feat(evidence): add table-free v3 and design guidance

2026-08-26 12:15:40 +02:00
parent 38f02cfd08
commit 9d4f994d3e
11 changed files with 983 additions and 67 deletions
@@ -221,6 +221,79 @@ def _domain_evidence_v2() -> CuratedEvidence:
    })


+def _domain_evidence_v3() -> CuratedEvidence:
+    return CuratedEvidence.model_validate({
+        **COMMON,
+        "schema_version": 3,
+        "title": "Dominio Ablazione",
+        "kind": "domain",
+        "purposes": ["disambiguation", "schema_linking", "rewriting"],
+        "applies_to": {
+            "concepts": ["ablazione", "studio elettrofisiologico", "SEE"],
+            "tables": ["clinical.fact_ablazione"],
+            "columns": ["clinical.fact_ablazione.patient_id"],
+        },
+        "payload": {
+            "rule": (
+                "Il dominio Ablazione rappresenta la procedura transcatetere.\n\n"
+                "La fact centrale è `clinical.fact_ablazione`."
+            ),
+        },
+    })
+
+
+def test_v3_curated_markdown_replaces_frontmatter_tables_with_readable_sections(tmp_path):
+    evidence = _domain_evidence_v3()
+    path = tmp_path / "curated" / "domain" / "dominio-ablazione.md"
+
+    text = dump_curated_markdown(evidence)
+    parsed = parse_curated_markdown(text, path=path)
+
+    assert text.startswith("<!-- tht:metadata:")
+    assert not text.startswith("---\n")
+    assert not any(line.startswith("|") for line in text.splitlines())
+    assert "> **Dominio** · Italiano" in text
+    assert "**Scopi:** Disambiguazione · Collegamento allo schema · Riscrittura" in text
+    assert "## Ambito di applicazione" in text
+    assert "### Concetti\n\n- ablazione\n- studio elettrofisiologico\n- SEE" in text
+    assert "### Tabelle\n\n- `clinical.fact_ablazione`" in text
+    assert "### Colonne\n\n- `clinical.fact_ablazione.patient_id`" in text
+    assert "<summary>Dettagli tecnici e provenienza</summary>" in text
+    assert parsed == evidence
+
+
+def test_v3_curated_markdown_renders_enum_values_as_a_list_instead_of_a_table(tmp_path):
+    evidence = CuratedEvidence.model_validate({
+        **COMMON,
+        "schema_version": 3,
+        "kind": "enum",
+        "payload": {
+            "column": "clinical.episode.discharge_status",
+            "values": {"D": "dimesso", "T": "trasferito | altra struttura"},
+        },
+    })
+
+    text = dump_curated_markdown(evidence)
+
+    assert "## Valori\n\n- `D`: dimesso\n- `T`: trasferito | altra struttura" in text
+    assert not any(line.startswith("|") for line in text.splitlines())
+    assert parse_curated_markdown(
+        text,
+        path=tmp_path / "curated" / "enum" / "discharge-status.md",
+    ) == evidence
+
+
+def test_v3_curated_markdown_rejects_visible_metadata_that_drifted_from_canonical_data():
+    text = dump_curated_markdown(_domain_evidence_v3()).replace(
+        "**Scopi:** Disambiguazione",
+        "**Scopi:** Testo alterato",
+        1,
+    )
+
+    with pytest.raises(ValueError, match="not canonical"):
+        parse_curated_markdown(text)
+
+
 def test_v2_curated_markdown_renders_domain_content_in_the_markdown_body(tmp_path):
    evidence = _domain_evidence_v2()
    path = tmp_path / "curated" / "domain" / "dominio-ablazione.md"