def download_if_missing(self):
super().download_if_missing()
full_topic_file = super().topic_file
assert full_topic_file.exists()
if self.topic_file.exists():
return
title = load_trec_topics(full_topic_file)["title"]
with open(self.topic_file, "w") as f:
for qid, full_topic in title:
kw_topic = self.tokenizer.tokenize(full_topic)
f.write(topic_to_trectxt(qid, kw_topic))
def test_searcher_query(tmpdir_as_cache, tmpdir, dummy_index, searcher_name):
topics_fn = DummyBenchmark.topic_file
query = list(load_trec_topics(topics_fn)["title"].values())[0]
nhits = 1
searcher = Searcher.create(searcher_name, config={"hits": nhits}, provide={"index": dummy_index})
results = searcher.query(query)
if searcher_name == "SPL":
# if searcher_name != "BM25":
return
print(results.values())
if isinstance(list(results.values())[0], dict):
assert all(len(d) == nhits for d in results.values())
else:
assert len(results) == nhits
def topics(self):
if not hasattr(self, "_topics"):
self._topics = load_trec_topics(self.config["topics"]["path"])
return self._topics
def set_topics(self, path):
self._topics = load_trec_topics(path)
def topics(self):
if not hasattr(self, "_topics"):
self._topics = load_trec_topics(self.topic_file)
return self._topics