summaryrefslogtreecommitdiff
path: root/crocoite/task.py
diff options
context:
space:
mode:
authorLars-Dominik Braun <lars@6xq.net>2018-07-28 20:25:49 +0200
committerLars-Dominik Braun <lars@6xq.net>2018-08-04 14:11:31 +0200
commit3deded13df1339ef59a760c188804adffd9ed902 (patch)
tree5eaf69ee38389073e7323585c6afdbbf5eeab487 /crocoite/task.py
parent33a137f2d7c04468038d689b53a70fb534297f55 (diff)
downloadcrocoite-3deded13df1339ef59a760c188804adffd9ed902.tar.gz
crocoite-3deded13df1339ef59a760c188804adffd9ed902.tar.bz2
crocoite-3deded13df1339ef59a760c188804adffd9ed902.zip
Reintroduce WARC logging
Commit 7730e0d64ec895091a0dd7eb0e3c6ce2ed02d981 removed logging to WARC files. Add it again, but with a different implementation.. Credits to structlog for inspiration.
Diffstat (limited to 'crocoite/task.py')
-rw-r--r--crocoite/task.py9
1 files changed, 5 insertions, 4 deletions
diff --git a/crocoite/task.py b/crocoite/task.py
index 6b3c9d1..06dd022 100644
--- a/crocoite/task.py
+++ b/crocoite/task.py
@@ -22,7 +22,7 @@
Celery distributed tasks
"""
-import os, logging
+import os
from urllib.parse import urlsplit
from datetime import datetime
@@ -113,8 +113,8 @@ class DistributedRecursiveController (RecursiveController):
__slots__ = ('concurrency', 'stats')
- def __init__ (self, url, service=ChromeService (), behavior=behavior.available, \
- logger=logging.getLogger(__name__), settings=defaultSettings,
+ def __init__ (self, url, logger, service=ChromeService (), behavior=behavior.available, \
+ settings=defaultSettings,
recursionPolicy=DepthLimit (0), concurrency=1):
super ().__init__ (url, None, service, behavior, logger, settings, recursionPolicy)
self.concurrency = concurrency
@@ -134,10 +134,11 @@ class DistributedRecursiveController (RecursiveController):
def controller (self, url, settings, enabledBehaviorNames, recursive, concurrency):
""" Recursive controller """
+ logger = Logger (consumer=[DatetimeConsumer (), JsonPrintConsumer ()])
recursionPolicy = parseRecursive (recursive, url)
enabledBehavior = list (filter (lambda x: x.name in enabledBehaviorNames, behavior.available))
settings = ControllerSettings (**settings)
- c = DistributedRecursiveController (url, None, behavior=enabledBehavior,
+ c = DistributedRecursiveController (url, None, logger=logger, behavior=enabledBehavior,
settings=settings, recursionPolicy=recursionPolicy, concurrency=concurrency)
c.run ()
return dict (c.stats)