def test_smoke_test_scrape(smoke_test_language): """A smoke test for scrape().""" n = 10 # number of word-pron pairs to scrape config = config_factory(key=smoke_test_language.key, **smoke_test_language.config_params) assert config.language == smoke_test_language.wik_name pairs = [] for i, (word, pron) in enumerate(scrape(config)): if i >= n: break pairs.append((word, pron)) assert len(pairs) == n assert all(word and pron for (word, pron) in pairs)
def _scrape_and_write(config: Config) -> None: for i, (word, pron) in enumerate(scrape(config), 1): print(f"{word}\t{pron}") if i % 100 == 0: logging.info("%d pronunciations scraped", i)