def padding_data(data, dictionary, params, type):
if type == 'msg':
pad_msg = padding_message(data=data, max_length=params.msg_length)
pad_msg = mapping_dict_msg(pad_msg=pad_msg, dict_msg=dictionary)
return pad_msg
elif type == 'code':
pad_code = padding_commit_code(data=data, max_line=params.code_line, max_length=params.code_length)
pad_code = mapping_dict_code(pad_code=pad_code, dict_code=dictionary)
return pad_code
else:
print('Your type is incorrect -- please correct it')
exit()
obj = pickle.load(f)
f.close()
messages, codes, labels, ids = obj
labels = convert_label(labels)
codes = clean_and_reformat_code(data=codes)
info_label(data=labels)
print('Number of instances in commit message %i and commit code %i ' % (len(messages), len(codes)))
print('Labels: %i' % (len(labels)))
dict_msg, dict_code = dictionary_commit(data=messages, type_data='msg'), dictionary_commit(data=codes,
type_data='code')
pad_msg = padding_message(data=messages, max_length=256)
added_code, removed_code = codes
pad_added_code = padding_commit_code(data=added_code, max_file=3, max_line=10, max_length=256)
pad_removed_code = padding_commit_code(data=removed_code, max_file=3, max_line=10, max_length=256)
pad_msg = mapping_dict_msg(pad_msg=pad_msg, dict_msg=dict_msg)
pad_added_code = mapping_dict_code(pad_code=pad_added_code, dict_code=dict_code)
pad_removed_code = mapping_dict_code(pad_code=pad_removed_code, dict_code=dict_code)
data = (pad_msg, pad_added_code, pad_removed_code, labels, dict_msg, dict_code, ids)
print('Dictionary message: %i -- Dictionary code: %i' % (len(dict_msg), len(dict_code)))
print('Shape of commit message:', pad_msg.shape)
print('Shape of added code:', pad_added_code.shape)
print('Shape of removed code:', pad_removed_code.shape)
print('Shape of labels:', labels.shape)
print('Ids of projects:', project, len(ids))
write_data = open('../data/jit_' + project + '.pkl', 'wb')
pickle.dump(data, write_data)