fix: complete P2 Task 2 schema machine contracts

2026-08-11 05:38:25 +02:00
parent b235132d89
commit 554de4ff2f
4 changed files with 356 additions and 30 deletions
@@ -221,3 +221,34 @@ def test_vector_index_schema_json_failure_is_safe(monkeypatch, tmp_path):
    payload = json.loads(response.stdout)
    assert payload == {"status": "failed", "code": "schema_index_failed"}
    assert "secret qdrant" not in response.stdout
+
+
+
+def test_vector_index_schema_human_missing_physical_has_original_error(tmp_path):
+    cfg = _qdrant_runtime_config(tmp_path)
+    _write_schema_artifacts(tmp_path)
+    physical = tmp_path / "artifacts" / "mschema" / "physical.yaml"
+    physical.unlink()
+    response = CliRunner().invoke(app, ["vector", "index-schema", "-c", str(cfg)])
+    assert response.exit_code == 1
+    assert "physical.yaml non trovato. Esegui prima `tht schema introspect`." in response.output
+
+
+def test_vector_index_schema_json_missing_physical_has_no_stderr_prose(tmp_path):
+    import json
+
+    cfg = _qdrant_runtime_config(tmp_path)
+    _write_schema_artifacts(tmp_path)
+    (tmp_path / "artifacts" / "mschema" / "physical.yaml").unlink()
+    response = CliRunner().invoke(app, ["vector", "index-schema", "--json", "-c", str(cfg)])
+    assert response.exit_code == 1
+    assert response.stderr == ""
+    assert json.loads(response.stdout) == {"status": "failed", "code": "physical_schema_missing"}
+
+
+def test_vector_index_schema_human_missing_config_has_original_error(tmp_path):
+    cfg = tmp_path / "missing.yaml"
+    response = CliRunner().invoke(app, ["vector", "index-schema", "-c", str(cfg)])
+    assert response.exit_code == 1
+    assert response.output
+    assert "ERRORE:" in response.output
@@ -281,3 +281,151 @@ def test_schema_check_json_reports_orphan_count_without_prose(tmp_path):
        "status": "failed", "code": "annotation_orphans", "orphan_count": 1,
        "orphans": ["gone"],
    }
+
+
+def test_suggest_fks_json_rejects_more_than_32_staged_sql_files(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    staged = tmp_path / "many"
+    staged.mkdir()
+    for i in range(33):
+        (staged / f"q{i:02d}.sql").write_text("SELECT 1")
+    response = CliRunner().invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(staged)]
+    )
+    assert response.exit_code == 1
+    assert json.loads(response.stdout) == {"status": "failed", "code": "staged_sql_too_many"}
+
+
+def test_staged_sql_size_limits_are_inclusive(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    staged = tmp_path / "boundary"
+    staged.mkdir()
+    # Exactly one MiB per file and exactly 16 MiB in aggregate are accepted.
+    body = "-- padding\n" + ("x" * (1 << 20))
+    body = body[: 1 << 20]
+    for i in range(16):
+        (staged / f"q{i:02d}.sql").write_bytes(body.encode())
+    response = CliRunner().invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(staged)]
+    )
+    assert response.exit_code == 0, response.output
+    assert json.loads(response.stdout)["staged_sql_count"] == 16
+
+
+def test_schema_check_json_success_and_reviewed_annotations_are_preserved(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    annotations_path = tmp_path / "artifacts" / "mschema" / "annotations.yaml"
+    reviewed = _annotations_with_fks()
+    reviewed.to_yaml(annotations_path)
+    before = annotations_path.read_text()
+
+    check = CliRunner().invoke(app, ["schema", "check", "--json", "-c", str(cfg)])
+    assert check.exit_code == 0, check.output
+    assert json.loads(check.stdout) == {
+        "status": "succeeded", "code": "ok", "orphan_count": 0, "orphans": []
+    }
+    suggest = CliRunner().invoke(app, ["schema", "suggest-fks", "--json", "-c", str(cfg)])
+    assert suggest.exit_code == 0, suggest.output
+    assert json.loads(suggest.stdout)["candidate_count"] == 0
+    assert annotations_path.read_text() == before
+
+
+def test_schema_human_check_and_suggest_report_missing_physical_schema(tmp_path):
+    cfg = _write_workspace(tmp_path)
+    physical = tmp_path / "artifacts" / "mschema" / "physical.yaml"
+    physical.unlink()
+    expected = "physical.yaml non trovato. Esegui prima `tht schema introspect`."
+    check = CliRunner().invoke(app, ["schema", "check", "-c", str(cfg)])
+    assert check.exit_code == 1
+    assert expected in check.output
+    suggest = CliRunner().invoke(app, ["schema", "suggest-fks", "-c", str(cfg)])
+    assert suggest.exit_code == 1
+    assert expected in suggest.output
+
+
+def test_schema_human_check_reports_ignored_columns(tmp_path):
+    cfg = _write_workspace(tmp_path)
+    physical = _physical()
+    physical.tables["fact_ablazione"].columns["esito"].eligible = False
+    physical.tables["fact_ablazione"].columns["esito"].eligibility_reason = "test ignored"
+    physical.to_yaml(tmp_path / "artifacts" / "mschema" / "physical.yaml")
+    response = CliRunner().invoke(app, ["schema", "check", "-c", str(cfg)])
+    assert response.exit_code == 0, response.output
+    assert "Colonne ignorate (testo ampio, 1):" in response.output
+    assert "fact_ablazione.esito (test ignored)" in response.output
+
+
+
+def test_staged_sql_file_size_over_one_mib_is_rejected(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    staged = tmp_path / "oversize"
+    staged.mkdir()
+    (staged / "too-large.sql").write_bytes(b"x" * ((1 << 20) + 1))
+    response = CliRunner().invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(staged)]
+    )
+    assert response.exit_code == 1
+    assert json.loads(response.stdout) == {"status": "failed", "code": "staged_sql_too_large"}
+
+
+def test_staged_sql_aggregate_over_16_mib_is_rejected(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    staged = tmp_path / "aggregate"
+    staged.mkdir()
+    for i in range(16):
+        (staged / f"q{i:02d}.sql").write_bytes(b"x" * (1 << 20))
+    (staged / "over.sql").write_bytes(b"x")
+    response = CliRunner().invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(staged)]
+    )
+    assert response.exit_code == 1
+    assert json.loads(response.stdout) == {"status": "failed", "code": "staged_sql_too_large"}
+
+
+def test_suggest_fks_candidate_order_is_stable_for_reordered_staged_inputs(tmp_path):
+    import json
+
+    cfg = _write_workspace(tmp_path)
+    first = tmp_path / "first"
+    second = tmp_path / "second"
+    first.mkdir()
+    second.mkdir()
+    (first / "join.sql").write_text(
+        "SELECT * FROM fact_ablazione f JOIN dim_patient p ON f.cod_paz = p.cod_paz"
+    )
+    (second / "join.sql").write_text(
+        "SELECT * FROM fact_ablazione f JOIN dim_time p ON f.data_time_key = p.day_key"
+    )
+    runner = CliRunner()
+    one = runner.invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(first), "--from-sql", str(second)]
+    )
+    two = runner.invoke(
+        app, ["schema", "suggest-fks", "--json", "-c", str(cfg), "--from-sql", str(second), "--from-sql", str(first)]
+    )
+    assert one.exit_code == two.exit_code == 0
+    assert json.loads(one.stdout) == json.loads(two.stdout)
+
+
+def test_schema_human_check_missing_config_has_original_error(tmp_path):
+    response = CliRunner().invoke(app, ["schema", "check", "-c", str(tmp_path / "missing.yaml")])
+    assert response.exit_code == 1
+    assert response.output
+    assert "ERRORE:" in response.output
+
+
+def test_schema_human_suggest_missing_config_has_original_error(tmp_path):
+    response = CliRunner().invoke(app, ["schema", "suggest-fks", "-c", str(tmp_path / "missing.yaml")])
+    assert response.exit_code == 1
+    assert response.output
+    assert "ERRORE:" in response.output