feat: complete evidence restructuring worktree

harness/tht/evidence/authoring.py

@@ -293,6 +293,15 @@ class EvidencePreparationReport:
     model_calls: int
 
 
+@dataclass(frozen=True)
+class EvidenceMigrationReport:
+    """Result of a deterministic Curated Evidence presentation-format migration."""
+
+    migrated: tuple[str, ...]
+    unchanged: tuple[str, ...]
+    findings: tuple[ValidationFinding, ...]
+
+
 @dataclass(frozen=True)
 class EvidenceResolutionReport:
     """The result of one curator-directed Evidence resolution."""
@@ -680,6 +689,48 @@ def prepare_workspace_evidence(
     )
 
 
+def migrate_workspace_evidence(
+    workspace_root: Path,
+    *,
+    git_status: Callable[[Path], tuple[str, ...]] | None = None,
+) -> EvidenceMigrationReport:
+    """Rewrite v1 Curated units as readable v2 Markdown without changing semantics."""
+    workspace_root = workspace_root.resolve()
+    evidence_root = workspace_root / "evidence"
+    _reject_dirty_authoring_state(workspace_root, git_status or _git_status)
+    try:
+        manifest = load_manifest(evidence_root / "manifest.yaml")
+        documents = load_curated_tree(evidence_root / "curated")
+    except (OSError, ValidationError, ValueError) as error:
+        raise EvidencePreparationError("authoring_state_invalid") from error
+
+    documents_by_id = {document.id: document for document in documents}
+    if len(documents_by_id) != len(documents):
+        raise EvidencePreparationError("duplicate_evidence_id")
+    migrated = tuple(sorted(
+        document.id for document in documents if document.schema_version == 1
+    ))
+    unchanged = tuple(sorted(
+        document.id for document in documents if document.schema_version == 2
+    ))
+    if not migrated:
+        return EvidenceMigrationReport(
+            migrated=(),
+            unchanged=unchanged,
+            findings=validate_workspace_evidence(workspace_root).findings,
+        )
+    upgraded = {
+        evidence_id: document.model_copy(update={"schema_version": 2})
+        for evidence_id, document in documents_by_id.items()
+    }
+    findings = _stage_and_apply_authoring_tree(workspace_root, upgraded, manifest)
+    return EvidenceMigrationReport(
+        migrated=migrated,
+        unchanged=unchanged,
+        findings=findings,
+    )
+
+
 def resolve_workspace_evidence(
     workspace_root: Path,
     evidence_id: str,
@@ -827,7 +878,7 @@ def _load_resolution_source(evidence_root: Path, source_file: str) -> str:
 
 def _git_status(workspace_root: Path) -> tuple[str, ...]:
     result = subprocess.run(
-        ["git", "status", "--porcelain"],
+        ["git", "status", "--porcelain", "--untracked-files=all"],
         cwd=workspace_root,
         check=False,
         capture_output=True,
@@ -835,7 +886,25 @@ def _git_status(workspace_root: Path) -> tuple[str, ...]:
     )
     if result.returncode != 0:
         raise EvidencePreparationError("canonical_git_worktree_required")
-    return tuple(line for line in result.stdout.splitlines() if line)
+    prefix_result = subprocess.run(
+        ["git", "rev-parse", "--show-prefix"],
+        cwd=workspace_root,
+        check=False,
+        capture_output=True,
+        text=True,
+    )
+    if prefix_result.returncode != 0:
+        raise EvidencePreparationError("canonical_git_worktree_required")
+    prefix = prefix_result.stdout.strip()
+    entries: list[str] = []
+    for line in result.stdout.splitlines():
+        if not line:
+            continue
+        path = line[3:] if len(line) > 3 else line
+        if prefix and path.startswith(prefix):
+            line = line[:3] + path.removeprefix(prefix)
+        entries.append(line)
+    return tuple(entries)
 
 
 def _reject_dirty_authoring_state(
@@ -934,7 +1003,11 @@ def _candidate_to_evidence(
     source_file: str,
     source_hash: str,
 ) -> CuratedEvidence:
-    data = candidate.model_dump(mode="json", exclude={"existing_id", "supporting_excerpts"})
+    data = candidate.model_dump(
+        mode="json",
+        exclude={"schema_version", "existing_id", "supporting_excerpts"},
+    )
+    data["schema_version"] = 2
     data["id"] = evidence_id
     data["provenance"] = {
         "source_file": source_file,
@@ -955,6 +1028,7 @@ def _unsupported_unit(
         message="The current source no longer supports this Evidence unit.",
     ),)
     return evidence.model_copy(update={
+        "schema_version": 2,
         "provenance": evidence.provenance.model_copy(update={
             "source_file": source_file,
             "source_sha256": source_hash,