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))
def add(self, guid, content=None, source=None):
print >>sys.stderr, "Adding '%s' (%s)" % (title, url)
try:
doc = Document.byGuid(guid)
except:
doc = Document(guid=guid, doc_type=self.doc_type)
Link(source=source, target=doc)
if content and not doc.data:
data, tags, vector = self.process(content)
doc.data = data
doc.vector = vector
doc.tags = tags
doc.sync()
return doc
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)
