def beautify_similar_files(self, similar_files, fo_list, similarity):
result_dict = {}
for match in similar_files:
match_dict = {fo.uid: None for fo in fo_list}
for file in match:
firm, sub = file.split(':')
match_dict[firm] = sub
if convert_uid_list_to_compare_id(match) in similarity.keys():
match_dict['similarity'] = similarity[convert_uid_list_to_compare_id(match)]
else:
match_dict['similarity'] = ''
result_dict[self._match_id(match)] = match_dict
return result_dict
def _calculate_compare_result_id(compare_result):
general_dict = compare_result['general']
uid_set = set()
for key in general_dict:
uid_set.update(list(general_dict[key].keys()))
comp_id = convert_uid_list_to_compare_id(list(uid_set))
return comp_id
def _app_show_start_compare(self):
if 'uids_for_comparison' not in session or not isinstance(
session['uids_for_comparison'],
list) or len(session['uids_for_comparison']) < 2:
return render_template('compare/error.html',
error='No UIDs found for comparison')
compare_id = convert_uid_list_to_compare_id(
session['uids_for_comparison'])
session['uids_for_comparison'] = None
redo = True if request.args.get('force_recompare') else None
with ConnectTo(CompareDbInterface, self._config) as sc:
compare_exists = sc.compare_result_is_in_db(compare_id)
if compare_exists and not redo:
return redirect(
url_for('/compare/<compare_id>', compare_id=compare_id))
try:
with ConnectTo(CompareDbInterface, self._config) as sc:
sc.check_objects_exist(compare_id)
except FactCompareException as exception:
return render_template('compare/error.html',
error=exception.get_message())
with ConnectTo(InterComFrontEndBinding, self._config) as sc:
sc.add_compare_task(compare_id, force=redo)
return render_template('compare/wait.html', compare_id=compare_id)
def _get_similar_files(self, fo_list: List[FileObject], exclusive_files: Dict[str, List[str]]) -> Tuple[List[list], dict]:
similar_files = []
similarity = {}
for parent_one, parent_two in combinations(fo_list, 2):
for file_one in exclusive_files[parent_one.uid]:
for similar_file_pair, value in self._find_similar_file_for(file_one, parent_one.uid, parent_two):
similar_files.append(similar_file_pair)
similarity[convert_uid_list_to_compare_id(similar_file_pair)] = value
similarity_sets = generate_similarity_sets(remove_duplicates_from_unhashable(similar_files))
return similarity_sets, similarity
def _get_similarity_value(group_of_similar_files: List[str], similarity_dict: Dict[str, str]) -> str:
similarities_list = []
for id_tuple in combinations(group_of_similar_files, 2):
similar_file_pair_id = convert_uid_list_to_compare_id(id_tuple)
if similar_file_pair_id in similarity_dict:
similarities_list.append(similarity_dict[similar_file_pair_id])
if not similarities_list:
return ''
if len(similarities_list) == 1:
return similarities_list.pop()
similarities_list = [int(v) for v in similarities_list]
return '{} ‒ {}'.format(min(similarities_list), max(similarities_list))
def _get_similar_files(self, fo_list, exclusive_files):
similars = list()
similarity = dict()
for index, _ in enumerate(fo_list):
tmp_list = deepcopy(fo_list)
parent_one = tmp_list.pop(index)
for parent_two in tmp_list:
for file_one in exclusive_files[parent_one.uid]:
for item, value in self._find_similar_file_for(file=file_one, parent_id=parent_one.uid, potential_matches=parent_two):
similars.append(item)
similarity[convert_uid_list_to_compare_id(item)] = value
similarity_sets = self.produce_similarity_sets(remove_duplicates_from_list_of_lists(similars))
remove_subsets_from_list_of_sets(similarity_sets)
return remove_duplicates_from_list_of_lists(list_of_sets_to_list_of_lists(similarity_sets)), similarity