Git-Mirrors/brozzler
1
0
Fork 0
mirror of https://github.com/internetarchive/brozzler.git synced 2025-08-08 06:22:23 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

Configurable JS templates location

Browse source 
Brozzler has hard-coded the JS templates logic in  ``brozzler/behaviors.yaml``
and ``brozzler/js-templates/`` locations. With this change, you can use
the optional ``behaviors_dir`` ``browser.browse_page`` parameter to set a
custom location and use any potential JS behaviors.
This commit is contained in:
Vangelis Banos 2018-01-04 17:37:02 +00:00
parent 503771d653
commit dacfba330c
2 changed files with 24 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

7
brozzler/browser.py
View file

@ -377,7 +377,7 @@ class Browser:
def browse_page(
self, page_url, extra_headers=None,
user_agent=None, behavior_parameters=None,
user_agent=None, behavior_parameters=None, behaviors_dir=None,
on_request=None, on_response=None, on_screenshot=None,
username=None, password=None, hashtags=None,
skip_extract_outlinks=False, skip_visit_hashtags=False,
@ -397,6 +397,8 @@ class Browser:
supplied (default None)
behavior_parameters: dict of parameters for populating the
javascript behavior template (default None)
behaviors_dir: Directory containing behaviors.yaml and JS templates
(default None loads Brozzler default JS behaviors)
on_request: callback to invoke on every Network.requestWillBeSent
event, takes one argument, the json-decoded message (default
None)
@ -447,7 +449,8 @@ class Browser:
jpeg_bytes = self.screenshot()
on_screenshot(jpeg_bytes)
behavior_script = brozzler.behavior_script(
page_url, behavior_parameters)
page_url, behavior_parameters,
behaviors_dir=behaviors_dir)
self.run_behavior(behavior_script, timeout=behavior_timeout)
if skip_extract_outlinks:
outlinks = []