catloop.pt

#!/usr/bin/python
# -*- coding: utf-8 -*-
import oursql, os, codecs, pywikibot
#import cPickle, codecs

connection = oursql.connect(db='ptwiki_p', host='ptwiki.labsdb', read_default_file=os.path.expanduser('~/replica.my.cnf'))
c = connection.cursor()
c.execute(u'''SELECT
 cl_to,
 page_title
 FROM categorylinks
 INNER JOIN page ON cl_from = page_id
 WHERE page_namespace = 14''')

cats = {}
for cat, subcat in c:
  cat, subcat = cat.decode('utf-8'), subcat.decode('utf-8')
  if cat in cats:
    cats[cat].add(subcat)
  else:
    cats[cat] = {subcat}

# cats é um dicionário em que a chave é o nome da categoria
# e o valor é um set com os nomes de suas subcategorias

#with open('catsdump.pkl', 'r') as f:
#  cats = cPickle.load(f)

loops = []
loopssets = []
path = []
checked = {}
def check(cat):
  global cats, path, loops, loopssets, deep
  if cat in cats:
    if cat in path:
      loop = path[path.index(cat):] + [cat]
      if set(loop) not in loopssets:
        loops.append(loop)
        loopssets.append(set(loop))
        #print 'LOOP!', u'>>'.join(loop)
    elif cat not in checked:
      path.append(cat)
      for subcat in cats[cat]:
        check(subcat)
      if deep < len(path):
        deep = len(path)
      path.remove(cat)
      checked[cat] = deep
    elif deep < len(path) + checked[cat]:
      deep = len(path) + checked[cat]

for cat in cats:
  deep = 0
  check(cat)

with codecs.open('catdeep.txt', 'w', 'utf-8') as f:
  f.write(u'\n'.join(u'{} -> {}'.format(cat, deep) for cat, deep in sorted(checked.items(), key=lambda i:i[1], reverse=True)))

lista = u'\n'.join(u'# ' + u' &gt; '.join(u'[[:Categoria:{0}|{0}]]'.format(cat) for cat in loop) for loop in sorted(loops, key=lambda x:len(x)))
pywikibot.Page(pywikibot.Site(), u'Usuário:Danilo.mac/Loops de categorias').put(u'Lista de loops de categorias\n\nÚltima atualização: ~~~~~\n\n' + lista, comment=u'Bot: Atualizando lista')