def _load_run_base_infer(self, run_dir, rankfn, ranktopk, fmt='dict'):
best_candidate_run, best_topk = None, None
for candidate_run in glob(os.path.join(run_dir, f'{rankfn}.*.run')):
c_topk = int(candidate_run.split('.')[-2])
if c_topk > ranktopk and (best_topk is None or c_topk < best_topk):
best_candidate_run = candidate_run
best_topk = c_topk
if best_candidate_run is not None:
with self.logger.duration(f'loading {rankfn}:{ranktopk} from larger batch {rankfn}:{best_topk}'):
return trec.read_run_fmt(best_candidate_run, fmt, top=ranktopk)
return None
def _load_run_base_query(self, index, subset, rankfn, ranktopk, run_path,
fmt):
queries = self._load_queries_base(subset).items()
index.batch_query(queries, rankfn, ranktopk, destf=run_path)
return trec.read_run_fmt(run_path, fmt)
def _load_run_base_direct(self, run_path, rankfn, ranktopk, fmt='dict'):
if os.path.exists(run_path):
with self.logger.duration(f'reading {rankfn}:{ranktopk}'):
return trec.read_run_fmt(run_path, fmt=fmt)
return None