* master:
avoid "Uncaught TypeError: Cannot read property 'querySelectorAll' of undefined" from outlinks script
little readme fix
for vagrant, static ansible inventory file, add brozzler-webconsole
add info to display of jobless sites in brozzler-webconsole; fix creation of "least_hops" index on the rethinkdb table "pages"
add arguments --webconsole-address --webconsole-port --pywb-address and change default ports
list jobless sites on brozzler-webconsole front page
run brozzler-webconsole inside brozzler-easy
add section about brozzler-easy to the readme
add --help to brozzler-webconsole
* master:
fix exception happening now that we have binary data in rethinkdb (the cookie db) "TypeError: <binary, 7168 bytes, '53 51 4c 69 74 65...'> is not JSON serializable"
dev version number again
another version for pypi
avoid "Uncaught RangeError: Maximum call stack size exceeded" compiling outlinks
back to a dev version number
bump version to 1.1b4 for pypi upload
logging tweak
install brozzler.webconsole package
* master:
add docstring to _chain_chrome_messages, remove debug logging, tweak name of websock thread
add a timeout to the one post-behavior step that didn't already have one (getting a screenshot), and majorly refactored the post-behavior code to incorporate timeouts automatically into each step, and hopefully make it easier to follow
logging tweaks
reduce log level of messages from chrome, since it spews stuff that looks bad but usually isn't
back to a dev version number
1.1b3 for upload to pypi
* master:
pass behavior template parameters on to behavior - fixes umbra's ability to log in with parameters received from amqp
Changing EnvironmentError to OSError
Fix naming conventions.
Create cookie directory if it doesn't exist. Add debug messages for cookie db read/write.
Read/Write Cookie DB file when creating and stopping browser instance.
brozzler[easy] requires warcprox>=2.0b1
look for a sensible default chromium/chrome executable
tweak thread names
convert domain specific rule url prefixes to our style of surt
have pywb support loading warc records from warc files still being written (look for foo.warc.gz.open)
install flash plugin for chromium
make state dumping signal handler more robust (now you can kill -QUIT a thousand times in a row without causing problems)
handle case where websocket connection is unexpectedly closed during the post-behavior phase
implement timeout and retries to work around issue where sometimes we receive no result message after requesting outlinks
forgot to commit easy.py, add pywb.py with support for pywb rethinkdb index, and make brozzler-easy also run pywb
working on brozzler-easy, single process with brozzler-worker and warcprox working together (pywb to be added)
twirldown for site yaml on site page
give master a version number considered later than the one up on pypi (1.1b3.dev45 > 1.1b2)
in vagrant/ansible, install brozzler from this checkout instead of from github master
option to save list of outlinks (categorized as "accepted", "blocked" (by robots), or "rejected") per page in rethinkdb (to be used by archive-it for out-of-scope reporting)
oops didn't mean to leave that windows-only subprocess flag
remove accidentally committed playbook.retry
vagrant setup (unfinished)
do not send more than one SIGTERM when shutting down browser process, because on recent chromium on linux, the second sigterm abruptly ends the process, and sometimes leaves orphan subprocesses; also send TERM/KILL signals to the whole process group, another measure to avoid orphans; and adjust logging levels for captured chrome output
command line utility brozzler-ensure-tables, creates rethinkdb tables if they don't already exist... brozzler normally creates them on demand at startup, but if multiple instances are starting up at the same time, you can end up with duplicate broken tables, so it's a good idea to use this utility when spinning up a cluster
