bacow/AV_Data_Capture
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update 3.7-finaly

Browse Source
This commit is contained in:
root
2020-08-15 17:14:38 +08:00
parent e687035722
commit 43b0bf7c34
1 changed files with 1 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
WebCrawler/avsox.py
View File

@@ -1,5 +1,5 @@
import sys import sys
sys.path.append('../') sys.path.append('..')
import re import re
from lxml import etree from lxml import etree
import json import json
@@ -88,12 +88,10 @@ def main(number):
result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']") result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']")
if result1 == '' or result1 == 'null' or result1 == 'None': if result1 == '' or result1 == 'null' or result1 == 'None':
a = get_html('https://avsox.host/cn/search/' + number.replace('-', '_')) a = get_html('https://avsox.host/cn/search/' + number.replace('-', '_'))
print(a)
html = etree.fromstring(a, etree.HTMLParser()) # //table/tr[1]/td[1]/text() html = etree.fromstring(a, etree.HTMLParser()) # //table/tr[1]/td[1]/text()
result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']") result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']")
if result1 == '' or result1 == 'null' or result1 == 'None': if result1 == '' or result1 == 'null' or result1 == 'None':
a = get_html('https://avsox.host/cn/search/' + number.replace('_', '')) a = get_html('https://avsox.host/cn/search/' + number.replace('_', ''))
print(a)
html = etree.fromstring(a, etree.HTMLParser()) # //table/tr[1]/td[1]/text() html = etree.fromstring(a, etree.HTMLParser()) # //table/tr[1]/td[1]/text()
result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']") result1 = str(html.xpath('//*[@id="waterfall"]/div/a/@href')).strip(" ['']")
web = get_html(result1) web = get_html(result1)