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 slughifi import slughifi
  12 from catalogue.models import Book
  13
  14
  15 def common_prefix(texts):
  16     common = []
  17
  18     min_len = min(len(text) for text in texts)
  19     for i in range(min_len):
  20         chars = list(set([text[i] for text in texts]))
  21         if len(chars) > 1:
  22             break
  23         common.append(chars[0])
  24     return "".join(common)
  25
  26
  27 class Command(BaseCommand):
  28     option_list = BaseCommand.option_list + (
  29         make_option('-s', '--slug', dest='new_slug', metavar='SLUG',
  30             help='New slug of the merged book (defaults to common part of all slugs).'),
  31         make_option('-t', '--title', dest='new_title', metavar='TITLE',
  32             help='New title of the merged book (defaults to common part of all titles).'),
  33         make_option('-q', '--quiet', action='store_false', dest='verbose', default=True,
  34             help='Less output'),
  35         make_option('-g', '--guess', action='store_true', dest='guess', default=False,
  36             help='Try to guess what merges are needed (but do not apply them).'),
  37         make_option('-d', '--dry-run', action='store_true', dest='dry_run', default=False,
  38             help='Dry run: do not actually change anything.'),
  39         make_option('-f', '--force', action='store_true', dest='force', default=False,
  40             help='On slug conflict, hide the original book to archive.'),
  41     )
  42     help = 'Merges multiple books into one.'
  43     args = '[slug]...'
  44
  45
  46     def print_guess(self, dry_run=True, force=False):
  47         from collections import defaultdict
  48         from pipes import quote
  49         import re
  50
  51         def read_slug(slug):
  52             res = []
  53             res.append((re.compile(ur'__?(przedmowa)$'), -1))
  54             res.append((re.compile(ur'__?(cz(esc)?|ksiega|rozdzial)__?(?P<n>\d*)$'), None))
  55             res.append((re.compile(ur'__?(rozdzialy__?)?(?P<n>\d*)-'), None))
  56
  57             for r, default in res:
  58                 m = r.search(slug)
  59                 if m:
  60                     start = m.start()
  61                     try:
  62                         return int(m.group('n')), slug[:start]
  63                     except IndexError:
  64                         return default, slug[:start]
  65             return None, slug
  66
  67         def file_to_title(fname):
  68             """ Returns a title-like version of a filename. """
  69             parts = (p.replace('_', ' ').title() for p in fname.split('__'))
  70             return ' / '.join(parts)
  71
  72         merges = defaultdict(list)
  73         slugs = []
  74         for b in Book.objects.all():
  75             slugs.append(b.slug)
  76             n, ns = read_slug(b.slug)
  77             if n is not None:
  78                 merges[ns].append((n, b))
  79
  80         conflicting_slugs = []
  81         for slug in sorted(merges.keys()):
  82             merge_list = sorted(merges[slug])
  83             if len(merge_list) < 2:
  84                 continue
  85
  86             merge_slugs = [b.slug for i, b in merge_list]
  87             if slug in slugs and slug not in merge_slugs:
  88                 conflicting_slugs.append(slug)
  89
  90             title = file_to_title(slug)
  91             print "./manage.py merge_books %s%s--title=%s --slug=%s \\\n    %s\n" % (
  92                 '--dry-run ' if dry_run else '',
  93                 '--force ' if force else '',
  94                 quote(title), slug,
  95                 " \\\n    ".join(merge_slugs)
  96                 )
  97
  98         if conflicting_slugs:
  99             if force:
 100                 print self.style.NOTICE('# These books will be archived:')
 101             else:
 102                 print self.style.ERROR('# ERROR: Conflicting slugs:')
 103             for slug in conflicting_slugs:
 104                 print '#', slug
 105
 106
 107     def handle(self, *slugs, **options):
 108
 109         self.style = color_style()
 110
 111         force = options.get('force')
 112         guess = options.get('guess')
 113         dry_run = options.get('dry_run')
 114         new_slug = options.get('new_slug').decode('utf-8')
 115         new_title = options.get('new_title').decode('utf-8')
 116         verbose = options.get('verbose')
 117
 118         if guess:
 119             if slugs:
 120                 print "Please specify either slugs, or --guess."
 121                 return
 122             else:
 123                 self.print_guess(dry_run, force)
 124                 return
 125         if not slugs:
 126             print "Please specify some book slugs"
 127             return
 128
 129         # Start transaction management.
 130         transaction.commit_unless_managed()
 131         transaction.enter_transaction_management()
 132         transaction.managed(True)
 133
 134         books = [Book.objects.get(slug=slug) for slug in slugs]
 135         common_slug = common_prefix(slugs)
 136         common_title = common_prefix([b.title for b in books])
 137
 138         if not new_title:
 139             new_title = common_title
 140         elif common_title.startswith(new_title):
 141             common_title = new_title
 142
 143         if not new_slug:
 144             new_slug = common_slug
 145         elif common_slug.startswith(new_slug):
 146             common_slug = new_slug
 147
 148         if slugs[0] != new_slug and Book.objects.filter(slug=new_slug).exists():
 149             self.style.ERROR('Book already exists, skipping!')
 150
 151
 152         if dry_run and verbose:
 153             print self.style.NOTICE('DRY RUN: nothing will be changed.')
 154             print
 155
 156         if verbose:
 157             print "New title:", self.style.NOTICE(new_title)
 158             print "New slug:", self.style.NOTICE(new_slug)
 159             print
 160
 161         for i, book in enumerate(books):
 162             chunk_titles = []
 163             chunk_slugs = []
 164
 165             book_title = book.title[len(common_title):].replace(' / ', ' ').lstrip()
 166             book_slug = book.slug[len(common_slug):].replace('__', '_').lstrip('-_')
 167             for j, chunk in enumerate(book):
 168                 if j:
 169                     new_chunk_title = book_title + '_%d' % j
 170                     new_chunk_slug = book_slug + '_%d' % j
 171                 else:
 172                     new_chunk_title, new_chunk_slug = book_title, book_slug
 173
 174                 chunk_titles.append(new_chunk_title)
 175                 chunk_slugs.append(new_chunk_slug)
 176
 177                 if verbose:
 178                     print "title: %s // %s  -->\n       %s // %s\nslug: %s / %s  -->\n      %s / %s" % (
 179                         book.title, chunk.title,
 180                         new_title, new_chunk_title,
 181                         book.slug, chunk.slug,
 182                         new_slug, new_chunk_slug)
 183                     print
 184
 185             if not dry_run:
 186                 try:
 187                     conflict = Book.objects.get(slug=new_slug)
 188                 except Book.DoesNotExist:
 189                     conflict = None
 190                 else:
 191                     if conflict == books[0]:
 192                         conflict = None
 193
 194                 if conflict:
 195                     if force:
 196                         # FIXME: there still may be a conflict
 197                         conflict.slug = '.' + conflict.slug
 198                         conflict.save()
 199                         print self.style.NOTICE('Book with slug "%s" moved to "%s".' % (new_slug, conflict.slug))
 200                     else:
 201                         print self.style.ERROR('ERROR: Book with slug "%s" exists.' % new_slug)
 202                         return
 203
 204                 if i:
 205                     books[0].append(books[i], slugs=chunk_slugs, titles=chunk_titles)
 206                 else:
 207                     book.title = new_title
 208                     book.slug = new_slug
 209                     book.save()
 210                     for j, chunk in enumerate(book):
 211                         chunk.title = chunk_titles[j]
 212                         chunk.slug = chunk_slugs[j]
 213                         chunk.save()
 214
 215
 216         transaction.commit()
 217         transaction.leave_transaction_management()
 218