def schedule (self):
self.refresh = datetime.now() + timedelta(seconds=Loader.Reload_Min)
self.queue = []
for doc in Document.select(Document.q.interval > 0):
if doc.fetched_at:
event = doc.fetched_at + timedelta(seconds=doc.interval)
else:
event = datetime.now()
loader = self.loader[doc.doc_type]
heappush(self.queue, (event, doc, loader.load))
log.msg("%d events scheduled." % len(self.queue))
from store import Document, sqlhub, connectionForURI
from vector import Vector, VectorSet
import sys, os
DB_URI = "sqlite:/tmp/memex.db"
sqlhub.processConnection = connectionForURI(DB_URI)
index = VectorSet()
for doc in Document.select():
if doc.vector:
index[doc.url] = Vector(map(int, doc.vector.split(",")))
for dist, url in index.nearest(index[sys.argv[1]]):
print "%.3f\t%s" % (dist, url)
