def _get_resumable_chunked_iterator(dbaccessor, iteration_key, log_prefix): total_docs = 0 for db in dbaccessor.sql_db_aliases: total_docs += dbaccessor.get_approximate_doc_count(db) iterable = resumable_sql_model_iterator(iteration_key, dbaccessor, chunk_size=CHUNK_SIZE, transform=lambda x: x) progress = ProgressManager(iterable, total=total_docs, reset=False, chunk_size=CHUNK_SIZE, logger=ProcessorProgressLogger( log_prefix, sys.stderr)) with progress: for chunk in chunked(iterable, CHUNK_SIZE): yield chunk progress.add(len(chunk))
def get_iterator(self, deleted_doc_ids=None, chunk_size=2): reindex_accessor = SimulateDeleteReindexAccessor(self.reindex_accessor, deleted_doc_ids) return resumable_sql_model_iterator(self.iteration_key, reindex_accessor, chunk_size)