def es_index(text, meta, doctype=None, options=None): es = get_elasticsearch(options) index_name = 'frisc_{}'.format(doctype) document_id = meta.get('filesystem_absolute_path', '') try: result = es.index( index=index_name, doc_type="{}_doc".format(doctype), id=document_id, body=meta ) except es_exceptions.TransportError as es_error: print(es_error) return print(result) document_id = result.get('_id') count = 1 result for page in text: try: result = es.index( index=index_name, doc_type="{}_page".format(doctype), parent=document_id, id="{}_page{}".format(document_id, count), body={"content": page} ) except es_exceptions.TransportError as es_error: print(es_error) continue print(result) count += 1
def __init__(self, options=None): self.options = options or {} self.es = get_elasticsearch(self.options) self.esc = IndicesClient(self.es) self.conf_dir = sys.path[0]