from app.email_identity_extraction_service import _sanitize_company_mention, is_plausible_company_mention from app.fiscal_enrichment_service import _company_name_matches_mention def test_rejects_tld_and_domain_fragments_as_company_mentions(): for value in ["pt", "com", "www", "sementenergias.pt", "http", "email"]: assert _sanitize_company_mention(value) == "" assert not is_plausible_company_mention(value) def test_keeps_valid_portuguese_company_mentions(): assert _sanitize_company_mention("SEMENTENERGIAS, UNIPESSOAL, LDA") assert is_plausible_company_mention("Dietimport S.A") def test_short_mentions_do_not_match_inside_company_names(): assert not _company_name_matches_mention("pt", "planetoption lda") assert not _company_name_matches_mention("pt", "sementenergias unipessoal lda") assert _company_name_matches_mention("sementenergias unipessoal lda", "sementenergias unipessoal lda") assert _company_name_matches_mention("dietimport s a", "dietimport s a")