def transform_to_distributive_context(self, context):
"""Generate distributive context from current
Can add relations I(j,m) but can't destroy one
"""
# print('Generate the distributive context \''+context.context_name+'_d\' of the context \''+context.context_name+'\'')
if context.mode == Context.distributive:
return context
distributive_context = Context(context.context_name + '_d')
distributive_context.J = copy(context.J)
for j in distributive_context.J:
j_prime = context.get_j_prime(j)
j_filters = set()
X = set()
for k in distributive_context.J:
i_prime = context.get_j_prime(k)
if j_prime >= i_prime:
j_filters.add(k)
else:
X.add(k)
mj = 'm_' + j
distributive_context.add_m(mj)
for x in X:
distributive_context.add_i(x, mj)
return distributive_context
def import_context_from_file(self, file):
"""Import context values from a file.
Read line by line the file to create J, M and I sets.
Keyword arguments:
file_path -- the file path
"""
# print('Generate the context \'' + file.stem + '\' from \'' + str(file) + '\'')
context = Context(file.stem)
with file.open() as csv_file:
lecteur = csv.reader(csv_file, delimiter='\t', quotechar='|')
M_list = next(lecteur)
for m in M_list:
context.add_m(m)
l = 1
for line in lecteur:
c = 0
j = 'o' + str(l)
context.add_j(j)
for colonne in line:
if (colonne != '') and (colonne != '0'):
context.add_i(j, M_list[c])
c += 1
l += 1
return context
def step_transform_to_median_context(self, context, atoms):
global_context = Context(context.context_name)
# For each first filters
for atom in atoms:
# Create a new working context
filter_context = Context(context.context_name + '_' + atom)
for m in context.get_j_prime(atom):
filter_context.M.add(m)
for j in context.get_m_prime(m):
if j != atom:
filter_context.add_j(j)
filter_context.add_i(j, m)
# Generate the distributive first filter context
distributive_filter_context = self.transform_to_distributive_context(
filter_context)
if not len(distributive_filter_context.I):
for j in distributive_filter_context.J:
m_label = 'c' + str(atom)
distributive_filter_context.add_m(m_label)
distributive_filter_context.add_i(j, m_label)
# Fusion with previous distributive first filter context (no merge node atm)
global_context.J.update(distributive_filter_context.J)
# Add first filter if not exist
# TODO: utile ?
if not atom in global_context.J:
global_context.add_j(atom)
# If no attribute, create new one
if not len(distributive_filter_context.M):
new_name_m = 'a' + str(global_context.name_next_variable)
global_context.add_m(new_name_m)
global_context.add_i(atom, new_name_m)
global_context.name_next_variable += 1
# Add all M with a new name
for m in distributive_filter_context.M:
new_name_m = 'a' + str(global_context.name_next_variable)
global_context.add_m(new_name_m)
for i in distributive_filter_context.I:
if i[1] == m:
global_context.add_i(i[0], new_name_m)
global_context.add_i(atom, new_name_m)
global_context.name_next_variable += 1
return global_context
def extract_atoms_contexts(self, context, atoms):
list_contexts = set()
std = self.transform_to_standard_context(context)
# For each first filters
for atom in atoms:
# Create a new working context
atom_context = Context(std.context_name + '_' + atom)
for m in std.get_j_prime(atom):
atom_context.add_m(m)
for j in std.get_m_prime(m):
atom_context.add_j(j)
atom_context.add_i(j, m)
list_contexts.add(atom_context)
return list_contexts
