Use proper logging from cronjobs

This makes it reasonable to redirect it to a file (it now has timestamps),
and also adds the -debug parameter to get more logging when actually
testing something out

2 files changed, 28 insertions, 3 deletions

diff --git a/tools/commitfest/check_patches_in_archives.py b/tools/commitfest/check_patches_in_archives.py
index 2429e27..ddbcb0a 100755
--- a/tools/commitfest/check_patches_in_archives.py
+++ b/tools/commitfest/check_patches_in_archives.py
@@ -12,6 +12,7 @@ import sys
 import socket
 import httplib
 import magic
+import logging
 
 # Set up for accessing django
 from django.core.management import setup_environ
@@ -24,10 +25,18 @@ from django.db import connection
 from commitfest.models import MailThreadAttachment
 
 if __name__ == "__main__":
+	debug = "--debug" in sys.argv
+
+	# Logging always done to stdout, but we can turn on/off how much
+	logging.basicConfig(format='%(asctime)s %(levelname)s: %(msg)s',
+						level=debug and logging.DEBUG or logging.INFO)
+
 	socket.setdefaulttimeout(settings.ARCHIVES_TIMEOUT)
 	mag = magic.open(magic.MIME)
 	mag.load()
-	
+
+	logging.debug("Updating attachment metadata from archives")
+
 	# Try to fetch/scan all attachments that haven't already been scanned.
 	# If they have already been scanned, we don't bother.
 	# We will hit the archives without delay when doing this, but that
@@ -35,6 +44,8 @@ if __name__ == "__main__":
 	# downloading a lot...
 	for a in MailThreadAttachment.objects.filter(ispatch=None):
 		url = "/message-id/attachment/%s/attach" % a.attachmentid
+		logging.debug("Checking attachment %s" % a.attachmentid)
+
 		h = httplib.HTTPConnection(settings.ARCHIVES_SERVER,
 								   settings.ARCHIVES_PORT,
 								   True,
@@ -44,7 +55,7 @@ if __name__ == "__main__":
 			})
 		resp = h.getresponse()
 		if resp.status != 200:
-			print "Failed to get %s: %s" % (url, resp.status)
+			logging.error("Failed to get %s: %s" % (url, resp.status))
 			continue
 
 		contents = resp.read()
@@ -53,6 +64,7 @@ if __name__ == "__main__":
 
 		# Attempt to identify the file using magic information
 		mtype = mag.buffer(contents)
+		logging.debug("Detected MIME type is %s" % mtype)
 
 		# We don't support gzipped or tar:ed patches or anything like
 		# that at this point - just plain patches.
@@ -60,6 +72,8 @@ if __name__ == "__main__":
 			a.ispatch = True
 		else:
 			a.ispatch = False
+		logging.info("Attachment %s is patch: %s" % (a.id, a.ispatch))
 		a.save()
 
 	connection.close()
+	logging.debug("Done.")
diff --git a/tools/commitfest/update_archive_threads.py b/tools/commitfest/update_archive_threads.py
index ace46f4..bc42085 100755
--- a/tools/commitfest/update_archive_threads.py
+++ b/tools/commitfest/update_archive_threads.py
@@ -8,6 +8,7 @@
 
 import os
 import sys
+import logging
 
 # Set up for accessing django
 from django.core.management import setup_environ
@@ -21,11 +22,19 @@ from commitfest.models import MailThread
 from commitfest.ajax import _archivesAPI, parse_and_add_attachments
 
 if __name__ == "__main__":
+	debug = "--debug" in sys.argv
+
+	# Logging always done to stdout, but we can turn on/off how much
+	logging.basicConfig(format='%(asctime)s %(levelname)s: %(msg)s',
+						level=debug and logging.DEBUG or logging.INFO)
+
+	logging.debug("Checking for updated mail threads in the archives")
 	for thread in MailThread.objects.filter(patches__commitfests__status__in=(1,2,3)).distinct():
+		logging.debug("Checking %s in the archives" % thread.messageid)
 		r = sorted(_archivesAPI('/message-id.json/%s' % thread.messageid), key=lambda x: x['date'])
 		if thread.latestmsgid != r[-1]['msgid']:
 			# There is now a newer mail in the thread!
-			print "Thread %s updated" % thread.messageid
+			logging.info("Thread %s updated" % thread.messageid)
 			thread.latestmsgid = r[-1]['msgid']
 			thread.latestmessage = r[-1]['date']
 			thread.latestauthor = r[-1]['from']
@@ -35,7 +44,9 @@ if __name__ == "__main__":
 			# Potentially update the last mail date - if there wasn't already a mail on each patch
 			# from a *different* thread that had an earlier date.
 			for p in thread.patches.filter(lastmail__lt=thread.latestmessage):
+				logging.debug("Last mail time updated for %s" % thread.messageid)
 				p.lastmail = thread.latestmessage
 				p.save()
 
 	connection.close()
+	logging.debug("Done.")