[packaging] we should add schemas to include_dirs so sql extensions files are properly installed
# copyright 2003-2010 LOGILAB S.A. (Paris, FRANCE), all rights reserved.
# contact http://www.logilab.fr/ -- mailto:contact@logilab.fr
#
# This file is part of CubicWeb.
#
# CubicWeb is free software: you can redistribute it and/or modify it under the
# terms of the GNU Lesser General Public License as published by the Free
# Software Foundation, either version 2.1 of the License, or (at your option)
# any later version.
#
# CubicWeb 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 Lesser General Public License for more
# details.
#
# You should have received a copy of the GNU Lesser General Public License along
# with CubicWeb. If not, see <http://www.gnu.org/licenses/>.
"""
"""
class Schema :
def __init__(self, schema) :
self._schema = schema
def get_attrs(self, entity) :
return self._schema[entity][0]
def get_relations(self, entity) :
return self._schema[entity][1]
def get_attr_index(self, entity, attr) :
return list(self._schema[entity][0]).index(attr)
SCHEMA = Schema({'societe': ( ('nom','ville'),
[('concerne_par','affaire'),
] ),
'affaire': ( ('ref',),
[('concerne','societe'),
('concerne_par', 'document')
] ),
'document':( ('fichier', 'annee','mois','jour','type'),
[('concerne','affaire'),
] ),
})
DATA = { 'societe': [ ('CETIAD', 'Dijon'),
('EDF_R&D', 'Clamart'),
('Logilab', 'Paris'),
],
'affaire': [ ('CTIA01', 'CETIAD'),
('EDFR01', 'EDF_R&D'),
('EDFR02', 'EDF_R&D'),
],
'document':[ ('CTIA01-040906-PRE-1-01.pdf','2004','09','06','PRE','CTIA01'),
('EDFR01-050201-CLI-1-01.pdf','2005','02','01','CLI','EDFR01'),
('EDFR01-050322-OFR-1-01.pdf','2005','03','22','OFR','EDFR01'),
],
}
def get_data(entity, where=[]) :
for value in DATA[entity] :
for index, val in where :
if value[index] != val :
break
else :
yield value
class PathParser :
def __init__(self, schema, path) :
self.schema = schema
self.path = path
self._components = iter([comp for comp in self.path.split('/') if comp])
self._entity = None
self._attr = None
self._rel = None
self._restrictions = []
def parse(self) :
self._entity = self._components.next()
try:
self.process_entity()
except StopIteration :
pass
def process_entity(self) :
_next = self._components.next()
if _next in self.schema.get_attrs(self._entity) :
self._attr = _next
_next = self._components.next()
self._restrictions.append( (self._entity, self._attr, _next) )
self._attr = None
self._rel = None
self.process_entity()
def get_list(self) :
if self._rel :
return
elif self._attr :
where = []
for e,a,v in self._restrictions :
i = self.schema.get_attr_index(e, a)
where.append( (i,v) )
i = self.schema.get_attr_index(self._entity, self._attr)
for values in get_data(self._entity,where) :
yield values[i]+'/'
else :
attr_restrict = [a for e,a,v in self._restrictions]
for attr in self.schema.get_attrs(self._entity) :
if attr not in attr_restrict :
yield attr+'/'
for data in DATA[self._entity]:
yield data[0]
for nom, entity in self.schema.get_relations(self._entity) :
yield nom+'/'
yield entity+'/'
def ls(path) :
p = PathParser(SCHEMA,path)
p.parse()
return list(p.get_list())
class SytPathParser :
def __init__(self, schema, path) :
self.schema = schema
self.path = path
self._components = iter([comp for comp in self.path.split('/') if comp])
self._e_type = None
self._restrictions = []
self._alphabet = list('ABCDEFGHIJKLMNOPQRSTUVWXYZ')
def parse(self):
self._var = self._alphabet.pop(0)
self._e_type = self._components.next()
e_type = self._e_type.capitalize()
self._restrictions.append('%s is %s' % (self._var, e_type))
try:
self.process_entity()
except StopIteration :
pass
return 'Any %s WHERE %s' % (self._var, ', '.join(self._restrictions))
def process_entity(self) :
_next = self._components.next()
if _next in self.schema.get_attrs(self._e_type) :
attr = _next
try:
_next = self._components.next()
self._restrictions.append('%s %s %s' % (self._var, attr, _next))
except StopIteration:
a_var = self._alphabet.pop(0)
self._restrictions.append('%s %s %s' % (self._var, attr, a_var) )
self._var = a_var
raise
elif _next in [r for r,e in self.schema.get_relations(self._e_type)]:
rel = _next
r_var = self._alphabet.pop(0)
self._restrictions.append('%s %s %s' % (self._var, rel, r_var))
self._var = r_var
try:
_next = self._components.next()
self._restrictions.append('%s is %s' % (r_var, _next.capitalize()))
except StopIteration:
raise
self.process_entity()
def to_rql(path) :
p = SytPathParser(SCHEMA,path)
return p.parse()