apps/catalogue/management/commands/import_wl.py

   1 # -*- coding: utf-8 -*-
   2
   3 from collections import defaultdict
   4 import json
   5 from optparse import make_option
   6 import urllib2
   7
   8 from django.core.management.base import BaseCommand
   9 from django.core.management.color import color_style
  10 from django.db import transaction
  11 from librarian.dcparser import BookInfo
  12 from librarian import ParseError, ValidationError
  13
  14 from catalogue.models import Book
  15
  16
  17 WL_API = 'http://www.wolnelektury.pl/api/books/'
  18
  19
  20 class Command(BaseCommand):
  21     option_list = BaseCommand.option_list + (
  22         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  23             help='Less output'),
  24     )
  25     help = 'Imports XML files from WL.'
  26
  27     def handle(self, *args, **options):
  28
  29         self.style = color_style()
  30
  31         verbose = options.get('verbose')
  32
  33         # Start transaction management.
  34         transaction.commit_unless_managed()
  35         transaction.enter_transaction_management()
  36         transaction.managed(True)
  37
  38         if verbose:
  39             print 'Reading currently managed files (skipping hidden ones).'
  40         slugs = defaultdict(list)
  41         for b in Book.objects.exclude(slug__startswith='.').all():
  42             if verbose:
  43                 print b.slug
  44             text = b.materialize().encode('utf-8')
  45             try:
  46                 info = BookInfo.from_string(text)
  47             except (ParseError, ValidationError):
  48                 pass
  49             else:
  50                 slugs[info.slug].append(b)
  51
  52         #~ conflicts = []
  53         #~ for slug, book_list in slugs.items():
  54             #~ if len(book_list) > 1:
  55                 #~ conflicts.append((slug, book_list))
  56         #~ if conflicts:
  57             #~ print self.style.ERROR("There is more than one book "
  58                     #~ "with the same slug in dc:url. "
  59                     #~ "Merge or hide them before proceeding.")
  60             #~ for slug, book_list in sorted(conflicts):
  61                 #~ print slug
  62                 #~ print "\n".join(b.slug for b in book_list)
  63                 #~ print
  64             #~ return
  65
  66         book_count = 0
  67         commit_args = {
  68             "author_name": 'Platforma',
  69             "description": 'Automatycznie zaimportowane z Wolnych Lektur',
  70             "publishable": True,
  71         }
  72
  73         if verbose:
  74             print 'Opening books list'
  75         for book in json.load(urllib2.urlopen(WL_API))[:10]:
  76             book_detail = json.load(urllib2.urlopen(book['href']))
  77             xml_text = urllib2.urlopen(book_detail['xml']).read()
  78             info = BookInfo.from_string(xml_text)
  79             previous_books = slugs.get(info.slug)
  80             if previous_books:
  81                 if len(previous_books) > 1:
  82                     print self.style.ERROR("There is more than one book "
  83                         "with slug %s:"),
  84                 previous_book = previous_books[0]
  85                 comm = previous_book.slug
  86             else:
  87                 previous_book = None
  88                 comm = '*'
  89             print book_count, info.slug , '-->', comm
  90             Book.import_xml_text(xml_text, title=info.title,
  91                 slug=info.slug, previous_book=previous_book,
  92                 commit_args=commit_args)
  93             book_count += 1
  94
  95         # Print results
  96         print
  97         print "Results:"
  98         print "Imported %d books from WL:" % (
  99                 book_count, )
 100         print
 101
 102
 103         transaction.commit()
 104         transaction.leave_transaction_management()
 105