def add_sniffed(self, href, label, sniffed_from):
if not self.ignore(href):
uri = resource(href)
label = literal(label)
if not self.exists(uri, None, None):
print "ADDED:", href, label
self.add(uri, LABEL, label)
self.add(uri, TYPE, SNIFFED)
timestamp = literal(date_time())
self.add(uri, SNIFFED_ON, timestamp)
self.add(uri, SNIFFED_FROM, resource(sniffed_from))
def add_statement(self, subject, predicate, object, literal_object=None, anonymous_subject=None, anonymous_object=None):
s = resource(subject, anonymous_subject)
p = resource(predicate)
if literal_object:
o = literal(object)
else:
o = resource(object, anonymous_object)
self.add(s, p, o)
def __check(self, url):
every = int(self.get_first_value(url, SNIFF_EVERY, str(SECONDS_IN_DAY)))
# TODO: take the oldest value
last = self.get_first_value(url, SNIFF_LAST, None)
if not last or time() - parse_date_time(last) > every:
self.remove(url, SNIFF_LAST, None)
self.add(url, SNIFF_LAST, literal(date_time()))
self.sniff(url)
def __init__(self, host, username, password):
self.store = TripleStore()
M = poplib.POP3(host)
M.user(username)
M.pass_(password)
numMessages = len(M.list()[1])
self.messages = {}
for i in range(numMessages):
uidl = string.split(M.uidl(i+1))[2]
msg = M.retr(i+1)[1]
lineFile = LineFile(msg)
message = rfc822.Message(lineFile)
bodyStartLine = lineFile.count
headers = message.dict
body = string.join(msg[bodyStartLine:], "\n")
subject = resource(POP_URI + uidl)
for header in headers.keys():
predicate = resource(RFC_822_URI + header)
object = literal(headers[header])
self.store.add(subject, predicate, object)
self.store.add(subject, BODY, literal(body))
self.store.add(subject, TYPE, MESSAGE)
M.quit()
def get_triple(self, text):
parts = text.split(" ", 2)
subject = self.process_resource(parts[0])
if subject == -1:
return None # error
property = self.process_resource(parts[1])
if property == -1:
return None # error
if parts[2][0] == '"' and parts[2][-1] == '"':
value = literal(parts[2][1:-1])
else:
value = self.process_resource(parts[2])
if value == -1:
return None # error
return (subject, property, value)
