feat: complete evidence restructuring worktree

This commit is contained in:
Codex
2026-08-26 11:39:02 +02:00
parent a54d4769dd
commit 38f02cfd08
56 changed files with 1981 additions and 1801 deletions
+148
View File
@@ -206,6 +206,154 @@ def _formula_evidence() -> CuratedEvidence:
})
def _domain_evidence_v2() -> CuratedEvidence:
return CuratedEvidence.model_validate({
**COMMON,
"schema_version": 2,
"title": "Dominio Ablazione",
"kind": "domain",
"payload": {
"rule": (
"Il dominio Ablazione rappresenta la procedura transcatetere.\n\n"
"La fact centrale è `clinical.fact_ablazione`."
),
},
})
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"
text = dump_curated_markdown(evidence)
frontmatter = text.split("---\n", 2)[1]
parsed = parse_curated_markdown(text, path=path)
assert "domain:" not in frontmatter
assert "supporting_excerpts:" not in frontmatter
assert "review_items:" not in frontmatter
assert "# Dominio Ablazione\n" in text
assert "## Regola\n\nIl dominio Ablazione" in text
assert "## Estratti di supporto\n\n> I pazienti sotto i 18 anni sono pediatrici." in text
assert parsed == evidence
@pytest.mark.parametrize(("kind", "payload", "rendered"), [
("glossary", {
"definition": "Un paziente con età inferiore a 18 anni.",
"synonyms": ["minore"],
"variants": ["pediatrico"],
}, "## Sinonimi\n\n- minore"),
("enum", {
"column": "clinical.episode.discharge_status",
"values": {"D": "dimesso", "T": "trasferito | altra struttura"},
}, "| `D` | dimesso |"),
("example", {
"question": "Come riconosco un paziente pediatrico?",
"interpretation": "Applicare la formula della fascia pediatrica.",
}, "## Domanda\n\nCome riconosco un paziente pediatrico?"),
("mapping", {
"concept": "fascia pediatrica",
"tables": ["clinical.patient"],
"columns": ["clinical.patient.birth_date"],
}, "## Tabelle\n\n- `clinical.patient`"),
("normalization", {
"input": "PEDS",
"output": "pediatrico",
"rule": "Converte il codice abbreviato nella forma canonica.",
}, "## Output\n\npediatrico"),
("formula", {
"concept": "fascia pediatrica",
"columns": ["clinical.patient.birth_date"],
"sql": "CASE WHEN age < 18 THEN 'pediatrica' ELSE 'adulta' END",
}, "```sql\nCASE WHEN age < 18"),
("reference", {
"url": "https://example.test/linea-guida",
"label": "Linea guida",
"description": "Criteri clinici di riferimento.",
}, "## URL\n\n<https://example.test/linea-guida>"),
])
def test_v2_curated_markdown_renders_and_round_trips_every_typed_payload(
tmp_path, kind, payload, rendered,
):
evidence = CuratedEvidence.model_validate({
**COMMON,
"schema_version": 2,
"kind": kind,
"payload": payload,
})
path = tmp_path / "curated" / kind / "fascia-pediatrica.md"
text = dump_curated_markdown(evidence)
assert rendered in text
assert parse_curated_markdown(text, path=path) == evidence
def test_v2_curated_markdown_renders_review_items_as_readable_blocks(tmp_path):
evidence = CuratedEvidence.model_validate({
**COMMON,
"schema_version": 2,
"kind": "domain",
"review_items": [{
"code": "ambiguous_source_statement",
"message": "Il sorgente non chiarisce la data di riferimento.",
"field": "domain.rule",
}],
"payload": {"rule": "L'età è calcolata alla data di ricovero."},
})
path = tmp_path / "curated" / "domain" / "fascia-pediatrica.md"
text = dump_curated_markdown(evidence)
assert "## Elementi da rivedere" in text
assert "### `ambiguous_source_statement`" in text
assert "Il sorgente non chiarisce la data di riferimento." in text
assert "**Campo:** `domain.rule`" in text
assert parse_curated_markdown(text, path=path) == evidence
@pytest.mark.parametrize("legacy_field", [
"review_items: []\n",
"payload:\n rule: should-not-be-ignored\n",
])
def test_v2_curated_markdown_rejects_body_owned_fields_in_frontmatter(legacy_field):
text = dump_curated_markdown(_domain_evidence_v2()).replace(
"kind: domain\n",
f"kind: domain\n{legacy_field}",
)
with pytest.raises(ValueError, match="frontmatter"):
parse_curated_markdown(text)
def test_v2_curated_markdown_rejects_unstructured_body_content():
text = dump_curated_markdown(_domain_evidence_v2()) + "should-not-be-ignored\n"
with pytest.raises(ValueError, match="unstructured"):
parse_curated_markdown(text)
def test_v2_curated_markdown_round_trips_an_empty_enum_as_an_explicit_empty_state(tmp_path):
evidence = CuratedEvidence.model_validate({
**COMMON,
"schema_version": 2,
"kind": "enum",
"payload": {
"column": "clinical.episode.discharge_status",
"values": {},
},
})
text = dump_curated_markdown(evidence)
assert "Nessun elemento" in text
assert parse_curated_markdown(
text,
path=tmp_path / "curated" / "enum" / "discharge-status.md",
) == evidence
def test_curated_markdown_round_trip_uses_the_kind_specific_key(tmp_path):
evidence = _formula_evidence()
path = tmp_path / "curated" / "formula" / "fascia-pediatrica.md"