def parse_file(filename):
print u"*** Parse %s file ***" % filename
reader = Sax2.Reader()
fh = open(filename, "r")
s = fh.read()
fh.close()
d = reader.fromString(s)
type = "jabber" if filename.find("JabberProtocol") != -1 else "icq"
head = d.getElementsByTagName("head")[0]
date_xml = head.getElementsByTagName("date")[0]
my_from = ""
with_account = None
for c in head.getElementsByTagName("contact"):
if c.hasAttribute("type"):
if c.getAttribute("type") == "myself":
my_from = c.getAttribute("contactId")
else:
with_account = c.getAttribute("contactId")
year = int(date_xml.getAttribute("year"))
month = int(date_xml.getAttribute("month"))
chat = Chat(account=my_from, type=type, with_account=with_account)
chat.save()
for msg_xml in d.getElementsByTagName("msg"):
day_time = msg_xml.getAttribute("time").split()
day = int(day_time[0])
time = day_time[1].split(":")
hour = int(time[0])
minutes = int(time[1])
seconds = int(time[2])
from_user = msg_xml.getAttribute("from")
from_nick = msg_xml.getAttribute("nick")
myself = True if from_user == my_from else False
date = datetime(year=year, month=month, day=day, hour=hour, minute=minutes, second=seconds)
if msg_xml.hasChildNodes():
text = msg_xml.childNodes[0].nodeValue
else:
test = u""
msg = Message(date=date, text=text, from_user=from_user, from_nick=from_nick, myself=myself, chat=chat)
msg.save()
print msg
def split_chats():
""" Split chats by day """
for c in list(Chat.objects.all()):
msg_qs = Message.objects.filter(chat=c)
stats = msg_qs.aggregate(Max("date"), Min("date"))
if not stats["date__min"] or not stats["date__max"]:
continue
d = stats["date__min"]
d = d.replace(microsecond=0, second=0, hour=0, minute=0)
first = True
while d < stats["date__max"]:
d2 = d + timedelta(days=1)
if first:
first = False
else:
new_chat = Chat(account=c.account, type=c.type, with_account=c.with_account)
new_chat.save()
msg_qs.filter(date__gte=d, date__lte=d2).update(chat=new_chat)
d = d2
