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

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

tests/test_chunks.py

@@ -0,0 +1,72 @@
+"""로직 조각 후처리(summarize/chunks.py) — 줄 앵커 검증·보정, 파서 사실 채움, 버림 규칙."""
+from __future__ import annotations
+
+from summarize.chunks import numbered_code, parser_hints, resolve_chunks
+from summarize.schemas import LogicChunk
+
+CODE = """REPORT ztest.
+FORM select_data.
+  CLEAR gt_t001.
+  SELECT bukrs waers FROM t001
+    INTO TABLE gt_t001
+    WHERE bukrs IN s_bukrs.
+  SORT gt_t001 BY bukrs.
+  CALL FUNCTION 'BAPI_CURRENCY_CONV_TO_EXTERNAL'
+    EXPORTING amount = lv_amt.
+ENDFORM.""".split("\n")
+UNIT = (2, 10)
+
+
+def _chunk(**kw) -> LogicChunk:
+    base = {"line_start": 4, "line_end": 7, "first_line": "SELECT bukrs waers FROM t001",
+            "kind": "sql_select", "purpose_ko": "회사코드별 통화 조회", "confidence": 0.8}
+    base.update(kw)
+    return LogicChunk(**base)
+
+
+def test_exact_anchor_and_parser_facts():
+    chunks, dropped = resolve_chunks([_chunk()], CODE, *UNIT, "ZTEST", known_symbols={"GT_T001"})
+    assert dropped == []
+    c = chunks[0]
+    assert (c.line_start, c.line_end, c.seq) == (4, 7, 1)
+    assert c.tables_read == ["T001"]          # LLM 이 아니라 파서가 채운 값
+    assert c.code.startswith("  SELECT bukrs")
+    assert len(c.code_hash) == 64
+
+
+def test_anchor_shift_corrects_line_numbers():
+    """LLM 이 줄 번호를 2줄 어긋나게 줘도 first_line 으로 보정된다."""
+    chunks, dropped = resolve_chunks([_chunk(line_start=6, line_end=9)], CODE, *UNIT, "ZTEST", set())
+    assert dropped == []
+    assert (chunks[0].line_start, chunks[0].line_end) == (4, 7)
+
+
+def test_unfindable_anchor_is_dropped():
+    chunks, dropped = resolve_chunks(
+        [_chunk(first_line="SELECT * FROM nowhere_table_xyz")], CODE, *UNIT, "ZTEST", set())
+    assert chunks == [] and len(dropped) == 1
+
+
+def test_out_of_unit_range_dropped_and_end_clamped():
+    chunks, dropped = resolve_chunks(
+        [_chunk(line_start=1, line_end=3, first_line="REPORT ztest."),          # unit 밖 → 버림
+         _chunk(line_start=8, line_end=40, first_line="CALL FUNCTION 'BAPI_CURRENCY_CONV_TO_EXTERNAL'",
+                kind="fm_call", purpose_ko="환율 변환 BAPI 호출")],                # 끝은 unit 끝으로 클램프
+        CODE, *UNIT, "ZTEST", set())
+    assert len(dropped) == 1 and dropped[0].startswith("L1-L3")
+    assert (chunks[0].line_start, chunks[0].line_end) == (8, 10)
+    assert "BAPI_CURRENCY_CONV_TO_EXTERNAL" in chunks[0].calls
+
+
+def test_unknown_kind_falls_back_and_duplicates_dropped():
+    chunks, dropped = resolve_chunks([_chunk(kind="weird"), _chunk()], CODE, *UNIT, "ZTEST", set())
+    assert len(chunks) == 1 and chunks[0].kind == "other"
+    assert any("중복" in d for d in dropped)
+
+
+def test_numbered_code_and_hints():
+    txt = numbered_code(CODE, 2, 4)
+    assert txt.splitlines()[0].startswith("    2| FORM select_data.")
+    hints = parser_hints(CODE, *UNIT)
+    assert any(h.startswith("L4 SELECT") for h in hints)
+    assert any("CALL FUNCTION" in h for h in hints)