comparison hgext/largefiles/lfutil.py @ 43076:2372284d9457

formatting: blacken the codebase

This is using my patch to black
(https://github.com/psf/black/pull/826) so we don't un-wrap collection
literals.

Done with:

  hg files 'set:**.py - mercurial/thirdparty/** - "contrib/python-zstandard/**"' | xargs black -S

# skip-blame mass-reformatting only

# no-check-commit reformats foo_bar functions

Differential Revision: https://phab.mercurial-scm.org/D6971

43075:57875cf423c9

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 error.Abort(_('largefiles: size must be number (not %s)\n')
                             % lfsize)
    if lfsize is None:
        raise error.Abort(_('minimum size for largefiles must be specified'))
    return lfsize

def link(src, dest):
    """Try to create hardlink - if that fails, efficiently make a copy."""
    util.makedirs(os.path.dirname(dest))
    try:

        with open(src, 'rb') as srcf, util.atomictempfile(dest) as dstf:
            for chunk in util.filechunkiter(srcf):
                dstf.write(chunk)
        os.chmod(dest, os.stat(src).st_mode)

def usercachepath(ui, hash):
    '''Return the correct location in the "global" largefiles cache for a file
    with the given hash.
    This cache is used for sharing of largefiles across repositories - both
    to preserve download bandwidth and storage space.'''
    return os.path.join(_usercachedir(ui), hash)

def _usercachedir(ui, name=longname):
    '''Return the location of the "global" largefiles cache.'''
    path = ui.configpath(name, 'usercache')
    if path:
        return path
    if pycompat.iswindows:
        appdata = encoding.environ.get('LOCALAPPDATA',
                                       encoding.environ.get('APPDATA'))
        if appdata:
            return os.path.join(appdata, name)
    elif pycompat.isdarwin:
        home = encoding.environ.get('HOME')
        if home:

            return os.path.join(path, name)
        home = encoding.environ.get('HOME')
        if home:
            return os.path.join(home, '.cache', name)
    else:
        raise error.Abort(_('unknown operating system: %s\n')
                          % pycompat.osname)
    raise error.Abort(_('unknown %s usercache location') % name)

def inusercache(ui, hash):
    path = usercachepath(ui, hash)
    return os.path.exists(path)

def findfile(repo, hash):
    '''Return store path of the largefile with the specified hash.
    As a side effect, the file might be linked from user cache.
    Return None if the file can't be found locally.'''

        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 write(self, tr=False):
        # (1) disable PENDING mode always
        #     (lfdirstate isn't yet managed as a part of the transaction)
        # (2) avoid develwarn 'use dirstate.write with ....'
        super(largefilesdirstate, self).write(None)

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.
    '''
    vfs = repo.vfs
    lfstoredir = longname
    opener = vfsmod.vfs(vfs.join(lfstoredir))
    lfdirstate = largefilesdirstate(opener, ui, repo.root,
                                    repo.dirstate._validate,
                                    lambda: sparse.matcher(repo))

    # 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 vfs.exists(vfs.join(lfstoredir, 'dirstate')):
        matcher = getstandinmatcher(repo)
        standins = repo.dirstate.walk(matcher, subrepos=[], unknown=False,
                                      ignored=False)

        if len(standins) > 0:
            vfs.makedirs(lfstoredir)

        for standin in standins:
            lfile = splitstandin(standin)
            lfdirstate.normallookup(lfile)
    return lfdirstate

def lfdirstatestatus(lfdirstate, repo):
    pctx = repo['.']
    match = matchmod.always()
    unsure, s = lfdirstate.status(match, subrepos=[], ignored=False,
                                  clean=False, unknown=False)
    modified, clean = s.modified, s.clean
    for lfile in unsure:
        try:
            fctx = pctx[standin(lfile)]
        except LookupError:

        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, forcelocal=False):
    '''Return true if a largefile with the given hash exists in the store'''
    return os.path.exists(storepath(repo, hash, forcelocal))

def storepath(repo, hash, forcelocal=False):
    '''Return the correct location in the repository largefiles store for a
    file with the given hash.'''
    if not forcelocal and repo.shared():
        return repo.vfs.reljoin(repo.sharedpath, longname, hash)
    return repo.vfs.join(longname, hash)

def findstorepath(repo, hash):
    '''Search through the local store path(s) to find the file for the given
    hash.  If the file is not found, its path in the primary store is returned.
    The return value is a tuple of (path, exists(path)).

        return (path, True)
    elif repo.shared() and instore(repo, hash, True):
        return storepath(repo, hash, True), True

    return (path, False)

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:

        return False
    wvfs.makedirs(wvfs.dirname(wvfs.join(filename)))
    # The write may fail before the file is fully written, but we
    # don't use atomic writes in the working copy.
    with open(path, 'rb') as srcfd, wvfs(filename, 'wb') as destfd:
        gothash = copyandhash(
            util.filechunkiter(srcfd), destfd)
    if gothash != hash:
        repo.ui.warn(_('%s: data corruption in %s with hash %s\n')
                     % (filename, path, gothash))
        wvfs.unlink(filename)
        return False
    return True

def copytostore(repo, ctx, file, fstandin):
    wvfs = repo.wvfs
    hash = readasstandin(ctx[fstandin])
    if instore(repo, hash):
        return
    if wvfs.exists(file):
        copytostoreabsolute(repo, wvfs.join(file), hash)
    else:
        repo.ui.warn(_("%s: largefile %s not available from local store\n") %
                     (file, hash))

def copyalltostore(repo, node):
    '''Copy all largefiles in a given revision to the store'''

    ctx = repo[node]
    for filename in ctx.files():
        realfile = splitstandin(filename)
        if realfile is not None and filename in ctx.manifest():
            copytostore(repo, ctx, realfile, filename)

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)))
        with open(file, 'rb') as srcf:
            with util.atomictempfile(storepath(repo, hash),
                                     createmode=repo.store.createmode) as dstf:
                for chunk in util.filechunkiter(srcf):
                    dstf.write(chunk)
        linktousercache(repo, hash)

def linktousercache(repo, hash):
    '''Link / copy the largefile with the specified hash from the store
    to the cache.'''
    path = usercachepath(repo.ui, hash)
    link(storepath(repo, hash), path)

def getstandinmatcher(repo, rmatcher=None):
    '''Return a match object that applies rmatcher to the standin directory'''
    wvfs = repo.wvfs
    standindir = shortname

    else:
        # no patterns: relative to repo root
        match = scmutil.match(repo[None], [wvfs.join(standindir)], badfn=badfn)
    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)
    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:

    # 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).

    if len(bits) == 2 and bits[0] == shortname:
        return bits[1]
    else:
        return None

def updatestandin(repo, lfile, standin):
    """Re-calculate hash value of lfile and write it into standin

    This assumes that "lfutil.standin(lfile) == standin", for efficiency.
    """

        executable = getexecutable(file)
        writestandin(repo, standin, hash, executable)
    else:
        raise error.Abort(_('%s: file not found!') % lfile)

def readasstandin(fctx):
    '''read hex hash from given filectx of standin file

    This encapsulates how "standin" data is stored into storage layer.'''
    return fctx.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 = hashlib.sha1('')
    for data in instream:
        hasher.update(data)
        outfile.write(data)
    return hex(hasher.digest())

def hashfile(file):
    if not os.path.exists(file):
        return ''
    with open(file, 'rb') as fd:
        return hexsha1(fd)

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 += '/'

    url = join(first, second)
    for a in arg:
        url = join(url, a)
    return url

def hexsha1(fileobj):
    """hexsha1 returns the hex-encoded sha1 sum of the data in the file-like
    object data"""
    h = hashlib.sha1()
    for chunk in util.filechunkiter(fileobj):
        h.update(chunk)
    return hex(h.digest())

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):
    '''Return true if the repo is a largefile repo.'''
    if ('largefiles' in repo.requirements and
            any(shortnameslash in f[0] for f in repo.store.datafiles())):
        return True

    return any(openlfdirstate(repo.ui, repo, False))

class storeprotonotcapable(Exception):
    def __init__(self, storetypes):
        self.storetypes = storetypes

def getstandinsstate(repo):
    standins = []
    matcher = getstandinmatcher(repo)
    wctx = repo[None]
    for standin in repo.dirstate.walk(matcher, subrepos=[], unknown=False,
                                      ignored=False):
        lfile = splitstandin(standin)
        try:
            hash = readasstandin(wctx[standin])
        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 or
            not repo.wvfs.exists(lfile)):
            # state 'n' doesn't ensure 'clean' in this case
            lfdirstate.normallookup(lfile)
        else:
            lfdirstate.normal(lfile)
    elif state == 'm':

    elif state == 'a':
        lfdirstate.add(lfile)
    elif state == '?':
        lfdirstate.drop(lfile)

def markcommitted(orig, ctx, node):
    repo = ctx.repo()

    orig(node)

    # lookup while copyalltostore(), but can omit redundant check for
    # files comming from the 2nd parent, which should exist in store
    # at merging.
    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):
    makeprogress = repo.ui.makeprogress
    with makeprogress(_('finding outgoing largefiles'),
                      unit=_('revisions'), total=len(missing)) as progress:
        for i, n in enumerate(missing):
            progress.update(i)
            parents = [p for p in repo[n].parents() if p != node.nullid]

            oldlfstatus = repo.lfstatus

                        files.add(f)
            for fn in files:
                if isstandin(fn) and fn in ctx:
                    addfunc(fn, readasstandin(ctx[fn]))

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.

        # It can cost a lot of time (several seconds)
        # otherwise to update all standins if the largefiles are
        # large.
        lfdirstate = openlfdirstate(ui, repo)
        dirtymatch = matchmod.always()
        unsure, s = lfdirstate.status(dirtymatch, subrepos=[], ignored=False,
                                      clean=False, unknown=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:

    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, subrepos=[], unknown=False,
                                  ignored=False)

    # No matching big files: get out of the way and pass control to
    # the usual commit() method.
    if not standins:
        return match

    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

    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

    '''
    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

43076:2372284d9457

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 error.Abort(
                _('largefiles: size must be number (not %s)\n') % lfsize
            )
    if lfsize is None:
        raise error.Abort(_('minimum size for largefiles must be specified'))
    return lfsize


def link(src, dest):
    """Try to create hardlink - if that fails, efficiently make a copy."""
    util.makedirs(os.path.dirname(dest))
    try:

        with open(src, 'rb') as srcf, util.atomictempfile(dest) as dstf:
            for chunk in util.filechunkiter(srcf):
                dstf.write(chunk)
        os.chmod(dest, os.stat(src).st_mode)


def usercachepath(ui, hash):
    '''Return the correct location in the "global" largefiles cache for a file
    with the given hash.
    This cache is used for sharing of largefiles across repositories - both
    to preserve download bandwidth and storage space.'''
    return os.path.join(_usercachedir(ui), hash)


def _usercachedir(ui, name=longname):
    '''Return the location of the "global" largefiles cache.'''
    path = ui.configpath(name, 'usercache')
    if path:
        return path
    if pycompat.iswindows:
        appdata = encoding.environ.get(
            'LOCALAPPDATA', encoding.environ.get('APPDATA')
        )
        if appdata:
            return os.path.join(appdata, name)
    elif pycompat.isdarwin:
        home = encoding.environ.get('HOME')
        if home:

            return os.path.join(path, name)
        home = encoding.environ.get('HOME')
        if home:
            return os.path.join(home, '.cache', name)
    else:
        raise error.Abort(_('unknown operating system: %s\n') % pycompat.osname)
    raise error.Abort(_('unknown %s usercache location') % name)


def inusercache(ui, hash):
    path = usercachepath(ui, hash)
    return os.path.exists(path)


def findfile(repo, hash):
    '''Return store path of the largefile with the specified hash.
    As a side effect, the file might be linked from user cache.
    Return None if the file can't be found locally.'''

        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 write(self, tr=False):
        # (1) disable PENDING mode always
        #     (lfdirstate isn't yet managed as a part of the transaction)
        # (2) avoid develwarn 'use dirstate.write with ....'
        super(largefilesdirstate, self).write(None)


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.
    '''
    vfs = repo.vfs
    lfstoredir = longname
    opener = vfsmod.vfs(vfs.join(lfstoredir))
    lfdirstate = largefilesdirstate(
        opener,
        ui,
        repo.root,
        repo.dirstate._validate,
        lambda: sparse.matcher(repo),
    )

    # 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 vfs.exists(vfs.join(lfstoredir, 'dirstate')):
        matcher = getstandinmatcher(repo)
        standins = repo.dirstate.walk(
            matcher, subrepos=[], unknown=False, ignored=False
        )

        if len(standins) > 0:
            vfs.makedirs(lfstoredir)

        for standin in standins:
            lfile = splitstandin(standin)
            lfdirstate.normallookup(lfile)
    return lfdirstate


def lfdirstatestatus(lfdirstate, repo):
    pctx = repo['.']
    match = matchmod.always()
    unsure, s = lfdirstate.status(
        match, subrepos=[], ignored=False, clean=False, unknown=False
    )
    modified, clean = s.modified, s.clean
    for lfile in unsure:
        try:
            fctx = pctx[standin(lfile)]
        except LookupError:

        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, forcelocal=False):
    '''Return true if a largefile with the given hash exists in the store'''
    return os.path.exists(storepath(repo, hash, forcelocal))


def storepath(repo, hash, forcelocal=False):
    '''Return the correct location in the repository largefiles store for a
    file with the given hash.'''
    if not forcelocal and repo.shared():
        return repo.vfs.reljoin(repo.sharedpath, longname, hash)
    return repo.vfs.join(longname, hash)


def findstorepath(repo, hash):
    '''Search through the local store path(s) to find the file for the given
    hash.  If the file is not found, its path in the primary store is returned.
    The return value is a tuple of (path, exists(path)).

        return (path, True)
    elif repo.shared() and instore(repo, hash, True):
        return storepath(repo, hash, True), True

    return (path, False)


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:

        return False
    wvfs.makedirs(wvfs.dirname(wvfs.join(filename)))
    # The write may fail before the file is fully written, but we
    # don't use atomic writes in the working copy.
    with open(path, 'rb') as srcfd, wvfs(filename, 'wb') as destfd:
        gothash = copyandhash(util.filechunkiter(srcfd), destfd)
    if gothash != hash:
        repo.ui.warn(
            _('%s: data corruption in %s with hash %s\n')
            % (filename, path, gothash)
        )
        wvfs.unlink(filename)
        return False
    return True


def copytostore(repo, ctx, file, fstandin):
    wvfs = repo.wvfs
    hash = readasstandin(ctx[fstandin])
    if instore(repo, hash):
        return
    if wvfs.exists(file):
        copytostoreabsolute(repo, wvfs.join(file), hash)
    else:
        repo.ui.warn(
            _("%s: largefile %s not available from local store\n")
            % (file, hash)
        )


def copyalltostore(repo, node):
    '''Copy all largefiles in a given revision to the store'''

    ctx = repo[node]
    for filename in ctx.files():
        realfile = splitstandin(filename)
        if realfile is not None and filename in ctx.manifest():
            copytostore(repo, ctx, realfile, filename)


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)))
        with open(file, 'rb') as srcf:
            with util.atomictempfile(
                storepath(repo, hash), createmode=repo.store.createmode
            ) as dstf:
                for chunk in util.filechunkiter(srcf):
                    dstf.write(chunk)
        linktousercache(repo, hash)


def linktousercache(repo, hash):
    '''Link / copy the largefile with the specified hash from the store
    to the cache.'''
    path = usercachepath(repo.ui, hash)
    link(storepath(repo, hash), path)


def getstandinmatcher(repo, rmatcher=None):
    '''Return a match object that applies rmatcher to the standin directory'''
    wvfs = repo.wvfs
    standindir = shortname

    else:
        # no patterns: relative to repo root
        match = scmutil.match(repo[None], [wvfs.join(standindir)], badfn=badfn)
    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)
    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:

    # 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).

    if len(bits) == 2 and bits[0] == shortname:
        return bits[1]
    else:
        return None


def updatestandin(repo, lfile, standin):
    """Re-calculate hash value of lfile and write it into standin

    This assumes that "lfutil.standin(lfile) == standin", for efficiency.
    """

        executable = getexecutable(file)
        writestandin(repo, standin, hash, executable)
    else:
        raise error.Abort(_('%s: file not found!') % lfile)


def readasstandin(fctx):
    '''read hex hash from given filectx of standin file

    This encapsulates how "standin" data is stored into storage layer.'''
    return fctx.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 = hashlib.sha1('')
    for data in instream:
        hasher.update(data)
        outfile.write(data)
    return hex(hasher.digest())


def hashfile(file):
    if not os.path.exists(file):
        return ''
    with open(file, 'rb') as fd:
        return hexsha1(fd)


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 += '/'

    url = join(first, second)
    for a in arg:
        url = join(url, a)
    return url


def hexsha1(fileobj):
    """hexsha1 returns the hex-encoded sha1 sum of the data in the file-like
    object data"""
    h = hashlib.sha1()
    for chunk in util.filechunkiter(fileobj):
        h.update(chunk)
    return hex(h.digest())


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):
    '''Return true if the repo is a largefile repo.'''
    if 'largefiles' in repo.requirements and any(
        shortnameslash in f[0] for f in repo.store.datafiles()
    ):
        return True

    return any(openlfdirstate(repo.ui, repo, False))


class storeprotonotcapable(Exception):
    def __init__(self, storetypes):
        self.storetypes = storetypes


def getstandinsstate(repo):
    standins = []
    matcher = getstandinmatcher(repo)
    wctx = repo[None]
    for standin in repo.dirstate.walk(
        matcher, subrepos=[], unknown=False, ignored=False
    ):
        lfile = splitstandin(standin)
        try:
            hash = readasstandin(wctx[standin])
        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 or not repo.wvfs.exists(lfile):
            # state 'n' doesn't ensure 'clean' in this case
            lfdirstate.normallookup(lfile)
        else:
            lfdirstate.normal(lfile)
    elif state == 'm':

    elif state == 'a':
        lfdirstate.add(lfile)
    elif state == '?':
        lfdirstate.drop(lfile)


def markcommitted(orig, ctx, node):
    repo = ctx.repo()

    orig(node)

    # lookup while copyalltostore(), but can omit redundant check for
    # files comming from the 2nd parent, which should exist in store
    # at merging.
    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):
    makeprogress = repo.ui.makeprogress
    with makeprogress(
        _('finding outgoing largefiles'),
        unit=_('revisions'),
        total=len(missing),
    ) as progress:
        for i, n in enumerate(missing):
            progress.update(i)
            parents = [p for p in repo[n].parents() if p != node.nullid]

            oldlfstatus = repo.lfstatus

                        files.add(f)
            for fn in files:
                if isstandin(fn) and fn in ctx:
                    addfunc(fn, readasstandin(ctx[fn]))


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.

        # It can cost a lot of time (several seconds)
        # otherwise to update all standins if the largefiles are
        # large.
        lfdirstate = openlfdirstate(ui, repo)
        dirtymatch = matchmod.always()
        unsure, s = lfdirstate.status(
            dirtymatch, subrepos=[], ignored=False, clean=False, unknown=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:

    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, subrepos=[], unknown=False, ignored=False
    )

    # No matching big files: get out of the way and pass control to
    # the usual commit() method.
    if not standins:
        return match

    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

    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

    '''
    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