feat(index): project review_status into the search index (engine#94)

scripts/generate_search_index.py

@@ -25,6 +25,9 @@
   * ``tags``        — from the manifest, else ``[]``
   * ``visibility``  — consumer-display hint (engine schema 1.8);
     absent or out-of-enum normalizes to ``"visible"``
+  * ``review_status`` - three-state review standing (engine schema 1.9),
+    derived from ORIGIN; absent or out-of-enum normalizes to ``"authored"``.
+    Consumers derive "advertisable" as ``review_status != "generated"``
   * ``ai_validated``— ``true`` if the set/lesson carries an
     ``ai_validation`` block
   * ``trust_level`` — from ``recommended-repos.json``, else ``1``
@@ -47,6 +50,7 @@
 SCHEMA_VERSION = "1.0"
 DEFAULT_TRUST_LEVEL = 1
 VISIBILITY_VALUES = ("visible", "hidden")
+REVIEW_STATUS_VALUES = ("authored", "generated", "reviewed")
 
 
 # --------------------------------------------------------------------------- #
@@ -128,6 +132,15 @@ def load_recommended_trust() -> dict[str, int]:
     return trust
 
 
+def normalize_review_status(raw_status: object) -> str:
+    """Engine-parity projection of the manifest ``review_status`` flag.
+
+    Absent or out-of-enum folds back to ``"authored"`` (legacy hand-written
+    content), so consumers filter without their own defaulting.
+    """
+    return raw_status if raw_status in REVIEW_STATUS_VALUES else "authored"
+
+
 def normalize_visibility(raw_visibility: object) -> str:
     """Engine-parity projection of the manifest ``visibility`` flag.
 
@@ -207,6 +220,7 @@ def build_set_entry(root_set: dict) -> tuple[dict, list[str]]:
         "card_count": card_count,
         "tags": merged.get("tags") or [],
         "visibility": normalize_visibility(merged.get("visibility")),
+        "review_status": normalize_review_status(merged.get("review_status")),
         "ai_validated": has_ai_validation(set_manifest, lessons),
         "trust_level": None,  # filled in by caller (repo-level)
         "book": merged.get("book"),
@@ -270,6 +284,7 @@ def _now_iso() -> str:
     "lesson_count",
     "card_count",
     "visibility",
+    "review_status",
 )
 
 

search-index.json

@@ -1,6 +1,6 @@
 {
   "repo": "astrapi69/alc-programming",
-  "generated": "2026-07-31T15:48:40Z",
+  "generated": "2026-07-31T16:08:57Z",
   "schema_version": "1.0",
   "sets": [
     {
@@ -19,6 +19,7 @@
         "python"
       ],
       "visibility": "visible",
+      "review_status": "authored",
       "ai_validated": false,
       "trust_level": 1,
       "book": null,
@@ -41,6 +42,7 @@
         "grundlagen"
       ],
       "visibility": "visible",
+      "review_status": "authored",
       "ai_validated": false,
       "trust_level": 1,
       "book": null,
@@ -64,6 +66,7 @@
         "multiple-choice"
       ],
       "visibility": "visible",
+      "review_status": "authored",
       "ai_validated": false,
       "trust_level": 1,
       "book": null,

tests/test_generate_search_index.py

@@ -85,3 +85,27 @@ def test_every_index_entry_carries_visibility() -> None:
     assert index["sets"], "index carries no sets"
     for entry in index["sets"]:
         assert entry["visibility"] in ("visible", "hidden")
+
+
+def test_absent_review_status_defaults_to_authored() -> None:
+    assert gsi.normalize_review_status(None) == "authored"
+
+
+def test_review_status_states_pass_through() -> None:
+    for state in ("authored", "generated", "reviewed"):
+        assert gsi.normalize_review_status(state) == state
+
+
+def test_out_of_enum_review_status_normalizes_to_authored() -> None:
+    assert gsi.normalize_review_status("verified") == "authored"
+
+
+def test_every_index_entry_carries_review_status() -> None:
+    """The field has to reach the INDEX, not just the manifest: the index is
+    what consumers read, so a badge counting advertisable sets would
+    otherwise count every set (engine#94)."""
+    index, build_errors = gsi.build_index()
+    assert not build_errors
+    assert index["sets"], "index carries no sets"
+    for entry in index["sets"]:
+        assert entry["review_status"] in ("authored", "generated", "reviewed")