test(usage): close round-5 nits (empty doc_types, consistency tidy-ups)

Round-5 claude-review (merge-ready; all nits):

- 🟡 Added test_empty_doc_types_normalizes_to_null pinning doc_types=[] → None
  in record_search_usage metadata (matches the None case).
- 🟡 record_search_usage docstring now notes nc_semantic_search_answer always
  meters with doc_types=None (it exposes no doc_types parameter).
- 🟢 BM25HybridSearchAlgorithm.__init__ now sets query_embedding /
  query_token_count alongside _embedded_query, so all three cache fields are
  instance attributes from construction (was relying on the class-level
  SearchAlgorithm defaults).
- 🟢 Ollama embed_batch_with_usage caches _dimension inline (mirrors
  OpenAI/Mistral), so the dimension is set via any embed path.
- 🟢 record_indexing_usage documents the independent-record / partial-failure
  semantics under SUM aggregation.

Deck #67.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

nextcloud_mcp_server/providers/ollama.py

@@ -159,6 +159,11 @@ class OllamaProvider(Provider):
             data = response.json()
             all_embeddings.extend(data["embeddings"])
 
+            # Cache the dimension inline (mirrors OpenAI/Mistral) so it is set
+            # via any embed path, not only an explicit _detect_dimension() call.
+            if self._dimension is None and data["embeddings"]:
+                self._dimension = len(data["embeddings"][0])
+
             prompt_eval = data.get("prompt_eval_count")
             total_tokens += (
                 round(prompt_eval)

nextcloud_mcp_server/search/bm25_hybrid.py

@@ -56,9 +56,13 @@ class BM25HybridSearchAlgorithm(SearchAlgorithm):
         self.score_threshold = score_threshold
         self.fusion = models.Fusion.RRF if fusion == "rrf" else models.Fusion.DBSF
         self.fusion_name = fusion
-        # The query string whose dense embedding is cached in
-        # ``self.query_embedding`` — lets repeated search() calls on this
-        # per-request instance (the doc_types loop) reuse one embedding.
+        # Per-request query-embedding cache. ``_embedded_query`` is the query
+        # string whose dense embedding is held in ``query_embedding`` — repeated
+        # search() calls on this instance (the doc_types loop) reuse it. These
+        # shadow the class-level defaults on SearchAlgorithm; set here so all
+        # three cache fields are instance attributes from construction.
+        self.query_embedding: list[float] | None = None
+        self.query_token_count: int | None = None
         self._embedded_query: str | None = None
 
     @property

nextcloud_mcp_server/server/semantic.py

@@ -70,7 +70,9 @@ async def record_search_usage(
     estimated) — the unit upstream providers bill on, and the same metric the
     indexing path records for chunk embeddings (Deck #67). ``nc_semantic_search``
     and ``nc_semantic_search_answer`` (which reuses it) both flow through here —
-    do not add a second hook.
+    do not add a second hook. ``nc_semantic_search_answer`` exposes no
+    ``doc_types`` parameter, so its searches always meter with
+    ``doc_types=None``.
 
     Best-effort and flag-gated: a metering failure is logged and never breaks
     the search. Unlike the indexing path's chunk-count guard, a 0-token query is

nextcloud_mcp_server/vector/processor.py

@@ -153,6 +153,10 @@ async def record_indexing_usage(
         store = await UsageEventStore.shared()
         # enabled=True: the guard above already confirmed the flag, so the store
         # skips a second uncached Settings build per record (ADR-024).
+        # record_usage_event swallows its own write failures, so the two records
+        # are independent; if pages_chunks somehow raised mid-way, embeddings_-
+        # queries would be skipped, leaving an unmatched pages_chunks row —
+        # acceptable under the (day, metric) SUM-aggregation billing model.
         await store.record_usage_event(
             metric="pages_chunks", value=chunk_count, metadata=metadata, enabled=True
         )

tests/unit/server/test_semantic_metering.py

@@ -75,6 +75,21 @@ async def test_none_token_count_records_zero(store_spy):
     assert kwargs["metadata"]["doc_types"] is None
 
 
+@pytest.mark.unit
+async def test_empty_doc_types_normalizes_to_null(store_spy):
+    """An empty doc_types list normalizes to None, same as a None input, so a
+    metadata->'doc_types' IS NULL query counts the all-types case consistently."""
+    await semantic.record_search_usage(
+        enabled=True,
+        user_id="alice",
+        fusion="rrf",
+        doc_types=[],
+        token_count=5,
+    )
+    kwargs = store_spy.record_usage_event.await_args.kwargs
+    assert kwargs["metadata"]["doc_types"] is None
+
+
 @pytest.mark.unit
 async def test_doc_types_metadata_is_bounded(store_spy):
     """A large doc_types list is truncated to the metadata cap."""