bookie-python/xml_import.py

#!/usr/bin/env python3.4

import os, re
from flask.ext.mongoengine import MongoEngine
import datetime
import base64
from bs4 import BeautifulSoup as BS
from xml.sax.saxutils import unescape


from bookie import *

app.config["MONGODB_DB"] = "bookie"
db = MongoEngine(app)

date_marker = datetime.datetime.now()
count = 0

Bookmark.objects.all().delete()
Tag.objects.all().delete()
ArchivedText.objects.all().delete()
ArchivedImage.objects.all().delete()

with open("import.xml", "r") as importfile:
    data = importfile.read()

soup = BS(data)
count = 0
for i in soup(["bookmark"]):
    b = Bookmark()
    b.title = unescape(i.find("title").get_text())
    b.short = unescape(i.find("short").get_text())
    b.url = unescape(i.find("url").get_text())
    b.note = i.find("note").get_text()
    b.created_at = datetime.datetime(*map(int, re.split('[^\d]', unescape(i.find("created_at").get_text()))[:-1]))
    if i.find("unread").get_text() == "True":
        b.unread = True
    else:
        b.unread = False
    b.tags = []
    for t in i.find("tags").get_text().split(" "):
        tag = Tag.objects.get_or_create(name=t)[0].save()
        b.tags.append(tag)
    count += 1
    print(str(count)+" - "+b.short)
    print(b.title)
    print(b.url)
    b.factor = b.get_factor()
    b.save()