull(orig, ui, repo, source=None, **opts):
     revsprepull = len(repo)
     if not source:
         source = 'default'
     repo.lfpullsource = source
     result = orig(ui, repo, source, **opts)
     revspostpull = len(repo)
     lfrevs = opts.get('lfrev', [])
     if opts.get('all_largefiles'):
         lfrevs.append('pulled()')
     if lfrevs and revspostpull > revsprepull:
         numcached = 0
         repo.firstpulled = revsprepull # for pulled() revset expression
         try:
             for rev in scmutil.revrange(repo, lfrevs):
                 ui.note(_('pulling largefiles for revision %s\n') % rev)
                 (cached, missing) = lfcommands.cachelfiles(ui, repo, rev)
                 numcached += len(cached)
         finally:
             del repo.firstpulled
         ui.status(_("%d largefiles cached\n") % numcached)
     return result
 
 def pulledrevsetsymbol(repo, subset, x):
     """``pulled()``
     Changesets that just has been pulled.
 
     Only available with largefiles from pull --lfrev expressions.
 
     .. container:: verbose
 
       Some examples:
 
       - pull largefiles for all new changesets::
 
           hg pull -lfrev "pulled()"
 
       - pull largefiles for all new branch heads::
 
           hg pull -lfrev "head(pulled()) and not closed()"
 
     """
 
     try:
         firstpulled = repo.firstpulled
     except AttributeError:
         raise util.Abort(_("pulled() only available in --lfrev"))
     return revset.baseset([r for r in subset if r >= firstpulled])
 
 def overrideclone(orig, ui, source, dest=None, **opts):
     d = dest
     if d is None:
         d = hg.defaultdest(source)
     if opts.get('all_largefiles') and not hg.islocal(d):
             raise util.Abort(_(
             '--all-largefiles is incompatible with non-local destination %s') %
             d)
 
     return orig(ui, source, dest, **opts)
 
 def hgclone(orig, ui, opts, *args, **kwargs):
     result = orig(ui, opts, *args, **kwargs)
 
     if result is not None:
         sourcerepo, destrepo = result
         repo = destrepo.local()
 
         # Caching is implicitly limited to 'rev' option, since the dest repo was
         # truncated at that point.  The user may expect a download count with
         # this option, so attempt whether or not this is a largefile repo.
         if opts.get('all_largefiles'):
             success, missing = lfcommands.downloadlfiles(ui, repo, None)
 
             if missing != 0:
                 return None
 
     return result
 
 def overriderebase(orig, ui, repo, **opts):
+    if not util.safehasattr(repo, '_largefilesenabled'):
+        return orig(ui, repo, **opts)
+
     resuming = opts.get('continue')
     repo._lfcommithooks.append(lfutil.automatedcommithook(resuming))
     repo._lfstatuswriters.append(lambda *msg, **opts: None)
     try:
         return orig(ui, repo, **opts)
     finally:
         repo._lfstatuswriters.pop()
         repo._lfcommithooks.pop()
 
 def overridearchive(orig, repo, dest, node, kind, decode=True, matchfn=None,
             prefix=None, mtime=None, subrepos=None):
     # No need to lock because we are only reading history and
     # largefile caches, neither of which are modified.
     lfcommands.cachelfiles(repo.ui, repo, node)
 
     if kind not in archival.archivers:
         raise util.Abort(_("unknown archive type '%s'") % kind)
 
     ctx = repo[node]
 
     if kind == 'files':
         if prefix:
             raise util.Abort(
                 _('cannot give prefix when archiving to files'))
     else:
         prefix = archival.tidyprefix(dest, kind, prefix)
 
     def write(name, mode, islink, getdata):
         if matchfn and not matchfn(name):
             return
         data = getdata()
         if decode:
             data = repo.wwritedata(name, data)
         archiver.addfile(prefix + name, mode, islink, data)
 
     archiver = archival.archivers[kind](dest, mtime or ctx.date()[0])
 
     if repo.ui.configbool("ui", "archivemeta", True):
         def metadata():
             base = 'repo: %s\nnode: %s\nbranch: %s\n' % (
                 hex(repo.changelog.node(0)), hex(node), ctx.branch())
 
             tags = ''.join('tag: %s\n' % t for t in ctx.tags()
                            if repo.tagtype(t) == 'global')
             if not tags:
                 repo.ui.pushbuffer()
                 opts = {'template': '{latesttag}\n{latesttagdistance}',
                         'style': '', 'patch': None, 'git': None}
                 cmdutil.show_changeset(repo.ui, repo, opts).show(ctx)
                 ltags, dist = repo.ui.popbuffer().split('\n')
                 tags = ''.join('latesttag: %s\n' % t for t in ltags.split(':'))
                 tags += 'latesttagdistance: %s\n' % dist
 
             return base + tags
 
         write('.hg_archival.txt', 0644, False, metadata)
 
     for f in ctx:
         ff = ctx.flags(f)
         getdata = ctx[f].data
         if lfutil.isstandin(f):
             path = lfutil.findfile(repo, getdata().strip())
             if path is None:
                 raise util.Abort(
                     _('largefile %s not found in repo store or system cache')
                     % lfutil.splitstandin(f))
             f = lfutil.splitstandin(f)
 
             def getdatafn():
                 fd = None
                 try:
                     fd = open(path, 'rb')
                     return fd.read()
                 finally:
                     if fd:
                         fd.close()
 
             getdata = getdatafn
         write(f, 'x' in ff and 0755 or 0644, 'l' in ff, getdata)
 
     if subrepos:
         for subpath in sorted(ctx.substate):
             sub = ctx.sub(subpath)
             submatch = match_.narrowmatcher(subpath, matchfn)
             sub.archive(archiver, prefix, submatch)
 
     archiver.done()
 
 def hgsubrepoarchive(orig, repo, archiver, prefix, match=None):
     repo._get(repo._state + ('hg',))
     rev = repo._state[1]
     ctx = repo._repo[rev]
 
     lfcommands.cachelfiles(repo.ui, repo._repo, ctx.node())
 
     def write(name, mode, islink, getdata):
         # At this point, the standin has been replaced with the largefile name,
         # so the normal matcher works here without the lfutil variants.
         if match and not match(f):
             return
         data = getdata()
 
         archiver.addfile(prefix + repo._path + '/' + name, mode, islink, data)
 
     for f in ctx:
         ff = ctx.flags(f)
         getdata = ctx[f].data
         if lfutil.isstandin(f):
             path = lfutil.findfile(repo._repo, getdata().strip())
             if path is None:
                 raise util.Abort(
                     _('largefile %s not found in repo store or system cache')
                     % lfutil.splitstandin(f))
             f = lfutil.splitstandin(f)
 
             def getdatafn():
                 fd = None
                 try:
                     fd = open(os.path.join(prefix, path), 'rb')
                     return fd.read()
                 finally:
                     if fd:
                         fd.close()
 
             getdata = getdatafn
 
         write(f, 'x' in ff and 0755 or 0644, 'l' in ff, getdata)
 
     for subpath in sorted(ctx.substate):
         sub = ctx.sub(subpath)
         submatch = match_.narrowmatcher(subpath, match)
         sub.archive(archiver, os.path.join(prefix, repo._path) + '/', submatch)
 
 # If a largefile is modified, the change is not reflected in its
 # standin until a commit. cmdutil.bailifchanged() raises an exception
 # if the repo has uncommitted changes. Wrap it to also check if
 # largefiles were changed. This is used by bisect, backout and fetch.
 def overridebailifchanged(orig, repo):
     orig(repo)
     repo.lfstatus = True
     s = repo.status()
     repo.lfstatus = False
     if s.modified or s.added or s.removed or s.deleted:
         raise util.Abort(_('uncommitted changes'))
 
 def cmdutilforget(orig, ui, repo, match, prefix, explicitonly):
     normalmatcher = composenormalfilematcher(match, repo[None].manifest())
     bad, forgot = orig(ui, repo, normalmatcher, prefix, explicitonly)
     m = composelargefilematcher(match, repo[None].manifest())
 
     try:
         repo.lfstatus = True
         s = repo.status(match=m, clean=True)
     finally:
         repo.lfstatus = False
     forget = sorted(s.modified + s.added + s.deleted + s.clean)
     forget = [f for f in forget if lfutil.standin(f) in repo[None].manifest()]
 
     for f in forget:
         if lfutil.standin(f) not in repo.dirstate and not \
                 repo.wvfs.isdir(lfutil.standin(f)):
             ui.warn(_('not removing %s: file is already untracked\n')
                     % m.rel(f))
             bad.append(f)
 
     for f in forget:
         if ui.verbose or not m.exact(f):
             ui.status(_('removing %s\n') % m.rel(f))
 
     # Need to lock because standin files are deleted then removed from the
     # repository and we could race in-between.
     wlock = repo.wlock()
     try:
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         for f in forget:
             if lfdirstate[f] == 'a':
                 lfdirstate.drop(f)
             else:
                 lfdirstate.remove(f)
         lfdirstate.write()
         standins = [lfutil.standin(f) for f in forget]
         for f in standins:
             util.unlinkpath(repo.wjoin(f), ignoremissing=True)
         rejected = repo[None].forget(standins)
     finally:
         wlock.release()
 
     bad.extend(f for f in rejected if f in m.files())
     forgot.extend(f for f in forget if f not in rejected)
     return bad, forgot
 
 def _getoutgoings(repo, other, missing, addfunc):
     """get pairs of filename and largefile hash in outgoing revisions
     in 'missing'.
 
     largefiles already existing on 'other' repository are ignored.
 
     'addfunc' is invoked with each unique pairs of filename and
     largefile hash value.
     """
     knowns = set()
     lfhashes = set()
     def dedup(fn, lfhash):
         k = (fn, lfhash)
         if k not in knowns:
             knowns.add(k)
             lfhashes.add(lfhash)
     lfutil.getlfilestoupload(repo, missing, dedup)
     if lfhashes:
         lfexists = basestore._openstore(repo, other).exists(lfhashes)
         for fn, lfhash in knowns:
             if not lfexists[lfhash]: # lfhash doesn't exist on "other"
                 addfunc(fn, lfhash)
 
 def outgoinghook(ui, repo, other, opts, missing):
     if opts.pop('large', None):
         lfhashes = set()
         if ui.debugflag:
             toupload = {}
             def addfunc(fn, lfhash):
                 if fn not in toupload:
                     toupload[fn] = []
                 toupload[fn].append(lfhash)
                 lfhashes.add(lfhash)
             def showhashes(fn):
                 for lfhash in sorted(toupload[fn]):
                     ui.debug('    %s\n' % (lfhash))
         else:
             toupload = set()
             def addfunc(fn, lfhash):
                 toupload.add(fn)
                 lfhashes.add(lfhash)
             def showhashes(fn):
                 pass
         _getoutgoings(repo, other, missing, addfunc)
 
         if not toupload:
             ui.status(_('largefiles: no files to upload\n'))
         else:
             ui.status(_('largefiles to upload (%d entities):\n')
                       % (len(lfhashes)))
             for file in sorted(toupload):
                 ui.status(lfutil.splitstandin(file) + '\n')
                 showhashes(file)
             ui.status('\n')
 
 def summaryremotehook(ui, repo, opts, changes):
     largeopt = opts.get('large', False)
     if changes is None:
         if largeopt:
             return (False, True) # only outgoing check is needed
         else:
             return (False, False)
     elif largeopt:
         url, branch, peer, outgoing = changes[1]
         if peer is None:
             # i18n: column positioning for "hg summary"
             ui.status(_('largefiles: (no remote repo)\n'))
             return
 
         toupload = set()
         lfhashes = set()
         def addfunc(fn, lfhash):
             toupload.add(fn)
             lfhashes.add(lfhash)
         _getoutgoings(repo, peer, outgoing.missing, addfunc)
 
         if not toupload:
             # i18n: column positioning for "hg summary"
             ui.status(_('largefiles: (no files to upload)\n'))
         else:
             # i18n: column positioning for "hg summary"
             ui.status(_('largefiles: %d entities for %d files to upload\n')
                       % (len(lfhashes), len(toupload)))
 
 def overridesummary(orig, ui, repo, *pats, **opts):
     try:
         repo.lfstatus = True
         orig(ui, repo, *pats, **opts)
     finally:
         repo.lfstatus = False
 
 def scmutiladdremove(orig, repo, matcher, prefix, opts={}, dry_run=None,
                      similarity=None):
     if not lfutil.islfilesrepo(repo):
         return orig(repo, matcher, prefix, opts, dry_run, similarity)
     # Get the list of missing largefiles so we can remove them
     lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
     unsure, s = lfdirstate.status(match_.always(repo.root, repo.getcwd()), [],
                                   False, False, False)
 
     # Call into the normal remove code, but the removing of the standin, we want
     # to have handled by original addremove.  Monkey patching here makes sure
     # we don't remove the standin in the largefiles code, preventing a very
     # confused state later.
     if s.deleted:
         m = copy.copy(matcher)
 
         # The m._files and m._map attributes are not changed to the deleted list
         # because that affects the m.exact() test, which in turn governs whether
         # or not the file name is printed, and how.  Simply limit the original
         # matches to those in the deleted status list.
         matchfn = m.matchfn
         m.matchfn = lambda f: f in s.deleted and matchfn(f)
 
         removelargefiles(repo.ui, repo, True, m, **opts)
     # Call into the normal add code, and any files that *should* be added as
     # largefiles will be
     added, bad = addlargefiles(repo.ui, repo, True, matcher, **opts)
     # Now that we've handled largefiles, hand off to the original addremove
     # function to take care of the rest.  Make sure it doesn't do anything with
     # largefiles by passing a matcher that will ignore them.
     matcher = composenormalfilematcher(matcher, repo[None].manifest(), added)
     return orig(repo, matcher, prefix, opts, dry_run, similarity)
 
 # Calling purge with --all will cause the largefiles to be deleted.
 # Override repo.status to prevent this from happening.
 def overridepurge(orig, ui, repo, *dirs, **opts):
     # XXX Monkey patching a repoview will not work. The assigned attribute will
     # be set on the unfiltered repo, but we will only lookup attributes in the
     # unfiltered repo if the lookup in the repoview object itself fails. As the
     # monkey patched method exists on the repoview class the lookup will not
     # fail. As a result, the original version will shadow the monkey patched
     # one, defeating the monkey patch.
     #
     # As a work around we use an unfiltered repo here. We should do something
     # cleaner instead.
     repo = repo.unfiltered()
     oldstatus = repo.status
     def overridestatus(node1='.', node2=None, match=None, ignored=False,
                         clean=False, unknown=False, listsubrepos=False):
         r = oldstatus(node1, node2, match, ignored, clean, unknown,
                       listsubrepos)
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         unknown = [f for f in r.unknown if lfdirstate[f] == '?']
         ignored = [f for f in r.ignored if lfdirstate[f] == '?']
         return scmutil.status(r.modified, r.added, r.removed, r.deleted,
                               unknown, ignored, r.clean)
     repo.status = overridestatus
     orig(ui, repo, *dirs, **opts)
     repo.status = oldstatus
 def overriderollback(orig, ui, repo, **opts):
     wlock = repo.wlock()
     try:
         before = repo.dirstate.parents()
         orphans = set(f for f in repo.dirstate
                       if lfutil.isstandin(f) and repo.dirstate[f] != 'r')
         result = orig(ui, repo, **opts)
         after = repo.dirstate.parents()
         if before == after:
             return result # no need to restore standins
 
         pctx = repo['.']
         for f in repo.dirstate:
             if lfutil.isstandin(f):
                 orphans.discard(f)
                 if repo.dirstate[f] == 'r':
                     repo.wvfs.unlinkpath(f, ignoremissing=True)
                 elif f in pctx:
                     fctx = pctx[f]
                     repo.wwrite(f, fctx.data(), fctx.flags())
                 else:
                     # content of standin is not so important in 'a',
                     # 'm' or 'n' (coming from the 2nd parent) cases
                     lfutil.writestandin(repo, f, '', False)
         for standin in orphans:
             repo.wvfs.unlinkpath(standin, ignoremissing=True)
 
         lfdirstate = lfutil.openlfdirstate(ui, repo)
         orphans = set(lfdirstate)
         lfiles = lfutil.listlfiles(repo)
         for file in lfiles:
             lfutil.synclfdirstate(repo, lfdirstate, file, True)
             orphans.discard(file)
         for lfile in orphans:
             lfdirstate.drop(lfile)
         lfdirstate.write()
     finally:
         wlock.release()
     return result
 
 def overridetransplant(orig, ui, repo, *revs, **opts):
     resuming = opts.get('continue')
     repo._lfcommithooks.append(lfutil.automatedcommithook(resuming))
     repo._lfstatuswriters.append(lambda *msg, **opts: None)
     try:
         result = orig(ui, repo, *revs, **opts)
     finally:
         repo._lfstatuswriters.pop()
         repo._lfcommithooks.pop()
     return result
 
 def overridecat(orig, ui, repo, file1, *pats, **opts):
     ctx = scmutil.revsingle(repo, opts.get('rev'))
     err = 1
     notbad = set()
     m = scmutil.match(ctx, (file1,) + pats, opts)
     origmatchfn = m.matchfn
     def lfmatchfn(f):
         if origmatchfn(f):
             return True
         lf = lfutil.splitstandin(f)
         if lf is None:
             return False
         notbad.add(lf)
         return origmatchfn(lf)
     m.matchfn = lfmatchfn
     origbadfn = m.bad
     def lfbadfn(f, msg):
         if not f in notbad:
             origbadfn(f, msg)
     m.bad = lfbadfn
     for f in ctx.walk(m):
         fp = cmdutil.makefileobj(repo, opts.get('output'), ctx.node(),
                                  pathname=f)
         lf = lfutil.splitstandin(f)
         if lf is None or origmatchfn(f):
             # duplicating unreachable code from commands.cat
             data = ctx[f].data()
             if opts.get('decode'):
                 data = repo.wwritedata(f, data)
             fp.write(data)
         else:
             hash = lfutil.readstandin(repo, lf, ctx.rev())
             if not lfutil.inusercache(repo.ui, hash):
                 store = basestore._openstore(repo)
                 success, missing = store.get([(lf, hash)])
                 if len(success) != 1:
                     raise util.Abort(
                         _('largefile %s is not in cache and could not be '
                           'downloaded')  % lf)
             path = lfutil.usercachepath(repo.ui, hash)
             fpin = open(path, "rb")
             for chunk in util.filechunkiter(fpin, 128 * 1024):
                 fp.write(chunk)
             fpin.close()
         fp.close()
         err = 0
     return err
 
 def mergeupdate(orig, repo, node, branchmerge, force, partial,
                 *args, **kwargs):
     wlock = repo.wlock()
     try:
         # branch |       |         |
         #  merge | force | partial | action
         # -------+-------+---------+--------------
         #    x   |   x   |    x    | linear-merge
         #    o   |   x   |    x    | branch-merge
         #    x   |   o   |    x    | overwrite (as clean update)
         #    o   |   o   |    x    | force-branch-merge (*1)
         #    x   |   x   |    o    |   (*)
         #    o   |   x   |    o    |   (*)
         #    x   |   o   |    o    | overwrite (as revert)
         #    o   |   o   |    o    |   (*)
         #
         # (*) don't care
         # (*1) deprecated, but used internally (e.g: "rebase --collapse")
 
         linearmerge = not branchmerge and not force and not partial
 
         if linearmerge or (branchmerge and force and not partial):
             # update standins for linear-merge or force-branch-merge,
             # because largefiles in the working directory may be modified
             lfdirstate = lfutil.openlfdirstate(repo.ui, repo)
             unsure, s = lfdirstate.status(match_.always(repo.root,
                                                         repo.getcwd()),
                                           [], False, False, False)
             pctx = repo['.']
             for lfile in unsure + s.modified:
                 lfileabs = repo.wvfs.join(lfile)
                 if not os.path.exists(lfileabs):
                     continue
                 lfhash = lfutil.hashrepofile(repo, lfile)
                 standin = lfutil.standin(lfile)
                 lfutil.writestandin(repo, standin, lfhash,
                                     lfutil.getexecutable(lfileabs))
                 if (standin in pctx and
                     lfhash == lfutil.readstandin(repo, lfile, '.')):
                     lfdirstate.normal(lfile)
             for lfile in s.added:
                 lfutil.updatestandin(repo, lfutil.standin(lfile))
             lfdirstate.write()
 
         if linearmerge:
             # Only call updatelfiles on the standins that have changed
             # to save time
             oldstandins = lfutil.getstandinsstate(repo)
 
         result = orig(repo, node, branchmerge, force, partial, *args, **kwargs)
 
         filelist = None
         if linearmerge:
             newstandins = lfutil.getstandinsstate(repo)
             filelist = lfutil.getlfilestoupdate(oldstandins, newstandins)
 
         lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                 normallookup=partial, checked=linearmerge)
 
         return result
     finally:
         wlock.release()
 
 def scmutilmarktouched(orig, repo, files, *args, **kwargs):
     result = orig(repo, files, *args, **kwargs)
 
     filelist = [lfutil.splitstandin(f) for f in files if lfutil.isstandin(f)]
     if filelist:
         lfcommands.updatelfiles(repo.ui, repo, filelist=filelist,
                                 printmessage=False, normallookup=True)
 
     return result

hgext/largefiles/reposetup.py

@@ -1,371 +1,374 b''
 # Copyright 2009-2010 Gregory P. Ward
 # Copyright 2009-2010 Intelerad Medical Systems Incorporated
 # Copyright 2010-2011 Fog Creek Software
 # Copyright 2010-2011 Unity Technologies
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 '''setup for largefiles repositories: reposetup'''
 import copy
 import os
 
 from mercurial import error, manifest, match as match_, util
 from mercurial.i18n import _
 from mercurial import scmutil, localrepo
 
 import lfcommands
 import lfutil
 
 def reposetup(ui, repo):
     # wire repositories should be given new wireproto functions
     # by "proto.wirereposetup()" via "hg.wirepeersetupfuncs"
     if not repo.local():
         return
 
     class lfilesrepo(repo.__class__):
+        # the mark to examine whether "repo" object enables largefiles or not
+        _largefilesenabled = True
+
         lfstatus = False
         def status_nolfiles(self, *args, **kwargs):
             return super(lfilesrepo, self).status(*args, **kwargs)
 
         # When lfstatus is set, return a context that gives the names
         # of largefiles instead of their corresponding standins and
         # identifies the largefiles as always binary, regardless of
         # their actual contents.
         def __getitem__(self, changeid):
             ctx = super(lfilesrepo, self).__getitem__(changeid)
             if self.lfstatus:
                 class lfilesmanifestdict(manifest.manifestdict):
                     def __contains__(self, filename):
                         orig = super(lfilesmanifestdict, self).__contains__
                         return orig(filename) or orig(lfutil.standin(filename))
                 class lfilesctx(ctx.__class__):
                     def files(self):
                         filenames = super(lfilesctx, self).files()
                         return [lfutil.splitstandin(f) or f for f in filenames]
                     def manifest(self):
                         man1 = super(lfilesctx, self).manifest()
                         man1.__class__ = lfilesmanifestdict
                         return man1
                     def filectx(self, path, fileid=None, filelog=None):
                         orig = super(lfilesctx, self).filectx
                         try:
                             if filelog is not None:
                                 result = orig(path, fileid, filelog)
                             else:
                                 result = orig(path, fileid)
                         except error.LookupError:
                             # Adding a null character will cause Mercurial to
                             # identify this as a binary file.
                             if filelog is not None:
                                 result = orig(lfutil.standin(path), fileid,
                                               filelog)
                             else:
                                 result = orig(lfutil.standin(path), fileid)
                             olddata = result.data
                             result.data = lambda: olddata() + '\0'
                         return result
                 ctx.__class__ = lfilesctx
             return ctx
 
         # Figure out the status of big files and insert them into the
         # appropriate list in the result. Also removes standin files
         # from the listing. Revert to the original status if
         # self.lfstatus is False.
         # XXX large file status is buggy when used on repo proxy.
         # XXX this needs to be investigated.
         @localrepo.unfilteredmethod
         def status(self, node1='.', node2=None, match=None, ignored=False,
                 clean=False, unknown=False, listsubrepos=False):
             listignored, listclean, listunknown = ignored, clean, unknown
             orig = super(lfilesrepo, self).status
             if not self.lfstatus:
                 return orig(node1, node2, match, listignored, listclean,
                             listunknown, listsubrepos)
 
             # some calls in this function rely on the old version of status
             self.lfstatus = False
             ctx1 = self[node1]
             ctx2 = self[node2]
             working = ctx2.rev() is None
             parentworking = working and ctx1 == self['.']
 
             if match is None:
                 match = match_.always(self.root, self.getcwd())
 
             wlock = None
             try:
                 try:
                     # updating the dirstate is optional
                     # so we don't wait on the lock
                     wlock = self.wlock(False)
                 except error.LockError:
                     pass
 
                 # First check if paths or patterns were specified on the
                 # command line.  If there were, and they don't match any
                 # largefiles, we should just bail here and let super
                 # handle it -- thus gaining a big performance boost.
                 lfdirstate = lfutil.openlfdirstate(ui, self)
                 if not match.always():
                     for f in lfdirstate:
                         if match(f):
                             break
                     else:
                         return orig(node1, node2, match, listignored, listclean,
                                     listunknown, listsubrepos)
 
                 # Create a copy of match that matches standins instead
                 # of largefiles.
                 def tostandins(files):
                     if not working:
                         return files
                     newfiles = []
                     dirstate = self.dirstate
                     for f in files:
                         sf = lfutil.standin(f)
                         if sf in dirstate:
                             newfiles.append(sf)
                         elif sf in dirstate.dirs():
                             # Directory entries could be regular or
                             # standin, check both
                             newfiles.extend((f, sf))
                         else:
                             newfiles.append(f)
                     return newfiles
 
                 m = copy.copy(match)
                 m._files = tostandins(m._files)
 
                 result = orig(node1, node2, m, ignored, clean, unknown,
                               listsubrepos)
                 if working:
 
                     def sfindirstate(f):
                         sf = lfutil.standin(f)
                         dirstate = self.dirstate
                         return sf in dirstate or sf in dirstate.dirs()
 
                     match._files = [f for f in match._files
                                     if sfindirstate(f)]
                     # Don't waste time getting the ignored and unknown
                     # files from lfdirstate
                     unsure, s = lfdirstate.status(match, [], False, listclean,
                                                   False)
                     (modified, added, removed, clean) = (s.modified, s.added,
                                                          s.removed, s.clean)
                     if parentworking:
                         for lfile in unsure:
                             standin = lfutil.standin(lfile)
                             if standin not in ctx1:
                                 # from second parent
                                 modified.append(lfile)
                             elif ctx1[standin].data().strip() \
                                     != lfutil.hashfile(self.wjoin(lfile)):
                                 modified.append(lfile)
                             else:
                                 if listclean:
                                     clean.append(lfile)
                                 lfdirstate.normal(lfile)
                     else:
                         tocheck = unsure + modified + added + clean
                         modified, added, clean = [], [], []
                         checkexec = self.dirstate._checkexec
 
                         for lfile in tocheck:
                             standin = lfutil.standin(lfile)
                             if standin in ctx1:
                                 abslfile = self.wjoin(lfile)
                                 if ((ctx1[standin].data().strip() !=
                                      lfutil.hashfile(abslfile)) or
                                     (checkexec and
                                      ('x' in ctx1.flags(standin)) !=
                                      bool(lfutil.getexecutable(abslfile)))):
                                     modified.append(lfile)
                                 elif listclean:
                                     clean.append(lfile)
                             else:
                                 added.append(lfile)
 
                         # at this point, 'removed' contains largefiles
                         # marked as 'R' in the working context.
                         # then, largefiles not managed also in the target
                         # context should be excluded from 'removed'.
                         removed = [lfile for lfile in removed
                                    if lfutil.standin(lfile) in ctx1]
 
                     # Standins no longer found in lfdirstate has been
                     # removed
                     for standin in ctx1.walk(lfutil.getstandinmatcher(self)):
                         lfile = lfutil.splitstandin(standin)
                         if not match(lfile):
                             continue
                         if lfile not in lfdirstate:
                             removed.append(lfile)
 
                     # Filter result lists
                     result = list(result)
 
                     # Largefiles are not really removed when they're
                     # still in the normal dirstate. Likewise, normal
                     # files are not really removed if they are still in
                     # lfdirstate. This happens in merges where files
                     # change type.
                     removed = [f for f in removed
                                if f not in self.dirstate]
                     result[2] = [f for f in result[2]
                                  if f not in lfdirstate]
 
                     lfiles = set(lfdirstate._map)
                     # Unknown files
                     result[4] = set(result[4]).difference(lfiles)
                     # Ignored files
                     result[5] = set(result[5]).difference(lfiles)
                     # combine normal files and largefiles
                     normals = [[fn for fn in filelist
                                 if not lfutil.isstandin(fn)]
                                for filelist in result]
                     lfstatus = (modified, added, removed, s.deleted, [], [],
                                 clean)
                     result = [sorted(list1 + list2)
                               for (list1, list2) in zip(normals, lfstatus)]
                 else: # not against working directory
                     result = [[lfutil.splitstandin(f) or f for f in items]
                               for items in result]
 
                 if wlock:
                     lfdirstate.write()
 
             finally:
                 if wlock:
                     wlock.release()
 
             self.lfstatus = True
             return scmutil.status(*result)
 
         def commitctx(self, ctx, *args, **kwargs):
             node = super(lfilesrepo, self).commitctx(ctx, *args, **kwargs)
             class lfilesctx(ctx.__class__):
                 def markcommitted(self, node):
                     orig = super(lfilesctx, self).markcommitted
                     return lfutil.markcommitted(orig, self, node)
             ctx.__class__ = lfilesctx
             return node
 
         # Before commit, largefile standins have not had their
         # contents updated to reflect the hash of their largefile.
         # Do that here.
         def commit(self, text="", user=None, date=None, match=None,
                 force=False, editor=False, extra={}):
             orig = super(lfilesrepo, self).commit
 
             wlock = self.wlock()
             try:
                 lfcommithook = self._lfcommithooks[-1]
                 match = lfcommithook(self, match)
                 result = orig(text=text, user=user, date=date, match=match,
                                 force=force, editor=editor, extra=extra)
                 return result
             finally:
                 wlock.release()
 
         def push(self, remote, force=False, revs=None, newbranch=False):
             if remote.local():
                 missing = set(self.requirements) - remote.local().supported
                 if missing:
                     msg = _("required features are not"
                             " supported in the destination:"
                             " %s") % (', '.join(sorted(missing)))
                     raise util.Abort(msg)
             return super(lfilesrepo, self).push(remote, force=force, revs=revs,
                 newbranch=newbranch)
 
         # TODO: _subdirlfs should be moved into "lfutil.py", because
         # it is referred only from "lfutil.updatestandinsbymatch"
         def _subdirlfs(self, files, lfiles):
             '''
             Adjust matched file list
             If we pass a directory to commit whose only committable files
             are largefiles, the core commit code aborts before finding
             the largefiles.
             So we do the following:
             For directories that only have largefiles as matches,
             we explicitly add the largefiles to the match list and remove
             the directory.
             In other cases, we leave the match list unmodified.
             '''
             actualfiles = []
             dirs = []
             regulars = []
 
             for f in files:
                 if lfutil.isstandin(f + '/'):
                     raise util.Abort(
                         _('file "%s" is a largefile standin') % f,
                         hint=('commit the largefile itself instead'))
                 # Scan directories
                 if os.path.isdir(self.wjoin(f)):
                     dirs.append(f)
                 else:
                     regulars.append(f)
 
             for f in dirs:
                 matcheddir = False
                 d = self.dirstate.normalize(f) + '/'
                 # Check for matched normal files
                 for mf in regulars:
                     if self.dirstate.normalize(mf).startswith(d):
                         actualfiles.append(f)
                         matcheddir = True
                         break
                 if not matcheddir:
                     # If no normal match, manually append
                     # any matching largefiles
                     for lf in lfiles:
                         if self.dirstate.normalize(lf).startswith(d):
                             actualfiles.append(lf)
                             if not matcheddir:
                                 # There may still be normal files in the dir, so
                                 # make sure a directory is in the list, which
                                 # forces status to walk and call the match
                                 # function on the matcher.  Windows does NOT
                                 # require this.
                                 actualfiles.append('.')
                                 matcheddir = True
                 # Nothing in dir, so readd it
                 # and let commit reject it
                 if not matcheddir:
                     actualfiles.append(f)
 
             # Always add normal files
             actualfiles += regulars
             return actualfiles
 
     repo.__class__ = lfilesrepo
 
     # stack of hooks being executed before committing.
     # only last element ("_lfcommithooks[-1]") is used for each committing.
     repo._lfcommithooks = [lfutil.updatestandinsbymatch]
 
     # Stack of status writer functions taking "*msg, **opts" arguments
     # like "ui.status()". Only last element ("_lfstatuswriters[-1]")
     # is used to write status out.
     repo._lfstatuswriters = [ui.status]
 
     def prepushoutgoinghook(local, remote, outgoing):
         if outgoing.missing:
             toupload = set()
             addfunc = lambda fn, lfhash: toupload.add(lfhash)
             lfutil.getlfilestoupload(local, outgoing.missing, addfunc)
             lfcommands.uploadlfiles(ui, local, remote, toupload)
     repo.prepushoutgoinghooks.add("largefiles", prepushoutgoinghook)
 
     def checkrequireslfiles(ui, repo, **kwargs):
         if 'largefiles' not in repo.requirements and util.any(
                 lfutil.shortname+'/' in f[0] for f in repo.store.datafiles()):
             repo.requirements.add('largefiles')
             repo._writerequirements()
 
     ui.setconfig('hooks', 'changegroup.lfiles', checkrequireslfiles,
                  'largefiles')
     ui.setconfig('hooks', 'commit.lfiles', checkrequireslfiles, 'largefiles')

tests/test-largefiles-misc.t

@@ -1,942 +1,1013 b''
 This file contains testcases that tend to be related to special cases or less
 common commands affecting largefile.
 
 Each sections should be independent of each others.
 
   $ USERCACHE="$TESTTMP/cache"; export USERCACHE
   $ mkdir "${USERCACHE}"
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > largefiles=
   > purge=
   > rebase=
   > transplant=
   > [phases]
   > publish=False
   > [largefiles]
   > minsize=2
   > patterns=glob:**.dat
   > usercache=${USERCACHE}
   > [hooks]
   > precommit=sh -c "echo \\"Invoking status precommit hook\\"; hg status"
   > EOF
 
 
 
 Test copies and moves from a directory other than root (issue3516)
 =========================================================================
 
   $ hg init lf_cpmv
   $ cd lf_cpmv
   $ mkdir dira
   $ mkdir dira/dirb
   $ touch dira/dirb/largefile
   $ hg add --large dira/dirb/largefile
   $ hg commit -m "added"
   Invoking status precommit hook
   A dira/dirb/largefile
   $ cd dira
   $ hg cp dirb/largefile foo/largefile
   $ hg ci -m "deep copy"
   Invoking status precommit hook
   A dira/foo/largefile
   $ find . | sort
   .
   ./dirb
   ./dirb/largefile
   ./foo
   ./foo/largefile
   $ hg mv foo/largefile baz/largefile
   $ hg ci -m "moved"
   Invoking status precommit hook
   A dira/baz/largefile
   R dira/foo/largefile
   $ find . | sort
   .
   ./baz
   ./baz/largefile
   ./dirb
   ./dirb/largefile
   $ cd ..
   $ hg mv dira dirc
   moving .hglf/dira/baz/largefile to .hglf/dirc/baz/largefile (glob)
   moving .hglf/dira/dirb/largefile to .hglf/dirc/dirb/largefile (glob)
   $ find * | sort
   dirc
   dirc/baz
   dirc/baz/largefile
   dirc/dirb
   dirc/dirb/largefile
 
   $ hg clone -q . ../fetch
   $ hg --config extensions.fetch= fetch ../fetch
   abort: uncommitted changes
   [255]
   $ hg up -qC
   $ cd ..
 
 Clone a local repository owned by another user
 ===================================================
 
 #if unix-permissions
 
 We have to simulate that here by setting $HOME and removing write permissions
   $ ORIGHOME="$HOME"
   $ mkdir alice
   $ HOME="`pwd`/alice"
   $ cd alice
   $ hg init pubrepo
   $ cd pubrepo
   $ dd if=/dev/zero bs=1k count=11k > a-large-file 2> /dev/null
   $ hg add --large a-large-file
   $ hg commit -m "Add a large file"
   Invoking status precommit hook
   A a-large-file
   $ cd ..
   $ chmod -R a-w pubrepo
   $ cd ..
   $ mkdir bob
   $ HOME="`pwd`/bob"
   $ cd bob
   $ hg clone --pull ../alice/pubrepo pubrepo
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   updating to branch default
   getting changed largefiles
   1 largefiles updated, 0 removed
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cd ..
   $ chmod -R u+w alice/pubrepo
   $ HOME="$ORIGHOME"
 
 #endif
 
 
 Symlink to a large largefile should behave the same as a symlink to a normal file
 =====================================================================================
 
 #if symlink
 
   $ hg init largesymlink
   $ cd largesymlink
   $ dd if=/dev/zero bs=1k count=10k of=largefile 2>/dev/null
   $ hg add --large largefile
   $ hg commit -m "commit a large file"
   Invoking status precommit hook
   A largefile
   $ ln -s largefile largelink
   $ hg add largelink
   $ hg commit -m "commit a large symlink"
   Invoking status precommit hook
   A largelink
   $ rm -f largelink
   $ hg up >/dev/null
   $ test -f largelink
   [1]
   $ test -L largelink
   [1]
   $ rm -f largelink # make next part of the test independent of the previous
   $ hg up -C >/dev/null
   $ test -f largelink
   $ test -L largelink
   $ cd ..
 
 #endif
 
 
 test for pattern matching on 'hg status':
 ==============================================
 
 
 to boost performance, largefiles checks whether specified patterns are
 related to largefiles in working directory (NOT to STANDIN) or not.
 
   $ hg init statusmatch
   $ cd statusmatch
 
   $ mkdir -p a/b/c/d
   $ echo normal > a/b/c/d/e.normal.txt
   $ hg add a/b/c/d/e.normal.txt
   $ echo large > a/b/c/d/e.large.txt
   $ hg add --large a/b/c/d/e.large.txt
   $ mkdir -p a/b/c/x
   $ echo normal > a/b/c/x/y.normal.txt
   $ hg add a/b/c/x/y.normal.txt
   $ hg commit -m 'add files'
   Invoking status precommit hook
   A a/b/c/d/e.large.txt
   A a/b/c/d/e.normal.txt
   A a/b/c/x/y.normal.txt
 
 (1) no pattern: no performance boost
   $ hg status -A
   C a/b/c/d/e.large.txt
   C a/b/c/d/e.normal.txt
   C a/b/c/x/y.normal.txt
 
 (2) pattern not related to largefiles: performance boost
   $ hg status -A a/b/c/x
   C a/b/c/x/y.normal.txt
 
 (3) pattern related to largefiles: no performance boost
   $ hg status -A a/b/c/d
   C a/b/c/d/e.large.txt
   C a/b/c/d/e.normal.txt
 
 (4) pattern related to STANDIN (not to largefiles): performance boost
   $ hg status -A .hglf/a
   C .hglf/a/b/c/d/e.large.txt
 
 (5) mixed case: no performance boost
   $ hg status -A a/b/c/x a/b/c/d
   C a/b/c/d/e.large.txt
   C a/b/c/d/e.normal.txt
   C a/b/c/x/y.normal.txt
 
 verify that largefiles doesn't break filesets
 
   $ hg log --rev . --exclude "set:binary()"
   changeset:   0:41bd42f10efa
   tag:         tip
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     add files
   
 verify that large files in subrepos handled properly
   $ hg init subrepo
   $ echo "subrepo = subrepo" > .hgsub
   $ hg add .hgsub
   $ hg ci -m "add subrepo"
   Invoking status precommit hook
   A .hgsub
   ? .hgsubstate
   $ echo "rev 1" > subrepo/large.txt
   $ hg add --large subrepo/large.txt
   $ hg sum
   parent: 1:8ee150ea2e9c tip
    add subrepo
   branch: default
   commit: 1 subrepos
   update: (current)
   $ hg st
   $ hg st -S
   A subrepo/large.txt
   $ hg ci -S -m "commit top repo"
   committing subrepository subrepo
   Invoking status precommit hook
   A large.txt
   Invoking status precommit hook
   M .hgsubstate
 # No differences
   $ hg st -S
   $ hg sum
   parent: 2:ce4cd0c527a6 tip
    commit top repo
   branch: default
   commit: (clean)
   update: (current)
   $ echo "rev 2" > subrepo/large.txt
   $ hg st -S
   M subrepo/large.txt
   $ hg sum
   parent: 2:ce4cd0c527a6 tip
    commit top repo
   branch: default
   commit: 1 subrepos
   update: (current)
   $ hg ci -m "this commit should fail without -S"
   abort: uncommitted changes in subrepo subrepo
   (use --subrepos for recursive commit)
   [255]
 
 Add a normal file to the subrepo, then test archiving
 
   $ echo 'normal file' > subrepo/normal.txt
   $ touch large.dat
   $ mv subrepo/large.txt subrepo/renamed-large.txt
   $ hg addremove -S --dry-run
   adding large.dat as a largefile
   removing subrepo/large.txt
   adding subrepo/normal.txt
   adding subrepo/renamed-large.txt
   $ hg status -S
   ! subrepo/large.txt
   ? large.dat
   ? subrepo/normal.txt
   ? subrepo/renamed-large.txt
 
   $ hg addremove --dry-run subrepo
   removing subrepo/large.txt (glob)
   adding subrepo/normal.txt (glob)
   adding subrepo/renamed-large.txt (glob)
   $ hg status -S
   ! subrepo/large.txt
   ? large.dat
   ? subrepo/normal.txt
   ? subrepo/renamed-large.txt
   $ cd ..
 
   $ hg -R statusmatch addremove --dry-run statusmatch/subrepo
   removing statusmatch/subrepo/large.txt (glob)
   adding statusmatch/subrepo/normal.txt (glob)
   adding statusmatch/subrepo/renamed-large.txt (glob)
   $ hg -R statusmatch status -S
   ! subrepo/large.txt
   ? large.dat
   ? subrepo/normal.txt
   ? subrepo/renamed-large.txt
 
   $ hg -R statusmatch addremove --dry-run -S
   adding large.dat as a largefile
   removing subrepo/large.txt
   adding subrepo/normal.txt
   adding subrepo/renamed-large.txt
   $ cd statusmatch
 
   $ mv subrepo/renamed-large.txt subrepo/large.txt
   $ hg addremove subrepo
   adding subrepo/normal.txt (glob)
   $ hg forget subrepo/normal.txt
 
   $ hg addremove -S
   adding large.dat as a largefile
   adding subrepo/normal.txt
   $ rm large.dat
 
   $ hg addremove subrepo
   $ hg addremove -S
   removing large.dat
 
 Lock in subrepo, otherwise the change isn't archived
 
   $ hg ci -S -m "add normal file to top level"
   committing subrepository subrepo
   Invoking status precommit hook
   M large.txt
   A normal.txt
   Invoking status precommit hook
   M .hgsubstate
   $ hg archive -S ../lf_subrepo_archive
   $ find ../lf_subrepo_archive | sort
   ../lf_subrepo_archive
   ../lf_subrepo_archive/.hg_archival.txt
   ../lf_subrepo_archive/.hgsub
   ../lf_subrepo_archive/.hgsubstate
   ../lf_subrepo_archive/a
   ../lf_subrepo_archive/a/b
   ../lf_subrepo_archive/a/b/c
   ../lf_subrepo_archive/a/b/c/d
   ../lf_subrepo_archive/a/b/c/d/e.large.txt
   ../lf_subrepo_archive/a/b/c/d/e.normal.txt
   ../lf_subrepo_archive/a/b/c/x
   ../lf_subrepo_archive/a/b/c/x/y.normal.txt
   ../lf_subrepo_archive/subrepo
   ../lf_subrepo_archive/subrepo/large.txt
   ../lf_subrepo_archive/subrepo/normal.txt
 
 Test update with subrepos.
 
   $ hg update 0
   getting changed largefiles
   0 largefiles updated, 1 removed
   0 files updated, 0 files merged, 2 files removed, 0 files unresolved
   $ hg status -S
   $ hg update tip
   getting changed largefiles
   1 largefiles updated, 0 removed
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg status -S
 # modify a large file
   $ echo "modified" > subrepo/large.txt
   $ hg st -S
   M subrepo/large.txt
 # update -C should revert the change.
   $ hg update -C
   getting changed largefiles
   1 largefiles updated, 0 removed
   getting changed largefiles
   0 largefiles updated, 0 removed
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg status -S
 
   $ hg rm -v subrepo/large.txt
   removing subrepo/large.txt (glob)
   $ hg revert -R subrepo subrepo/large.txt
   $ rm subrepo/large.txt
   $ hg addremove -S
   removing subrepo/large.txt
   $ hg st -S
   R subrepo/large.txt
 
 Test archiving a revision that references a subrepo that is not yet
 cloned (see test-subrepo-recursion.t):
 
   $ hg clone -U . ../empty
   $ cd ../empty
   $ hg archive --subrepos -r tip ../archive.tar.gz
   cloning subrepo subrepo from $TESTTMP/statusmatch/subrepo
   $ cd ..
 
 
 
 
 
 
 Test addremove, forget and others
 ==============================================
 
 Test that addremove picks up largefiles prior to the initial commit (issue3541)
 
   $ hg init addrm2
   $ cd addrm2
   $ touch large.dat
   $ touch large2.dat
   $ touch normal
   $ hg add --large large.dat
   $ hg addremove -v
   adding large2.dat as a largefile
   adding normal
 
 Test that forgetting all largefiles reverts to islfilesrepo() == False
 (addremove will add *.dat as normal files now)
   $ hg forget large.dat
   $ hg forget large2.dat
   $ hg addremove -v
   adding large.dat
   adding large2.dat
 
 Test commit's addremove option prior to the first commit
   $ hg forget large.dat
   $ hg forget large2.dat
   $ hg add --large large.dat
   $ hg ci -Am "commit"
   adding large2.dat as a largefile
   Invoking status precommit hook
   A large.dat
   A large2.dat
   A normal
   $ find .hglf | sort
   .hglf
   .hglf/large.dat
   .hglf/large2.dat
 
 Test actions on largefiles using relative paths from subdir
 
   $ mkdir sub
   $ cd sub
   $ echo anotherlarge > anotherlarge
   $ hg add --large anotherlarge
   $ hg st
   A sub/anotherlarge
   $ hg st anotherlarge
   A anotherlarge
   $ hg commit -m anotherlarge anotherlarge
   Invoking status precommit hook
   A sub/anotherlarge
   $ hg log anotherlarge
   changeset:   1:9627a577c5e9
   tag:         tip
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     anotherlarge
   
   $ hg log -G anotherlarge
   @  changeset:   1:9627a577c5e9
   |  tag:         tip
   |  user:        test
   |  date:        Thu Jan 01 00:00:00 1970 +0000
   |  summary:     anotherlarge
   |
   $ echo more >> anotherlarge
   $ hg st .
   M anotherlarge
   $ hg cat anotherlarge
   anotherlarge
   $ hg revert anotherlarge
   $ hg st
   ? sub/anotherlarge.orig
   $ cd ..
 
   $ cd ..
 
 Check error message while exchange
 =========================================================
 
 issue3651: summary/outgoing with largefiles shows "no remote repo"
 unexpectedly
 
   $ mkdir issue3651
   $ cd issue3651
 
   $ hg init src
   $ echo a > src/a
   $ hg -R src add --large src/a
   $ hg -R src commit -m '#0'
   Invoking status precommit hook
   A a
 
 check messages when no remote repository is specified:
 "no remote repo" route for "hg outgoing --large" is not tested here,
 because it can't be reproduced easily.
 
   $ hg init clone1
   $ hg -R clone1 -q pull src
   $ hg -R clone1 -q update
   $ hg -R clone1 paths | grep default
   [1]
 
   $ hg -R clone1 summary --large
   parent: 0:fc0bd45326d3 tip
    #0
   branch: default
   commit: (clean)
   update: (current)
   largefiles: (no remote repo)
 
 check messages when there is no files to upload:
 
   $ hg -q clone src clone2
   $ hg -R clone2 paths | grep default
   default = $TESTTMP/issue3651/src (glob)
 
   $ hg -R clone2 summary --large
   parent: 0:fc0bd45326d3 tip
    #0
   branch: default
   commit: (clean)
   update: (current)
   largefiles: (no files to upload)
   $ hg -R clone2 outgoing --large
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   no changes found
   largefiles: no files to upload
   [1]
 
   $ hg -R clone2 outgoing --large --graph --template "{rev}"
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   no changes found
   largefiles: no files to upload
 
 check messages when there are files to upload:
 
   $ echo b > clone2/b
   $ hg -R clone2 add --large clone2/b
   $ hg -R clone2 commit -m '#1'
   Invoking status precommit hook
   A b
   $ hg -R clone2 summary --large
   parent: 1:1acbe71ce432 tip
    #1
   branch: default
   commit: (clean)
   update: (current)
   largefiles: 1 entities for 1 files to upload
   $ hg -R clone2 outgoing --large
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   changeset:   1:1acbe71ce432
   tag:         tip
   user:        test
   date:        Thu Jan 01 00:00:00 1970 +0000
   summary:     #1
   
   largefiles to upload (1 entities):
   b
   
   $ hg -R clone2 outgoing --large --graph --template "{rev}"
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   @  1
   
   largefiles to upload (1 entities):
   b
   
 
   $ cp clone2/b clone2/b1
   $ cp clone2/b clone2/b2
   $ hg -R clone2 add --large clone2/b1 clone2/b2
   $ hg -R clone2 commit -m '#2: add largefiles referring same entity'
   Invoking status precommit hook
   A b1
   A b2
   $ hg -R clone2 summary --large
   parent: 2:6095d0695d70 tip
    #2: add largefiles referring same entity
   branch: default
   commit: (clean)
   update: (current)
   largefiles: 1 entities for 3 files to upload
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n"
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   1:1acbe71ce432
   2:6095d0695d70
   largefiles to upload (1 entities):
   b
   b1
   b2
   
   $ hg -R clone2 cat -r 1 clone2/.hglf/b
   89e6c98d92887913cadf06b2adb97f26cde4849b
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n" --debug
   comparing with $TESTTMP/issue3651/src (glob)
   query 1; heads
   searching for changes
   all remote heads known locally
   1:1acbe71ce432
   2:6095d0695d70
   finding outgoing largefiles: 0/2 revision (0.00%)
   finding outgoing largefiles: 1/2 revision (50.00%)
   largefiles to upload (1 entities):
   b
       89e6c98d92887913cadf06b2adb97f26cde4849b
   b1
       89e6c98d92887913cadf06b2adb97f26cde4849b
   b2
       89e6c98d92887913cadf06b2adb97f26cde4849b
   
 
   $ echo bbb > clone2/b
   $ hg -R clone2 commit -m '#3: add new largefile entity as existing file'
   Invoking status precommit hook
   M b
   $ echo bbbb > clone2/b
   $ hg -R clone2 commit -m '#4: add new largefile entity as existing file'
   Invoking status precommit hook
   M b
   $ cp clone2/b1 clone2/b
   $ hg -R clone2 commit -m '#5: refer existing largefile entity again'
   Invoking status precommit hook
   M b
   $ hg -R clone2 summary --large
   parent: 5:036794ea641c tip
    #5: refer existing largefile entity again
   branch: default
   commit: (clean)
   update: (current)
   largefiles: 3 entities for 3 files to upload
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n"
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   1:1acbe71ce432
   2:6095d0695d70
   3:7983dce246cc
   4:233f12ada4ae
   5:036794ea641c
   largefiles to upload (3 entities):
   b
   b1
   b2
   
   $ hg -R clone2 cat -r 3 clone2/.hglf/b
   c801c9cfe94400963fcb683246217d5db77f9a9a
   $ hg -R clone2 cat -r 4 clone2/.hglf/b
   13f9ed0898e315bf59dc2973fec52037b6f441a2
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n" --debug
   comparing with $TESTTMP/issue3651/src (glob)
   query 1; heads
   searching for changes
   all remote heads known locally
   1:1acbe71ce432
   2:6095d0695d70
   3:7983dce246cc
   4:233f12ada4ae
   5:036794ea641c
   finding outgoing largefiles: 0/5 revision (0.00%)
   finding outgoing largefiles: 1/5 revision (20.00%)
   finding outgoing largefiles: 2/5 revision (40.00%)
   finding outgoing largefiles: 3/5 revision (60.00%)
   finding outgoing largefiles: 4/5 revision (80.00%)
   largefiles to upload (3 entities):
   b
       13f9ed0898e315bf59dc2973fec52037b6f441a2
       89e6c98d92887913cadf06b2adb97f26cde4849b
       c801c9cfe94400963fcb683246217d5db77f9a9a
   b1
       89e6c98d92887913cadf06b2adb97f26cde4849b
   b2
       89e6c98d92887913cadf06b2adb97f26cde4849b
   
 
 Pushing revision #1 causes uploading entity 89e6c98d9288, which is
 shared also by largefiles b1, b2 in revision #2 and b in revision #5.
 
 Then, entity 89e6c98d9288 is not treated as "outgoing entity" at "hg
 summary" and "hg outgoing", even though files in outgoing revision #2
 and #5 refer it.
 
   $ hg -R clone2 push -r 1 -q
   $ hg -R clone2 summary --large
   parent: 5:036794ea641c tip
    #5: refer existing largefile entity again
   branch: default
   commit: (clean)
   update: (current)
   largefiles: 2 entities for 1 files to upload
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n"
   comparing with $TESTTMP/issue3651/src (glob)
   searching for changes
   2:6095d0695d70
   3:7983dce246cc
   4:233f12ada4ae
   5:036794ea641c
   largefiles to upload (2 entities):
   b
   
   $ hg -R clone2 outgoing --large -T "{rev}:{node|short}\n" --debug
   comparing with $TESTTMP/issue3651/src (glob)
   query 1; heads
   searching for changes
   all remote heads known locally
   2:6095d0695d70
   3:7983dce246cc
   4:233f12ada4ae
   5:036794ea641c
   finding outgoing largefiles: 0/4 revision (0.00%)
   finding outgoing largefiles: 1/4 revision (25.00%)
   finding outgoing largefiles: 2/4 revision (50.00%)
   finding outgoing largefiles: 3/4 revision (75.00%)
   largefiles to upload (2 entities):
   b
       13f9ed0898e315bf59dc2973fec52037b6f441a2
       c801c9cfe94400963fcb683246217d5db77f9a9a
   
 
   $ cd ..
 
 merge action 'd' for 'local renamed directory to d2/g' which has no filename
 ==================================================================================
 
   $ hg init merge-action
   $ cd merge-action
   $ touch l
   $ hg add --large l
   $ mkdir d1
   $ touch d1/f
   $ hg ci -Aqm0
   Invoking status precommit hook
   A d1/f
   A l
   $ echo > d1/f
   $ touch d1/g
   $ hg ci -Aqm1
   Invoking status precommit hook
   M d1/f
   A d1/g
   $ hg up -qr0
   $ hg mv d1 d2
   moving d1/f to d2/f (glob)
   $ hg ci -qm2
   Invoking status precommit hook
   A d2/f
   R d1/f
   $ hg merge
   merging d2/f and d1/f to d2/f
   getting changed largefiles
   0 largefiles updated, 0 removed
   1 files updated, 1 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ cd ..
 
 
 Merge conflicts:
 =====================
 
   $ hg init merge
   $ cd merge
   $ echo 0 > f-different
   $ echo 0 > f-same
   $ echo 0 > f-unchanged-1
   $ echo 0 > f-unchanged-2
   $ hg add --large *
   $ hg ci -m0
   Invoking status precommit hook
   A f-different
   A f-same
   A f-unchanged-1
   A f-unchanged-2
   $ echo tmp1 > f-unchanged-1
   $ echo tmp1 > f-unchanged-2
   $ echo tmp1 > f-same
   $ hg ci -m1
   Invoking status precommit hook
   M f-same
   M f-unchanged-1
   M f-unchanged-2
   $ echo 2 > f-different
   $ echo 0 > f-unchanged-1
   $ echo 1 > f-unchanged-2
   $ echo 1 > f-same
   $ hg ci -m2
   Invoking status precommit hook
   M f-different
   M f-same
   M f-unchanged-1
   M f-unchanged-2
   $ hg up -qr0
   $ echo tmp2 > f-unchanged-1
   $ echo tmp2 > f-unchanged-2
   $ echo tmp2 > f-same
   $ hg ci -m3
   Invoking status precommit hook
   M f-same
   M f-unchanged-1
   M f-unchanged-2
   created new head
   $ echo 1 > f-different
   $ echo 1 > f-unchanged-1
   $ echo 0 > f-unchanged-2
   $ echo 1 > f-same
   $ hg ci -m4
   Invoking status precommit hook
   M f-different
   M f-same
   M f-unchanged-1
   M f-unchanged-2
   $ hg merge
   largefile f-different has a merge conflict
   ancestor was 09d2af8dd22201dd8d48e5dcfcaed281ff9422c7
   keep (l)ocal e5fa44f2b31c1fb553b6021e7360d07d5d91ff5e or
   take (o)ther 7448d8798a4380162d4b56f9b452e2f6f9e24e7a? l
   getting changed largefiles
   1 largefiles updated, 0 removed
   0 files updated, 4 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ cat f-different
   1
   $ cat f-same
   1
   $ cat f-unchanged-1
   1
   $ cat f-unchanged-2
   1
   $ cd ..
 
 Test largefile insulation (do not enabled a side effect
 ========================================================
 
 Check whether "largefiles" feature is supported only in repositories
 enabling largefiles extension.
 
   $ mkdir individualenabling
   $ cd individualenabling
 
   $ hg init enabledlocally
   $ echo large > enabledlocally/large
   $ hg -R enabledlocally add --large enabledlocally/large
   $ hg -R enabledlocally commit -m '#0'
   Invoking status precommit hook
   A large
 
   $ hg init notenabledlocally
   $ echo large > notenabledlocally/large
   $ hg -R notenabledlocally add --large notenabledlocally/large
   $ hg -R notenabledlocally commit -m '#0'
   Invoking status precommit hook
   A large
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > # disable globally
   > largefiles=!
   > EOF
   $ cat >> enabledlocally/.hg/hgrc <<EOF
   > [extensions]
   > # enable locally
   > largefiles=
   > EOF
   $ hg -R enabledlocally root
   $TESTTMP/individualenabling/enabledlocally (glob)
   $ hg -R notenabledlocally root
   abort: repository requires features unknown to this Mercurial: largefiles!
   (see http://mercurial.selenic.com/wiki/MissingRequirement for more information)
   [255]
 
   $ hg init push-dst
   $ hg -R enabledlocally push push-dst
   pushing to push-dst
   abort: required features are not supported in the destination: largefiles
   [255]
 
   $ hg init pull-src
   $ hg -R pull-src pull enabledlocally
   pulling from enabledlocally
   abort: required features are not supported in the destination: largefiles
   [255]
 
   $ hg clone enabledlocally clone-dst
   abort: repository requires features unknown to this Mercurial: largefiles!
   (see http://mercurial.selenic.com/wiki/MissingRequirement for more information)
   [255]
   $ test -d clone-dst
   [1]
   $ hg clone --pull enabledlocally clone-pull-dst
   abort: required features are not supported in the destination: largefiles
   [255]
   $ test -d clone-pull-dst
   [1]
 
 #if serve
 
 Test largefiles specific peer setup, when largefiles is enabled
 locally (issue4109)
 
   $ hg showconfig extensions | grep largefiles
   extensions.largefiles=!
   $ mkdir -p $TESTTMP/individualenabling/usercache
 
   $ hg serve -R enabledlocally -d -p $HGPORT --pid-file hg.pid
   $ cat hg.pid >> $DAEMON_PIDS
 
   $ hg init pull-dst
   $ cat > pull-dst/.hg/hgrc <<EOF
   > [extensions]
   > # enable locally
   > largefiles=
   > [largefiles]
   > # ignore system cache to force largefiles specific wire proto access
   > usercache=$TESTTMP/individualenabling/usercache
   > EOF
   $ hg -R pull-dst -q pull -u http://localhost:$HGPORT
 
   $ "$TESTDIR/killdaemons.py" $DAEMON_PIDS
 #endif
 
+Test overridden functions work correctly even for repos disabling
+largefiles (issue4547)
+
+  $ hg showconfig extensions | grep largefiles
+  extensions.largefiles=!
+
+(test updating implied by clone)
+
+  $ hg init enabled-but-no-largefiles
+  $ echo normal1 > enabled-but-no-largefiles/normal1
+  $ hg -R enabled-but-no-largefiles add enabled-but-no-largefiles/normal1
+  $ hg -R enabled-but-no-largefiles commit -m '#0@enabled-but-no-largefiles'
+  Invoking status precommit hook
+  A normal1
+  $ cat >> enabled-but-no-largefiles/.hg/hgrc <<EOF
+  > [extensions]
+  > # enable locally
+  > largefiles=
+  > EOF
+  $ hg clone -q enabled-but-no-largefiles no-largefiles
+
+(test rebasing implied by pull: precommit while rebasing unexpectedly
+shows "normal3" as "?", because lfdirstate isn't yet written out at
+that time)
+
+  $ echo normal2 > enabled-but-no-largefiles/normal2
+  $ hg -R enabled-but-no-largefiles add enabled-but-no-largefiles/normal2
+  $ hg -R enabled-but-no-largefiles commit -m '#1@enabled-but-no-largefiles'
+  Invoking status precommit hook
+  A normal2
+
+  $ echo normal3 > no-largefiles/normal3
+  $ hg -R no-largefiles add no-largefiles/normal3
+  $ hg -R no-largefiles commit -m '#1@no-largefiles'
+  Invoking status precommit hook
+  A normal3
+
+  $ hg -R no-largefiles -q pull --rebase
+  Invoking status precommit hook
+  ? normal3
+
+(test reverting)
+
+  $ hg init subrepo-root
+  $ cat >> subrepo-root/.hg/hgrc <<EOF
+  > [extensions]
+  > # enable locally
+  > largefiles=
+  > EOF
+  $ echo large > subrepo-root/large
+  $ hg -R subrepo-root add --large subrepo-root/large
+  $ hg clone -q no-largefiles subrepo-root/no-largefiles
+  $ cat > subrepo-root/.hgsub <<EOF
+  > no-largefiles = no-largefiles
+  > EOF
+  $ hg -R subrepo-root add subrepo-root/.hgsub
+  $ hg -R subrepo-root commit -m '#0'
+  Invoking status precommit hook
+  A .hgsub
+  A large
+  ? .hgsubstate
+  $ echo dirty >> subrepo-root/large
+  $ echo dirty >> subrepo-root/no-largefiles/normal1
+  $ hg -R subrepo-root status -S
+  M large
+  M no-largefiles/normal1
+  $ hg -R subrepo-root revert --all
+  reverting subrepo-root/.hglf/large (glob)
+  reverting subrepo no-largefiles
+  reverting subrepo-root/no-largefiles/normal1 (glob)
+
   $ cd ..
 
 
 Test "pull --rebase" when rebase is enabled before largefiles (issue3861)
 =========================================================================
 
   $ hg showconfig extensions | grep largefiles
   extensions.largefiles=!
 
   $ mkdir issue3861
   $ cd issue3861
   $ hg init src
   $ hg clone -q src dst
   $ echo a > src/a
   $ hg -R src commit -Aqm "#0"
   Invoking status precommit hook
   A a
 
   $ cat >> dst/.hg/hgrc <<EOF
   > [extensions]
   > largefiles=
   > EOF
   $ hg -R dst pull --rebase
   pulling from $TESTTMP/issue3861/src (glob)
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   nothing to rebase - working directory parent is already an ancestor of destination bf5e395ced2c
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
 
   $ cd ..