Git-Mirrors/brozzler
1
0
Fork 0
mirror of https://github.com/internetarchive/brozzler.git synced 2025-08-08 22:42:32 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

fancier prioritization takes into account hops from seed, path depth; and clean shutdown

Browse source
This commit is contained in:
Noah Levitt 2015-07-09 22:35:37 -07:00
parent 5f3c247e0c
commit fcc63b6675
2 changed files with 69 additions and 12 deletions
Download patch file Download diff file Expand all files Collapse all files

5
umbra/browser.py
View file

@ -53,6 +53,9 @@ class BrowserPool:
class BrowsingException(Exception):
pass
class BrowsingAborted(BrowsingException):
pass
class Browser:
"""Runs chrome/chromium to synchronously browse one page at a time using
worker.browse_page(). Currently the implementation starts up a new instance
@ -162,7 +165,7 @@ class Browser:
self.logger.info("finished browsing page, reached hard timeout of {} seconds url={}".format(Browser.HARD_TIMEOUT_SECONDS, self.url))
return True
elif self._abort_browse_page:
raise BrowsingException("browsing page aborted")
raise BrowsingAborted("browsing page aborted")
def send_to_chrome(self, suppress_logging=False, **kwargs):
msg_id = next(self.command_id)