feat(ingest): index generation and stamped-replacement semantics (I2)

tests/test_ingest_materialize.py

@@ -250,3 +250,105 @@ def test_sql_source_has_no_connector_in_i2(tmp_path: Path) -> None:
     )
     with pytest.raises(IngestError):
         materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+
+
+# --- Step 4: index generation + stamped-replacement semantics ---
+
+_TWO_EXTRACTIONS = [
+    {"id": "costs", "title": "Project costs", "query": "costs.csv", "okf_type": "dataset", "max_rows": 100},
+    {"id": "meta", "title": "Catalogue metadata", "query": "meta.csv", "okf_type": "reference", "max_rows": 10},
+]
+_TWO_FILES = {
+    "costs.csv": b"item,cost\nled,120\n",
+    "meta.csv": b"key,value\nowner,anlegg\n",
+}
+
+
+def _bundle_bytes(bundle_dir: Path) -> dict[str, bytes]:
+    return {p.name: p.read_bytes() for p in sorted(bundle_dir.iterdir()) if p.is_file()}
+
+
+def test_fresh_dir_gets_index_with_bundle_summary_and_links_in_manifest_order(
+    tmp_path: Path,
+) -> None:
+    manifest_path, bundle_dir = _project(
+        tmp_path, extractions=copy.deepcopy(_TWO_EXTRACTIONS), files=dict(_TWO_FILES)
+    )
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    index = (bundle_dir / "index.md").read_text(encoding="utf-8")
+    assert index.startswith("Cost extracts from the project archive.\n")
+    link_lines = [line for line in index.splitlines() if line.startswith("- [")]
+    assert link_lines == [
+        "- [Project costs](ingest-costs.md)",
+        "- [Catalogue metadata](ingest-meta.md)",
+    ]
+
+
+def test_curated_index_line_preserved_byte_for_byte(tmp_path: Path) -> None:
+    manifest_path, bundle_dir = _project(tmp_path)
+    bundle_dir.mkdir()
+    curated = "Hand-written summary.\n\n- [Curated note](note.md)\n"
+    (bundle_dir / "index.md").write_bytes(curated.encode("utf-8"))
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    index = (bundle_dir / "index.md").read_text(encoding="utf-8")
+    assert index.startswith(curated)  # §6: every unmanaged line preserved verbatim
+    assert "- [Project costs](ingest-costs.md)" in index
+
+
+def test_removed_extraction_drops_file_and_index_link(tmp_path: Path) -> None:
+    manifest_path, bundle_dir = _project(
+        tmp_path, extractions=copy.deepcopy(_TWO_EXTRACTIONS), files=dict(_TWO_FILES)
+    )
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    # Re-materialize with `meta` dropped from the manifest.
+    manifest_path2, _ = _project(tmp_path, files=dict(_TWO_FILES))
+    materialize(manifest_path2, bundle_dir, ingested_at=_INGESTED_AT)
+    assert not (bundle_dir / "ingest-meta.md").exists()
+    index = (bundle_dir / "index.md").read_text(encoding="utf-8")
+    assert "](ingest-meta.md)" not in index  # §6: link to removed ingest file is gone
+    assert "- [Project costs](ingest-costs.md)" in index  # sibling link intact
+
+
+def test_changed_title_refreshes_label_in_place(tmp_path: Path) -> None:
+    manifest_path, bundle_dir = _project(
+        tmp_path, extractions=copy.deepcopy(_TWO_EXTRACTIONS), files=dict(_TWO_FILES)
+    )
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    renamed = copy.deepcopy(_TWO_EXTRACTIONS)
+    renamed[0]["title"] = "Renamed costs"
+    manifest_path2, _ = _project(tmp_path, extractions=renamed, files=dict(_TWO_FILES))
+    materialize(manifest_path2, bundle_dir, ingested_at=_INGESTED_AT)
+    link_lines = [
+        line
+        for line in (bundle_dir / "index.md").read_text(encoding="utf-8").splitlines()
+        if line.startswith("- [")
+    ]
+    # §6: the label IS the title — refreshed in place, position preserved (no stale label,
+    # no move-to-end).
+    assert link_lines == [
+        "- [Renamed costs](ingest-costs.md)",
+        "- [Catalogue metadata](ingest-meta.md)",
+    ]
+
+
+def test_unstamped_collision_fails_and_leaves_bundle_unmodified(tmp_path: Path) -> None:
+    manifest_path, bundle_dir = _project(tmp_path)
+    bundle_dir.mkdir()
+    (bundle_dir / "index.md").write_bytes(b"Curated.\n")
+    # A curated file occupying a generated filename WITHOUT the ingest stamp (§3): never
+    # overwrite curated content.
+    (bundle_dir / "ingest-costs.md").write_bytes(b"---\ntype: note\n---\n\nCurated body.\n")
+    before = _bundle_bytes(bundle_dir)
+    with pytest.raises(IngestError):
+        materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    assert _bundle_bytes(bundle_dir) == before
+
+
+def test_rematerialization_twice_is_byte_identical_including_index(tmp_path: Path) -> None:
+    manifest_path, bundle_dir = _project(
+        tmp_path, extractions=copy.deepcopy(_TWO_EXTRACTIONS), files=dict(_TWO_FILES)
+    )
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    first = _bundle_bytes(bundle_dir)
+    materialize(manifest_path, bundle_dir, ingested_at=_INGESTED_AT)
+    assert _bundle_bytes(bundle_dir) == first  # §10, index.md included