def is_valid_repo_group(repo_group_name, base_path, skip_path_check=False):

    """

    Returns True if given path is a repository group False otherwise

    :param repo_name:

    :param base_path:

    """

    full_path = os.path.join(safe_str(base_path), safe_str(repo_group_name))

    # check if it's not a repo

    if is_valid_repo(repo_group_name, base_path):

        return False

    try:

        # we need to check bare git repos at higher level

        # since we might match branches/hooks/info/objects or possible

        # other things inside bare git repo

        get_scm(os.path.dirname(full_path))

        return False

    except VCSError:

        pass

    # check if it's a valid path

    if skip_path_check or os.path.isdir(full_path):

        return True

    return False

# propagated from mercurial documentation

ui_sections = ['alias', 'auth',

                'decode/encode', 'defaults',

                'diff', 'email',

                'extensions', 'format',

                'merge-patterns', 'merge-tools',

                'hooks', 'http_proxy',

                'smtp', 'patch',

                'paths', 'profiling',

                'server', 'trusted',

                'ui', 'web', ]

def make_ui(repo_path=None):

    """

    Create an Mercurial 'ui' object based on database Ui settings, possibly

    augmenting with content from a hgrc file.

    """

    baseui = ui.ui()

    # clean the baseui object

    baseui._ocfg = config.config()

    baseui._ucfg = config.config()

    baseui._tcfg = config.config()

    sa = meta.Session()

    for ui_ in sa.query(Ui).all():

        if ui_.ui_active:

            ui_val = '' if ui_.ui_value is None else safe_str(ui_.ui_value)

            log.debug('config from db: [%s] %s=%r', ui_.ui_section,

                      ui_.ui_key, ui_val)

            baseui.setconfig(safe_str(ui_.ui_section), safe_str(ui_.ui_key),

                             ui_val)

    # force set push_ssl requirement to False, Kallithea handles that

    baseui.setconfig('web', 'push_ssl', False)

    baseui.setconfig('web', 'allow_push', '*')

    # prevent interactive questions for ssh password / passphrase

    ssh = baseui.config('ui', 'ssh', default='ssh')

    baseui.setconfig('ui', 'ssh', '%s -oBatchMode=yes -oIdentitiesOnly=yes' % ssh)

    # push / pull hooks

    baseui.setconfig('hooks', 'changegroup.kallithea_log_push_action', 'python:kallithea.lib.hooks.log_push_action')

    baseui.setconfig('hooks', 'outgoing.kallithea_log_pull_action', 'python:kallithea.lib.hooks.log_pull_action')

    if repo_path is not None:

        hgrc_path = os.path.join(repo_path, '.hg', 'hgrc')

        if os.path.isfile(hgrc_path):

            log.debug('reading hgrc from %s', hgrc_path)

            cfg = config.config()

            cfg.read(hgrc_path)

            for section in ui_sections:

                for k, v in cfg.items(section):

                    log.debug('config from file: [%s] %s=%s', section, k, v)

                    baseui.setconfig(safe_str(section), safe_str(k), safe_str(v))

        else:

            log.debug('hgrc file is not present at %s, skipping...', hgrc_path)

    return baseui

def set_app_settings(config):

    """

    Updates app config with new settings from database

    :param config:

    """

    hgsettings = Setting.get_app_settings()

    for k, v in hgsettings.items():

        config[k] = v

def set_vcs_config(config):

    """

    Patch VCS config with some Kallithea specific stuff

    :param config: kallithea.CONFIG

    """

    settings.BACKENDS = {

        'hg': 'kallithea.lib.vcs.backends.hg.MercurialRepository',

        'git': 'kallithea.lib.vcs.backends.git.GitRepository',

    }

    settings.GIT_EXECUTABLE_PATH = config.get('git_path', 'git')

    settings.GIT_REV_FILTER = config.get('git_rev_filter', '--all').strip()

    settings.DEFAULT_ENCODINGS = aslist(config.get('default_encoding',

                                                        'utf-8'), sep=',')

def set_indexer_config(config):

    """

    Update Whoosh index mapping

    :param config: kallithea.CONFIG

    """

    log.debug('adding extra into INDEX_EXTENSIONS')

    kallithea.config.conf.INDEX_EXTENSIONS.extend(re.split(r'\s+', config.get('index.extensions', '')))

    log.debug('adding extra into INDEX_FILENAMES')

    kallithea.config.conf.INDEX_FILENAMES.extend(re.split(r'\s+', config.get('index.filenames', '')))

def map_groups(path):

    """

    Given a full path to a repository, create all nested groups that this

    repo is inside. This function creates parent-child relationships between

    groups and creates default perms for all new groups.

    :param paths: full path to repository

    """

    from kallithea.model.repo_group import RepoGroupModel

    sa = meta.Session()

    groups = path.split(Repository.url_sep())

    parent = None

    group = None

    # last element is repo in nested groups structure

    groups = groups[:-1]

    rgm = RepoGroupModel()

    owner = User.get_first_admin()

    for lvl, group_name in enumerate(groups):

        group_name = u'/'.join(groups[:lvl] + [group_name])

        group = RepoGroup.get_by_group_name(group_name)

        desc = '%s group' % group_name

        # skip folders that are now removed repos

        if REMOVED_REPO_PAT.match(group_name):

            break

        if group is None:

            log.debug('creating group level: %s group_name: %s',

                      lvl, group_name)

            group = RepoGroup(group_name, parent)

            group.group_description = desc

            group.owner = owner

            sa.add(group)

            rgm._create_default_perms(group)

            sa.flush()

        parent = group

    return group

def repo2db_mapper(initial_repo_list, remove_obsolete=False,

                   install_git_hooks=False, user=None, overwrite_git_hooks=False):

    """

    maps all repos given in initial_repo_list, non existing repositories

    are created, if remove_obsolete is True it also check for db entries

    that are not in initial_repo_list and removes them.

    :param initial_repo_list: list of repositories found by scanning methods

    :param remove_obsolete: check for obsolete entries in database

    :param install_git_hooks: if this is True, also check and install git hook

        for a repo if missing

    :param overwrite_git_hooks: if this is True, overwrite any existing git hooks

        that may be encountered (even if user-deployed)

    """

    from kallithea.model.repo import RepoModel

    from kallithea.model.scm import ScmModel

    sa = meta.Session()

    repo_model = RepoModel()

    if user is None:

        user = User.get_first_admin()

    added = []

    if _str in (True, False):

        return _str

    _str = str(_str).strip().lower()

    return _str in ('t', 'true', 'y', 'yes', 'on', '1')

def aslist(obj, sep=None, strip=True):

    """

    Returns given string separated by sep as list

    :param obj:

    :param sep:

    :param strip:

    """

    if isinstance(obj, (basestring)):

        lst = obj.split(sep)

        if strip:

            lst = [v.strip() for v in lst]

        return lst

    elif isinstance(obj, (list, tuple)):

        return obj

    elif obj is None:

        return []

    else:

        return [obj]

def convert_line_endings(line, mode):

    """

    Converts a given line  "line end" according to given mode

    Available modes are::

        0 - Unix

        1 - Mac

        2 - DOS

    :param line: given line to convert

    :param mode: mode to convert to

    :rtype: str

    :return: converted line according to mode

    """

    from string import replace

    if mode == 0:

        line = replace(line, '\r\n', '\n')

        line = replace(line, '\r', '\n')

    elif mode == 1:

        line = replace(line, '\r\n', '\r')

        line = replace(line, '\n', '\r')

    elif mode == 2:

        line = re.sub("\r(?!\n)|(?<!\r)\n", "\r\n", line)

    return line

def detect_mode(line, default):

    """

    Detects line break for given line, if line break couldn't be found

    given default value is returned

    :param line: str line

    :param default: default

    :rtype: int

    :return: value of line end on of 0 - Unix, 1 - Mac, 2 - DOS

    """

    if line.endswith('\r\n'):

        return 2

    elif line.endswith('\n'):

        return 0

    elif line.endswith('\r'):

        return 1

    else:

        return default

def generate_api_key():

    """

    Generates a random (presumably unique) API key.

    This value is used in URLs and "Bearer" HTTP Authorization headers,

    which in practice means it should only contain URL-safe characters

    (RFC 3986):

        unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~"

    """

    # Hexadecimal certainly qualifies as URL-safe.

    return binascii.hexlify(os.urandom(20))

def safe_int(val, default=None):

    """

    Returns int() of val if val is not convertable to int use default

    instead

    :param val:

    :param default:

    """

    try:

        val = int(val)

    except (ValueError, TypeError):

        val = default

    return val

def remove_suffix(s, suffix):

    if s.endswith(suffix):

        s = s[:-1 * len(suffix)]

    return s

def remove_prefix(s, prefix):

    if s.startswith(prefix):

        s = s[len(prefix):]

    return s

def age(prevdate, show_short_version=False, now=None):

    """

    turns a datetime into an age string.

    If show_short_version is True, then it will generate a not so accurate but shorter string,

    example: 2days ago, instead of 2 days and 23 hours ago.

    :param prevdate: datetime object

    :param show_short_version: if it should approximate the date and return a shorter string

    :rtype: unicode

    :returns: unicode words describing age

    """

    now = now or datetime.datetime.now()

    order = ['year', 'month', 'day', 'hour', 'minute', 'second']

    deltas = {}

    future = False

    if prevdate > now:

        now, prevdate = prevdate, now

        future = True

    if future:

        prevdate = prevdate.replace(microsecond=0)

    # Get date parts deltas

    from dateutil import relativedelta

    for part in order:

        d = relativedelta.relativedelta(now, prevdate)

        deltas[part] = getattr(d, part + 's')

    # Fix negative offsets (there is 1 second between 10:59:59 and 11:00:00,

    # not 1 hour, -59 minutes and -59 seconds)

    for num, length in [(5, 60), (4, 60), (3, 24)]:  # seconds, minutes, hours

        part = order[num]

        carry_part = order[num - 1]

        if deltas[part] < 0:

            deltas[part] += length

            deltas[carry_part] -= 1

    # Same thing for days except that the increment depends on the (variable)

    # number of days in the month

    month_lengths = [31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]

    if deltas['day'] < 0:

        if prevdate.month == 2 and (prevdate.year % 4 == 0 and

            (prevdate.year % 100 != 0 or prevdate.year % 400 == 0)

        ):

            deltas['day'] += 29

        else:

            deltas['day'] += month_lengths[prevdate.month - 1]

        deltas['month'] -= 1

    if deltas['month'] < 0:

        deltas['month'] += 12

        deltas['year'] -= 1

    # In short version, we want nicer handling of ages of more than a year

    if show_short_version:

        if deltas['year'] == 1:

            # ages between 1 and 2 years: show as months

            deltas['month'] += 12

            deltas['year'] = 0

        if deltas['year'] >= 2:

            # ages 2+ years: round

            if deltas['month'] > 6:

                deltas['year'] += 1

                deltas['month'] = 0

    # Format the result

    fmt_funcs = {

        'year': lambda d: ungettext(u'%d year', '%d years', d) % d,

        'month': lambda d: ungettext(u'%d month', '%d months', d) % d,

        'day': lambda d: ungettext(u'%d day', '%d days', d) % d,

        'hour': lambda d: ungettext(u'%d hour', '%d hours', d) % d,

        'minute': lambda d: ungettext(u'%d minute', '%d minutes', d) % d,

        'second': lambda d: ungettext(u'%d second', '%d seconds', d) % d,

    }

    for i, part in enumerate(order):

        value = deltas[part]

        if value == 0:

            continue

        gitenv = os.environ

        # need to clean fix GIT_DIR !

        if 'GIT_DIR' in gitenv:

            del gitenv['GIT_DIR']

        gitenv['GIT_CONFIG_NOGLOBAL'] = '1'

        _git_path = settings.GIT_EXECUTABLE_PATH

        cmd = [_git_path] + _copts + cmd

        try:

            _opts = dict(

                env=gitenv,

                shell=False,

            )

            _opts.update(opts)

            p = subprocessio.SubprocessIOChunker(cmd, **_opts)

        except (EnvironmentError, OSError) as err:

            tb_err = ("Couldn't run git command (%s).\n"

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

            log.error(tb_err)

            if safe_call:

                return '', err

            else:

                raise RepositoryError(tb_err)

        try:

            return ''.join(p.output), ''.join(p.error)

        finally:

            p.close()

    def run_git_command(self, cmd):

        opts = {}

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

            opts['cwd'] = self.path

        return self._run_git_command(cmd, **opts)

    @classmethod

    def _check_url(cls, url):

        """

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

        link. Sometimes it may happened that git 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, exception is also thrown

        when the return code is non 200

        """

        # check first if it's not an local url

        if os.path.isdir(url) or url.startswith('file:'):

            return True

        if url.startswith('git://'):

            return True

        if '+' in url[:url.find('://')]:

            url = url[url.find('+') + 1:]

        handlers = []

        url_obj = hg_url(url)

        test_uri, authinfo = url_obj.authinfo()

        url_obj.passwd = '*****'

        cleaned_uri = str(url_obj)

        if not test_uri.endswith('info/refs'):

            test_uri = test_uri.rstrip('/') + '/info/refs'

        if authinfo:

            # create a password manager

            passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()

            passmgr.add_password(*authinfo)

            handlers.extend((httpbasicauthhandler(passmgr),

                             httpdigestauthhandler(passmgr)))

        o = urllib2.build_opener(*handlers)

        o.addheaders = [('User-Agent', 'git/1.7.8.0')]  # fake some git

        q = {"service": 'git-upload-pack'}

        qs = '?%s' % urllib.urlencode(q)

        cu = "%s%s" % (test_uri, qs)

        req = urllib2.Request(cu, None, {})

        try:

            resp = o.open(req)

            if resp.code != 200:

                raise Exception('Return Code is not 200')

        except Exception as e:

            # means it cannot be cloned

            raise urllib2.URLError("[%s] org_exc: %s" % (cleaned_uri, e))

        # now detect if it's proper git repo

        gitdata = resp.read()

        if 'service=git-upload-pack' not in gitdata:

            raise urllib2.URLError(

        return True

    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:

                GitRepository._check_url(src_url)

                self.clone(src_url, update_after_clone, bare)

                return Repo(self.path)

            elif create:

                os.makedirs(self.path)

                if bare:

                    return Repo.init_bare(self.path)

                else:

                    return Repo.init(self.path)

            else:

                return Repo(self.path)

        except (NotGitRepository, OSError) as err:

            raise RepositoryError(err)

    def _get_all_revisions(self):

        # we must check if this repo is not empty, since later command

        # fails if it is. And it's cheaper to ask than throw the subprocess

        # errors

        try:

            self._repo.head()

        except KeyError:

            return []

        rev_filter = settings.GIT_REV_FILTER

        cmd = ['rev-list', rev_filter, '--reverse', '--date-order']

        try:

            so, se = self.run_git_command(cmd)

        except RepositoryError:

            # Can be raised for empty repositories

            return []

        return so.splitlines()

    def _get_all_revisions2(self):

        # alternate implementation using dulwich

        includes = [x[1][0] for x in self._parsed_refs.iteritems()

                    if x[1][1] != 'T']

        return [c.commit.id for c in self._repo.get_walker(include=includes)]

    def _get_revision(self, revision):

        """

        For git backend we always return integer here. This way we ensure

        that changeset's revision attribute would become integer.

        """

        is_null = lambda o: len(o) == revision.count('0')

        if self._empty:

            raise EmptyRepositoryError("There are no changesets yet")

        if revision in (None, '', 'tip', 'HEAD', 'head', -1):

            return self.revisions[-1]

        is_bstr = isinstance(revision, (str, unicode))

        if ((is_bstr and revision.isdigit() and len(revision) < 12)

            or isinstance(revision, int) or is_null(revision)

        ):

            try:

                revision = self.revisions[int(revision)]

            except IndexError:

                msg = ("Revision %s does not exist for %s" % (revision, self))

                raise ChangesetDoesNotExistError(msg)

        elif is_bstr:

            # get by branch/tag name

            _ref_revision = self._parsed_refs.get(revision)

            if _ref_revision:  # and _ref_revision[1] in ['H', 'RH', 'T']:

                return _ref_revision[0]

            _tags_shas = self.tags.values()

            # maybe it's a tag ? we don't have them in self.revisions

            if revision in _tags_shas:

                return _tags_shas[_tags_shas.index(revision)]

            elif not SHA_PATTERN.match(revision) or revision not in self.revisions:

                msg = ("Revision %s does not exist for %s" % (revision, self))

                raise ChangesetDoesNotExistError(msg)

        # Ensure we return full id

        if not SHA_PATTERN.match(str(revision)):

            raise ChangesetDoesNotExistError("Given revision %s not recognized"

                % revision)

        return revision

    def get_ref_revision(self, ref_type, ref_name):

    In a way, this is a "communicate()" replacement with a twist.

    - We are multithreaded. Writing in and reading out, err are all sep threads.

    - We support concurrent (in and out) stream processing.

    - The output is not a stream. It's a queue of read string (bytes, not unicode)

      chunks. The object behaves as an iterable. You can "for chunk in obj:" us.

    - We are non-blocking in more respects than communicate()

      (reading from subprocess out pauses when internal buffer is full, but

       does not block the parent calling code. On the flip side, reading from

       slow-yielding subprocess may block the iteration until data shows up. This

       does not block the parallel inpipe reading occurring parallel thread.)

    The purpose of the object is to allow us to wrap subprocess interactions into

    an iterable that can be passed to a WSGI server as the application's return

    value. Because of stream-processing-ability, WSGI does not have to read ALL

    of the subprocess's output and buffer it, before handing it to WSGI server for

    HTTP response. Instead, the class initializer reads just a bit of the stream

    to figure out if error occurred or likely to occur and if not, just hands the

    further iteration over subprocess output to the server for completion of HTTP

    response.

    The real or perceived subprocess error is trapped and raised as one of

    EnvironmentError family of exceptions

    Example usage:

    #    try:

    #        answer = SubprocessIOChunker(

    #            cmd,

    #            input,

    #            buffer_size = 65536,

    #            chunk_size = 4096

    #            )

    #    except (EnvironmentError) as e:

    #        print str(e)

    #        raise e

    #

    #    return answer

    """

    def __init__(self, cmd, inputstream=None, buffer_size=65536,

                 chunk_size=4096, starting_values=None, **kwargs):

        """

        Initializes SubprocessIOChunker

        :param cmd: A Subprocess.Popen style "cmd". Can be string or array of strings

        :param inputstream: (Default: None) A file-like, string, or file pointer.

        :param buffer_size: (Default: 65536) A size of total buffer per stream in bytes.

        :param chunk_size: (Default: 4096) A max size of a chunk. Actual chunk may be smaller.

        :param starting_values: (Default: []) An array of strings to put in front of output que.

        """

        starting_values = starting_values or []

        if inputstream:

            input_streamer = StreamFeeder(inputstream)

            input_streamer.start()

            inputstream = input_streamer.output

        # Note: fragile cmd mangling has been removed for use in Kallithea

        assert isinstance(cmd, list), cmd

        _p = subprocess.Popen(cmd, bufsize=-1,

                              stdin=inputstream,

                              stdout=subprocess.PIPE,

                              stderr=subprocess.PIPE,

                              **kwargs)

        bg_out = BufferedGenerator(_p.stdout, buffer_size, chunk_size,

                                   starting_values)

        bg_err = BufferedGenerator(_p.stderr, 16000, 1, bottomless=True)

        while not bg_out.done_reading and not bg_out.reading_paused:

            # doing this until we reach either end of file, or end of buffer.

            bg_out.data_added_event.wait(1)

            bg_out.data_added_event.clear()

        # at this point it's still ambiguous if we are done reading or just full buffer.

        # Either way, if error (returned by ended process, or implied based on

        # presence of stuff in stderr output) we error out.

        # Else, we are happy.

        returncode = _p.poll()

        if (returncode is not None # process has terminated

            and returncode != 0

        ): # and it failed

            bg_out.stop()

            out = ''.join(bg_out)

            bg_err.stop()

            err = ''.join(bg_err)

            if (err.strip() == 'fatal: The remote end hung up unexpectedly' and

                out.startswith('0034shallow ')

            ):

                # hack inspired by https://github.com/schacon/grack/pull/7

                bg_out = iter([out])

                _p = None

            elif err:

            else:

                raise EnvironmentError(

                    "Subprocess exited with non 0 ret code: %s" % returncode)

        self.process = _p

        self.output = bg_out

        self.error = bg_err

        self.inputstream = inputstream

    def __iter__(self):

        return self

    def next(self):

        if self.process:

            returncode = self.process.poll()

            if (returncode is not None # process has terminated

                and returncode != 0

            ): # and it failed

                self.output.stop()

                self.error.stop()

                err = ''.join(self.error)

                raise EnvironmentError("Subprocess exited due to an error:\n" + err)

        return self.output.next()

    def throw(self, type, value=None, traceback=None):

        if self.output.length or not self.output.done_reading:

            raise type(value)

    def close(self):

        try:

            self.process.terminate()

        except:

            pass

        try:

            self.output.close()

        except:

            pass

        try:

            self.error.close()

        except:

            pass

        try:

            os.close(self.inputstream)

        except:

            pass

"""

This module provides some useful tools for ``vcs`` like annotate/diff html

output. It also includes some internal helpers.

"""

import datetime

import re

import time

def makedate():

    lt = time.localtime()

    if lt[8] == 1 and time.daylight:

        tz = time.altzone

    else:

        tz = time.timezone

    return time.mktime(lt), tz

def aslist(obj, sep=None, strip=True):

    """

    Returns given string separated by sep as list

    :param obj:

    :param sep:

    :param strip:

    """

        lst = obj.split(sep)

        if strip:

            lst = [v.strip() for v in lst]

        return lst

    elif isinstance(obj, (list, tuple)):

        return obj

    elif obj is None:

        return []

    else:

        return [obj]

def date_fromtimestamp(unixts, tzoffset=0):

    """

    Makes a local datetime object out of unix timestamp

    :param unixts:

    :param tzoffset:

    """

    return datetime.datetime.fromtimestamp(float(unixts))

def safe_int(val, default=None):

    """

    Returns int() of val if val is not convertible to int use default

    instead

    :param val:

    :param default:

    """

    try:

        val = int(val)

    except (ValueError, TypeError):

        val = default

    return val

def safe_unicode(s):

    """

    Safe unicode function. Use a few tricks to turn s into unicode string:

    In case of UnicodeDecodeError with configured default encodings, try to

    detect encoding with chardet library, then fall back to first encoding with

    errors replaced.

    """

    if isinstance(s, unicode):

        return s

    if not isinstance(s, str):  # use __str__ / __unicode__ and don't expect UnicodeDecodeError

        return unicode(s)

    from kallithea.lib.vcs.conf import settings

    for enc in settings.DEFAULT_ENCODINGS:

        try:

            return unicode(s, enc)

        except UnicodeDecodeError:

            pass

    try:

        import chardet

        encoding = chardet.detect(s)['encoding']

        if encoding is not None:

            return s.decode(encoding)

    except (ImportError, UnicodeDecodeError):

        pass

    return unicode(s, settings.DEFAULT_ENCODINGS[0], 'replace')

def safe_str(s):

    """

    Safe str function. Use a few tricks to turn s into bytes string:

    In case of UnicodeEncodeError with configured default encodings, fall back

    to first configured encoding with errors replaced.

    """

    if isinstance(s, str):

        return s

    assert isinstance(s, unicode), s  # don't use safe_str to coerce non-strings

    from kallithea.lib.vcs.conf import settings

    for enc in settings.DEFAULT_ENCODINGS:

        try:

            return s.encode(enc)

        except UnicodeEncodeError:

            pass

    return s.encode(settings.DEFAULT_ENCODINGS[0], 'replace')

# Regex taken from http://www.regular-expressions.info/email.html

email_re = re.compile(

    r"""[a-z0-9!#$%&'*+/=?^_`{|}~-]+(?:\.[a-z0-9!#$%&'*+/=?^_`{|}~-]+)*@"""

    r"""(?:[a-z0-9](?:[a-z0-9-]*[a-z0-9])?\.)+[a-z0-9](?:[a-z0-9-]*[a-z0-9])?""",