panleicim/appointment_tool
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

disable no used comments

Browse Source
This commit is contained in:
Lei PAN
2024-03-20 21:10:52 +01:00
parent 32748762d6
commit f2627e102f
1 changed files with 13 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/person_name/extract_name_with_pinyinlist.py
+13 -13
View File
@@ -13,7 +13,7 @@ from src.utils.contacts.generate_random_passport_id import get_random_passport_i
def get_ordered_combins(stuff): def get_ordered_combins(stuff):
list_to_return = [] list_to_return = []
for i, j in itertools.combinations(range(len(stuff) + 3), 2): for i, j in itertools.combinations(range(len(stuff) + 3), 2):
print(stuff[i:j]) # print(stuff[i:j])
list_to_return.append(stuff[i:j]) list_to_return.append(stuff[i:j])
return list_to_return return list_to_return
@@ -35,7 +35,7 @@ def read_pinyin_list_from_file() -> list:
count = 0 count = 0
for line in lines: for line in lines:
count += 1 count += 1
print("Line{}: {}".format(count, line.strip())) # print("Line{}: {}".format(count, line.strip()))
name_list.append(line.strip()) name_list.append(line.strip())
return name_list return name_list
@@ -47,7 +47,7 @@ def read_pinyin_first_name_from_file() -> list:
count = 0 count = 0
for line in lines: for line in lines:
count += 1 count += 1
print("Line{}: {}".format(count, line.strip())) # print("Line{}: {}".format(count, line.strip()))
name_list.append(line.strip()) name_list.append(line.strip())
return name_list return name_list
@@ -59,7 +59,7 @@ def read_pinyin_last_name_from_file() -> list:
count = 0 count = 0
for line in lines: for line in lines:
count += 1 count += 1
print("Line{}: {}".format(count, line.strip())) # print("Line{}: {}".format(count, line.strip()))
name_list.append(line.strip()) name_list.append(line.strip())
return name_list return name_list
@@ -97,8 +97,8 @@ def get_maximum_length():
for first_name in all_first_name: for first_name in all_first_name:
if len(first_name) > max_first_name_lenghth: if len(first_name) > max_first_name_lenghth:
max_first_name_lenghth = len(first_name) max_first_name_lenghth = len(first_name)
print("max_last_name_lenghth :" + str(max_last_name_lenghth)) # print("max_last_name_lenghth :" + str(max_last_name_lenghth))
print("max_first_name_lenghth :" + str(max_first_name_lenghth)) # print("max_first_name_lenghth :" + str(max_first_name_lenghth))
def has_numbers(inputString): def has_numbers(inputString):
@@ -130,18 +130,18 @@ def generate_name_from_email(mail_address, pinyin_name_list):
# key_words = HanLP.extractKeyword(mail_address, 2) # key_words = HanLP.extractKeyword(mail_address, 2)
# print(key_words) # print(key_words)
# setence = "".join(key_words) # setence = "".join(key_words)
print("generate for " + mail_address) # print("generate for " + mail_address)
all_combins = get_ordered_combins(mail_address) all_combins = get_ordered_combins(mail_address)
all_combins.sort(key=len, reverse=True) all_combins.sort(key=len, reverse=True)
no_duplicated_list = [] no_duplicated_list = []
for word in all_combins: for word in all_combins:
if word not in no_duplicated_list: if word not in no_duplicated_list:
no_duplicated_list.append(word) no_duplicated_list.append(word)
print(all_combins) # print(all_combins)
for i in no_duplicated_list: for i in no_duplicated_list:
word_to_test = "".join(i) word_to_test = "".join(i)
print("word to test is " + word_to_test) # print("word to test is " + word_to_test)
# if len(word_to_test) >= 11: # if len(word_to_test) >= 11:
# found_name = check_name(word_to_test=word_to_test, pinyin_name_list=pinyin_name_list) # found_name = check_name(word_to_test=word_to_test, pinyin_name_list=pinyin_name_list)
# if found_name is not None: # if found_name is not None:
@@ -250,17 +250,17 @@ def filter_already_validated_contacts(_list_to_extract):
if __name__ == '__main__': if __name__ == '__main__':
start_position = 14500 start_position = 1
end_position = 15084 end_position = 502
# x = range(start_position, end_position, 100) # x = range(start_position, end_position, 100)
step = 100 step = 500
step_list = range(start_position, end_position, step) step_list = range(start_position, end_position, step)
print(step_list[-1]) print(step_list[-1])
for x in step_list: for x in step_list:
print(x) print(x)
_from_position = x _from_position = x
_end_position = x + step _end_position = x + step
print("_end_position is {}".format(_end_position)) print("_from_position is {},_end_position is {}".format(_from_position, _end_position))
if _end_position <= step_list[-1]: if _end_position <= step_list[-1]:
print("start extraction from {} to {}".format(_from_position, _end_position)) print("start extraction from {} to {}".format(_from_position, _end_position))
generate_contact_from_mail_list(_from_position, _end_position) generate_contact_from_mail_list(_from_position, _end_position)