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

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

tests/test_parser.py

@@ -0,0 +1,76 @@
+"""Stage 2 검증 (계획서 §4.6) — 정규화 산출물이 있을 때 ZFIR10070 실측 + 단위 테스트."""
+from pathlib import Path
+
+import pytest
+
+from parser.run import parse_program
+from parser.statements import split_statements
+from parser.tokenizer import split_comment, tokenize_code
+
+ROOT = Path(__file__).resolve().parent.parent
+PROG_DIR = ROOT / "data" / "normalized" / "ZFIR10070"
+
+
+def test_split_comment():
+    assert split_comment("* full comment") == ("", "full comment", True)
+    code, comment, full = split_comment("DATA lv_x TYPE i. \" inline")
+    assert comment == "inline" and not full and "DATA" in code
+
+
+def test_string_literal_quote_not_comment():
+    code, comment, _ = split_comment("WRITE 'has \" inside'.")
+    assert comment == "" and "'has \" inside'" in code
+
+
+def test_chain_expansion():
+    sts, _ = split_statements("DATA: a TYPE i,\n b TYPE i.", "T")
+    assert len(sts) == 2
+    assert sts[0].upper[:2] == ["DATA", "A"]
+    assert sts[1].upper[:2] == ["DATA", "B"]
+
+
+def test_tokenizer_identifiers():
+    toks = tokenize_code("zcl_fi_common=>f4_bukrs( ) gs_head-belnr TEXT-004 <fs>")
+    assert "zcl_fi_common=>f4_bukrs" in toks
+    assert "gs_head-belnr" in toks
+    assert "TEXT-004" in toks
+    assert "<fs>" in toks
+
+
+@pytest.fixture(scope="module")
+def parsed():
+    if not PROG_DIR.exists():
+        pytest.skip("정규화 산출물 없음 — python -m ingest.normalize 먼저 실행")
+    return parse_program(PROG_DIR)
+
+
+def test_f01_has_73_forms(parsed):
+    forms = [u for u in parsed["units"] if u["unit_type"] == "FORM" and u["include"] == "ZFIR10070_F01"]
+    assert len(forms) == 73
+
+
+def test_gt_acdoct_declared_in_top(parsed):
+    decls = [s for s in parsed["symbols"] if s["name"] == "GT_ACDOCT"]
+    assert decls and decls[0]["decl_include"] == "ZFIR10070_TOP"
+    assert decls[0]["scope"] == "global"
+
+
+def test_select_data_calls_select_gl_list(parsed):
+    sd = next(u for u in parsed["units"] if u["name"] == "SELECT_DATA")
+    targets = [c["target"] for c in sd["refs"]["calls"]]
+    assert "SELECT_GL_LIST" in targets
+
+
+def test_external_perform_zfir00010(parsed):
+    ext = [e for e in parsed["call_edges"]
+           if e["external_name"] and "ZFIR00010" in str(e["external_name"])]
+    assert any("CHECK_BUKRS" in e["external_name"] for e in ext)
+
+
+def test_text_symbol_004(parsed):
+    t = next(x for x in parsed["text_symbols"] if x["symbol"] == "004")
+    assert t["text"] == "회계단위 간편선택"
+
+
+def test_unknown_ratio_below_3pct(parsed):
+    assert parsed["stats"]["unknown_ratio"] < 0.03