fix(tolaria): C5F Phase B — search-source indexability path-independent (valid C3 object_id suffices; LEGACY kept; out-of-scope dirs no longer auto-exclude indexable knowledge)

tolaria/c5-sync-service/rq_c5d.py

@@ -81,6 +81,7 @@ from rq_c5b import (
     content_hash,
     detect_secret,
     KnowledgeScope,
+    LEGACY_SPECIAL_PATHS,
     SCOPE_IN_SCOPE,
     SCOPE_LEGACY_SPECIAL,
 )
@@ -281,6 +282,17 @@ class SearchSourceBuilder:
         Liest den Vault (read-only) und klassifiziert die indexierbaren Objekte.
         Rueckgabe: {"indexable": [{path, id, content, fm, body}],
                      "excluded": [...], "vault_object_count": int}
+
+        INDEXIERBARKEIT ist PFAD-INDEPENDENT (C5F Phase B):
+        Ein Vault-Objekt ist indexierbar, wenn es
+          * eine gueltige C3 object/<FULL_UUID>-ID im Frontmatter traegt
+            (extract_object_id) ODER
+          * ein expliziter LEGACY_SPECIAL-Sonderfall ist (README.md x2, vps.md;
+            ohne C3-ID, KEEP_DISTINCT / DEFERRED).
+        Der Pfad (z.B. notes/trading/second-brain/, notes/ai-agents/) ist dabei
+        NICHT mehr ausschlaggebend — jedes Markdown mit gueltiger object_id ist
+        ein Knowledge-Objekt und wird indexiert. Objekte OHNE gueltige object_id
+        (ausser Legacy) sind nicht indexierbar -> excluded (HUMAN_REVIEW/Orphan).
         """
         entries = self.tolaria.list(VAULT_PREFIX) or []
         indexable: List[Dict[str, Any]] = []
@@ -294,18 +306,15 @@ class SearchSourceBuilder:
             if content is None:
                 excluded.append({"path": rel, "reason": "unreadable"})
                 continue
-            cls = self.scope.classify_with_content(rel, content)
-            sc = cls.get("scope")
-            if sc == SCOPE_IN_SCOPE:
-                oid = cls.get("object_id")
+            # Pfad-unabhaengige Indexierbarkeit
+            if rel in LEGACY_SPECIAL_PATHS:
+                indexable.append({"path": rel, "id": None, "content": content})
+                continue
+            oid = extract_object_id(content)
             if oid:
                 indexable.append({"path": rel, "id": oid, "content": content})
                 continue
             excluded.append({"path": rel, "reason": "no_object_id"})
-            elif sc == SCOPE_LEGACY_SPECIAL:
-                indexable.append({"path": rel, "id": None, "content": content})
-            else:
-                excluded.append({"path": rel, "reason": sc})
         return {"indexable": indexable, "excluded": excluded,
                 "vault_object_count": len(entries)}
 

tolaria/c5-sync-service/test_c5d_source_pipeline.py

@@ -678,5 +678,70 @@ class TestRealisticIntegration(unittest.TestCase):
             self.assertGreater(base["total"], 0)
 
 
+class TestPhaseB_PathIndependentScope(unittest.TestCase):
+    """C5F Phase B: Indexierbarkeit ist pfad-unabhaengig (gültige C3-ID reicht)."""
+
+    def _collect(self, fake, src=None):
+        import tempfile as _tf
+        if src is None:
+            src = os.path.join(_tf.mkdtemp(prefix="c5fb_"), "source.json")
+        builder = _make_builder(fake, src)
+        return builder._collect_indexable()
+
+    def test_object_under_second_brain_with_valid_id_indexed(self):
+        # Objekt unter notes/trading/second-brain/ (vorher OUT_OF_SCOPE via Pfad)
+        # mit gültiger C3-ID muss jetzt indexiert werden.
+        fake = FakeTolaria()
+        fake.vault[_vault_path("notes/trading/second-brain/regelwerk.md")] = (
+            "---\nkid: 1\nid: " + UUID_A + "\ntitle: Regelwerk\n---\nbody")
+        r = self._collect(fake)
+        paths = [o["path"] for o in r["indexable"]]
+        ids = [o["id"] for o in r["indexable"]]
+        self.assertIn("notes/trading/second-brain/regelwerk.md", paths)
+        self.assertIn(UUID_A, ids)
+
+    def test_object_under_ai_agents_with_valid_id_indexed(self):
+        fake = FakeTolaria()
+        fake.vault[_vault_path("notes/ai-agents/reference.md")] = (
+            "---\nkid: 1\nid: " + UUID_B + "\ntitle: Ref\n---\nbody")
+        r = self._collect(fake)
+        paths = [o["path"] for o in r["indexable"]]
+        self.assertIn("notes/ai-agents/reference.md", paths)
+
+    def test_unknown_file_without_id_excluded(self):
+        # Unbekannte Nicht-Knowledge-Datei OHNE gültige C3-ID -> ausgeschlossen,
+        # auch wenn sie .md ist.
+        fake = FakeTolaria()
+        fake.vault[_vault_path("notes/inbox/notizen.md")] = "kein frontmatter id"
+        r = self._collect(fake)
+        paths = [o["path"] for o in r["indexable"]]
+        excl = [e["path"] for e in r["excluded"]]
+        self.assertNotIn("notes/inbox/notizen.md", paths)
+        self.assertIn("notes/inbox/notizen.md", excl)
+
+    def test_legacy_special_indexed_without_id(self):
+        # LEGACY_SPECIAL (z.B. vps.md) bleibt ohne C3-ID indexierbar.
+        fake = FakeTolaria()
+        fake.vault[_vault_path("vps.md")] = "---\ntitle: VPS\n---\nbody"
+        r = self._collect(fake)
+        paths = [o["path"] for o in r["indexable"]]
+        legacy = [o for o in r["indexable"] if o["id"] is None]
+        self.assertIn("vps.md", paths)
+        self.assertTrue(any(o["path"] == "vps.md" for o in legacy))
+
+    def test_mixed_scope_total_count(self):
+        # Mixed: Root mit ID + second-brain mit ID + Notiz ohne ID + Legacy.
+        fake = FakeTolaria()
+        fake.vault[_vault_path("modul-09.md")] = (
+            "---\nkid: 1\nid: " + UUID_A + "\ntitle: M\n---\nbody")
+        fake.vault[_vault_path("notes/trading/second-brain/x.md")] = (
+            "---\nkid: 1\nid: " + UUID_B + "\ntitle: X\n---\nbody")
+        fake.vault[_vault_path("notes/inbox/notiz.md")] = "kein fm id"
+        fake.vault[_vault_path("README.md")] = "---\ntitle: R\n---\nbody"
+        r = self._collect(fake)
+        self.assertEqual(len(r["indexable"]), 3)  # 2 mit ID + 1 Legacy
+        self.assertEqual(len(r["excluded"]), 1)   # Notiz ohne ID
+
+
 if __name__ == "__main__":
     unittest.main()