src/documents/management/commands/merge_books.py

   1 # This file is part of FNP-Redakcja, licensed under GNU Affero GPLv3 or later.
   2 # Copyright © Fundacja Nowoczesna Polska. See NOTICE for more information.
   3 #
   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 documents.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     help = 'Merges multiple books into one.'
  28     args = '[slug]...'
  29
  30     def add_arguments(self, parser):
  31         self.add_argument(
  32             '-s', '--slug', dest='new_slug', metavar='SLUG',
  33             help='New slug of the merged book (defaults to common part of all slugs).')
  34         self.add_argument(
  35             '-t', '--title', dest='new_title', metavar='TITLE',
  36             help='New title of the merged book (defaults to common part of all titles).')
  37         self.add_argument(
  38             '-q', '--quiet', action='store_false', dest='verbose', default=True,
  39             help='Less output')
  40         self.add_argument(
  41             '-g', '--guess', action='store_true', dest='guess', default=False,
  42             help='Try to guess what merges are needed (but do not apply them).')
  43         self.add_argument(
  44             '-d', '--dry-run', action='store_true', dest='dry_run', default=False,
  45             help='Dry run: do not actually change anything.')
  46         self.add_argument(
  47             '-f', '--force', action='store_true', dest='force', default=False,
  48             help='On slug conflict, hide the original book to archive.')
  49
  50     def print_guess(self, dry_run=True, force=False):
  51         from collections import defaultdict
  52         from pipes import quote
  53         import re
  54
  55         def read_slug(slug):
  56             res = []
  57             res.append((re.compile(r'__?(przedmowa)$'), -1))
  58             res.append((re.compile(r'__?(cz(esc)?|ksiega|rozdzial)__?(?P<n>\d*)$'), None))
  59             res.append((re.compile(r'__?(rozdzialy__?)?(?P<n>\d*)-'), None))
  60
  61             for r, default in res:
  62                 m = r.search(slug)
  63                 if m:
  64                     start = m.start()
  65                     try:
  66                         return int(m.group('n')), slug[:start]
  67                     except IndexError:
  68                         return default, slug[:start]
  69             return None, slug
  70
  71         def file_to_title(fname):
  72             """ Returns a title-like version of a filename. """
  73             parts = (p.replace('_', ' ').title() for p in fname.split('__'))
  74             return ' / '.join(parts)
  75
  76         merges = defaultdict(list)
  77         slugs = []
  78         for b in Book.objects.all():
  79             slugs.append(b.slug)
  80             n, ns = read_slug(b.slug)
  81             if n is not None:
  82                 merges[ns].append((n, b))
  83
  84         conflicting_slugs = []
  85         for slug in sorted(merges.keys()):
  86             merge_list = sorted(merges[slug])
  87             if len(merge_list) < 2:
  88                 continue
  89
  90             merge_slugs = [b.slug for i, b in merge_list]
  91             if slug in slugs and slug not in merge_slugs:
  92                 conflicting_slugs.append(slug)
  93
  94             title = file_to_title(slug)
  95             print("./manage.py merge_books %s%s--title=%s --slug=%s \\\n    %s\n" % (
  96                 '--dry-run ' if dry_run else '',
  97                 '--force ' if force else '',
  98                 quote(title), slug,
  99                 " \\\n    ".join(merge_slugs)
 100                 ))
 101
 102         if conflicting_slugs:
 103             if force:
 104                 print(self.style.NOTICE('# These books will be archived:'))
 105             else:
 106                 print(self.style.ERROR('# ERROR: Conflicting slugs:'))
 107             for slug in conflicting_slugs:
 108                 print('#', slug)
 109
 110
 111     def handle(self, *slugs, **options):
 112
 113         self.style = color_style()
 114
 115         force = options.get('force')
 116         guess = options.get('guess')
 117         dry_run = options.get('dry_run')
 118         new_slug = options.get('new_slug').decode('utf-8')
 119         new_title = options.get('new_title').decode('utf-8')
 120         verbose = options.get('verbose')
 121
 122         if guess:
 123             if slugs:
 124                 print("Please specify either slugs, or --guess.")
 125                 return
 126             else:
 127                 self.print_guess(dry_run, force)
 128                 return
 129         if not slugs:
 130             print("Please specify some book slugs")
 131             return
 132
 133         # Start transaction management.
 134         transaction.enter_transaction_management()
 135
 136         books = [Book.objects.get(slug=slug) for slug in slugs]
 137         common_slug = common_prefix(slugs)
 138         common_title = common_prefix([b.title for b in books])
 139
 140         if not new_title:
 141             new_title = common_title
 142         elif common_title.startswith(new_title):
 143             common_title = new_title
 144
 145         if not new_slug:
 146             new_slug = common_slug
 147         elif common_slug.startswith(new_slug):
 148             common_slug = new_slug
 149
 150         if slugs[0] != new_slug and Book.objects.filter(slug=new_slug).exists():
 151             self.style.ERROR('Book already exists, skipping!')
 152
 153
 154         if dry_run and verbose:
 155             print(self.style.NOTICE('DRY RUN: nothing will be changed.'))
 156             print()
 157
 158         if verbose:
 159             print("New title:", self.style.NOTICE(new_title))
 160             print("New slug:", self.style.NOTICE(new_slug))
 161             print()
 162
 163         for i, book in enumerate(books):
 164             chunk_titles = []
 165             chunk_slugs = []
 166
 167             book_title = book.title[len(common_title):].replace(' / ', ' ').lstrip()
 168             book_slug = book.slug[len(common_slug):].replace('__', '_').lstrip('-_')
 169             for j, chunk in enumerate(book):
 170                 if j:
 171                     new_chunk_title = book_title + '_%d' % j
 172                     new_chunk_slug = book_slug + '_%d' % j
 173                 else:
 174                     new_chunk_title, new_chunk_slug = book_title, book_slug
 175
 176                 chunk_titles.append(new_chunk_title)
 177                 chunk_slugs.append(new_chunk_slug)
 178
 179                 if verbose:
 180                     print("title: %s // %s  -->\n       %s // %s\nslug: %s / %s  -->\n      %s / %s" % (
 181                         book.title, chunk.title,
 182                         new_title, new_chunk_title,
 183                         book.slug, chunk.slug,
 184                         new_slug, new_chunk_slug))
 185                     print()
 186
 187             if not dry_run:
 188                 try:
 189                     conflict = Book.objects.get(slug=new_slug)
 190                 except Book.DoesNotExist:
 191                     conflict = None
 192                 else:
 193                     if conflict == books[0]:
 194                         conflict = None
 195
 196                 if conflict:
 197                     if force:
 198                         # FIXME: there still may be a conflict
 199                         conflict.slug = '.' + conflict.slug
 200                         conflict.save()
 201                         print(self.style.NOTICE('Book with slug "%s" moved to "%s".' % (new_slug, conflict.slug)))
 202                     else:
 203                         print(self.style.ERROR('ERROR: Book with slug "%s" exists.' % new_slug))
 204                         return
 205
 206                 if i:
 207                     books[0].append(books[i], slugs=chunk_slugs, titles=chunk_titles)
 208                 else:
 209                     book.title = new_title
 210                     book.slug = new_slug
 211                     book.save()
 212                     for j, chunk in enumerate(book):
 213                         chunk.title = chunk_titles[j]
 214                         chunk.slug = chunk_slugs[j]
 215                         chunk.save()
 216
 217
 218         transaction.commit()
 219         transaction.leave_transaction_management()
 220