apps/catalogue/management/commands/merge_books.py

   1 # -*- coding: utf-8 -*-
   2
   3 from optparse import make_option
   4 import sys
   5
   6 from django.contrib.auth.models import User
   7 from django.core.management.base import BaseCommand
   8 from django.core.management.color import color_style
   9 from django.db import transaction
  10
  11 from catalogue.models import Book
  12
  13
  14 def common_prefix(texts):
  15     common = []
  16
  17     min_len = min(len(text) for text in texts)
  18     for i in range(min_len):
  19         chars = list(set([text[i] for text in texts]))
  20         if len(chars) > 1:
  21             break
  22         common.append(chars[0])
  23     return "".join(common)
  24
  25
  26 class Command(BaseCommand):
  27     option_list = BaseCommand.option_list + (
  28         make_option('-s', '--slug', dest='new_slug', metavar='SLUG',
  29             help='New slug of the merged book (defaults to common part of all slugs).'),
  30         make_option('-t', '--title', dest='new_title', metavar='TITLE',
  31             help='New title of the merged book (defaults to common part of all titles).'),
  32         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  33             help='Less output'),
  34         make_option('-g', '--guess', action='store_true', dest='guess', default=False,
  35             help='Try to guess what merges are needed (but do not apply them).'),
  36         make_option('-d', '--dry-run', action='store_true', dest='dry_run', default=False,
  37             help='Dry run: do not actually change anything.'),
  38         make_option('-f', '--force', action='store_true', dest='force', default=False,
  39             help='On slug conflict, hide the original book to archive.'),
  40     )
  41     help = 'Merges multiple books into one.'
  42     args = '[slug]...'
  43
  44
  45     def print_guess(self, dry_run=True, force=False):
  46         from collections import defaultdict
  47         from pipes import quote
  48         import re
  49
  50         def read_slug(slug):
  51             res = []
  52             res.append((re.compile(ur'__?(przedmowa)$'), -1))
  53             res.append((re.compile(ur'__?(cz(esc)?|ksiega|rozdzial)__?(?P<n>\d*)$'), None))
  54             res.append((re.compile(ur'__?(rozdzialy__?)?(?P<n>\d*)-'), None))
  55
  56             for r, default in res:
  57                 m = r.search(slug)
  58                 if m:
  59                     start = m.start()
  60                     try:
  61                         return int(m.group('n')), slug[:start]
  62                     except IndexError:
  63                         return default, slug[:start]
  64             return None, slug
  65
  66         def file_to_title(fname):
  67             """ Returns a title-like version of a filename. """
  68             parts = (p.replace('_', ' ').title() for p in fname.split('__'))
  69             return ' / '.join(parts)
  70
  71         merges = defaultdict(list)
  72         slugs = []
  73         for b in Book.objects.all():
  74             slugs.append(b.slug)
  75             n, ns = read_slug(b.slug)
  76             if n is not None:
  77                 merges[ns].append((n, b))
  78
  79         conflicting_slugs = []
  80         for slug in sorted(merges.keys()):
  81             merge_list = sorted(merges[slug])
  82             if len(merge_list) < 2:
  83                 continue
  84
  85             merge_slugs = [b.slug for i, b in merge_list]
  86             if slug in slugs and slug not in merge_slugs:
  87                 conflicting_slugs.append(slug)
  88
  89             title = file_to_title(slug)
  90             print "./manage.py merge_books %s%s--title=%s --slug=%s \\\n    %s\n" % (
  91                 '--dry-run ' if dry_run else '',
  92                 '--force ' if force else '',
  93                 quote(title), slug,
  94                 " \\\n    ".join(merge_slugs)
  95                 )
  96
  97         if conflicting_slugs:
  98             if force:
  99                 print self.style.NOTICE('# These books will be archived:')
 100             else:
 101                 print self.style.ERROR('# ERROR: Conflicting slugs:')
 102             for slug in conflicting_slugs:
 103                 print '#', slug
 104
 105
 106     def handle(self, *slugs, **options):
 107
 108         self.style = color_style()
 109
 110         force = options.get('force')
 111         guess = options.get('guess')
 112         dry_run = options.get('dry_run')
 113         new_slug = options.get('new_slug').decode('utf-8')
 114         new_title = options.get('new_title').decode('utf-8')
 115         verbose = options.get('verbose')
 116
 117         if guess:
 118             if slugs:
 119                 print "Please specify either slugs, or --guess."
 120                 return
 121             else:
 122                 self.print_guess(dry_run, force)
 123                 return
 124         if not slugs:
 125             print "Please specify some book slugs"
 126             return
 127
 128         # Start transaction management.
 129         transaction.commit_unless_managed()
 130         transaction.enter_transaction_management()
 131         transaction.managed(True)
 132
 133         books = [Book.objects.get(slug=slug) for slug in slugs]
 134         common_slug = common_prefix(slugs)
 135         common_title = common_prefix([b.title for b in books])
 136
 137         if not new_title:
 138             new_title = common_title
 139         elif common_title.startswith(new_title):
 140             common_title = new_title
 141
 142         if not new_slug:
 143             new_slug = common_slug
 144         elif common_slug.startswith(new_slug):
 145             common_slug = new_slug
 146
 147         if slugs[0] != new_slug and Book.objects.filter(slug=new_slug).exists():
 148             self.style.ERROR('Book already exists, skipping!')
 149
 150
 151         if dry_run and verbose:
 152             print self.style.NOTICE('DRY RUN: nothing will be changed.')
 153             print
 154
 155         if verbose:
 156             print "New title:", self.style.NOTICE(new_title)
 157             print "New slug:", self.style.NOTICE(new_slug)
 158             print
 159
 160         for i, book in enumerate(books):
 161             chunk_titles = []
 162             chunk_slugs = []
 163
 164             book_title = book.title[len(common_title):].replace(' / ', ' ').lstrip()
 165             book_slug = book.slug[len(common_slug):].replace('__', '_').lstrip('-_')
 166             for j, chunk in enumerate(book):
 167                 if j:
 168                     new_chunk_title = book_title + '_%d' % j
 169                     new_chunk_slug = book_slug + '_%d' % j
 170                 else:
 171                     new_chunk_title, new_chunk_slug = book_title, book_slug
 172
 173                 chunk_titles.append(new_chunk_title)
 174                 chunk_slugs.append(new_chunk_slug)
 175
 176                 if verbose:
 177                     print "title: %s // %s  -->\n       %s // %s\nslug: %s / %s  -->\n      %s / %s" % (
 178                         book.title, chunk.title,
 179                         new_title, new_chunk_title,
 180                         book.slug, chunk.slug,
 181                         new_slug, new_chunk_slug)
 182                     print
 183
 184             if not dry_run:
 185                 try:
 186                     conflict = Book.objects.get(slug=new_slug)
 187                 except Book.DoesNotExist:
 188                     conflict = None
 189                 else:
 190                     if conflict == books[0]:
 191                         conflict = None
 192
 193                 if conflict:
 194                     if force:
 195                         # FIXME: there still may be a conflict
 196                         conflict.slug = '.' + conflict.slug
 197                         conflict.save()
 198                         print self.style.NOTICE('Book with slug "%s" moved to "%s".' % (new_slug, conflict.slug))
 199                     else:
 200                         print self.style.ERROR('ERROR: Book with slug "%s" exists.' % new_slug)
 201                         return
 202
 203                 if i:
 204                     books[0].append(books[i], slugs=chunk_slugs, titles=chunk_titles)
 205                 else:
 206                     book.title = new_title
 207                     book.slug = new_slug
 208                     book.save()
 209                     for j, chunk in enumerate(book):
 210                         chunk.title = chunk_titles[j]
 211                         chunk.slug = chunk_slugs[j]
 212                         chunk.save()
 213
 214
 215         transaction.commit()
 216         transaction.leave_transaction_management()
 217