Convert Apple webloc files (such as DEVONthink bookmarks) to org-mode entries and preserve filetags and timestamps

webloc2orgmode.py

"""
webloc2orgmode.py

This script will take Apple webloc files (such as bookmarks made in DEVONthink) and export them as org-mode entries.
It will also save the timestamp and tag information for each webloc

Usage:
$ pip3 install lxml biplist xattr bs4
$ touch ~/bookmarks.org
$ mkdir done
$ mkdir weblocs
(move all webloc files to the weblocs directory)
$ python3 webloc2orgmode.py

All parsed webloc's will have moved to the done/ directory
"""

import biplist
import glob
import lxml.html
import os
import requests
import shutil
import time
import xattr
from bs4 import BeautifulSoup
from lxml import etree
from struct import unpack
from time import sleep

for filename in glob.glob('bookmarks/*.webloc'):
   tree = etree.parse(filename).getroot()
   dict = tree.getchildren()[0]
   url = dict.getchildren()[1].text

   tags = []

   try:
     tagsPlist = xattr.getxattr(filename, 'com.apple.metadata:_kMDItemUserTags')
     tags = biplist.readPlistFromString(tagsPlist)
   except:
     print("tagerror")

   formattedTags = []

   if len(tags) > 0:
     for tag in tags:
       formattedTags.append(tag.replace('\n0', ''))

   stat = os.stat(filename)
   date = time.strftime('[%Y-%m-%d %a %H:%M:%S %z]', time.localtime(stat.st_birthtime))

   title = 'Undefined'

   try:
      request = requests.get(url)
      soup = BeautifulSoup(request.content, "lxml")

      if soup.title is not None:
         title = soup.title.string

   except:
     print("url error")

   line1 = ''

   if title and len(formattedTags) > 0:
     line1 = "* [[%s][%s]] :%s:" % (url, title, ':'.join(formattedTags))
   else:
     line1 = "* [[%s][%s]]" % (url, title)


   properties = ":PROPERTIES:\n:CREATED: %s\n:END:" % date

   homedir = os.path.expanduser('~')
   print(line1, file=open(homedir + "/bookmarks.org", "a"))
   print(properties, file=open(homedir + "/bookmarks.org", "a"))
   shutil.move(filename, "done/")