# 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 contextlib

import copy

import os

import stat

from mercurial.i18n import _

from mercurial.node import hex

from mercurial.pycompat import open

from mercurial import (

    dirstate,

    encoding,

    error,

    httpconnection,

    match as matchmod,

    pycompat,

    requirements,

    scmutil,

    sparse,

    util,

    vfs as vfsmod,

)

from mercurial.utils import hashutil

from mercurial.dirstateutils import timestamp

shortname = b'.hglf'

shortnameslash = shortname + b'/'

longname = b'largefiles'

# -- Private worker functions ------------------------------------------

@contextlib.contextmanager

def lfstatus(repo, value=True):

    oldvalue = getattr(repo, 'lfstatus', False)

    repo.lfstatus = value

    try:

        yield

    finally:

        repo.lfstatus = oldvalue

def getminsize(ui, assumelfiles, opt, default=10):

    lfsize = opt

    if not lfsize and assumelfiles:

        lfsize = ui.config(longname, b'minsize', default=default)

    if lfsize:

        try:

            lfsize = float(lfsize)

        except ValueError:

            raise error.Abort(

                _(b'largefiles: size must be number (not %s)\n') % lfsize

            )

    if lfsize is None:

        raise error.Abort(_(b'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:

        util.oslink(src, dest)

    except OSError:

        # if hardlinks fail, fallback on atomic copy

        with open(src, b'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, b'usercache')

    if path:

        return path

    hint = None

    if pycompat.iswindows:

        appdata = encoding.environ.get(

            b'LOCALAPPDATA', encoding.environ.get(b'APPDATA')

        )

        if appdata:

            return os.path.join(appdata, name)

        hint = _(b"define %s or %s in the environment, or set %s.usercache") % (

            b"LOCALAPPDATA",

            b"APPDATA",

            name,

        )

    elif pycompat.isdarwin:

        home = encoding.environ.get(b'HOME')

        if home:

            return os.path.join(home, b'Library', b'Caches', name)

        hint = _(b"define %s in the environment, or set %s.usercache") % (

            b"HOME",

            name,

        )

    elif pycompat.isposix:

        path = encoding.environ.get(b'XDG_CACHE_HOME')

        if path:

            return os.path.join(path, name)

        home = encoding.environ.get(b'HOME')

        if home:

            return os.path.join(home, b'.cache', name)

        hint = _(b"define %s or %s in the environment, or set %s.usercache") % (

            b"XDG_CACHE_HOME",

            b"HOME",

            name,

        )

    else:

        raise error.Abort(

            _(b'unknown operating system: %s\n') % pycompat.osname

        )

    raise error.Abort(_(b'unknown %s usercache location') % name, hint=hint)

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, exists = findstorepath(repo, hash)

    if exists:

        repo.ui.note(_(b'found %s in store\n') % hash)

        return path

    elif inusercache(repo.ui, hash):

        repo.ui.note(_(b'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 set_tracked(self, f):

        return super(largefilesdirstate, self).set_tracked(unixpath(f))

    def set_untracked(self, f):

        return super(largefilesdirstate, self).set_untracked(unixpath(f))

    def normal(self, f, parentfiledata=None):

        # not sure if we should pass the `parentfiledata` down or throw it

        # away. So throwing it away to stay on the safe side.

        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):

        # (1) disable PENDING mode always

        #     (lfdirstate isn't yet managed as a part of the transaction)

        # (2) avoid develwarn 'use dirstate.write with ....'

        if tr:

            tr.addbackup(b'largefiles/dirstate', location=b'plain')

        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))

    use_dirstate_v2 = requirements.DIRSTATE_V2_REQUIREMENT in repo.requirements

    lfdirstate = largefilesdirstate(

        opener,

        ui,

        repo.root,

        repo.dirstate._validate,

        lambda: sparse.matcher(repo),

        repo.nodeconstants,

        use_dirstate_v2,

    )

    # 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, b'dirstate')):

        matcher = getstandinmatcher(repo)

        standins = repo.dirstate.walk(

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

        )

        if len(standins) > 0:

            vfs.makedirs(lfstoredir)

        with lfdirstate.parentchange():

            for standin in standins:

                lfile = splitstandin(standin)

                lfdirstate.update_file(

                    lfile, p1_tracked=True, wc_tracked=True, possibly_dirty=True

                )

    return lfdirstate

def lfdirstatestatus(lfdirstate, repo):

    pctx = repo[b'.']

    match = matchmod.always()

    unsure, s, mtime_boundary = lfdirstate.status(

        match, subrepos=[], ignored=False, clean=False, unknown=False

    )

    modified, clean = s.modified, s.clean

    wctx = repo[None]

    for lfile in unsure:

        try:

            fctx = pctx[standin(lfile)]

        except LookupError:

            fctx = None

        if not fctx or readasstandin(fctx) != hashfile(repo.wjoin(lfile)):

            modified.append(lfile)

        else:

            clean.append(lfile)

            st = wctx[lfile].lstat()

            mode = st.st_mode

            size = st.st_size

            mtime = timestamp.reliable_mtime_of(st, mtime_boundary)

            if mtime is not None:

                cache_data = (mode, size, mtime)

                lfdirstate.set_clean(lfile, cache_data)

    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.get_entry(f).any_tracked

    ]

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)).

    """

    # For shared repos, the primary store is in the share source.  But for

    # backward compatibility, force a lookup in the local store if it wasn't

    # found in the share source.

    path = storepath(repo, hash, False)

    if instore(repo, hash):

        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:

    this is meant to be called only after ensuring that the needed

    largefile exists in the cache)."""

    wvfs = repo.wvfs

    path = findfile(repo, hash)

    if path is None:

        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, b'rb') as srcfd, wvfs(filename, b'wb') as destfd:

        gothash = copyandhash(util.filechunkiter(srcfd), destfd)

    if gothash != hash:

        repo.ui.warn(

            _(b'%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(

            _(b"%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, b'rb') as srcf: