Fix every real lint finding and drop degenerate splice fragments

ingestion/tests/test_segment_vocab.py

@@ -0,0 +1,83 @@
+from ingestion.segment.vocab import match_section, match_section_with_inline_value
+
+
+def test_exact_label_match_with_trailing_colon():
+    d = match_section("Tên chung quốc tế:")
+    assert d is not None and d.key == "ten_chung_quoc_te"
+
+
+def test_exact_label_match_without_trailing_colon():
+    d = match_section("Chỉ định")
+    assert d is not None and d.key == "chi_dinh"
+
+
+def test_inline_value_combined_span_confirmed_real_amitriptylin_case():
+    # AMITRIPTYLIN's real "Mã ATC:" field is one non-bold span combining
+    # label and value: "Mã ATC: N06AA09." — see outlier item 20.
+    result = match_section_with_inline_value("Mã ATC: N06AA09.")
+    assert result is not None
+    section_def, value = result
+    assert section_def.key == "ma_atc"
+    assert value == "N06AA09."
+
+
+def test_inline_value_not_matched_when_no_colon_follows():
+    assert match_section_with_inline_value("Mã ATC something else entirely") is None
+
+
+def test_inline_value_does_not_confuse_plain_body_text():
+    assert match_section_with_inline_value("Bệnh nhân cần theo dõi chặt chẽ.") is None
+
+
+def test_exact_match_takes_priority_over_prefix_for_label_only_span():
+    d = match_section("Mã ATC:")
+    assert d is not None and d.key == "ma_atc"
+
+
+def test_real_spelling_variants_found_in_the_book_all_match():
+    # measured whole-corpus: 42 distinct near-miss heading strings, 542
+    # occurrences, none of which matched before aliases were added. The
+    # heaviest is "Thông tin qui chế" (469x) — the book prints "qui" where
+    # its own documented template says "quy", which cost 586 of 682
+    # monographs their thong_tin_quy_che section entirely.
+    from ingestion.segment.vocab import match_section
+    cases = {
+        "Thông tin qui chế": "thong_tin_quy_che",
+        "Thông tin về qui chế": "thong_tin_quy_che",
+        "Thông tin và quy chế": "thong_tin_quy_che",
+        "Mã ACT": "ma_atc",
+        "Chống chỉ đinh": "chong_chi_dinh",
+        "Thời kì mang thai": "thoi_ky_mang_thai",
+        "Thời kì cho con bú": "thoi_ky_cho_con_bu",
+        "Dược lí và cơ chế tác dụng": "duoc_ly_va_co_che_tac_dung",
+        "Hướng dẫn cách sử trí ADR": "huong_dan_xu_tri_adr",
+        "Quá liều và xử lý": "qua_lieu_va_xu_tri",
+        "Lọai thuốc": "loai_thuoc",
+    }
+    for text, expected_key in cases.items():
+        matched = match_section(text)
+        assert matched is not None, f"{text!r} should match a section"
+        assert matched.key == expected_key
+
+
+def test_typesetting_noise_is_folded_without_needing_an_alias_each():
+    # missing/extra spaces and the Ð/Đ look-alike are handled by the lookup
+    # key, not enumerated per-variant
+    from ingestion.segment.vocab import match_section
+    assert match_section("Chỉđịnh").key == "chi_dinh"
+    assert match_section("Chống chỉđịnh").key == "chong_chi_dinh"
+    assert match_section("Độổn định và bảo quản").key == "do_on_dinh_va_bao_quan"
+    assert match_section("Ðộ ổn định và bảo quản").key == "do_on_dinh_va_bao_quan"
+    assert match_section("H ướng dẫn cách xử trí ADR").key == "huong_dan_xu_tri_adr"
+    assert match_section("Tư ơng kỵ").key == "tuong_ky"
+    assert match_section("Tác dụng khôngmong muốn (ADR)").key == "tac_dung_khong_mong_muon"
+    assert match_section("Thận trọng.").key == "than_trong"
+
+
+def test_near_misses_that_are_not_sections_stay_unmatched():
+    # "Thể trọng" is body weight, not "Thận trọng" (caution) — a 0.84
+    # similarity that must NOT become an alias; the opioid string is a
+    # drug-specific sub-heading inside a section, not the section itself
+    from ingestion.segment.vocab import match_section
+    assert match_section("Thể trọng") is None
+    assert match_section("Tác dụng không mong muốn của opioid") is None