apps/catalogue/management/commands/import_wl.py

   1 # -*- coding: utf-8 -*-
   2
   3 import json
   4 from optparse import make_option
   5 import urllib2
   6
   7 from django.core.management.base import BaseCommand
   8 from django.core.management.color import color_style
   9 from django.db import transaction
  10 from librarian.dcparser import BookInfo
  11 from librarian import ParseError, ValidationError
  12
  13 from catalogue.models import Book
  14
  15
  16 WL_API = 'http://www.wolnelektury.pl/api/books/'
  17
  18
  19 class Command(BaseCommand):
  20     option_list = BaseCommand.option_list + (
  21         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  22             help='Less output'),
  23     )
  24     help = 'Imports XML files from WL.'
  25
  26     def handle(self, *args, **options):
  27
  28         self.style = color_style()
  29
  30         verbose = options.get('verbose')
  31
  32         # Start transaction management.
  33         transaction.commit_unless_managed()
  34         transaction.enter_transaction_management()
  35         transaction.managed(True)
  36
  37         if verbose:
  38             print 'Reading currently managed files.'
  39         slugs = {}
  40         for b in Book.objects.all():
  41             if verbose:
  42                 print b.slug
  43             text = b.materialize().encode('utf-8')
  44             try:
  45                 info = BookInfo.from_string(text)
  46             except (ParseError, ValidationError):
  47                 pass
  48             else:
  49                 slugs[info.slug] = b
  50
  51         book_count = 0
  52         commit_args = {
  53             "author_name": 'Platforma',
  54             "description": 'Import from WL',
  55         }
  56
  57         if verbose:
  58             print 'Opening books list'
  59         for book in json.load(urllib2.urlopen(WL_API)):
  60             book_detail = json.load(urllib2.urlopen(book['href']))
  61             xml_text = urllib2.urlopen(book_detail['xml']).read()
  62             info = BookInfo.from_string(xml_text)
  63             previous_book = slugs.get(info.slug, None)
  64             if previous_book:
  65                 comm = previous_book.slug
  66             else:
  67                 comm = '*'
  68             print book_count, info.slug , '-->', comm
  69             Book.import_xml_text(xml_text, title=info.title,
  70                 slug=info.slug, previous_book=slugs.get(info.slug, None))
  71             book_count += 1
  72
  73         # Print results
  74         print
  75         print "Results:"
  76         print "Imported %d books from WL:" % (
  77                 book_count, )
  78         print
  79
  80
  81         transaction.commit()
  82         transaction.leave_transaction_management()
  83