Initial commit: ABAP indexing pipeline (ingest, parser, summarize, index, query, wiki)

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>

tests/test_expand.py

@@ -0,0 +1,69 @@
+"""질의 확장 검증 (수정사항 4번)."""
+from config.glossary import parse_glossary, synonym_map
+from index.db import fts_or, query_tokens
+from query import expand
+
+
+def test_parse_glossary_ignores_nested_and_comments():
+    g = parse_glossary(
+        "# 주석\n"
+        "총계정원장: [G/L, GL, ACDOCT]\n"
+        "tags:\n"
+        "  - 전표\n"
+        "입고: [GR, 101]\n"
+    )
+    assert g == {"총계정원장": ["G/L", "GL", "ACDOCT"], "입고": ["GR", "101"]}
+
+
+def test_synonym_map_is_bidirectional(tmp_path):
+    p = tmp_path / "g.yaml"
+    p.write_text("입고: [GR, MSEG, 101]\n", encoding="utf-8")
+    m = synonym_map(p)
+    # 대표어로 물어도, 동의어로 물어도 나머지가 나온다
+    assert set(m["입고"]) == {"GR", "MSEG", "101"}
+    assert "입고" in m["GR"] and "MSEG" in m["GR"]
+    assert "입고" in m["101"]
+
+
+def test_expansion_terms_excludes_query_itself():
+    terms = expand.expansion_terms("총계정원장")
+    assert terms, "사전에 있는 용어는 동의어가 나와야 한다"
+    assert "총계정원장" not in terms
+    assert "ACDOCT" in [t.upper() for t in terms]
+
+
+def test_expansion_terms_empty_for_unknown_word():
+    assert expand.expansion_terms("zzz알수없는말") == []
+
+
+def test_match_exprs_separates_primary_and_expanded():
+    primary, expanded = expand.match_exprs("총계정원장")
+    assert '"총계정원장"' in primary
+    # 동의어는 별 식으로 나와야 한다 — 같은 OR 버킷에 섞으면 정밀도가 떨어진다
+    assert expanded is not None
+    assert "ACDOCT" in expanded.upper()
+    assert "ACDOCT" not in primary.upper()
+
+
+def test_no_expansion_returns_none():
+    _, expanded = expand.match_exprs("zzz알수없는말")
+    assert expanded is None
+
+
+def test_expanded_weight_is_a_penalty():
+    assert 0 < expand.EXPANDED_WEIGHT < 1, "동의어 히트는 원질의 히트보다 낮게 점수화돼야 한다"
+
+
+def test_fts_or_quotes_and_adds_bigrams():
+    expr = fts_or(["총계정원장"])
+    assert '"총계정원장"' in expr
+    assert '"총계' in expr and " OR " in expr
+
+
+def test_fts_or_strips_quotes_to_keep_match_valid():
+    assert '""' not in fts_or(['A"B']).replace('"A', "").replace('B"', "")
+    assert fts_or([]) == '""'
+
+
+def test_query_tokens_drops_punctuation():
+    assert query_tokens("총계정원장, 잔액 조회!") == ["총계정원장", "잔액", "조회"]