mirrors/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2025-05-13 22:54:27 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

After a timeout, chrome will leave behind a SingletonLock, which prevents future instances of chrome from starting. When an extractor fails due to a timeout, remove this file.

Browse source
This commit is contained in:
spresse1 2023-08-28 17:27:03 +02:00
parent 00ecf57b0f
commit 603ce7ec10
4 changed files with 18 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

12
archivebox/util.py
View file

@ -17,6 +17,8 @@ from requests.exceptions import RequestException, ReadTimeout
from .vendor.base32_crockford import encode as base32_encode # type: ignore
from w3lib.encoding import html_body_declared_encoding, http_content_type_encoding
from os.path import lexists
from os import remove as remove_file
try:
import chardet
@ -272,6 +274,16 @@ def chrome_args(**options) -> List[str]:
return cmd_args
def chrome_cleanup():
"""
Cleans up any state or runtime files that chrome leaves behind when killed by
a timeout or other error
"""
from .config import IN_DOCKER
if IN_DOCKER and lexists("/home/archivebox/.config/chromium/SingletonLock"):
remove_file("/home/archivebox/.config/chromium/SingletonLock")
def ansi_to_html(text):
"""