Browse Source

Using pathlib.Path.parent instead of os.path.join(__fille,".."

master
Raphael Roberts 7 years ago
parent
commit
23aab706d2
  1. 1
      .gitignore
  2. 16
      restscrape/scraping/browser.py

1
.gitignore

@ -120,3 +120,4 @@ nppBackup
# HTML files for testing # HTML files for testing
*.html *.html
!**/uBlock/* !**/uBlock/*
/restscrape_venv/*

16
restscrape/scraping/browser.py

@ -1,6 +1,7 @@
import asyncio import asyncio
import os import os
import time import time
from pathlib import Path
import pyppeteer import pyppeteer
@ -23,7 +24,8 @@ class Browser:
def connect(self, socket_address): def connect(self, socket_address):
self.browser_instance = run( self.browser_instance = run(
pyppeteer.launcher.connect(browserWSEndpoint=socket_address))
pyppeteer.launcher.connect(browserWSEndpoint=socket_address)
)
self.address = socket_address self.address = socket_address
self.page: pyppeteer.page.Page = run(self.browser_instance.pages())[0] self.page: pyppeteer.page.Page = run(self.browser_instance.pages())[0]
# self.page: pyppeteer.page.Page = run(b.browser_instance.newPage()) # self.page: pyppeteer.page.Page = run(b.browser_instance.newPage())
@ -32,16 +34,16 @@ class Browser:
opts = {} opts = {}
opts.update(self.launch_opts) opts.update(self.launch_opts)
if use_adblock: if use_adblock:
adblock_path = os.path.join(__file__, "..", "uBlock")
opts.setdefault('args', []).extend(
adblock_path = Path(__file__).parent / "uBlock"
opts.setdefault("args", []).extend(
[ [
'--disable-extensions-except='+adblock_path,
'--load-extension='+adblock_path
"--disable-extensions-except={}".format(adblock_path),
"--load-extension={}".format(adblock_path),
] ]
) )
if proxy is not None: if proxy is not None:
opts.setdefault('args', []).extend(['--proxy-server='+proxy])
opts.setdefault('args', []).append('about:blank')
opts.setdefault("args", []).extend(["--proxy-server=" + proxy])
opts.setdefault("args", []).append("about:blank")
self.browser_instance = run(pyppeteer.launch(**opts)) self.browser_instance = run(pyppeteer.launch(**opts))
self.address = self.browser_instance.wsEndpoint self.address = self.browser_instance.wsEndpoint
self.page: pyppeteer.page.Page = run(self.browser_instance.pages())[0] self.page: pyppeteer.page.Page = run(self.browser_instance.pages())[0]

Loading…
Cancel
Save