def stop_running_job(frequency, heritrix):
"""Stops a running job, notifies RabbitMQ and cleans up the directory."""
launchid = heritrix.launchid(frequency)
message = "%s/%s" % (frequency, launchid)
job = W3actJob.from_directory("%s/%s" % (settings.HERITRIX_JOBS, frequency), heritrix=heritrix)
job.stop()
logger.info("Sending SIP message: %s" % message)
send_message(settings.QUEUE_HOST, settings.SIP_QUEUE_NAME, settings.SIP_QUEUE_KEY, message)
logger.info("Sending QA message: %s" % message)
send_message(settings.QUEUE_HOST, settings.QA_QUEUE_NAME, settings.QA_QUEUE_KEY, message)
remove_action_files(frequency)
if settings.SLACK:
stats = generate_log_stats(glob("%s/%s/%s/crawl.log*" % (settings.HERITRIX_LOGS, frequency, launchid)))
send_slack_messages(stats, frequency)
def check_watched_targets(jobname, heritrix):
"""If there are any Watched Targets, send a message."""
timestamp = heritrix.launchid(jobname)
if not os.path.exists("%s/%s/%s/w3act-info.json" % (settings.HERITRIX_JOBS, jobname, timestamp)):
return
with open("%s/%s/%s/w3act-info.json" % (settings.HERITRIX_JOBS, jobname, timestamp), "rb") as i:
info = i.read()
for job in json.loads(info):
if job["watched"]:
logger.info("Found a Watched Target in %s/%s." % (jobname, timestamp))
send_message(
settings.QUEUE_HOST,
settings.WATCHED_QUEUE_NAME,
settings.WATCHED_QUEUE_KEY,
"%s/%s" % (jobname, timestamp),
)