def test_doesnt_have_subtitles(ucache: URLCache) -> None:
summ_resp = ucache.get(youtube_without_cc)
# shouldnt match, is the 'corrected' preprocessed URL
assert summ_resp.url != youtube_without_cc
# make sure this parsed the youtube id
assert "xvQUiX26RfE" == get_yt_video_id(youtube_without_cc)
assert summ_resp.data is not None
# deleted for youtube by the site-specific extractor
assert summ_resp.html_summary is None
assert "subtitles" not in summ_resp.data
dir_full_path = ucache.summary_cache.dir_cache.get(
ucache.preprocess_url(youtube_without_cc))
assert not os.path.exists(
os.path.join(dir_full_path, "data", "subtitles.srt"))
assert os.path.exists(os.path.join(dir_full_path, "metadata.json"))
# this deletes the summary files on purpose, since theyre somewhat useless
assert not os.path.exists(os.path.join(dir_full_path, "html_summary.html"))
def test_youtube_has_subtitles(ucache: URLCache) -> None:
# make sure subtitles download to file
assert not ucache.in_cache(youtube_with_cc)
summ_resp: Summary = ucache.get(youtube_with_cc)
assert ucache.in_cache(youtube_with_cc)
assert isinstance(summ_resp, Summary)
assert summ_resp is not None
assert summ_resp.data is not None and "subtitles" in summ_resp.data
assert "trade-off between space" in summ_resp.data["subtitles"]
# make sure corresponding file exists
dcache = ucache.summary_cache.dir_cache
assert isinstance(dcache, DirCache)
dir_full_path = dcache.get(ucache.preprocess_url(youtube_with_cc))
assert dir_full_path.endswith(
"data/2/c/7/6284b2f664f381372fab3276449b2/000")
subtitles_file = Path(os.path.join(dir_full_path, "data", "subtitles.srt"))
assert subtitles_file.exists()
# make sure subtitle is in cache dir
assert "trade-off between space" in subtitles_file.read_text()
def test_youtube_preprocessor(ucache: URLCache) -> None:
assert youtube_without_cc != "https://www.youtube.com/watch?v=xvQUiX26RfE"
assert (ucache.preprocess_url(youtube_without_cc) ==
"https://www.youtube.com/watch?v=xvQUiX26RfE")