tsr/planet
1
0
Fork 0
Code Issues 3 Pull Requests Releases Wiki Activity

Reorder checks so that feeds with no dates aren't automatically flagged as

Browse Source 
inactive.
This commit is contained in:
Sam Ruby 2006-09-23 16:33:53 -04:00
parent b1dd3a33af
commit 4b0cd8d5d9
1 changed files with 26 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

52
planet/spider.py
View File

@ -170,32 +170,6 @@ def spiderFeed(feed):
for name, value in config.feed_options(feed).items(): for name, value in config.feed_options(feed).items():
data.feed['planet_'+name] = value data.feed['planet_'+name] = value
# identify inactive feeds
if config.activity_threshold(feed):
activity_horizon = \
time.gmtime(time.time()-86400*config.activity_threshold(feed))
updated = [entry.updated_parsed for entry in data.entries
if entry.has_key('updated_parsed')]
updated.sort()
if not updated or updated[-1] < activity_horizon:
msg = "no activity in %d days" % config.activity_threshold(feed)
log.info(msg)
data.feed['planet_message'] = msg
# report channel level errors
if data.status == 403:
data.feed['planet_message'] = "403: forbidden"
elif data.status == 404:
data.feed['planet_message'] = "404: not found"
elif data.status == 408:
data.feed['planet_message'] = "408: request timeout"
elif data.status == 410:
data.feed['planet_message'] = "410: gone"
elif data.status == 500:
data.feed['planet_message'] = "internal server error"
elif data.status >= 400:
data.feed['planet_message'] = "http status %s" % data.status
# perform user configured scrub operations on the data # perform user configured scrub operations on the data
scrub(feed, data) scrub(feed, data)
@ -237,6 +211,32 @@ def spiderFeed(feed):
write(output, cache_file) write(output, cache_file)
os.utime(cache_file, (mtime, mtime)) os.utime(cache_file, (mtime, mtime))
# identify inactive feeds
if config.activity_threshold(feed):
activity_horizon = \
time.gmtime(time.time()-86400*config.activity_threshold(feed))
updated = [entry.updated_parsed for entry in data.entries
if entry.has_key('updated_parsed')]
updated.sort()
if not updated or updated[-1] < activity_horizon:
msg = "no activity in %d days" % config.activity_threshold(feed)
log.info(msg)
data.feed['planet_message'] = msg
# report channel level errors
if data.status == 403:
data.feed['planet_message'] = "403: forbidden"
elif data.status == 404:
data.feed['planet_message'] = "404: not found"
elif data.status == 408:
data.feed['planet_message'] = "408: request timeout"
elif data.status == 410:
data.feed['planet_message'] = "410: gone"
elif data.status == 500:
data.feed['planet_message'] = "internal server error"
elif data.status >= 400:
data.feed['planet_message'] = "http status %s" % data.status
# write the feed info to the cache # write the feed info to the cache
if not os.path.exists(sources): os.makedirs(sources) if not os.path.exists(sources): os.makedirs(sources)
xdoc=minidom.parseString('''<feed xmlns:planet="%s" xdoc=minidom.parseString('''<feed xmlns:planet="%s"