def name_to_id(name, tr_pairs):
for pair in tr_pairs:
pair_name = pair[0][findStr(pair[0], "/", 1) +
1:findStr(pair[0], "/", 3)]
if name == pair_name:
return pair[1]
return -1
def get_apnn_from_id_with_dict(a_id, tr_pairs, all_pn_dict):
# 由id获取名字
for pair in tr_pairs:
pair_id = pair[1]
if pair_id == a_id:
path = pair[0]
name = path[findStr(path, "/", 1) + 1:findStr(path, "/", 3)]
break
# 从all_pn_dict中取出当前a的p和n list
p_list = all_pn_dict[name]['p_list']
n_list = all_pn_dict[name]['n_list']
# 从p和n的list中抽取一个p，两个n，并在training set中找到其对应的id
# 如果抽到的p或n不存在于training set，则重新抽取这个p或n
p_id = -1
n1_id = -1
n2_id = -1
n3_id = -1
while p_id == -1:
p_name = random.choice(p_list)
p_id = name_to_id(p_name, tr_pairs)
while n1_id == -1:
n1_name = random.choice(n_list)
n1_id = name_to_id(n1_name, tr_pairs)
while n2_id == -1:
n2_name = random.choice(n_list)
n2_id = name_to_id(n2_name, tr_pairs)
while n3_id == -1:
n3_name = random.choice(n_list)
n3_id = name_to_id(n3_name, tr_pairs)
return [a_id, p_id, n1_id, n2_id, n3_id]
def people_picker(a_index, tr_pairs):
# a_index是人的id
# 获取a的人名
a_path = ''
for pair in tr_pairs:
if pair[1] == a_index:
a_path = pair[0]
a_name = a_path[findStr(a_path, "/", 1) + 1:findStr(a_path, "/", 3)]
# 获取所有人的人名
people_name = []
for pair in tr_pairs:
path = pair[0]
people_name.append(path[findStr(path, "/", 1) +
1:findStr(path, "/", 3)])
people_name = list(set(people_name))
# 读取关系表
with open("new_label.csv") as f:
reader = csv.reader(f)
column1 = []
column2 = []
for row in reader:
column1.append(row[0])
column2.append(row[1])
# 找到和a有亲属关系的人的名称
p_list = []
for id, name in enumerate(column1):
if name == a_name:
p_list.append(column2[id])
for id, name in enumerate(column2):
if name == a_name:
p_list.append(column1[id])
# 找到和a没有亲属关系的人的名称
n_list = [x for x in people_name if x not in p_list]
# 随机抽取p，n1，n2的名称
p_name = random.choice(p_list)
n1_name = random.choice(n_list)
n2_name = random.choice(n_list)
n3_name = random.choice(n_list)
# print("a: ", a_name)
# print("p: ", p_name)
# print("n: ", n1_name)
# print("n: ", n2_name)
return a_name, p_name, n1_name, n2_name, n3_name
def all_pn_json():
tr_pairs = get_tr_pairs()
all_pn_dict = {}
# 获取所有人的人名
people_name = []
for pair in tr_pairs:
path = pair[0]
people_name.append(path[findStr(path, "/", 1) +
1:findStr(path, "/", 3)])
people_name = list(set(people_name))
print("total number: ", people_name.__len__())
# 读取关系表
with open("new_label.csv") as f:
reader = csv.reader(f)
column1 = []
column2 = []
for row in reader:
column1.append(row[0])
column2.append(row[1])
i = 1
for name in people_name:
print("current: ", i)
name_dict = {}
# 找到有亲属关系的人的名称
p_list = []
for id, table_name in enumerate(column1):
if table_name == name:
p_list.append(column2[id])
for id, table_name in enumerate(column2):
if table_name == name:
p_list.append(column1[id])
# 找到没有亲属关系的人的名称
n_list = [x for x in people_name if x not in p_list]
name_dict['p_list'] = p_list
name_dict['n_list'] = n_list
all_pn_dict[name] = name_dict
i += 1
# 存为json
dict_to_json("all_pn_json.json", all_pn_dict)
def data_clean(tr_pairs):
# 获取所有人的人名
people_name = []
for pair in tr_pairs:
path = pair[0]
people_name.append(path[findStr(path, "/", 1) +
1:findStr(path, "/", 3)])
people_name = list(set(people_name))
# 读取关系表
with open("new_label.csv") as f:
reader = csv.reader(f)
column1 = []
column2 = []
for row in reader:
column1.append(row[0])
column2.append(row[1])
for name in people_name:
if name not in column1 and name not in column2:
print(name)
def table_clean(tr_pairs):
# 获取所有人的人名
people_name = []
for pair in tr_pairs:
path = pair[0]
people_name.append(path[findStr(path, "/", 1) +
1:findStr(path, "/", 3)])
people_name = list(set(people_name))
# 读取关系表
with open("train_relationships.csv") as f:
reader = csv.reader(f)
column1 = []
column2 = []
for row in reader:
column1.append(row[0])
column2.append(row[1])
column1_new1 = []
column2_new1 = []
for id, name in enumerate(column1):
if name in people_name:
column1_new1.append(column1[id])
column2_new1.append(column2[id])
column1_new2 = []
column2_new2 = []
for id, name in enumerate(column2_new1):
if name in people_name:
column1_new2.append(column1_new1[id])
column2_new2.append(column2_new1[id])
table = []
for i in range(column1_new2.__len__()):
table.append([column1_new2[i], column2_new2[i]])
list_to_csv(table, "new_label.csv")