def handle(self, *args, **options):
contador = 0
total = 50
materias = Materia.objects.filter(corpo__icontains="saibamais")[:total]
# materias = [Materia.objects.get(id=863)]
i = 0
for m in materias:
print "\n", m.titulo
documento = get_document(m)
# html = lhtml.fromstring(m.corpo.decode('utf-8'))
# documento['relacionadas'] = set([ change_host(h.attrib['href']) for h in html.cssselect('.saibamais ul li a')])
recomendadas = [str(recomendada.permalink) for recomendada in relacionadas(documento, 4)]
for r in recomendadas:
urls = [u.origem for u in Analytics.objects.filter(destino=r).order_by("pageviews")]
documents = [(tf(m.corpo).keys(), "dentro") for m in Materia.objects.filter(permalink__in=urls)]
prob = 0
if documents:
documents += [
(tf(m.corpo).keys(), "fora")
for m in Materia.objects.exclude(permalink__in=urls)[: len(documents)]
]
prob = classifica(m, documents)
# print i, round(prob, 2)
if prob > 0.7:
print round(prob, 2), r
i += 1
def handle(self, *args, **options):
import time
inicio = time.time()
settings.CACHE = True
# if options['editoria']:
# folder = Folder.objects.get(name=options['editoria'])
# materias = Materia.objects.filter(status='T', folders=folder)
# else:
# materias = Materia.objects.filter(status='T')
# materias = materias[:options['total']]
materias = Materia.objects.filter(corpo__icontains='saibamais')[:100]
# editorias_id = [39,31,119,214,339,216,146,
# 8,133,101,94,20,42,76,105]
# for f in editorias_id:
# folder = Folder.objects.get(id=f)
# materias = Materia.objects.filter(status='T', folders=folder)[:total]
contador = 0
combinacoes = []
dict_combinacoes = {}
seq = options['sequential']
for tam_comb in range(len(seq)):
for comb in itertools.combinations(seq,tam_comb+1):
combinacoes.append("".join(comb))
for m in materias:
contador+=1
# print contador, time.time() -inicio
mfolder = m.primary_folder().name
if not dict_combinacoes.has_key(mfolder):
dict_combinacoes[mfolder] = {}
if dict_combinacoes[mfolder].has_key('_TotalMaterias'):
dict_combinacoes[mfolder]['_TotalMaterias'] +=1
else:
dict_combinacoes[mfolder]['_TotalMaterias']=1
for comb in combinacoes:
documento = monta_doc(m)
# import pdb; pdb.set_trace();
materiasSolr = relacionadas(documento, comb=comb, total=int(options['recomendadas']), similaridade=eval(options['similaridade']))
recomendadas = []
recomendadas = [str(recomendada.url) for (recomendada, score) in materiasSolr]
encontradas = documento['relacionadas'].intersection(recomendadas)
if any(encontradas):
if dict_combinacoes[mfolder].has_key(comb):
dict_combinacoes[mfolder][comb]+=1
else:
dict_combinacoes[mfolder][comb]=1
# print "total de acertos em %s matérias" % (contador)
dict_final={}
# import pdb; pdb.set_trace();
for dc in dict_combinacoes.keys():
print "\n", dc
for d in sorted(dict_combinacoes[dc].keys()):
print d,dict_combinacoes[dc][d]
if dict_final.has_key(d):
dict_final[d]+=dict_combinacoes[dc][d]
else:
dict_final[d]=dict_combinacoes[dc][d]
print "\nTotal Geral"
for f in sorted(dict_final.keys()):
print f, dict_final[f]
