self.revision = repository.revisions.index(revision)

        self.short_id = self.raw_id[:12]

        self.id = self.raw_id

        try:

            commit = self.repository._repo.get_object(self.raw_id)

        except KeyError:

            raise RepositoryError("Cannot get object with id %s" % self.raw_id)

        self._commit = commit

        self._tree_id = commit.tree

        try:

            self.message = safe_unicode(commit.message[:-1])

            # Always strip last eol

        except UnicodeDecodeError:

            self.message = commit.message[:-1].decode(commit.encoding

                or 'utf-8')

        #self.branch = None

        self.tags = []

        #tree = self.repository.get_object(self._tree_id)

        self.nodes = {}

        self._paths = {}

    @LazyProperty

    def author(self):

        return safe_unicode(self._commit.committer)

    @LazyProperty

    def date(self):

        return date_fromtimestamp(self._commit.commit_time,

                                  self._commit.commit_timezone)

    @LazyProperty

    def status(self):

        """

        Returns modified, added, removed, deleted files for current changeset

        """

        return self.changed, self.added, self.removed

    @LazyProperty

    def branch(self):

        heads = self.repository._heads(reverse=False)

        ref = heads.get(self.raw_id)

        if ref:

            return safe_unicode(ref)

    def _fix_path(self, path):

        """

        Paths are stored without trailing slash so we need to get rid off it if

        needed.

        """

        if path.endswith('/'):

            path = path.rstrip('/')

        return path

    def _get_id_for_path(self, path):

        # FIXME: Please, spare a couple of minutes and make those codes cleaner;

        if not path in self._paths:

            path = path.strip('/')

            # set root tree

            tree = self.repository._repo[self._commit.tree]

            if path == '':

                self._paths[''] = tree.id

                return tree.id

            splitted = path.split('/')

            dirs, name = splitted[:-1], splitted[-1]

            curdir = ''

            # initially extract things from root dir

            for item, stat, id in tree.iteritems():

                if curdir:

                    name = '/'.join((curdir, item))

                else:

                    name = item

                self._paths[name] = id

                self._stat_modes[name] = stat

            for dir in dirs:

                if curdir:

                    curdir = '/'.join((curdir, dir))

                else:

                    curdir = dir

                dir_id = None

                for item, stat, id in tree.iteritems():

                    if dir == item:

                        dir_id = id

                if dir_id:

                    # Update tree

                    tree = self.repository._repo[dir_id]

                    if not isinstance(tree, objects.Tree):

                        raise ChangesetError('%s is not a directory' % curdir)

                else:

                    raise ChangesetError('%s have not been found' % curdir)

                # cache all items from the given traversed tree

                for item, stat, id in tree.iteritems():

                    if curdir:

                        name = '/'.join((curdir, item))

                    else:

                        name = item

                    self._paths[name] = id

                    self._stat_modes[name] = stat

            if not path in self._paths:

                raise NodeDoesNotExistError("There is no file nor directory "

                    "at the given path %r at revision %r"

                    % (path, self.short_id))

        return self._paths[path]

    def _get_kind(self, path):

        id = self._get_id_for_path(path)

        obj = self.repository._repo[id]

        if isinstance(obj, objects.Blob):

            return NodeKind.FILE

        elif isinstance(obj, objects.Tree):

            return NodeKind.DIR

    def _get_file_nodes(self):

        return chain(*(t[2] for t in self.walk()))

    @LazyProperty

    def parents(self):

        """

        Returns list of parents changesets.

        """

        return [self.repository.get_changeset(parent)

            for parent in self._commit.parents]

    def next(self, branch=None):

        if branch and self.branch != branch:

            raise VCSError('Branch option used on changeset not belonging '

                           'to that branch')

        def _next(changeset, branch):

            try:

                next_ = changeset.revision + 1

                next_rev = changeset.repository.revisions[next_]

            except IndexError:

                raise ChangesetDoesNotExistError

            cs = changeset.repository.get_changeset(next_rev)

            if branch and branch != cs.branch:

                return _next(cs, branch)

            return cs

        return _next(self, branch)

    def prev(self, branch=None):

        if branch and self.branch != branch:

                                     "or Blob, is %r" % type(obj))

        nodes = dirnodes + filenodes

        for node in nodes:

            if not node.path in self.nodes:

                self.nodes[node.path] = node

        nodes.sort()

        return nodes

    def get_node(self, path):

        if isinstance(path, unicode):

            path = path.encode('utf-8')

        path = self._fix_path(path)

        if not path in self.nodes:

            try:

                id = self._get_id_for_path(path)

            except ChangesetError:

                raise NodeDoesNotExistError("Cannot find one of parents' "

                    "directories for a given path: %s" % path)

            obj = self.repository._repo.get_object(id)

            if isinstance(obj, objects.Tree):

                if path == '':

                    node = RootNode(changeset=self)

                else:

                    node = DirNode(path, changeset=self)

                node._tree = obj

            elif isinstance(obj, objects.Blob):

                node = FileNode(path, changeset=self)

                node._blob = obj

            else:

                raise NodeDoesNotExistError("There is no file nor directory "

                    "at the given path %r at revision %r"

                    % (path, self.short_id))

            # cache node

            self.nodes[path] = node

        return self.nodes[path]

    @LazyProperty

    def affected_files(self):

        """

        Get's a fast accessible file changes for given changeset

        """

        return self.added + self.changed

    @LazyProperty

    def _diff_name_status(self):

        output = []

        for parent in self.parents:

            so, se = self.repository.run_git_command(cmd)

            output.append(so.strip())

        return '\n'.join(output)

    def _get_paths_for_status(self, status):

        """

        Returns sorted list of paths for given ``status``.

        :param status: one of: *added*, *modified* or *deleted*

        """

        paths = set()

        char = status[0].upper()

        for line in self._diff_name_status.splitlines():

            if not line:

                continue

            if line.startswith(char):

                splitted = line.split(char,1)

                if not len(splitted) == 2:

                    raise VCSError("Couldn't parse diff result:\n%s\n\n and "

                        "particularly that line: %s" % (self._diff_name_status,

                        line))

        return sorted(paths)

    @LazyProperty

    def added(self):

        """

        Returns list of added ``FileNode`` objects.

        """

        if not self.parents:

            return list(self._get_file_nodes())

        return [self.get_node(path) for path in self._get_paths_for_status('added')]

    @LazyProperty

    def changed(self):

        """

        Returns list of modified ``FileNode`` objects.

        """

        if not self.parents:

            return []

        return [self.get_node(path) for path in self._get_paths_for_status('modified')]

    @LazyProperty

    def removed(self):

        """

        Returns list of removed ``FileNode`` objects.

        """

        if not self.parents:

            return []

        return [RemovedFileNode(path) for path in self._get_paths_for_status('deleted')]

import time

import datetime

import posixpath

from dulwich import objects

from dulwich.repo import Repo

from rhodecode.lib.vcs.backends.base import BaseInMemoryChangeset

from rhodecode.lib.vcs.exceptions import RepositoryError

class GitInMemoryChangeset(BaseInMemoryChangeset):

    def commit(self, message, author, parents=None, branch=None, date=None,

            **kwargs):

        """

        Performs in-memory commit (doesn't check workdir in any way) and

        returns newly created ``Changeset``. Updates repository's

        ``revisions``.

        :param message: message of the commit

        :param author: full username, i.e. "Joe Doe <joe.doe@example.com>"

        :param parents: single parent or sequence of parents from which commit

          would be derieved

        :param date: ``datetime.datetime`` instance. Defaults to

          ``datetime.datetime.now()``.

        :param branch: branch name, as string. If none given, default backend's

          branch would be used.

        :raises ``CommitError``: if any error occurs while committing

        """

        self.check_integrity(parents)

        from .repository import GitRepository

        if branch is None:

            branch = GitRepository.DEFAULT_BRANCH_NAME

        repo = self.repository._repo

        object_store = repo.object_store

        ENCODING = "UTF-8"

        DIRMOD = 040000

        # Create tree and populates it with blobs

        commit_tree = self.parents[0] and repo[self.parents[0]._commit.tree] or\

            objects.Tree()

        for node in self.added + self.changed:

            # Compute subdirs if needed

            dirpath, nodename = posixpath.split(node.path)

            dirnames = dirpath and dirpath.split('/') or []

            parent = commit_tree

            ancestors = [('', parent)]

            # Tries to dig for the deepest existing tree

            while dirnames:

                curdir = dirnames.pop(0)

                try:

                curtree[node_path] = node.mode, blob.id

                new_trees.append(curtree)

                for dirname in reversed_dirnames[:-1]:

                    newtree = objects.Tree()

                    #newtree.add(DIRMOD, dirname, curtree.id)

                    newtree[dirname] = DIRMOD, curtree.id

                    new_trees.append(newtree)

                    curtree = newtree

                parent[reversed_dirnames[-1]] = DIRMOD, curtree.id

            else:

                parent.add(node.mode, node_path, blob.id)

            new_trees.append(parent)

            # Update ancestors

            for parent, tree, path in reversed([(a[1], b[1], b[0]) for a, b in

                zip(ancestors, ancestors[1:])]):

                parent[path] = DIRMOD, tree.id

                object_store.add_object(tree)

            object_store.add_object(blob)

            for tree in new_trees:

                object_store.add_object(tree)

        for node in self.removed:

            paths = node.path.split('/')

            tree = commit_tree

            trees = [tree]

            # Traverse deep into the forest...

            for path in paths:

                try:

                    obj = self.repository._repo[tree[path][1]]

                    if isinstance(obj, objects.Tree):

                        trees.append(obj)

                        tree = obj

                except KeyError:

                    break

            # Cut down the blob and all rotten trees on the way back...

            for path, tree in reversed(zip(paths, trees)):

                del tree[path]

                if tree:

                    # This tree still has elements - don't remove it or any

                    # of it's parents

                    break

        object_store.add_object(commit_tree)

        # Create commit

        commit = objects.Commit()

        commit.tree = commit_tree.id

        commit.parents = [p._commit.id for p in self.parents if p]

        commit.encoding = ENCODING

        # Compute date

        if date is None:

            date = time.time()

        elif isinstance(date, datetime.datetime):

            date = time.mktime(date.timetuple())

        author_time = kwargs.pop('author_time', date)

        commit.commit_time = int(date)

        commit.author_time = int(author_time)

        tz = time.timezone

        author_tz = kwargs.pop('author_timezone', tz)

        commit.commit_timezone = tz

        commit.author_timezone = author_tz

        object_store.add_object(commit)

        ref = 'refs/heads/%s' % branch

        repo.refs[ref] = commit.id

        repo.refs.set_symbolic_ref('HEAD', ref)

        # Update vcs repository object & recreate dulwich repo

        self.repository.revisions.append(commit.id)

        self.repository._repo = Repo(self.repository.path)

        tip = self.repository.get_changeset()

        self.reset()

        return tip

    def _get_missing_trees(self, path, root_tree):

        """

        Creates missing ``Tree`` objects for the given path.

        :param path: path given as a string. It may be a path to a file node

          (i.e. ``foo/bar/baz.txt``) or directory path - in that case it must

          end with slash (i.e. ``foo/bar/``).

        :param root_tree: ``dulwich.objects.Tree`` object from which we start

          traversing (should be commit's root tree)

        """

        dirpath = posixpath.split(path)[0]

        dirs = dirpath.split('/')

        if not dirs or dirs == ['']:

            return []

        def get_tree_for_dir(tree, dirname):

            for name, mode, id in tree.iteritems():

                if name == dirname:

                    obj = self.repository._repo[id]

                    if isinstance(obj, objects.Tree):

from .inmemory import GitInMemoryChangeset

from .config import ConfigFile

class GitRepository(BaseRepository):

    """

    Git repository backend.

    """

    DEFAULT_BRANCH_NAME = 'master'

    scm = 'git'

    def __init__(self, repo_path, create=False, src_url=None,

                 update_after_clone=False, bare=False):

        self.path = abspath(repo_path)

        self._repo = self._get_repo(create, src_url, update_after_clone, bare)

        try:

            self.head = self._repo.head()

        except KeyError:

            self.head = None

        self._config_files = [

            bare and abspath(self.path, 'config') or abspath(self.path, '.git',

                'config'),

            abspath(get_user_home(), '.gitconfig'),

        ]

    @LazyProperty

    def revisions(self):

        """

        Returns list of revisions' ids, in ascending order.  Being lazy

        attribute allows external tools to inject shas from cache.

        """

        return self._get_all_revisions()

    def run_git_command(self, cmd):

        """

        Runs given ``cmd`` as git command and returns tuple

        (returncode, stdout, stderr).

        .. note::

           This method exists only until log/blame functionality is implemented

           at Dulwich (see https://bugs.launchpad.net/bugs/645142). Parsing

           os command's output is road to hell...

        :param cmd: git command to be executed

        """

        if isinstance(cmd, basestring):

        try:

            opts = dict(

                shell=isinstance(cmd, basestring),

                stdout=PIPE,

                stderr=PIPE)

            if os.path.isdir(self.path):

                opts['cwd'] = self.path

            p = Popen(cmd, **opts)

        except OSError, err:

            raise RepositoryError("Couldn't run git command (%s).\n"

                "Original error was:%s" % (cmd, err))

        so, se = p.communicate()

        if not se.startswith("fatal: bad default revision 'HEAD'") and \

            p.returncode != 0:

            raise RepositoryError("Couldn't run git command (%s).\n"

                "stderr:\n%s" % (cmd, se))

        return so, se

    def _check_url(self, url):

        """

        Functon will check given url and try to verify if it's a valid

        link. Sometimes it may happened that mercurial will issue basic

        auth request that can cause whole API to hang when used from python

        or other external calls.

        On failures it'll raise urllib2.HTTPError

        """

        #TODO: implement this

        pass

    def _get_repo(self, create, src_url=None, update_after_clone=False,

            bare=False):

        if create and os.path.exists(self.path):

            raise RepositoryError("Location already exist")

        if src_url and not create:

            raise RepositoryError("Create should be set to True if src_url is "

                                  "given (clone operation creates repository)")

        try:

            if create and src_url:

                self._check_url(src_url)

                self.clone(src_url, update_after_clone, bare)

                return Repo(self.path)

            elif create:

                os.mkdir(self.path)

                if bare:

                    return Repo.init_bare(self.path)

                else:

import datetime

import errno

from rhodecode.lib.vcs.backends.base import BaseInMemoryChangeset

from rhodecode.lib.vcs.exceptions import RepositoryError

class MercurialInMemoryChangeset(BaseInMemoryChangeset):

    def commit(self, message, author, parents=None, branch=None, date=None,

            **kwargs):

        """

        Performs in-memory commit (doesn't check workdir in any way) and

        returns newly created ``Changeset``. Updates repository's

        ``revisions``.

        :param message: message of the commit

        :param author: full username, i.e. "Joe Doe <joe.doe@example.com>"

        :param parents: single parent or sequence of parents from which commit

          would be derieved

        :param date: ``datetime.datetime`` instance. Defaults to

          ``datetime.datetime.now()``.

        :param branch: branch name, as string. If none given, default backend's

          branch would be used.

        :raises ``CommitError``: if any error occurs while committing

        """

        self.check_integrity(parents)

        from .repository import MercurialRepository

            raise RepositoryError('Given message and author needs to be '

        if branch is None:

            branch = MercurialRepository.DEFAULT_BRANCH_NAME

        kwargs['branch'] = branch

        def filectxfn(_repo, memctx, path):

            """

            Marks given path as added/changed/removed in a given _repo. This is

            for internal mercurial commit function.

            """

            # check if this path is removed

            if path in (node.path for node in self.removed):

                # Raising exception is a way to mark node for removal

                raise IOError(errno.ENOENT, '%s is deleted' % path)

            # check if this path is added

            for node in self.added:

                if node.path == path:

                    return memfilectx(path=node.path,

                        data=(node.content.encode('utf8')

                              if not node.is_binary else node.content),

                        islink=False,

                        isexec=node.is_executable,

                        copied=False)

            # or changed

            for node in self.changed:

                if node.path == path:

                    return memfilectx(path=node.path,

                        data=(node.content.encode('utf8')

                              if not node.is_binary else node.content),

                        islink=False,

                        isexec=node.is_executable,

                        copied=False)

            raise RepositoryError("Given path haven't been marked as added,"

                "changed or removed (%s)" % path)

        parents = [None, None]

        for i, parent in enumerate(self.parents):

            if parent is not None:

                parents[i] = parent._ctx.node()

        if date and isinstance(date, datetime.datetime):

            date = date.ctime()

        commit_ctx = memctx(repo=self.repository._repo,

            parents=parents,

            text='',

            files=self.get_paths(),

            filectxfn=filectxfn,

            user=author,

            date=date,

            extra=kwargs)

        # injecting given _repo params

        commit_ctx._date = date

        # TODO: Catch exceptions!

        n = self.repository._repo.commitctx(commit_ctx)

        # Returns mercurial node

        self._commit_ctx = commit_ctx  # For reference

        # Update vcs repository object & recreate mercurial _repo

        # new_ctx = self.repository._repo[node]

        # new_tip = self.repository.get_changeset(new_ctx.hex())

        new_id = hex(n)

        self.repository.revisions.append(new_id)

        self._repo = self.repository._get_repo(create=False)

        self.repository.branches = self.repository._get_branches()

        tip = self.repository.get_changeset()

        self.reset()

        return tip

class RemovedFileNodesGenerator(NodeGeneratorBase):

    """

    Class holding removed files for current changeset

    """

    def __iter__(self):

        for p in self.current_paths:

            yield RemovedFileNode(path=p)

    def __getslice__(self, i, j):

        for p in self.current_paths[i:j]:

            yield RemovedFileNode(path=p)

class Node(object):

    """

    Simplest class representing file or directory on repository.  SCM backends

    should use ``FileNode`` and ``DirNode`` subclasses rather than ``Node``

    directly.

    Node's ``path`` cannot start with slash as we operate on *relative* paths

    only. Moreover, every single node is identified by the ``path`` attribute,

    so it cannot end with slash, too. Otherwise, path could lead to mistakes.

    """

    def __init__(self, path, kind):

        if path.startswith('/'):

            raise NodeError("Cannot initialize Node objects with slash at "

                "the beginning as only relative paths are supported")

        self.path = path.rstrip('/')

        if path == '' and kind != NodeKind.DIR:

            raise NodeError("Only DirNode and its subclasses may be "

                            "initialized with empty path")

        self.kind = kind

        #self.dirs, self.files = [], []

        if self.is_root() and not self.is_dir():

            raise NodeError("Root node cannot be FILE kind")

    @LazyProperty

    def parent(self):

        parent_path = self.get_parent_path()

        if parent_path:

            if self.changeset:

                return self.changeset.get_node(parent_path)

            return DirNode(parent_path)

        return None

    @LazyProperty

    def name(self):

        """

        Returns name of the node so if its path

        then only last part is returned.

        """

        return safe_unicode(self.path.rstrip('/').split('/')[-1])

    def _get_kind(self):

        return self._kind

    def _set_kind(self, kind):

        if hasattr(self, '_kind'):

            raise NodeError("Cannot change node's kind")

        else:

            self._kind = kind

            # Post setter check (path's trailing slash)

            if self.path.endswith('/'):

                raise NodeError("Node's path cannot end with slash")

    kind = property(_get_kind, _set_kind)

    def __cmp__(self, other):

        """

        Comparator using name of the node, needed for quick list sorting.

        """

        kind_cmp = cmp(self.kind, other.kind)

        if kind_cmp:

            return kind_cmp

        return cmp(self.name, other.name)

    def __eq__(self, other):

        for attr in ['name', 'path', 'kind']:

            if getattr(self, attr) != getattr(other, attr):

                return False

        if self.is_file():

            if self.content != other.content:

                return False

        else:

            # For DirNode's check without entering each dir

            self_nodes_paths = list(sorted(n.path for n in self.nodes))

            other_nodes_paths = list(sorted(n.path for n in self.nodes))

            if self_nodes_paths != other_nodes_paths:

                return False

        return True

    def __nq__(self, other):

        return not self.__eq__(other)

from mercurial import archival, merge as hg_merge, patch, ui

from mercurial.commands import clone, nullid, pull

from mercurial.context import memctx, memfilectx

from mercurial.error import RepoError, RepoLookupError, Abort

from mercurial.hgweb.common import get_contact

from mercurial.localrepo import localrepository

from mercurial.match import match

from mercurial.mdiff import diffopts

from mercurial.node import hex

# -*- coding: utf-8 -*-

"""

    rhodecode.model.scm

    ~~~~~~~~~~~~~~~~~~~

    Scm model for RhodeCode

    :created_on: Apr 9, 2010

    :author: marcink

    :copyright: (C) 2010-2012 Marcin Kuzminski <marcin@python-works.com>

    :license: GPLv3, see COPYING for more details.

"""

# This program is free software: you can redistribute it and/or modify

# it under the terms of the GNU General Public License as published by

# the Free Software Foundation, either version 3 of the License, or

# (at your option) any later version.

#

# This program is distributed in the hope that it will be useful,

# but WITHOUT ANY WARRANTY; without even the implied warranty of

# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the

# GNU General Public License for more details.

#

# You should have received a copy of the GNU General Public License

# along with this program.  If not, see <http://www.gnu.org/licenses/>.

import os

import time

import traceback

import logging

import cStringIO

from rhodecode.lib.vcs import get_backend

from rhodecode.lib.vcs.exceptions import RepositoryError

from rhodecode.lib.vcs.utils.lazy import LazyProperty

from rhodecode.lib.vcs.nodes import FileNode

from rhodecode import BACKENDS

from rhodecode.lib import helpers as h

from rhodecode.lib.auth import HasRepoPermissionAny, HasReposGroupPermissionAny

from rhodecode.lib.utils import get_repos as get_filesystem_repos, make_ui, \

    action_logger, EmptyChangeset, REMOVED_REPO_PAT

from rhodecode.model import BaseModel

from rhodecode.model.db import Repository, RhodeCodeUi, CacheInvalidation, \

    UserFollowing, UserLog, User, RepoGroup

log = logging.getLogger(__name__)

class UserTemp(object):

    def __init__(self, user_id):

        self.user_id = user_id

    def __repr__(self):

        return "<%s('id:%s')>" % (self.__class__.__name__, self.user_id)

class RepoTemp(object):

    def __init__(self, repo_id):

        self.repo_id = repo_id

    def __repr__(self):

        return "<%s('id:%s')>" % (self.__class__.__name__, self.repo_id)

class CachedRepoList(object):

    def __init__(self, db_repo_list, repos_path, order_by=None):

        self.db_repo_list = db_repo_list

        self.repos_path = repos_path

        self.order_by = order_by

        self.reversed = (order_by or '').startswith('-')

    def __len__(self):

        return len(self.db_repo_list)

    def __repr__(self):

        return '<%s (%s)>' % (self.__class__.__name__, self.__len__())

    def __iter__(self):

        for dbr in self.db_repo_list:

            scmr = dbr.scm_instance_cached

            # check permission at this level

            if not HasRepoPermissionAny(

                'repository.read', 'repository.write', 'repository.admin'

            )(dbr.repo_name, 'get repo check'):

                continue

    def get_forks(self, repo_id):

        if not isinstance(repo_id, int):

            repo_id = getattr(Repository.get_by_repo_name(repo_id), 'repo_id')