Compare commits
5 Commits
efe45c1a4b
...
master
| Author | SHA1 | Date | |
|---|---|---|---|
| b14024f1a3 | |||
|
|
1f746f8177 | ||
| b60d1c6b8a | |||
|
|
bbb30868ca | ||
|
|
cca9c3024a |
2
HBD.py
2
HBD.py
@@ -1,5 +1,5 @@
|
|||||||
from pymongo import MongoClient
|
from pymongo import MongoClient
|
||||||
|
#wdawfqfqwfqwfqw
|
||||||
|
|
||||||
def connect_to_mongo():
|
def connect_to_mongo():
|
||||||
"""Подключение к MongoDB"""
|
"""Подключение к MongoDB"""
|
||||||
|
|||||||
15
Serch_H.py
15
Serch_H.py
@@ -2,12 +2,13 @@ from bs4 import BeautifulSoup as bs
|
|||||||
import requests
|
import requests
|
||||||
import HBD
|
import HBD
|
||||||
|
|
||||||
link = 'https://x3.h-chan.me/manga/newest?offset='
|
link = 'https://x8.h-chan.me/manga/newest?offset=' #https://x8.h-chan.me/manga/
|
||||||
result = HBD.find_doc_with_max_id(HBD.connect_to_mongo())
|
result = HBD.find_doc_with_max_id(HBD.connect_to_mongo())
|
||||||
|
|
||||||
max_id = '51196'
|
max_id = '51196'
|
||||||
max_num = 29915
|
max_num = 29915
|
||||||
|
|
||||||
|
|
||||||
def form_date(date_str):
|
def form_date(date_str):
|
||||||
months = {
|
months = {
|
||||||
"января": "01",
|
"января": "01",
|
||||||
@@ -47,7 +48,7 @@ def try_request(link, max_retries=50):
|
|||||||
except:
|
except:
|
||||||
retries += 1
|
retries += 1
|
||||||
|
|
||||||
def pars(link, flag):
|
def pars(link, flag, count):
|
||||||
data_hantai = {}
|
data_hantai = {}
|
||||||
|
|
||||||
response = try_request(link)
|
response = try_request(link)
|
||||||
@@ -93,16 +94,22 @@ def pars(link, flag):
|
|||||||
'original_id' : ID,
|
'original_id' : ID,
|
||||||
}
|
}
|
||||||
|
|
||||||
|
print(f'стр - {count}', data_hantai)
|
||||||
|
|
||||||
|
for i, j in data_hantai.items():
|
||||||
|
print('-'*10, i, j['manga_link'])
|
||||||
|
|
||||||
print(data_hantai)
|
|
||||||
return data_hantai, flag
|
return data_hantai, flag
|
||||||
|
|
||||||
def get_data():
|
def get_data():
|
||||||
data = {}
|
data = {}
|
||||||
|
count = 1
|
||||||
|
|
||||||
for i in range(0, 5000, 20):
|
for i in range(0, 5000, 20):
|
||||||
|
|
||||||
flag = False
|
flag = False
|
||||||
data_hantai, flag = pars(link + str(i), flag)
|
data_hantai, flag = pars(link + str(i), flag, count)
|
||||||
|
count +=1
|
||||||
|
|
||||||
data.update(data_hantai)
|
data.update(data_hantai)
|
||||||
|
|
||||||
|
|||||||
@@ -2,7 +2,7 @@ from bs4 import BeautifulSoup as bs
|
|||||||
import requests
|
import requests
|
||||||
import re
|
import re
|
||||||
|
|
||||||
#Поиск фулл страниц манги
|
#Поиск фулл страниц манги цауц
|
||||||
def process_hentai_data(hent_data, max_retries=50):
|
def process_hentai_data(hent_data, max_retries=50):
|
||||||
|
|
||||||
data_hantai = {}
|
data_hantai = {}
|
||||||
|
|||||||
Reference in New Issue
Block a user