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

clean up

Browse Source
This commit is contained in:
lededev
2022-04-20 12:48:38 +08:00
parent 5da99067c8
commit 0dff1a72c0
2 changed files with 7 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
WebCrawler/gcolle.py
View File

@@ -4,7 +4,6 @@ sys.path.append('../')
from WebCrawler.crawler import * from WebCrawler.crawler import *
from ADC_function import * from ADC_function import *
from lxml import etree from lxml import etree
from requests_html import HTMLSession
def main(number): def main(number):
config_file = config.getInstance() config_file = config.getInstance()
@@ -12,7 +11,6 @@ def main(number):
number = number.upper().replace('GCOLLE-','') number = number.upper().replace('GCOLLE-','')
session = get_html_session() session = get_html_session()
htmlcode = get_html_session('https://gcolle.net/product_info.php/products_id/' + number)
htmlcode = session.get('https://gcolle.net/product_info.php/products_id/' + number).text htmlcode = session.get('https://gcolle.net/product_info.php/products_id/' + number).text
html = etree.HTML(htmlcode) html = etree.HTML(htmlcode)
# R18 countinue # R18 countinue
@@ -66,9 +64,10 @@ def main(number):
return dic return dic
if __name__ == '__main__': if __name__ == '__main__':
main('840724') from pprint import pprint
main('840386') pprint(main('840724'))
main('838671') pprint(main('840386'))
main('814179') pprint(main('838671'))
main('834255') pprint(main('814179'))
main('814179') pprint(main('834255'))
pprint(main('814179'))

1
requirements.txt
View File

@@ -10,4 +10,3 @@ certifi==2020.12.5
MechanicalSoup==1.1.0 MechanicalSoup==1.1.0
opencc-python-reimplemented opencc-python-reimplemented
face_recognition face_recognition
requests_html