marcel/linkding
1
0
Fork 0
mirror of https://github.com/sissbruecker/linkding.git synced 2025-08-13 21:49:26 +02:00
Code Issues Packages Projects Releases Wiki Activity

Remove ads and cookie banners from HTML snapshots (#695)

Browse Source 
* integrate ublock with single-file

* reuse chromium profile
This commit is contained in:
Sascha Ißbrücker
2024-04-14 13:09:46 +02:00
committed by GitHub
parent 22a1fc80ad
commit 25470edb2c
8 changed files with 89 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
bookmarks/services/singlefile.py
View File

@@ -18,11 +18,12 @@ logger = logging.getLogger(__name__)
def create_snapshot(url: str, filepath: str):
singlefile_path = settings.LD_SINGLEFILE_PATH
# parse string to list of arguments
singlefile_options = shlex.split(settings.LD_SINGLEFILE_OPTIONS)
# parse options to list of arguments
ublock_options = shlex.split(settings.LD_SINGLEFILE_UBLOCK_OPTIONS)
custom_options = shlex.split(settings.LD_SINGLEFILE_OPTIONS)
temp_filepath = filepath + ".tmp"
# concat lists
args = [singlefile_path] + singlefile_options + [url, temp_filepath]
args = [singlefile_path] + ublock_options + custom_options + [url, temp_filepath]
try:
# Use start_new_session=True to create a new process group
process = subprocess.Popen(args, start_new_session=True)

14
bookmarks/tests/test_singlefile_service.py
View File

@@ -61,6 +61,10 @@ class SingleFileServiceTestCase(TestCase):
expected_args = [
"single-file",
'--browser-arg="--headless=new"',
'--browser-arg="--user-data-dir=./chromium-profile"',
'--browser-arg="--no-sandbox"',
'--browser-arg="--load-extension=uBlock0.chromium"',
"http://example.com",
self.html_filepath + ".tmp",
]
@@ -79,6 +83,10 @@ class SingleFileServiceTestCase(TestCase):
expected_args = [
"single-file",
'--browser-arg="--headless=new"',
'--browser-arg="--user-data-dir=./chromium-profile"',
'--browser-arg="--no-sandbox"',
'--browser-arg="--load-extension=uBlock0.chromium"',
"--some-option",
"some value",
"--another-option",
@@ -97,9 +105,9 @@ class SingleFileServiceTestCase(TestCase):
with mock.patch("subprocess.Popen", return_value=mock_process):
singlefile.create_snapshot("http://example.com", self.html_filepath)
mock_process.wait.assert_called_with(timeout=60)
mock_process.wait.assert_called_with(timeout=120)
@override_settings(LD_SINGLEFILE_TIMEOUT_SEC=120)
@override_settings(LD_SINGLEFILE_TIMEOUT_SEC=180)
def test_create_snapshot_custom_timeout_setting(self):
mock_process = mock.Mock()
mock_process.wait.return_value = 0
@@ -108,4 +116,4 @@ class SingleFileServiceTestCase(TestCase):
with mock.patch("subprocess.Popen", return_value=mock_process):
singlefile.create_snapshot("http://example.com", self.html_filepath)
mock_process.wait.assert_called_with(timeout=120)
mock_process.wait.assert_called_with(timeout=180)