def committee_evidence(): for document in Document.evidence\ .annotate(scount=models.Count('statement'))\ .exclude(scount__gt=0).exclude(skip_parsing=True).order_by('date').iterator(): print document parl_document.import_document(document, interactive=False) if document.statement_set.all().count(): document.save_activity()
def committee_evidence(): for document in Document.evidence\ .annotate(scount=models.Count('statement'))\ .exclude(scount__gt=0).exclude(skip_parsing=True).order_by('date').iterator(): try: print document parl_document.import_document(document, interactive=False) if document.statement_set.all().count(): document.save_activity() except Exception, e: logger.exception("Evidence parse failure on #%s: %r" % (document.id, e)) continue
def hansards_parse(): for hansard in Document.objects.filter(document_type=Document.DEBATE)\ .annotate(scount=models.Count('statement'))\ .exclude(scount__gt=0).exclude(skip_parsing=True).order_by('date').iterator(): with transaction.atomic(): try: with transaction.atomic(): parl_document.import_document(hansard, interactive=False) except Exception, e: logger.exception("Hansard parse failure on #%s: %r" % (hansard.id, e)) continue # now reload the Hansard to get the date hansard = Document.objects.get(pk=hansard.id) hansard.save_activity()