research_workflow_items: validate package.json manifest shape in ids (#6856)

scripts/research_workflow_items.py

@@ -949,15 +949,46 @@ def apply(state, main, items):
     return edit(package, apply)
 
 
+def manifest_segments(manifest):
+    """Return the validated segments list from a package.json manifest.
+
+    The manifest is shape-checked before use: the top level must be a
+    JSON object and segments (when present) a list of objects each
+    with a string id. Any violation — including a manifest that is
+    not JSON at all — fails via fail() naming package.json and the
+    field/expectation, never a bare Attribute/Key/Type error.
+    """
+    if not manifest.is_file():
+        return []
+    try:
+        data = json.loads(manifest.read_text())
+    except (json.JSONDecodeError, UnicodeDecodeError):
+        fail(
+            "package.json: manifest is not valid JSON; "
+            "expected a JSON object with a segments list"
+        )
+    if not isinstance(data, dict):
+        fail(
+            "package.json: manifest must be a JSON object; "
+            "expected an object with a segments list"
+        )
+    segments = data.get("segments", [])
+    if not isinstance(segments, list):
+        fail("package.json: segments must be a list of objects each with a string id")
+    for segment in segments:
+        if not isinstance(segment, dict) or not isinstance(segment.get("id"), str):
+            fail(
+                "package.json: segments entries must be objects "
+                "each with a string id"
+            )
+    return segments
+
+
 def list_ids(args, root, config, package):
     state, main, _ = load(package)
     items = parse_items(main, concept_ids(state), state.get("created_at"))
     manifest = package / "package.json"
-    segments = (
-        json.loads(manifest.read_text()).get("segments", [])
-        if manifest.is_file()
-        else []
-    )
+    segments = manifest_segments(manifest)
     return {
         "package": str(package),
         "concepts": [

scripts/tests/test_research_workflow_items.py

@@ -371,6 +371,34 @@ def test_ids_lists_concepts_items_notes_and_decisions(self):
         self.assertEqual(result["segments"], [])
         self.assertEqual(result["reserved"], ["G"])
 
+    def test_ids_validates_the_package_manifest_shape(self):
+        package = self.rendered()
+        manifest = package / "package.json"
+        # Controls: a well-formed manifest lists its segment ids, and a
+        # manifest without segments behaves exactly as no manifest does.
+        manifest.write_text(
+            json.dumps({"segments": [{"id": "S1"}, {"id": "S2"}]}) + "\n",
+            encoding="utf-8",
+        )
+        self.assertEqual(self.cli("ids", "--package", package)["segments"], ["S1", "S2"])
+        manifest.write_text("{}\n", encoding="utf-8")
+        self.assertEqual(self.cli("ids", "--package", package)["segments"], [])
+        # Wrong shapes fail via the tool's single-line error contract,
+        # naming package.json — never a bare Attribute/Key/Type message.
+        for content in (
+            "[]",
+            '"segments"',
+            '{"segments": "S1"}',
+            '{"segments": {"id": "S1"}}',
+            '{"segments": ["S1"]}',
+            '{"segments": [{"name": "S1"}]}',
+            '{"segments": [{"id": 3}]}',
+            "not json",
+        ):
+            manifest.write_text(content + "\n", encoding="utf-8")
+            error = self.cli("ids", "--package", package, ok=False)
+            self.assertIn("package.json", error, content)
+
     # ------------------------------------------------------------- render
 
     def test_render_is_idempotent_and_replaces_init_placeholders(self):