"""cubicweb ldap user source

this source is for now limited to a read-only EUser source

:organization: Logilab

:copyright: 2003-2008 LOGILAB S.A. (Paris, FRANCE), all rights reserved.

:contact: http://www.logilab.fr/ -- mailto:contact@logilab.fr

Part of the code is coming form Zope's LDAPUserFolder

Copyright (c) 2004 Jens Vagelpohl.

All Rights Reserved.

This software is subject to the provisions of the Zope Public License,

Version 2.1 (ZPL).  A copy of the ZPL should accompany this distribution.

THIS SOFTWARE IS PROVIDED "AS IS" AND ANY AND ALL EXPRESS OR IMPLIED

WARRANTIES ARE DISCLAIMED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED

WARRANTIES OF TITLE, MERCHANTABILITY, AGAINST INFRINGEMENT, AND FITNESS

FOR A PARTICULAR PURPOSE.

"""

from mx.DateTime import now, DateTimeDelta

from logilab.common.textutils import get_csv

from rql.nodes import Relation, VariableRef, Constant, Function

import ldap

from ldap.ldapobject import ReconnectLDAPObject

from ldap.filter import filter_format, escape_filter_chars

from ldapurl import LDAPUrl

from cubicweb.common import AuthenticationError, UnknownEid, RepositoryError

from cubicweb.server.sources import AbstractSource, TrFunc, GlobTrFunc, ConnectionWrapper

from cubicweb.server.utils import cartesian_product

# search scopes

BASE = ldap.SCOPE_BASE

ONELEVEL = ldap.SCOPE_ONELEVEL

SUBTREE = ldap.SCOPE_SUBTREE

# XXX only for edition ??

## password encryption possibilities

#ENCRYPTIONS = ('SHA', 'CRYPT', 'MD5', 'CLEAR') # , 'SSHA'

# mode identifier : (port, protocol)

MODES = {

    0: (389, 'ldap'),

    1: (636, 'ldaps'),

    2: (0,   'ldapi'),

    }

class TimedCache(dict):

    def __init__(self, ttlm, ttls=0):

        # time to live in minutes

        self.ttl = DateTimeDelta(0, 0, ttlm, ttls)

    def __setitem__(self, key, value):

        dict.__setitem__(self, key, (now(), value))

    def __getitem__(self, key):

        return dict.__getitem__(self, key)[1]

    def clear_expired(self):

        now_ = now()

        ttl = self.ttl

        for key, (timestamp, value) in self.items():

            if now_ - timestamp > ttl:

                del self[key]

class LDAPUserSource(AbstractSource):

    """LDAP read-only EUser source"""

    support_entities = {'EUser': False} 

    port = None

    cnx_mode = 0

    cnx_dn = ''

    cnx_pwd = ''

    options = (

        ('host',

         {'type' : 'string',

          'default': 'ldap',

          'help': 'ldap host',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('user-base-dn',

         {'type' : 'string',

          'default': 'ou=People,dc=logilab,dc=fr',

          'help': 'base DN to lookup for users',

          'group': 'ldap-source', 'inputlevel': 0,

          }),

        ('user-scope',

         {'type' : 'choice',

          'default': 'ONELEVEL',

          'choices': ('BASE', 'ONELEVEL', 'SUBTREE'),

          'help': 'user search scope',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('user-classes',

         {'type' : 'csv',

          'default': ('top', 'posixAccount'),

          'help': 'classes of user',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('user-login-attr',

         {'type' : 'string',

          'default': 'uid',

          'help': 'attribute used as login on authentication',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('user-default-group',

         {'type' : 'csv',

          'default': ('users',),

          'help': 'name of a group in which ldap users will be by default. \

You can set multiple groups by separating them by a comma.',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('user-attrs-map',

         {'type' : 'named',

          'default': {'uid': 'login', 'gecos': 'email'},

          'help': 'map from ldap user attributes to cubicweb attributes',

          'group': 'ldap-source', 'inputlevel': 1,

          }),

        ('synchronization-interval',

         {'type' : 'int',

          'default': 24*60*60,

          'help': 'interval between synchronization with the ldap \

directory (default to once a day).',

          'group': 'ldap-source', 'inputlevel': 2,

          }),

        ('cache-life-time',

         {'type' : 'int',

          'default': 2*60,

          'help': 'life time of query cache in minutes (default to two hours).',

          'group': 'ldap-source', 'inputlevel': 2,

          }),

    )

    def __init__(self, repo, appschema, source_config, *args, **kwargs):

        AbstractSource.__init__(self, repo, appschema, source_config,

                                *args, **kwargs)

        self.host = source_config['host']

        self.user_base_dn = source_config['user-base-dn']

        self.user_base_scope = globals()[source_config['user-scope']]

        self.user_classes = get_csv(source_config['user-classes'])

        self.user_login_attr = source_config['user-login-attr']

        self.user_default_groups = get_csv(source_config['user-default-group'])

        self.user_attrs = dict(v.split(':', 1) for v in get_csv(source_config['user-attrs-map']))

        self.user_rev_attrs = {'eid': 'dn'}

        for ldapattr, cwattr in self.user_attrs.items():

            self.user_rev_attrs[cwattr] = ldapattr

        self.base_filters = [filter_format('(%s=%s)', ('objectClass', o))

                              for o in self.user_classes]

        self._conn = None

        self._cache = {}

        ttlm = int(source_config.get('cache-life-type', 2*60))

        self._query_cache = TimedCache(ttlm)

        self._interval = int(source_config.get('synchronization-interval',

                                               24*60*60))

    def reset_caches(self):

        """method called during test to reset potential source caches"""

        self._query_cache = TimedCache(2*60)

    def init(self):

        """method called by the repository once ready to handle request"""

        self.repo.looping_task(self._interval, self.synchronize) 

        self.repo.looping_task(self._query_cache.ttl.seconds/10, self._query_cache.clear_expired) 

    def synchronize(self):

        """synchronize content known by this repository with content in the

        external repository

        """

        self.info('synchronizing ldap source %s', self.uri)

        try:

            ldap_emailattr = self.user_rev_attrs['email']

        except KeyError:

            return # no email in ldap, we're done

        session = self.repo.internal_session()

        try:

            cursor = session.system_sql("SELECT eid, extid FROM entities WHERE "

                                        "source='%s'" % self.uri)

            for eid, extid in cursor.fetchall():

                # if no result found, _search automatically delete entity information

                res = self._search(session, extid, BASE)

                if res: 

                    ldapemailaddr = res[0].get(ldap_emailattr)

                    if ldapemailaddr:

                        rset = session.execute('EmailAddress X,A WHERE '

                                               'U use_email X, U eid %(u)s',

                                               {'u': eid})

                        ldapemailaddr = unicode(ldapemailaddr)

                        for emaileid, emailaddr in rset:

                            if emailaddr == ldapemailaddr:

                                break

                        else:

                            self.info('updating email address of user %s to %s',

                                      extid, ldapemailaddr)

                            if rset:

                                session.execute('SET X address %(addr)s WHERE '

                                                'U primary_email X, U eid %(u)s',

                                                {'addr': ldapemailaddr, 'u': eid})

                            else:

                                # no email found, create it

                                _insert_email(session, ldapemailaddr, eid)

        finally:

            session.commit()

            session.close()

    def get_connection(self):

        """open and return a connection to the source"""

        if self._conn is None:

            self._connect()

        return ConnectionWrapper(self._conn)

    def authenticate(self, session, login, password):

        """return EUser eid for the given login/password if this account is

        defined in this source, else raise `AuthenticationError`

        two queries are needed since passwords are stored crypted, so we have

        to fetch the salt first

        """

        assert login, 'no login!'

        searchfilter = [filter_format('(%s=%s)', (self.user_login_attr, login))]

        searchfilter.extend([filter_format('(%s=%s)', ('objectClass', o))

                             for o in self.user_classes])

        searchstr = '(&%s)' % ''.join(searchfilter)

        # first search the user

        try:

            user = self._search(session, self.user_base_dn,

                                self.user_base_scope, searchstr)[0]

        except IndexError:

            # no such user

            raise AuthenticationError()

        # check password by establishing a (unused) connection

        try:

            self._connect(user['dn'], password)

        except:

            # Something went wrong, most likely bad credentials

            raise AuthenticationError()

        return self.extid2eid(user['dn'], 'EUser', session)

    def ldap_name(self, var):

        if var.stinfo['relations']:

            relname = iter(var.stinfo['relations']).next().r_type

            return self.user_rev_attrs.get(relname)

        return None

    def prepare_columns(self, mainvars, rqlst):

        """return two list describin how to build the final results

        from the result of an ldap search (ie a list of dictionnary)

        """

        columns = []

        global_transforms = []

        for i, term in enumerate(rqlst.selection):

            if isinstance(term, Constant):

                columns.append(term)

                continue

            if isinstance(term, Function): # LOWER, UPPER, COUNT...

                var = term.get_nodes(VariableRef)[0]

                var = var.variable

                try:

                    mainvar = var.stinfo['attrvar'].name

                except AttributeError: # no attrvar set

                    mainvar = var.name

                assert mainvar in mainvars

                trname = term.name

                ldapname = self.ldap_name(var)

                if trname in ('COUNT', 'MIN', 'MAX', 'SUM'):

                    global_transforms.append(GlobTrFunc(trname, i, ldapname))

                    columns.append((mainvar, ldapname))

                    continue

                if trname in ('LOWER', 'UPPER'):

                    columns.append((mainvar, TrFunc(trname, i, ldapname)))

                    continue

                raise NotImplementedError('no support for %s function' % trname)

            if term.name in mainvars:

                columns.append((term.name, 'dn'))

                continue

            var = term.variable

            mainvar = var.stinfo['attrvar'].name

            columns.append((mainvar, self.ldap_name(var)))

            #else:

            #    # probably a bug in rql splitting if we arrive here

            #    raise NotImplementedError

        return columns, global_transforms

    def syntax_tree_search(self, session, union,

                           args=None, cachekey=None, varmap=None, debug=0):

        """return result from this source for a rql query (actually from a rql 

        syntax tree and a solution dictionary mapping each used variable to a 

        possible type). If cachekey is given, the query necessary to fetch the

        results (but not the results themselves) may be cached using this key.

        """

        # XXX not handled : transform/aggregat function, join on multiple users...

        assert len(union.children) == 1, 'union not supported'

        rqlst = union.children[0]

        assert not rqlst.with_, 'subquery not supported'

        rqlkey = rqlst.as_string(kwargs=args)

        try:

            results = self._query_cache[rqlkey]

        except KeyError:

            results = self.rqlst_search(session, rqlst, args)

            self._query_cache[rqlkey] = results

        return results

    def rqlst_search(self, session, rqlst, args):

        mainvars = []

        for varname in rqlst.defined_vars:

            for sol in rqlst.solutions:

                if sol[varname] == 'EUser':

                    mainvars.append(varname)

                    break

        assert mainvars

        columns, globtransforms = self.prepare_columns(mainvars, rqlst)

        eidfilters = []

        allresults = []

        generator = RQL2LDAPFilter(self, session, args, mainvars)

        for mainvar in mainvars:

            # handle restriction

            try:

                eidfilters_, ldapfilter = generator.generate(rqlst, mainvar)

            except GotDN, ex:

                assert ex.dn, 'no dn!'

                try:

                    res = [self._cache[ex.dn]]

                except KeyError:

                    res = self._search(session, ex.dn, BASE)

            except UnknownEid, ex:

                # raised when we are looking for the dn of an eid which is not

                # coming from this source

                res = []

            else:

                eidfilters += eidfilters_

                res = self._search(session, self.user_base_dn,

                                   self.user_base_scope, ldapfilter)

            allresults.append(res)

        # 1. get eid for each dn and filter according to that eid if necessary

        for i, res in enumerate(allresults):

            filteredres = []

            for resdict in res:

                # get sure the entity exists in the system table

                eid = self.extid2eid(resdict['dn'], 'EUser', session)

                for eidfilter in eidfilters:

                    if not eidfilter(eid):

                        break

                else:

                    resdict['eid'] = eid

                    filteredres.append(resdict)

            allresults[i] = filteredres

        # 2. merge result for each "mainvar": cartesian product

        allresults = cartesian_product(allresults)

        # 3. build final result according to column definition

        result = []

        for rawline in allresults:

            rawline = dict(zip(mainvars, rawline))

            line = []

            for varname, ldapname in columns:

                if ldapname is None:

                    value = None # no mapping available

                elif ldapname == 'dn':

                    value = rawline[varname]['eid']

                elif isinstance(ldapname, Constant):

                    if ldapname.type == 'Substitute':

                        value = args[ldapname.value]

                    else:

                        value = ldapname.value

                elif isinstance(ldapname, TrFunc):

                    value = ldapname.apply(rawline[varname])

                else:

                    value = rawline[varname].get(ldapname)

                line.append(value)

            result.append(line)

        for trfunc in globtransforms:

            result = trfunc.apply(result)

        #print '--> ldap result', result

        return result

    def _connect(self, userdn=None, userpwd=None):

        port, protocol = MODES[self.cnx_mode]

        if protocol == 'ldapi':

            hostport = self.host

        else:

            hostport = '%s:%s' % (self.host, self.port or port)

        self.info('connecting %s://%s as %s', protocol, hostport,

                  userdn or 'anonymous')

        url = LDAPUrl(urlscheme=protocol, hostport=hostport)

        conn = ReconnectLDAPObject(url.initializeUrl())

        # Set the protocol version - version 3 is preferred

        try: