#!/usr/bin/env python
"""
This small script takes the URL to one of your Pownce Export XML-files 
and downloads it including (as far as I can tell) all your files into a 
folder, which you can specify with the -o option::
    
    $ python pownce-files.py <url_to_export>
 
Requirements: Python >= 2.5, < 3.0
"""
 
from __future__ import with_statement
from xml.etree import ElementTree as ET
import os.path, os, logging, urllib2, sys, optparse
 
def main(folder, export_url):
    in_note = False
    in_file = False
    current_url = None
    current_filename = None
 
    if not os.path.isdir(folder):
        logging.info("Storage folder doesn't exist yet. Creating it now")
        os.makedirs(folder)
 
    fp = urllib2.urlopen(export_url)
    export_file = os.path.join(folder, 'export.xml')
    with open(export_file, 'w+') as file_:
        while True:
            data = fp.read(500000)
            if data is "":
                break
            file_.write(data)
    fp.close()
    with open(export_file, 'r') as fp:
        for event, elem in ET.iterparse(fp, events=('start', 'end')):
            if event == 'start' and elem.tag == 'note':
                in_note = True
            if event == 'end' and elem.tag == 'note':
                in_note = False
                elem.clear()
            if in_note and event == 'start' and elem.tag == 'file':
                in_file = True
            if in_note and event == 'end' and elem.tag == 'file':
                in_file = False
                fpath = os.path.join(folder, current_filename)
                logging.info("Downloading %s into %s" % (current_filename,folder,))
                if os.path.exists(fpath):
                    logging.info("%s already downloaded" % (current_filename,))
                    continue
                input_ = urllib2.urlopen(current_url)
                out = open(fpath, 'wb+')
                try:
                    while True:
                        data = input_.read(500000)
                        if data is "":
                            break
                        out.write(data)
                    out.close()
                    input_.close()
                except:
                    out.close()
                    input_.close()
                    os.unlink(fpath)            
            if in_file and event == 'end' and elem.tag == 'storage_name':
                current_filename = elem.text
            if in_file and event == 'end' and elem.tag == 'url':
                current_url = elem.text
 
if __name__ == '__main__':
    opts = optparse.OptionParser()
    opts.add_option('-d', '--debug', action='store_true', dest='debug',
        default=False, help='Enable debug output')
    opts.add_option('-o', '--output-directory', action='store', 
        dest='directory', default='pownce-files', help='Directory used for all downloaded content')
    options, args = opts.parse_args()
    log_level = options.debug and logging.DEBUG or logging.INFO
    logging.basicConfig(level=log_level)
 
    if not len(args):
        logging.error("You have to specify at least one argument for the URL of the export file")
        sys.exit(1)
    input_url = args[0]
 
    main(options.directory, input_url)