# Copyright 2009-2010 Gregory P. Ward

# Copyright 2009-2010 Intelerad Medical Systems Incorporated

# Copyright 2010-2011 Fog Creek Software

# Copyright 2010-2011 Unity Technologies

#

# This software may be used and distributed according to the terms of the

# GNU General Public License version 2 or any later version.

'''largefiles utility code: must not import other modules in this package.'''

import os

import platform

import shutil

import stat

import copy

from mercurial import dirstate, httpconnection, match as match_, util, scmutil

from mercurial.i18n import _

shortname = '.hglf'

shortnameslash = shortname + '/'

longname = 'largefiles'

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

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

    lfsize = opt

    if not lfsize and assumelfiles:

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

    if lfsize:

        try:

            lfsize = float(lfsize)

        except ValueError:

                             % lfsize)

    if lfsize is None:

    return lfsize

def link(src, dest):

    util.makedirs(os.path.dirname(dest))

    try:

        util.oslink(src, dest)

    except OSError:

        # if hardlinks fail, fallback on atomic copy

        dst = util.atomictempfile(dest)

        for chunk in util.filechunkiter(open(src, 'rb')):

            dst.write(chunk)

        dst.close()

        os.chmod(dest, os.stat(src).st_mode)

def usercachepath(ui, hash):

    path = ui.configpath(longname, 'usercache', None)

    if path:

        path = os.path.join(path, hash)

    else:

        if os.name == 'nt':

            appdata = os.getenv('LOCALAPPDATA', os.getenv('APPDATA'))

            if appdata:

                path = os.path.join(appdata, longname, hash)

        elif platform.system() == 'Darwin':

            home = os.getenv('HOME')

            if home:

                path = os.path.join(home, 'Library', 'Caches',

                                    longname, hash)

        elif os.name == 'posix':

            path = os.getenv('XDG_CACHE_HOME')

            if path:

                path = os.path.join(path, longname, hash)

            else:

                home = os.getenv('HOME')

                if home:

                    path = os.path.join(home, '.cache', longname, hash)

        else:

    return path

def inusercache(ui, hash):

    path = usercachepath(ui, hash)

    return path and os.path.exists(path)

def findfile(repo, hash):

    path, exists = findstorepath(repo, hash)

    if exists:

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

        return path

    elif inusercache(repo.ui, hash):

        repo.ui.note(_('found %s in system cache\n') % hash)

        path = storepath(repo, hash)

        link(usercachepath(repo.ui, hash), path)

        return path

    return None

class largefilesdirstate(dirstate.dirstate):

    def __getitem__(self, key):

        return super(largefilesdirstate, self).__getitem__(unixpath(key))

    def normal(self, f):

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

    def remove(self, f):

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

    def add(self, f):

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

    def drop(self, f):

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

    def forget(self, f):

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

    def normallookup(self, f):

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

    def _ignore(self, f):

        return False

def openlfdirstate(ui, repo, create=True):

    '''

    Return a dirstate object that tracks largefiles: i.e. its root is

    the repo root, but it is saved in .hg/largefiles/dirstate.

    '''

    lfstoredir = repo.join(longname)

    opener = scmutil.opener(lfstoredir)

    lfdirstate = largefilesdirstate(opener, ui, repo.root,

                                     repo.dirstate._validate)

    # If the largefiles dirstate does not exist, populate and create

    # it. This ensures that we create it on the first meaningful

    # largefiles operation in a new clone.

    if create and not os.path.exists(os.path.join(lfstoredir, 'dirstate')):

        matcher = getstandinmatcher(repo)

        standins = repo.dirstate.walk(matcher, [], False, False)

        if len(standins) > 0:

            util.makedirs(lfstoredir)

        for standin in standins:

            lfile = splitstandin(standin)

            lfdirstate.normallookup(lfile)

    return lfdirstate

def lfdirstatestatus(lfdirstate, repo):

    wctx = repo['.']

    match = match_.always(repo.root, repo.getcwd())

    unsure, s = lfdirstate.status(match, [], False, False, False)

    modified, clean = s.modified, s.clean

    for lfile in unsure:

        try:

            fctx = wctx[standin(lfile)]

        except LookupError:

            fctx = None

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

            modified.append(lfile)

        else:

            clean.append(lfile)

            lfdirstate.normal(lfile)

    return s

def listlfiles(repo, rev=None, matcher=None):

    '''return a list of largefiles in the working copy or the

    specified changeset'''

    if matcher is None:

        matcher = getstandinmatcher(repo)

    # ignore unknown files in working directory

    return [splitstandin(f)

            for f in repo[rev].walk(matcher)

            if rev is not None or repo.dirstate[f] != '?']

def instore(repo, hash, forcelocal=False):

    return os.path.exists(storepath(repo, hash, forcelocal))

def storepath(repo, hash, forcelocal=False):

    if not forcelocal and repo.shared():

        return repo.vfs.reljoin(repo.sharedpath, longname, hash)

    return repo.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)

    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).'''

    path = findfile(repo, hash)

    if path is None:

        return False

    util.makedirs(os.path.dirname(repo.wjoin(filename)))

    # The write may fail before the file is fully written, but we

    # don't use atomic writes in the working copy.

    shutil.copy(path, repo.wjoin(filename))

    return True

def copytostore(repo, rev, file, uploaded=False):

    hash = readstandin(repo, file, rev)

    if instore(repo, hash):

        return

    copytostoreabsolute(repo, repo.wjoin(file), hash)

def copyalltostore(repo, node):

    '''Copy all largefiles in a given revision to the store'''

    ctx = repo[node]

    for filename in ctx.files():

        if isstandin(filename) and filename in ctx.manifest():

            realfile = splitstandin(filename)

            copytostore(repo, ctx.node(), realfile)

def copytostoreabsolute(repo, file, hash):

    if inusercache(repo.ui, hash):

        link(usercachepath(repo.ui, hash), storepath(repo, hash))

    else:

        util.makedirs(os.path.dirname(storepath(repo, hash)))

        dst = util.atomictempfile(storepath(repo, hash),

                                  createmode=repo.store.createmode)

        for chunk in util.filechunkiter(open(file, 'rb')):

            dst.write(chunk)

        dst.close()

        linktousercache(repo, hash)

def linktousercache(repo, hash):

    path = usercachepath(repo.ui, hash)

    if path:

        link(storepath(repo, hash), path)

def getstandinmatcher(repo, rmatcher=None):

    '''Return a match object that applies rmatcher to the standin directory'''

    standindir = repo.wjoin(shortname)

    # no warnings about missing files or directories

    badfn = lambda f, msg: None

    if rmatcher and not rmatcher.always():

        pats = [os.path.join(standindir, pat) for pat in rmatcher.files()]

        if not pats:

            pats = [standindir]

        match = scmutil.match(repo[None], pats, badfn=badfn)

        # if pats is empty, it would incorrectly always match, so clear _always

        match._always = False

    else:

        # no patterns: relative to repo root

        match = scmutil.match(repo[None], [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:

    # 1) Some callers want an absolute path, but for instance addlargefiles

    #    needs it repo-relative so it can be passed to repo[None].add().  So

    #    leave it up to the caller to use repo.wjoin() to get an absolute path.

    # 2) Join with '/' because that's what dirstate always uses, even on

    #    Windows. Change existing separator to '/' first in case we are

    #    passed filenames from an external source (like the command line).

    return shortnameslash + util.pconvert(filename)

def isstandin(filename):

    '''Return true if filename is a big file standin. filename must be

    in Mercurial's internal form (slash-separated).'''

    return filename.startswith(shortnameslash)

def splitstandin(filename):

    # Split on / because that's what dirstate always uses, even on Windows.

    # Change local separator to / first just in case we are passed filenames

    # from an external source (like the command line).

    bits = util.pconvert(filename).split('/', 1)

    if len(bits) == 2 and bits[0] == shortname:

        return bits[1]

    else:

        return None

def updatestandin(repo, standin):

    file = repo.wjoin(splitstandin(standin))

    if os.path.exists(file):

        hash = hashfile(file)

        executable = getexecutable(file)

        writestandin(repo, standin, hash, executable)

def readstandin(repo, filename, node=None):

    '''read hex hash from standin for filename at given node, or working

    directory if no node is given'''

    return repo[node][standin(filename)].data().strip()

def writestandin(repo, standin, hash, executable):

    '''write hash to <repo.root>/<standin>'''

    repo.wwrite(standin, hash + '\n', executable and 'x' or '')

def copyandhash(instream, outfile):

    '''Read bytes from instream (iterable) and write them to outfile,

    computing the SHA-1 hash of the data along the way. Return the hash.'''

    hasher = util.sha1('')

    for data in instream:

        hasher.update(data)

        outfile.write(data)

    return hasher.hexdigest()

def hashrepofile(repo, file):

    return hashfile(repo.wjoin(file))

def hashfile(file):

    if not os.path.exists(file):

        return ''

    hasher = util.sha1('')

    fd = open(file, 'rb')

    for data in util.filechunkiter(fd, 128 * 1024):

        hasher.update(data)

    fd.close()

    return hasher.hexdigest()

def getexecutable(filename):

    mode = os.stat(filename).st_mode

    return ((mode & stat.S_IXUSR) and

            (mode & stat.S_IXGRP) and

            (mode & stat.S_IXOTH))

def urljoin(first, second, *arg):

    def join(left, right):

        if not left.endswith('/'):

            left += '/'

        if right.startswith('/'):

            right = right[1:]

        return left + right

    url = join(first, second)

    for a in arg:

        url = join(url, a)

    return url

def hexsha1(data):

    """hexsha1 returns the hex-encoded sha1 sum of the data in the file-like

    object data"""

    h = util.sha1()

    for chunk in util.filechunkiter(data):

        h.update(chunk)

    return h.hexdigest()

def httpsendfile(ui, filename):

    return httpconnection.httpsendfile(ui, filename, 'rb')

def unixpath(path):