fix(evidence): validate curated runtime corpus

harness/tests/test_preprocess_cli.py

@@ -2,14 +2,18 @@ import json
 from pathlib import Path
 from types import SimpleNamespace
 
+import pytest
 from typer.testing import CliRunner
 
 from tht.cli import app
 
 
-def _runtime_config(tmp_path: Path, name: str = "workspace.yaml") -> Path:
+def _runtime_config(
+    tmp_path: Path, name: str = "workspace.yaml", evidence_schema_version: int | None = None,
+) -> Path:
     path = tmp_path / name
     (tmp_path / "evidence").mkdir(exist_ok=True)
+    evidence_version = "" if evidence_schema_version is None else f"\n  schema_version: {evidence_schema_version}"
     path.write_text(
         f"""
 runtime_identity:
@@ -28,6 +32,7 @@ embeddings:
   model: qwen3-embedding:0.6b
   dim: 1024
 evidence:
+{evidence_version}
   sources:
     - type: filesystem
       root: {tmp_path / 'evidence'}
@@ -40,6 +45,54 @@ roots:
     return path
 
 
+def test_v2_invalid_materialized_corpus_stops_before_vector_store_construction(monkeypatch, tmp_path):
+    import tht.cli.preprocess_cmd as command
+    from tht.evidence import ValidationFinding, ValidationReport
+
+    config = _runtime_config(tmp_path, evidence_schema_version=2)
+    vector_store_constructed = False
+    monkeypatch.setattr(
+        "tht.evidence.validate_workspace_evidence",
+        lambda root: ValidationReport((ValidationFinding(
+            "error", "manifest_missing", "manifest.yaml", "missing",
+        ),)),
+    )
+
+    def forbidden_vector_store(*args, **kwargs):
+        nonlocal vector_store_constructed
+        vector_store_constructed = True
+        raise AssertionError("invalid corpus must not reach vector upsert setup")
+
+    monkeypatch.setattr("tht.adapters.factory.build_vector_store", forbidden_vector_store)
+
+    with pytest.raises(RuntimeError, match="curated Evidence corpus is invalid"):
+        command.run_from_config(config)
+
+    assert vector_store_constructed is False
+
+
+def test_v2_valid_materialized_corpus_is_validated_before_preprocessing(monkeypatch, tmp_path):
+    import tht.cli.preprocess_cmd as command
+    from tht.evidence import ValidationReport
+
+    config = _runtime_config(tmp_path, evidence_schema_version=2)
+    calls = []
+
+    class FakePipeline:
+        def run_as_job(self, **kwargs):
+            calls.append(("run", kwargs))
+            return "completed"
+
+    monkeypatch.setattr("tht.evidence.validate_workspace_evidence", lambda root: calls.append(("validate", root)) or ValidationReport(()))
+    monkeypatch.setattr("tht.adapters.factory.build_vector_store", lambda cfg, require_write: calls.append(("vector", require_write)) or object())
+    monkeypatch.setattr("tht.cli.vector_cmd.make_embedder", lambda cfg: object())
+    monkeypatch.setattr("tht.evidence.build_sources", lambda evidence: [])
+    monkeypatch.setattr("tht.evidence.build_preprocessing_pipeline", lambda **kwargs: FakePipeline())
+
+    assert command.run_from_config(config) == "completed"
+    assert calls[:2] == [("validate", tmp_path), ("vector", True)]
+
+
 def test_preprocess_evidence_json_is_pristine(monkeypatch, tmp_path):
     import tht.cli.preprocess_cmd as command
 

harness/tht/cli/preprocess_cmd.py

@@ -75,6 +75,18 @@ def _candidate_evaluator(cfg, *, vector_store, embedder):
     return evaluate
 
 
+def _validate_materialized_curated_corpus(cfg) -> None:
+    """Fail closed on a v2 pinned filesystem corpus before any vector write is possible."""
+    if cfg.evidence is None or cfg.evidence.schema_version != 2:
+        return
+    if not any(source.type == "filesystem" for source in cfg.evidence.sources):
+        return
+    from tht.evidence import validate_workspace_evidence
+
+    if not validate_workspace_evidence(_evaluation_workspace_root(cfg)).publishable:
+        raise RuntimeError("curated Evidence corpus is invalid")
+
+
 def _evidence_json_payload(cfg, payload: dict, *, code: str, error: str | None = None) -> dict:
     value = {
         **payload,
@@ -158,6 +170,7 @@ def run_from_config(config: Path, *, dry_run: bool = False, resume: str | None =
     cfg = _load_config_or_exit(config)
     if cfg.embeddings is None:
         raise RuntimeError("embeddings are not configured")
+    _validate_materialized_curated_corpus(cfg)
     corpus_root = cfg.paths.artifacts.parent / "corpus"
     vector_store = build_vector_store(cfg, require_write=True)
     embedder = make_embedder(cfg.embeddings)

harness/tht/config.py

@@ -473,6 +473,8 @@ EvidenceSourceConfig = Annotated[
 
 
 class EvidenceSourcesConfig(BaseModel):
+    # Version 2 is the materialized source/curated authoring layout.
+    schema_version: Literal[1, 2] = 1
     # Legacy curated-tree configuration remains accepted during migration.
     source_root: Path | None = None
     # cartella curata a mano nell'ETL (relativa a source_root): unica fonte delle