largefiles: check whether specified patterns are related to largefiles strictly

current 'lfiles_repo.status()' implementation examines whether
specified patterns are related to largefiles in working directory (not
to STANDIN) or not by NOT-EMPTY-NESS of below list:

    [f for f in match.files() if f in lfdirstate]

but it can not be assumed that all in 'match.files()' are file itself
exactly, because user may only specify part of path to match whole
under subdirectories recursively.

above examination will mis-recognize such pattern as 'not related to
largefiles', and executes normal 'status()' procedure. so, 'hg status'
shows '?'(unknown) status for largefiles in working directory unexpectedly.

this patch examines relation of pattern to largefiles by applying
'match()' on each entries in lfdirstate and checking wheter there is
no matched entry.

it may increase cost of examination, because it causes of full scan of
entries in lfdirstate.

so this patch uses normal for-loop instead of list comprehensions, to
decrease cost when matching is found.

hgext/largefiles/reposetup.py

@@ -1,448 +1,450
 # 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 types
 import os
 
 from mercurial import context, error, manifest, match as match_, util
 from mercurial import node as node_
 from mercurial.i18n import _
 
 import lfcommands
 import proto
 import lfutil
 
 def reposetup(ui, repo):
     # wire repositories should be given new wireproto functions but not the
     # other largefiles modifications
     if not repo.local():
         return proto.wirereposetup(ui, repo)
 
     for name in ('status', 'commitctx', 'commit', 'push'):
         method = getattr(repo, name)
         if (isinstance(method, types.FunctionType) and
             method.func_name == 'wrap'):
             ui.warn(_('largefiles: repo method %r appears to have already been'
                     ' wrapped by another extension: '
                     'largefiles may behave incorrectly\n')
                     % name)
 
     class lfiles_repo(repo.__class__):
         lfstatus = False
         def status_nolfiles(self, *args, **kwargs):
             return super(lfiles_repo, 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(lfiles_repo, self).__getitem__(changeid)
             if self.lfstatus:
                 class lfiles_manifestdict(manifest.manifestdict):
                     def __contains__(self, filename):
                         if super(lfiles_manifestdict,
                                 self).__contains__(filename):
                             return True
                         return super(lfiles_manifestdict,
                             self).__contains__(lfutil.standin(filename))
                 class lfiles_ctx(ctx.__class__):
                     def files(self):
                         filenames = super(lfiles_ctx, self).files()
                         return [lfutil.splitstandin(f) or f for f in filenames]
                     def manifest(self):
                         man1 = super(lfiles_ctx, self).manifest()
                         man1.__class__ = lfiles_manifestdict
                         return man1
                     def filectx(self, path, fileid=None, filelog=None):
                         try:
                             result = super(lfiles_ctx, self).filectx(path,
                                 fileid, filelog)
                         except error.LookupError:
                             # Adding a null character will cause Mercurial to
                             # identify this as a binary file.
                             result = super(lfiles_ctx, self).filectx(
                                 lfutil.standin(path), fileid, filelog)
                             olddata = result.data
                             result.data = lambda: olddata() + '\0'
                         return result
                 ctx.__class__ = lfiles_ctx
             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.
         def status(self, node1='.', node2=None, match=None, ignored=False,
                 clean=False, unknown=False, listsubrepos=False):
             listignored, listclean, listunknown = ignored, clean, unknown
             if not self.lfstatus:
                 return super(lfiles_repo, self).status(node1, node2, match,
                     listignored, listclean, listunknown, listsubrepos)
             else:
                 # some calls in this function rely on the old version of status
                 self.lfstatus = False
                 if isinstance(node1, context.changectx):
                     ctx1 = node1
                 else:
                     ctx1 = repo[node1]
                 if isinstance(node2, context.changectx):
                     ctx2 = node2
                 else:
                     ctx2 = repo[node2]
                 working = ctx2.rev() is None
                 parentworking = working and ctx1 == self['.']
 
                 def inctx(file, ctx):
                     try:
                         if ctx.rev() is None:
                             return file in ctx.manifest()
                         ctx[file]
                         return True
                     except KeyError:
                         return False
 
                 if match is None:
                     match = match_.always(self.root, self.getcwd())
 
                 # First check if there were files specified on the
                 # command line.  If there were, and none of them were
                 # largefiles, we should just bail here and let super
                 # handle it -- thus gaining a big performance boost.
                 lfdirstate = lfutil.openlfdirstate(ui, self)
                 if match.files() and not match.anypats():
-                    matchedfiles = [f for f in match.files() if f in lfdirstate]
-                    if not matchedfiles:
+                    for f in lfdirstate:
+                        if match(f):
+                            break
+                    else:
                         return super(lfiles_repo, self).status(node1, node2,
                                 match, listignored, listclean,
                                 listunknown, listsubrepos)
 
                 # Create a copy of match that matches standins instead
                 # of largefiles.
                 def tostandin(file):
                     if inctx(lfutil.standin(file), ctx2):
                         return lfutil.standin(file)
                     return file
 
                 # Create a function that we can use to override what is
                 # normally the ignore matcher.  We've already checked
                 # for ignored files on the first dirstate walk, and
                 # unecessarily re-checking here causes a huge performance
                 # hit because lfdirstate only knows about largefiles
                 def _ignoreoverride(self):
                     return False
 
                 m = copy.copy(match)
                 m._files = [tostandin(f) for f in m._files]
 
                 # Get ignored files here even if we weren't asked for them; we
                 # must use the result here for filtering later
                 result = super(lfiles_repo, self).status(node1, node2, m,
                     True, clean, unknown, listsubrepos)
                 if working:
                     try:
                         # Any non-largefiles that were explicitly listed must be
                         # taken out or lfdirstate.status will report an error.
                         # The status of these files was already computed using
                         # super's status.
                         # Override lfdirstate's ignore matcher to not do
                         # anything
                         orig_ignore = lfdirstate._ignore
                         lfdirstate._ignore = _ignoreoverride
 
                         match._files = [f for f in match._files if f in
                             lfdirstate]
                         # Don't waste time getting the ignored and unknown
                         # files again; we already have them
                         s = lfdirstate.status(match, [], False,
                                 listclean, False)
                         (unsure, modified, added, removed, missing, unknown,
                                 ignored, clean) = s
                         # Replace the list of ignored and unknown files with
                         # the previously caclulated lists, and strip out the
                         # largefiles
                         lfiles = set(lfdirstate._map)
                         ignored = set(result[5]).difference(lfiles)
                         unknown = set(result[4]).difference(lfiles)
                         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:
                                     clean.append(lfile)
                                     lfdirstate.normal(lfile)
                         else:
                             tocheck = unsure + modified + added + clean
                             modified, added, clean = [], [], []
 
                             for lfile in tocheck:
                                 standin = lfutil.standin(lfile)
                                 if inctx(standin, ctx1):
                                     if ctx1[standin].data().strip() != \
                                             lfutil.hashfile(self.wjoin(lfile)):
                                         modified.append(lfile)
                                     else:
                                         clean.append(lfile)
                                 else:
                                     added.append(lfile)
                     finally:
                         # Replace the original ignore function
                         lfdirstate._ignore = orig_ignore
 
                     for standin in ctx1.manifest():
                         if not lfutil.isstandin(standin):
                             continue
                         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 it's still in
                     # lfdirstate. This happens in merges where files
                     # change type.
                     removed = [f for f in removed if f not in repo.dirstate]
                     result[2] = [f for f in result[2] if f not in lfdirstate]
 
                     # Unknown files
                     unknown = set(unknown).difference(ignored)
                     result[4] = [f for f in unknown
                                  if (repo.dirstate[f] == '?' and
                                      not lfutil.isstandin(f))]
                     # Ignored files were calculated earlier by the dirstate,
                     # and we already stripped out the largefiles from the list
                     result[5] = ignored
                     # combine normal files and largefiles
                     normals = [[fn for fn in filelist
                                 if not lfutil.isstandin(fn)]
                                for filelist in result]
                     lfiles = (modified, added, removed, missing, [], [], clean)
                     result = [sorted(list1 + list2)
                               for (list1, list2) in zip(normals, lfiles)]
                 else:
                     def toname(f):
                         if lfutil.isstandin(f):
                             return lfutil.splitstandin(f)
                         return f
                     result = [[toname(f) for f in items] for items in result]
 
                 if not listunknown:
                     result[4] = []
                 if not listignored:
                     result[5] = []
                 if not listclean:
                     result[6] = []
                 self.lfstatus = True
                 return result
 
         # As part of committing, copy all of the largefiles into the
         # cache.
         def commitctx(self, *args, **kwargs):
             node = super(lfiles_repo, self).commitctx(*args, **kwargs)
             lfutil.copyalltostore(self, node)
             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(lfiles_repo, self).commit
 
             wlock = repo.wlock()
             try:
                 # Case 0: Rebase or Transplant
                 # We have to take the time to pull down the new largefiles now.
                 # Otherwise, any largefiles that were modified in the
                 # destination changesets get overwritten, either by the rebase
                 # or in the first commit after the rebase or transplant.
                 # updatelfiles will update the dirstate to mark any pulled
                 # largefiles as modified
                 if getattr(repo, "_isrebasing", False) or \
                         getattr(repo, "_istransplanting", False):
                     lfcommands.updatelfiles(repo.ui, repo, filelist=None,
                                             printmessage=False)
                     result = orig(text=text, user=user, date=date, match=match,
                                     force=force, editor=editor, extra=extra)
                     return result
                 # 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
                     (not match.anypats() and not match.files())):
                     # 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 = lfutil.openlfdirstate(ui, self)
                     dirtymatch = match_.always(repo.root, repo.getcwd())
                     s = lfdirstate.status(dirtymatch, [], False, False, False)
                     modifiedfiles = []
                     for i in s:
                         modifiedfiles.extend(i)
                     lfiles = lfutil.listlfiles(self)
                     # this only loops through largefiles that exist (not
                     # removed/renamed)
                     for lfile in lfiles:
                         if lfile in modifiedfiles:
                             if os.path.exists(self.wjoin(lfutil.standin(lfile))):
                                 # this handles the case where a rebase is being
                                 # performed and the working copy is not updated
                                 # yet.
                                 if os.path.exists(self.wjoin(lfile)):
                                     lfutil.updatestandin(self,
                                         lfutil.standin(lfile))
                                     lfdirstate.normal(lfile)
                     for lfile in lfdirstate:
                         if lfile in modifiedfiles:
                             if not os.path.exists(
                                     repo.wjoin(lfutil.standin(lfile))):
                                 lfdirstate.drop(lfile)
 
                     result = orig(text=text, user=user, date=date, match=match,
                                     force=force, editor=editor, extra=extra)
                     # This needs to be after commit; otherwise precommit hooks
                     # get the wrong status
                     lfdirstate.write()
                     return result
 
                 for f in match.files():
                     if lfutil.isstandin(f):
                         raise util.Abort(
                             _('file "%s" is a largefile standin') % f,
                             hint=('commit the largefile itself instead'))
 
                 # Case 2: user calls commit with specified patterns: refresh
                 # any matching big files.
                 smatcher = lfutil.composestandinmatcher(self, match)
                 standins = lfutil.dirstate_walk(self.dirstate, smatcher)
 
                 # No matching big files: get out of the way and pass control to
                 # the usual commit() method.
                 if not standins:
                     return orig(text=text, user=user, date=date, match=match,
                                     force=force, editor=editor, extra=extra)
 
                 # 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 = lfutil.openlfdirstate(ui, self)
                 for standin in standins:
                     lfile = lfutil.splitstandin(standin)
                     if lfdirstate[lfile] <> 'r':
                         lfutil.updatestandin(self, standin)
                         lfdirstate.normal(lfile)
                     else:
                         lfdirstate.drop(lfile)
 
                 # 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.
                 lfiles = lfutil.listlfiles(repo)
                 match = copy.copy(match)
                 orig_matchfn = 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 = lfutil.standin(f)
 
                     # ignore known largefiles and standins
                     if f in lfiles or fstandin in standins:
                         continue
 
                     # append directory separator to avoid collisions
                     if not fstandin.endswith(os.sep):
                         fstandin += os.sep
 
                     # prevalidate matching standin directories
                     if util.any(st for st in match._files
                                    if st.startswith(fstandin)):
                         continue
                     actualfiles.append(f)
                 match._files = actualfiles
 
                 def matchfn(f):
                     if orig_matchfn(f):
                         return f not in lfiles
                     else:
                         return f in standins
 
                 match.matchfn = matchfn
                 result = orig(text=text, user=user, date=date, match=match,
                                 force=force, editor=editor, extra=extra)
                 # This needs to be after commit; otherwise precommit hooks
                 # get the wrong status
                 lfdirstate.write()
                 return result
             finally:
                 wlock.release()
 
         def push(self, remote, force=False, revs=None, newbranch=False):
             o = lfutil.findoutgoing(repo, remote, force)
             if o:
                 toupload = set()
                 o = repo.changelog.nodesbetween(o, revs)[0]
                 for n in o:
                     parents = [p for p in repo.changelog.parents(n)
                                if p != node_.nullid]
                     ctx = repo[n]
                     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)
 
                     toupload = toupload.union(
                         set([ctx[f].data().strip()
                              for f in files
                              if lfutil.isstandin(f) and f in ctx]))
                 lfcommands.uploadlfiles(ui, self, remote, toupload)
             return super(lfiles_repo, self).push(remote, force, revs,
                 newbranch)
 
     repo.__class__ = lfiles_repo
 
     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)
     ui.setconfig('hooks', 'commit.lfiles', checkrequireslfiles)

tests/test-largefiles.t

@@ -1,951 +1,997
   $ "$TESTDIR/hghave" symlink unix-permissions serve || exit 80
   $ USERCACHE=`pwd`/cache; export USERCACHE
   $ mkdir -p ${USERCACHE}
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > largefiles=
   > purge=
   > rebase=
   > transplant=
   > [phases]
   > publish=False
   > [largefiles]
   > minsize=2
   > patterns=glob:**.dat
   > usercache=${USERCACHE}
   > [hooks]
   > precommit=echo "Invoking status precommit hook"; hg status
   > EOF
 
 Create the repo with a couple of revisions of both large and normal
 files, testing that status correctly shows largefiles and that summary output
 is correct.
 
   $ hg init a
   $ cd a
   $ mkdir sub
   $ echo normal1 > normal1
   $ echo normal2 > sub/normal2
   $ echo large1 > large1
   $ echo large2 > sub/large2
   $ hg add normal1 sub/normal2
   $ hg add --large large1 sub/large2
   $ hg commit -m "add files"
   Invoking status precommit hook
   A large1
   A normal1
   A sub/large2
   A sub/normal2
   $ echo normal11 > normal1
   $ echo normal22 > sub/normal2
   $ echo large11 > large1
   $ echo large22 > sub/large2
   $ hg commit -m "edit files"
   Invoking status precommit hook
   M large1
   M normal1
   M sub/large2
   M sub/normal2
   $ hg sum --large
   parent: 1:ce8896473775 tip
    edit files
   branch: default
   commit: (clean)
   update: (current)
   largefiles: No remote repo
 
 Commit preserved largefile contents.
 
   $ cat normal1
   normal11
   $ cat large1
   large11
   $ cat sub/normal2
   normal22
   $ cat sub/large2
   large22
 
 Remove both largefiles and normal files.
  
   $ hg remove normal1 large1
   $ hg commit -m "remove files"
   Invoking status precommit hook
   R large1
   R normal1
   $ ls
   sub
   $ echo "testlargefile" > large1-test
   $ hg add --large large1-test
   $ hg st
   A large1-test
   $ hg rm large1-test
   not removing large1-test: file has been marked for add (use forget to undo)
   $ hg st
   A large1-test
   $ hg forget large1-test
   $ hg st
   ? large1-test
   $ rm large1-test
 
 Copy both largefiles and normal files (testing that status output is correct).
 
   $ hg cp sub/normal2 normal1
   $ hg cp sub/large2 large1
   $ hg commit -m "copy files"
   Invoking status precommit hook
   A large1
   A normal1
   $ cat normal1
   normal22
   $ cat large1
   large22
 
 Test moving largefiles and verify that normal files are also unaffected.
 
   $ hg mv normal1 normal3
   $ hg mv large1 large3
   $ hg mv sub/normal2 sub/normal4
   $ hg mv sub/large2 sub/large4
   $ hg commit -m "move files"
   Invoking status precommit hook
   A large3
   A normal3
   A sub/large4
   A sub/normal4
   R large1
   R normal1
   R sub/large2
   R sub/normal2
   $ cat normal3
   normal22
   $ cat large3
   large22
   $ cat sub/normal4
   normal22
   $ cat sub/large4
   large22
 
 Test archiving the various revisions.  These hit corner cases known with
 archiving.
 
   $ hg archive -r 0 ../archive0
   $ hg archive -r 1 ../archive1
   $ hg archive -r 2 ../archive2
   $ hg archive -r 3 ../archive3
   $ hg archive -r 4 ../archive4
   $ cd ../archive0
   $ cat normal1 
   normal1
   $ cat large1
   large1
   $ cat sub/normal2
   normal2
   $ cat sub/large2
   large2
   $ cd ../archive1
   $ cat normal1
   normal11
   $ cat large1
   large11
   $ cat sub/normal2
   normal22
   $ cat sub/large2
   large22
   $ cd ../archive2
   $ ls
   sub
   $ cat sub/normal2
   normal22
   $ cat sub/large2
   large22
   $ cd ../archive3
   $ cat normal1
   normal22
   $ cat large1
   large22
   $ cat sub/normal2
   normal22
   $ cat sub/large2
   large22
   $ cd ../archive4
   $ cat normal3
   normal22
   $ cat large3
   large22
   $ cat sub/normal4
   normal22
   $ cat sub/large4
   large22
 
 Commit corner case: specify files to commit.
 
   $ cd ../a
   $ echo normal3 > normal3
   $ echo large3 > large3
   $ echo normal4 > sub/normal4
   $ echo large4 > sub/large4
   $ hg commit normal3 large3 sub/normal4 sub/large4 -m "edit files again"
   Invoking status precommit hook
   M large3
   M normal3
   M sub/large4
   M sub/normal4
   $ cat normal3
   normal3
   $ cat large3
   large3
   $ cat sub/normal4
   normal4
   $ cat sub/large4
   large4
 
 One more commit corner case: commit from a subdirectory.
 
   $ cd ../a
   $ echo normal33 > normal3
   $ echo large33 > large3
   $ echo normal44 > sub/normal4
   $ echo large44 > sub/large4
   $ cd sub
   $ hg commit -m "edit files yet again"
   Invoking status precommit hook
   M large3
   M normal3
   M sub/large4
   M sub/normal4
   $ cat ../normal3
   normal33
   $ cat ../large3
   large33
   $ cat normal4
   normal44
   $ cat large4
   large44
 
 Committing standins is not allowed.
 
   $ cd ..
   $ echo large3 > large3
   $ hg commit .hglf/large3 -m "try to commit standin"
   abort: file ".hglf/large3" is a largefile standin
   (commit the largefile itself instead)
   [255]
 
 Corner cases for adding largefiles.
 
   $ echo large5 > large5
   $ hg add --large large5
   $ hg add --large large5
   large5 already a largefile
   $ mkdir sub2
   $ echo large6 > sub2/large6
   $ echo large7 > sub2/large7
   $ hg add --large sub2
   adding sub2/large6 as a largefile (glob)
   adding sub2/large7 as a largefile (glob)
   $ hg st
   M large3
   A large5
   A sub2/large6
   A sub2/large7
 
 Config settings (pattern **.dat, minsize 2 MB) are respected.
 
   $ echo testdata > test.dat
   $ dd bs=1k count=2k if=/dev/zero of=reallylarge > /dev/null 2> /dev/null
   $ hg add
   adding reallylarge as a largefile
   adding test.dat as a largefile
 
 Test that minsize and --lfsize handle float values;
 also tests that --lfsize overrides largefiles.minsize.
 (0.250 MB = 256 kB = 262144 B)
 
   $ dd if=/dev/zero of=ratherlarge bs=1024 count=256 > /dev/null 2> /dev/null
   $ dd if=/dev/zero of=medium bs=1024 count=128 > /dev/null 2> /dev/null
   $ hg --config largefiles.minsize=.25 add
   adding ratherlarge as a largefile
   adding medium
   $ hg forget medium
   $ hg --config largefiles.minsize=.25 add --lfsize=.125
   adding medium as a largefile
   $ dd if=/dev/zero of=notlarge bs=1024 count=127 > /dev/null 2> /dev/null
   $ hg --config largefiles.minsize=.25 add --lfsize=.125
   adding notlarge
   $ hg forget notlarge
 
 Test forget on largefiles.
 
   $ hg forget large3 large5 test.dat reallylarge ratherlarge medium
   $ hg commit -m "add/edit more largefiles"
   Invoking status precommit hook
   A sub2/large6
   A sub2/large7
   R large3
   ? large5
   ? medium
   ? notlarge
   ? ratherlarge
   ? reallylarge
   ? test.dat
   $ hg st
   ? large3
   ? large5
   ? medium
   ? notlarge
   ? ratherlarge
   ? reallylarge
   ? test.dat
 
 Purge with largefiles: verify that largefiles are still in the working
 dir after a purge.
 
   $ hg purge --all
   $ cat sub/large4
   large44
   $ cat sub2/large6
   large6
   $ cat sub2/large7
   large7
 
 Test addremove: verify that files that should be added as largfiles are added as
 such and that already-existing largfiles are not added as normal files by
 accident.
 
   $ rm normal3
   $ rm sub/large4
   $ echo "testing addremove with patterns" > testaddremove.dat
   $ echo "normaladdremove" > normaladdremove
   $ hg addremove
   removing sub/large4
   adding testaddremove.dat as a largefile
   removing normal3
   adding normaladdremove
 
 Clone a largefiles repo.
 
   $ hg clone . ../b
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
   $ cd ../b
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   7:daea875e9014  add/edit more largefiles
   6:4355d653f84f  edit files yet again
   5:9d5af5072dbd  edit files again
   4:74c02385b94c  move files
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat normal3
   normal33
   $ cat sub/normal4
   normal44
   $ cat sub/large4
   large44
   $ cat sub2/large6
   large6
   $ cat sub2/large7
   large7
   $ cd ..
   $ hg clone a -r 3 c
   adding changesets
   adding manifests
   adding file changes
   added 4 changesets with 10 changes to 4 files
   updating to branch default
   4 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   2 largefiles updated, 0 removed
   $ cd c
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat normal1
   normal22
   $ cat large1
   large22
   $ cat sub/normal2
   normal22
   $ cat sub/large2
   large22
 
 Old revisions of a clone have correct largefiles content (this also
 tests update).
 
   $ hg update -r 1 
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ cat large1
   large11
   $ cat sub/large2
   large22
 
 Rebasing between two repositories does not revert largefiles to old
 revisions (this was a very bad bug that took a lot of work to fix).
 
   $ cd ..
   $ hg clone a d
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
   $ cd b
   $ echo large4-modified > sub/large4
   $ echo normal3-modified > normal3
   $ hg commit -m "modify normal file and largefile in repo b"
   Invoking status precommit hook
   M normal3
   M sub/large4
   $ cd ../d
   $ echo large6-modified > sub2/large6
   $ echo normal4-modified > sub/normal4
   $ hg commit -m "modify normal file largefile in repo d"
   Invoking status precommit hook
   M sub/normal4
   M sub2/large6
   $ cd ..
   $ hg clone d e
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
   $ cd d
   $ hg pull --rebase ../b
   pulling from ../b
   searching for changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 2 changes to 2 files (+1 heads)
   Invoking status precommit hook
   M sub/normal4
   M sub2/large6
   saved backup bundle to $TESTTMP/d/.hg/strip-backup/f574fb32bb45-backup.hg
   nothing to rebase
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   9:598410d3eb9a  modify normal file largefile in repo d
   8:a381d2c8c80e  modify normal file and largefile in repo b
   7:daea875e9014  add/edit more largefiles
   6:4355d653f84f  edit files yet again
   5:9d5af5072dbd  edit files again
   4:74c02385b94c  move files
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat normal3
   normal3-modified
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-modified
   $ cat sub2/large6
   large6-modified
   $ cat sub2/large7
   large7
   $ cd ../e
   $ hg pull ../b
   pulling from ../b
   searching for changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 2 changes to 2 files (+1 heads)
   (run 'hg heads' to see heads, 'hg merge' to merge)
   caching new largefiles
   0 largefiles cached
   $ hg rebase
   Invoking status precommit hook
   M sub/normal4
   M sub2/large6
   saved backup bundle to $TESTTMP/e/.hg/strip-backup/f574fb32bb45-backup.hg
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   9:598410d3eb9a  modify normal file largefile in repo d
   8:a381d2c8c80e  modify normal file and largefile in repo b
   7:daea875e9014  add/edit more largefiles
   6:4355d653f84f  edit files yet again
   5:9d5af5072dbd  edit files again
   4:74c02385b94c  move files
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat normal3
   normal3-modified
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-modified
   $ cat sub2/large6
   large6-modified
   $ cat sub2/large7
   large7
 
 Rollback on largefiles.
 
   $ echo large4-modified-again > sub/large4 
   $ hg commit -m "Modify large4 again"
   Invoking status precommit hook
   M sub/large4
   $ hg rollback
   repository tip rolled back to revision 9 (undo commit)
   working directory now based on revision 9
   $ hg st
   M sub/large4
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   9:598410d3eb9a  modify normal file largefile in repo d
   8:a381d2c8c80e  modify normal file and largefile in repo b
   7:daea875e9014  add/edit more largefiles
   6:4355d653f84f  edit files yet again
   5:9d5af5072dbd  edit files again
   4:74c02385b94c  move files
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat sub/large4
   large4-modified-again
 
 "update --check" refuses to update with uncommitted changes.
   $ hg update --check 8
   abort: uncommitted local changes
   [255]
 
 "update --clean" leaves correct largefiles in working copy.
 
   $ hg update --clean 
   0 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ cat normal3
   normal3-modified
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-modified
   $ cat sub2/large6
   large6-modified
   $ cat sub2/large7
   large7
 
 Now "update check" is happy.
   $ hg update --check 8
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ hg update --check
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   1 largefiles updated, 0 removed
 
 Test removing empty largefiles directories on update
   $ test -d sub2 && echo "sub2 exists"
   sub2 exists
   $ hg update -q null
   $ test -d sub2 && echo "error: sub2 should not exist anymore"
   [1]
   $ hg update -q
 
 Test hg remove removes empty largefiles directories
   $ test -d sub2 && echo "sub2 exists"
   sub2 exists
   $ hg remove sub2/*
   $ test -d sub2 && echo "error: sub2 should not exist anymore"
   [1]
   $ hg revert sub2/large6 sub2/large7
 
 "revert" works on largefiles (and normal files too).
   $ echo hack3 >> normal3
   $ echo hack4 >> sub/normal4
   $ echo hack4 >> sub/large4
   $ rm sub2/large6
   $ hg revert sub2/large6
   $ hg rm sub2/large6
   $ echo new >> sub2/large8
   $ hg add --large sub2/large8
 # XXX we don't really want to report that we're reverting the standin;
 # that's just an implementation detail. But I don't see an obvious fix. ;-(
   $ hg revert sub
   reverting .hglf/sub/large4 (glob)
   reverting sub/normal4 (glob)
   $ hg status
   M normal3
   A sub2/large8
   R sub2/large6
   ? sub/large4.orig
   ? sub/normal4.orig
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-modified
   $ hg revert -a --no-backup
   undeleting .hglf/sub2/large6 (glob)
   forgetting .hglf/sub2/large8 (glob)
   reverting normal3
   $ hg status
   ? sub/large4.orig
   ? sub/normal4.orig
   ? sub2/large8
   $ cat normal3
   normal3-modified
   $ cat sub2/large6
   large6-modified
   $ rm sub/*.orig sub2/large8
 
 revert some files to an older revision
   $ hg revert --no-backup -r 8 sub2
   reverting .hglf/sub2/large6 (glob)
   $ cat sub2/large6
   large6
   $ hg revert --no-backup sub2
   reverting .hglf/sub2/large6 (glob)
   $ hg status
 
 "verify --large" actually verifies largefiles
 
   $ hg verify --large
   checking changesets
   checking manifests
   crosschecking files in changesets and manifests
   checking files
   10 files, 10 changesets, 28 total revisions
   searching 1 changesets for largefiles
   verified existence of 3 revisions of 3 largefiles
 
 Merging does not revert to old versions of largefiles and also check
 that merging after having pulled from a non-default remote works
 correctly.
 
   $ cd ..
   $ hg clone -r 7 e temp
   adding changesets
   adding manifests
   adding file changes
   added 8 changesets with 24 changes to 10 files
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
   $ hg clone temp f
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
 # Delete the largefiles in the largefiles system cache so that we have an
 # opportunity to test that caching after a pull works.
   $ rm ${USERCACHE}/*
   $ cd f
   $ echo "large4-merge-test" > sub/large4
   $ hg commit -m "Modify large4 to test merge"
   Invoking status precommit hook
   M sub/large4
   $ hg pull ../e
   pulling from ../e
   searching for changes
   adding changesets
   adding manifests
   adding file changes
   added 2 changesets with 4 changes to 4 files (+1 heads)
   (run 'hg heads' to see heads, 'hg merge' to merge)
   caching new largefiles
   2 largefiles cached
   $ hg merge
   merging sub/large4
   largefile sub/large4 has a merge conflict
   keep (l)ocal or take (o)ther? l
   3 files updated, 1 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ hg commit -m "Merge repos e and f"
   Invoking status precommit hook
   M normal3
   M sub/normal4
   M sub2/large6
   $ cat normal3
   normal3-modified
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-merge-test
   $ cat sub2/large6
   large6-modified
   $ cat sub2/large7
   large7
 
 Test status after merging with a branch that introduces a new largefile:
 
   $ echo large > large
   $ hg add --large large
   $ hg commit -m 'add largefile'
   Invoking status precommit hook
   A large
   $ hg update -q ".^"
   $ echo change >> normal3
   $ hg commit -m 'some change'
   Invoking status precommit hook
   M normal3
   created new head
   $ hg merge
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ hg status
   M large
 
 Test that a normal file and a largefile with the same name and path cannot
 coexist.
 
   $ rm sub2/large7
   $ echo "largeasnormal" > sub2/large7
   $ hg add sub2/large7
   sub2/large7 already a largefile
 
 Test that transplanting a largefile change works correctly.
 
   $ cd ..
   $ hg clone -r 8 d g
   adding changesets
   adding manifests
   adding file changes
   added 9 changesets with 26 changes to 10 files
   updating to branch default
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   3 largefiles updated, 0 removed
   $ cd g
   $ hg transplant -s ../d 598410d3eb9a 
   searching for changes
   searching for changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 2 changes to 2 files
   $ hg log --template '{rev}:{node|short}  {desc|firstline}\n'
   9:598410d3eb9a  modify normal file largefile in repo d
   8:a381d2c8c80e  modify normal file and largefile in repo b
   7:daea875e9014  add/edit more largefiles
   6:4355d653f84f  edit files yet again
   5:9d5af5072dbd  edit files again
   4:74c02385b94c  move files
   3:9e8fbc4bce62  copy files
   2:51a0ae4d5864  remove files
   1:ce8896473775  edit files
   0:30d30fe6a5be  add files
   $ cat normal3
   normal3-modified
   $ cat sub/normal4
   normal4-modified
   $ cat sub/large4
   large4-modified
   $ cat sub2/large6
   large6-modified
   $ cat sub2/large7
   large7
 
 Test that renaming a largefile results in correct output for status
 
   $ hg rename sub/large4 large4-renamed
   $ hg commit -m "test rename output"
   Invoking status precommit hook
   A large4-renamed
   R sub/large4
   $ cat large4-renamed
   large4-modified
   $ cd sub2
   $ hg rename large6 large6-renamed
   $ hg st
   A sub2/large6-renamed
   R sub2/large6
   $ cd ..
 
 Test --normal flag
 
   $ dd if=/dev/urandom bs=2k count=11k > new-largefile 2> /dev/null
   $ hg add --normal --large new-largefile
   abort: --normal cannot be used with --large
   [255]
   $ hg add --normal new-largefile
   new-largefile: up to 69 MB of RAM may be required to manage this file
   (use 'hg revert new-largefile' to cancel the pending addition)
   $ cd ..
 
 vanilla clients not locked out from largefiles servers on vanilla repos
   $ mkdir r1
   $ cd r1
   $ hg init
   $ echo c1 > f1
   $ hg add f1
   $ hg commit -m "m1"
   Invoking status precommit hook
   A f1
   $ cd ..
   $ hg serve -R r1 -d -p $HGPORT --pid-file hg.pid
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg --config extensions.largefiles=! clone http://localhost:$HGPORT r2
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
 
 largefiles clients still work with vanilla servers
   $ hg --config extensions.largefiles=! serve -R r1 -d -p $HGPORT1 --pid-file hg.pid
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg clone http://localhost:$HGPORT1 r3
   requesting all changes
   adding changesets
   adding manifests
   adding file changes
   added 1 changesets with 1 changes to 1 files
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
 
 vanilla clients locked out from largefiles http repos
   $ mkdir r4
   $ cd r4
   $ hg init
   $ echo c1 > f1
   $ hg add --large f1
   $ hg commit -m "m1"
   Invoking status precommit hook
   A f1
   $ cd ..
   $ hg serve -R r4 -d -p $HGPORT2 --pid-file hg.pid
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg --config extensions.largefiles=! clone http://localhost:$HGPORT2 r5
   abort: remote error:
   
   This repository uses the largefiles extension.
   
   Please enable it in your Mercurial config file.
   [255]
 
 used all HGPORTs, kill all daemons
   $ "$TESTDIR/killdaemons.py"
 
 vanilla clients locked out from largefiles ssh repos
   $ hg --config extensions.largefiles=! clone -e "python $TESTDIR/dummyssh" ssh://user@dummy/r4 r5
   abort: remote error:
   
   This repository uses the largefiles extension.
   
   Please enable it in your Mercurial config file.
   [255]
 
 largefiles clients refuse to push largefiles repos to vanilla servers
   $ mkdir r6
   $ cd r6
   $ hg init
   $ echo c1 > f1
   $ hg add f1
   $ hg commit -m "m1"
   Invoking status precommit hook
   A f1
   $ cat >> .hg/hgrc <<!
   > [web]
   > push_ssl = false
   > allow_push = *
   > !
   $ cd ..
   $ hg clone r6 r7
   updating to branch default
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ cd r7
   $ echo c2 > f2
   $ hg add --large f2
   $ hg commit -m "m2"
   Invoking status precommit hook
   A f2
   $ hg --config extensions.largefiles=! -R ../r6 serve -d -p $HGPORT --pid-file ../hg.pid
   $ cat ../hg.pid >> $DAEMON_PIDS
   $ hg push http://localhost:$HGPORT
   pushing to http://localhost:$HGPORT/
   searching for changes
   abort: http://localhost:$HGPORT/ does not appear to be a largefile store
   [255]
   $ cd ..
 
 putlfile errors are shown (issue3123)
 Corrupt the cached largefile in r7
   $ echo corruption > $USERCACHE/4cdac4d8b084d0b599525cf732437fb337d422a8
   $ hg init empty
   $ hg serve -R empty -d -p $HGPORT1 --pid-file hg.pid \
   >   --config 'web.allow_push=*' --config web.push_ssl=False
   $ cat hg.pid >> $DAEMON_PIDS
   $ hg push -R r7 http://localhost:$HGPORT1
   pushing to http://localhost:$HGPORT1/
   searching for changes
   remote: largefiles: failed to put 4cdac4d8b084d0b599525cf732437fb337d422a8 into store: largefile contents do not match hash
   abort: remotestore: could not put $TESTTMP/r7/.hg/largefiles/4cdac4d8b084d0b599525cf732437fb337d422a8 to remote store http://localhost:$HGPORT1/
   [255]
   $ rm -rf empty
 
 Clone a local repository owned by another user
 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/urandom 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
   1 files updated, 0 files merged, 0 files removed, 0 files unresolved
   getting changed largefiles
   1 largefiles updated, 0 removed
   $ cd ..
   $ chmod -R u+w alice/pubrepo
   $ HOME="$ORIGHOME"
 
 Symlink to a large largefile should behave the same as a symlink to a normal file
   $ 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 ..
 
+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
+
+  $ cd ..