"""classes to define schemas for CubicWeb

:organization: Logilab

:copyright: 2001-2010 LOGILAB S.A. (Paris, FRANCE), license is LGPL v2.

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

:license: GNU Lesser General Public License, v2.1 - http://www.gnu.org/licenses

"""

__docformat__ = "restructuredtext en"

_ = unicode

import re

from os.path import join

from logging import getLogger

from warnings import warn

from logilab.common.decorators import cached, clear_cache, monkeypatch

from logilab.common.logging_ext import set_log_methods

from logilab.common.deprecation import deprecated, class_moved

from logilab.common.graph import get_cycles

from logilab.common.compat import any

from yams import BadSchemaDefinition, buildobjs as ybo

from yams.schema import Schema, ERSchema, EntitySchema, RelationSchema, \

     RelationDefinitionSchema, PermissionMixIn

from yams.constraints import BaseConstraint, FormatConstraint

from yams.reader import (CONSTRAINTS, PyFileReader, SchemaLoader,

                         obsolete as yobsolete, cleanup_sys_modules)

from rql import parse, nodes, RQLSyntaxError, TypeResolverException

import cubicweb

from cubicweb import ETYPE_NAME_MAP, ValidationError, Unauthorized

PURE_VIRTUAL_RTYPES = set(('identity', 'has_text',))

VIRTUAL_RTYPES = set(('eid', 'identity', 'has_text',))

#  set of meta-relations available for every entity types

META_RTYPES = set((

    'owned_by', 'created_by', 'is', 'is_instance_of', 'identity',

    'eid', 'creation_date', 'modification_date', 'has_text', 'cwuri',

    ))

SYSTEM_RTYPES = set(('require_permission', 'custom_workflow', 'in_state', 'wf_info_for'))

#  set of entity and relation types used to build the schema

SCHEMA_TYPES = set((

    'CWEType', 'CWRType', 'CWAttribute', 'CWRelation',

    'CWConstraint', 'CWConstraintType', 'RQLExpression',

    'relation_type', 'from_entity', 'to_entity',

    'constrained_by', 'cstrtype',

    ))

WORKFLOW_TYPES = set(('Transition', 'State', 'TrInfo', 'Workflow',

INTERNAL_TYPES = set(('CWProperty', 'CWPermission', 'CWCache', 'ExternalUri'))

_LOGGER = getLogger('cubicweb.schemaloader')

# schema entities created from serialized schema have an eid rproperty

ybo.ETYPE_PROPERTIES += ('eid',)

ybo.RTYPE_PROPERTIES += ('eid',)

ybo.RDEF_PROPERTIES += ('eid',)

PUB_SYSTEM_ENTITY_PERMS = {

    'read':   ('managers', 'users', 'guests',),

    'add':    ('managers',),

    'delete': ('managers',),

    'update': ('managers',),

    }

PUB_SYSTEM_REL_PERMS = {

    'read':   ('managers', 'users', 'guests',),

    'add':    ('managers',),

    'delete': ('managers',),

    }

PUB_SYSTEM_ATTR_PERMS = {

    'read':   ('managers', 'users', 'guests',),

    'update':    ('managers',),

    }

RO_REL_PERMS = {

    'read':   ('managers', 'users', 'guests',),

    'add':    (),

    'delete': (),

    }

RO_ATTR_PERMS = {

    'read':   ('managers', 'users', 'guests',),

    'update': (),

    }

# XXX same algorithm as in reorder_cubes and probably other place,

# may probably extract a generic function

def order_eschemas(eschemas):

    """return entity schemas ordered such that entity types which specializes an

    other one appears after that one

    """

    graph = {}

    for eschema in eschemas:

        if eschema.specializes():

            graph[eschema] = set((eschema.specializes(),))

        else:

            graph[eschema] = set()

    cycles = get_cycles(graph)

    if cycles:

        cycles = '\n'.join(' -> '.join(cycle) for cycle in cycles)

        raise Exception('cycles in entity schema specialization: %s'

                        % cycles)

    eschemas = []

    while graph:

        # sorted to get predictable results

        for eschema, deps in sorted(graph.items()):

            if not deps:

                eschemas.append(eschema)

                del graph[eschema]

                for deps in graph.itervalues():

                    try:

                        deps.remove(eschema)

                    except KeyError:

                        continue

    return eschemas

def bw_normalize_etype(etype):

    if etype in ETYPE_NAME_MAP:

        msg = '%s has been renamed to %s, please update your code' % (

            etype, ETYPE_NAME_MAP[etype])

        warn(msg, DeprecationWarning, stacklevel=4)

        etype = ETYPE_NAME_MAP[etype]

    return etype

def display_name(req, key, form='', context=None):

    """return a internationalized string for the key (schema entity or relation

    name) in a given form

    """

    assert form in ('', 'plural', 'subject', 'object')

    if form == 'subject':

        form = ''

    if form:

        key = key + '_' + form

    # ensure unicode

    # .lower() in case no translation are available XXX done whatever a translation is there or not!

    if context is not None:

        return unicode(req.pgettext(context, key)).lower()

    else:

        return unicode(req._(key)).lower()

__builtins__['display_name'] = deprecated('[3.4] display_name should be imported from cubicweb.schema')(display_name)

# rql expression utilities function ############################################

def guess_rrqlexpr_mainvars(expression):

    defined = set(split_expression(expression))

    mainvars = []

    if 'S' in defined:

        mainvars.append('S')

    if 'O' in defined:

        mainvars.append('O')

    if 'U' in defined:

        mainvars.append('U')

    if not mainvars:

        raise Exception('unable to guess selection variables')

    return ','.join(mainvars)

def split_expression(rqlstring):

    for expr in rqlstring.split(','):

        for noparen in expr.split('('):

            for word in noparen.split():

                yield word

def normalize_expression(rqlstring):

    """normalize an rql expression to ease schema synchronization (avoid

    suppressing and reinserting an expression if only a space has been added/removed

    for instance)

    """

    return u', '.join(' '.join(expr.split()) for expr in rqlstring.split(','))

# Schema objects definition ###################################################

def ERSchema_display_name(self, req, form='', context=None):

    """return a internationalized string for the entity/relation type name in

    a given form

    """

    return display_name(req, self.type, form, context)

ERSchema.display_name = ERSchema_display_name

@cached

def get_groups(self, action):

    """return the groups authorized to perform <action> on entities of

    this type

    :type action: str

    :param action: the name of a permission

    :rtype: tuple

    :return: names of the groups with the given permission

    """

    assert action in self.ACTIONS, action

    #assert action in self._groups, '%s %s' % (self, action)

    try:

        return frozenset(g for g in self.permissions[action] if isinstance(g, basestring))

    except KeyError:

        return ()

PermissionMixIn.get_groups = get_groups

@cached

def get_rqlexprs(self, action):

    """return the rql expressions representing queries to check the user is allowed

    to perform <action> on entities of this type

    :type action: str

    :param action: the name of a permission

    :rtype: tuple

    :return: the rql expressions with the given permission

    """

    assert action in self.ACTIONS, action

    #assert action in self._rqlexprs, '%s %s' % (self, action)

    try:

        return tuple(g for g in self.permissions[action] if not isinstance(g, basestring))

    except KeyError:

        return ()

PermissionMixIn.get_rqlexprs = get_rqlexprs

orig_set_action_permissions = PermissionMixIn.set_action_permissions

def set_action_permissions(self, action, permissions):

    """set the groups and rql expressions allowing to perform <action> on

    entities of this type

    :type action: str

    :param action: the name of a permission

    :type permissions: tuple

    :param permissions: the groups and rql expressions allowing the given action

    """

    orig_set_action_permissions(self, action, tuple(permissions))

    clear_cache(self, 'get_rqlexprs')

    clear_cache(self, 'get_groups')

PermissionMixIn.set_action_permissions = set_action_permissions

def has_local_role(self, action):

    """return true if the action *may* be granted localy (eg either rql

    expressions or the owners group are used in security definition)

    XXX this method is only there since we don't know well how to deal with

    'add' action checking. Also find a better name would be nice.

    """

    assert action in self.ACTIONS, action

    if self.get_rqlexprs(action):

        return True

    if action in ('update', 'delete'):

        return 'owners' in self.get_groups(action)

    return False

PermissionMixIn.has_local_role = has_local_role

def may_have_permission(self, action, req):

    if action != 'read' and not (self.has_local_role('read') or

                                 self.has_perm(req, 'read')):

        return False

    return self.has_local_role(action) or self.has_perm(req, action)

PermissionMixIn.may_have_permission = may_have_permission

def has_perm(self, session, action, **kwargs):

    """return true if the action is granted globaly or localy"""

    try:

        self.check_perm(session, action, **kwargs)

        return True

    except Unauthorized:

        return False

PermissionMixIn.has_perm = has_perm

def check_perm(self, session, action, **kwargs):

    # NB: session may be a server session or a request object check user is

    # in an allowed group, if so that's enough internal sessions should

    # always stop there

    groups = self.get_groups(action)

    if session.user.matching_groups(groups):

        return

    # if 'owners' in allowed groups, check if the user actually owns this

    # object, if so that's enough

    if 'owners' in groups and (

          kwargs.get('creating')

          or ('eid' in kwargs and session.user.owns(kwargs['eid']))):

        return

    # else if there is some rql expressions, check them

    if any(rqlexpr.check(session, **kwargs)

           for rqlexpr in self.get_rqlexprs(action)):

        return

    raise Unauthorized(action, str(self))

PermissionMixIn.check_perm = check_perm

RelationDefinitionSchema._RPROPERTIES['eid'] = None

def rql_expression(self, expression, mainvars=None, eid=None):

    """rql expression factory"""

    if self.rtype.final:

        return ERQLExpression(expression, mainvars, eid)

    return RRQLExpression(expression, mainvars, eid)

RelationDefinitionSchema.rql_expression = rql_expression

orig_check_permission_definitions = RelationDefinitionSchema.check_permission_definitions

def check_permission_definitions(self):

    orig_check_permission_definitions(self)

    schema = self.subject.schema

    for action, groups in self.permissions.iteritems():

        for group_or_rqlexpr in groups:

            if action == 'read' and \

                   isinstance(group_or_rqlexpr, RQLExpression):

                msg = "can't use rql expression for read permission of %s"

                raise BadSchemaDefinition(msg % self)

            if self.final and isinstance(group_or_rqlexpr, RRQLExpression):

                msg = "can't use RRQLExpression on %s, use an ERQLExpression"

                raise BadSchemaDefinition(msg % self)

            if not self.final and isinstance(group_or_rqlexpr, ERQLExpression):

                msg = "can't use ERQLExpression on %s, use a RRQLExpression"

                raise BadSchemaDefinition(msg % self)

RelationDefinitionSchema.check_permission_definitions = check_permission_definitions

class CubicWebEntitySchema(EntitySchema):

    """a entity has a type, a set of subject and or object relations

    the entity schema defines the possible relations for a given type and some

    constraints on those relations

    """

    def __init__(self, schema=None, edef=None, eid=None, **kwargs):

        super(CubicWebEntitySchema, self).__init__(schema, edef, **kwargs)

        if eid is None and edef is not None:

            eid = getattr(edef, 'eid', None)

        self.eid = eid

    def check_permission_definitions(self):

        super(CubicWebEntitySchema, self).check_permission_definitions()

        for groups in self.permissions.itervalues():

            for group_or_rqlexpr in groups:

                if isinstance(group_or_rqlexpr, RRQLExpression):

                    msg = "can't use RRQLExpression on %s, use an ERQLExpression"

                    raise BadSchemaDefinition(msg % self.type)

    def attribute_definitions(self):

        """return an iterator on attribute definitions

        attribute relations are a subset of subject relations where the

        object's type is a final entity

        an attribute definition is a 2-uple :

        * name of the relation

        * schema of the destination entity type

        """

        iter = super(CubicWebEntitySchema, self).attribute_definitions()

        for rschema, attrschema in iter:

            if rschema.type == 'has_text':

                continue

            yield rschema, attrschema

    def main_attribute(self):

        """convenience method that returns the *main* (i.e. the first non meta)

        attribute defined in the entity schema

        """

        for rschema, _ in self.attribute_definitions():

            if not (rschema in META_RTYPES

                    or self.is_metadata(rschema)):

                return rschema

    def add_subject_relation(self, rschema):

        """register the relation schema as possible subject relation"""

        super(CubicWebEntitySchema, self).add_subject_relation(rschema)

        self._update_has_text()

    def del_subject_relation(self, rtype):

        super(CubicWebEntitySchema, self).del_subject_relation(rtype)

        self._update_has_text(True)

    def _update_has_text(self, deletion=False):

        may_need_has_text, has_has_text = False, False

        need_has_text = None

        for rschema in self.subject_relations():

            if rschema.final:

                if rschema == 'has_text':

                    has_has_text = True

                elif self.rdef(rschema).get('fulltextindexed'):

                    may_need_has_text = True

            elif rschema.fulltext_container:

                if rschema.fulltext_container == 'subject':

                    may_need_has_text = True

                else:

                    need_has_text = False

        for rschema in self.object_relations():

            if rschema.fulltext_container: