fix(node,client): query TMDB only once a language is chosen, in that languageHEADmain

TMDB fiches are fetched lazily, on browse, and the fetch used to run whatever
the moment it was first triggered — routinely before the operator had opened
settings and picked a language, so it queried in TMDB's English default. Then
the fiche was cached by tmdb_id alone, with no note of language and a 30-day
TTL, so switching to the intended language afterwards changed nothing: the
English fiche was served until it expired. The operator's only recourse was to
find and wipe the cache by hand (found live 2026-09-26: a whole library indexed
in English although "Français" had been chosen).

Two rules now, both there to make the first fetch the right language rather
than English-then-corrected, and to stop the doubled requests that eventually
get a node rate-limited:

- No language configured, no query. media_meta_req/season_meta_req answer
  confidence 0 and make no TMDB call while tmdb_language is unset; the fetch
  waits for the operator's choice, so the first (and only) query is in it.
  English is now a first-class choice (en-US), not the default of skipping the
  setting.
- Changing the language wipes the metadata cache (ops.set_tmdb_config), so the
  new language takes effect on an already-browsed library. The file->tmdb
  matches are language-independent and kept. The client refetches on the
  tmdb_config_ack that carries the new language, so the grid updates without a
  page reload.

docs/MESHBAY_DESIGN.md §9.7 states both rules; tests cover the gate and the
cache wipe, and two existing handler harnesses now declare a language.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

3 files changed, 78 insertions, 0 deletions

diff --git a/packages/meshbay-node/src/meshbay_node/media_cache.py b/packages/meshbay-node/src/meshbay_node/media_cache.py
index 9c692ca..17262bc 100644
--- a/packages/meshbay-node/src/meshbay_node/media_cache.py
+++ b/packages/meshbay-node/src/meshbay_node/media_cache.py
@@ -290,6 +290,33 @@ class MediaCache:
         await self._db.commit()
         return cur.rowcount
 
+    async def clear_tmdb_metadata(self) -> int:
+        """
+        Drop every cached TMDB fiche — show/movie details (`tmdb_meta`) and
+        per-season metadata (`season_meta`) — so the next `media_meta_req`
+        refetches each from TMDB. The file->tmdb *matches* (`file_tmdb`) are
+        language-independent and deliberately kept: the match is the same
+        title whatever language its blurb is in.
+
+        Called when the node's TMDB *language* changes (`ops.set_tmdb_config`).
+        A fiche is cached under `tmdb_id` alone, on purpose — there is only
+        ever one node-wide language, so a per-language key would be dead
+        weight — which is exactly why the language it was fetched in is not
+        recorded, and a fiche cached under the old language would otherwise be
+        served unchanged for its whole 30-day TTL after the operator switched.
+        Wiping them on the switch is what makes the new language actually take
+        effect on a library that has already been browsed. Returns the number
+        of rows removed.
+        """
+        if not self._db:
+            return 0
+        cur = await self._db.execute("DELETE FROM tmdb_meta")
+        removed = cur.rowcount
+        cur = await self._db.execute("DELETE FROM season_meta")
+        removed += cur.rowcount
+        await self._db.commit()
+        return removed
+
     # ── tmdb id -> metadata json ─────────────────────────────────────────────
 
     async def get_tmdb_meta(self, tmdb_id: str, media_type: str) -> dict | None:
diff --git a/packages/meshbay-node/src/meshbay_node/ops/apps.py b/packages/meshbay-node/src/meshbay_node/ops/apps.py
index 0e4f6dd..fbd984d 100644
--- a/packages/meshbay-node/src/meshbay_node/ops/apps.py
+++ b/packages/meshbay-node/src/meshbay_node/ops/apps.py
@@ -53,6 +53,13 @@ async def set_tmdb_config(state: dict, token: str | None = None,
     `language`.
     """
     roster = _roster(state)
+    # Whether the *language* actually changes decides whether the cached
+    # fiches must go (below) — read the old value before overwriting it.
+    # "" (default/English) and None (unset) are the same language here.
+    language_changed = False
+    if language is not None:
+        _, old_language = await roster.tmdb_config()
+        language_changed = (language or None) != (old_language or None)
     await roster.set_tmdb_config(token, language, set_by=state.get("node_user_id", ""))
     # `token=None` means "leave whatever was there" (§ set_tmdb_config's own
     # docstring) — so the customized flag only changes when a value (a real
@@ -61,6 +68,19 @@ async def set_tmdb_config(state: dict, token: str | None = None,
         state["tmdb_token_customized"] = bool(token)
     if language is not None:
         state["tmdb_language"] = language
+    # A cached TMDB fiche is stored under its tmdb_id alone and carries no note
+    # of the language it was fetched in (there is only one node-wide language),
+    # so changing the language leaves every fiche stale for its 30-day TTL.
+    # Drop the metadata cache here so the next media_meta_req refetches in the
+    # new language — this is what makes the setting take on a library that was
+    # already browsed, instead of the operator having to find a cache to clear.
+    # The matches (file_tmdb) are language-independent and kept.
+    if language_changed:
+        media_cache = state.get("media_cache")
+        if media_cache is not None:
+            removed = await media_cache.clear_tmdb_metadata()
+            log.info("TMDB language changed to %s: cleared %d cached fiche(s)",
+                     language or "(default)", removed)
     log.info("TMDB config: custom_token=%s language=%s",
              bool(token), language or state.get("tmdb_language", ""))
     return {
diff --git a/packages/meshbay-node/src/meshbay_node/transport/webrtc/apps/video_meta.py b/packages/meshbay-node/src/meshbay_node/transport/webrtc/apps/video_meta.py
index 9222ad8..834bac4 100644
--- a/packages/meshbay-node/src/meshbay_node/transport/webrtc/apps/video_meta.py
+++ b/packages/meshbay-node/src/meshbay_node/transport/webrtc/apps/video_meta.py
@@ -169,6 +169,15 @@ class VideoMetaMixin:
         await media_cache.put_thumb(thumb_hash, synthetic_id, content)
         return thumb_hash
 
+    def _tmdb_language(self) -> str:
+        """
+        The node-wide TMDB query language, or "" when the operator has not
+        chosen one yet. Read from the live daemon state (kept current by
+        tmdb_config_ack, same source the handshake ack reads), not the DB, so
+        it is a cheap in-memory lookup on the hot metadata path.
+        """
+        return (self._ctx.get("daemon_state") or {}).get("tmdb_language") or ""
+
     async def _do_media_meta_request(self, msg: dict) -> None:
         """
         docs/MESHBAY_DESIGN.md §9.7: TMDB metadata for one file, resolved from
@@ -250,6 +259,19 @@ class VideoMetaMixin:
                 self._send({"type": MNP.MEDIA_META_RESP, "v": MNP_VERSION,
                            "file_id": file_id, "confidence": 0})
                 return
+            if not self._tmdb_language():
+                # No query language chosen yet: hold off entirely rather than
+                # search now. TMDB would answer in its English default, which
+                # is both the wrong language and a wasted call — the whole
+                # library fetched now would be thrown away and refetched the
+                # moment a language is set, doubling the request count against
+                # TMDB's rate limit. Waiting until the operator has chosen one
+                # is what makes the first (and only) fetch the chosen language
+                # (docs/MESHBAY_DESIGN.md §9.7). The client refetches on the
+                # tmdb_config_ack that carries the new language.
+                self._send({"type": MNP.MEDIA_META_RESP, "v": MNP_VERSION,
+                           "file_id": file_id, "confidence": 0})
+                return
             result, ratio = await self._tmdb_search(tmdb_client, entry, is_show)
             if result is None or ratio < 0.6:
                 self._send({"type": MNP.MEDIA_META_RESP, "v": MNP_VERSION,
@@ -315,6 +337,15 @@ class VideoMetaMixin:
 
         details = await media_cache.get_season_meta(tmdb_id, season)
         if details is None:
+            if not self._tmdb_language():
+                # Same gate as media_meta_req above: no query language yet
+                # means no TMDB call (docs/MESHBAY_DESIGN.md §9.7). A show is
+                # only matched once a language is set, so this is normally
+                # unreachable, but a client holding a tmdb_id from an earlier
+                # session must not reopen an English fetch either.
+                self._send({"type": MNP.SEASON_META_RESP, "v": MNP_VERSION,
+                           "tmdb_id": tmdb_id, "season": season, "confidence": 0})
+                return
             fetched = await tmdb_client.tv_season(tmdb_id, season)
             if fetched is None:
                 self._send({"type": MNP.SEASON_META_RESP, "v": MNP_VERSION,