kryten-webqueue 0.48.0__tar.gz → 0.49.0__tar.gz

{kryten_webqueue-0.48.0 → kryten_webqueue-0.49.0}/CHANGELOG.md

@@ -1,5 +1,55 @@
 # Changelog
 
+## [0.49.0] - 2026-09-23
+
+### Added
+
+- **Postgres repository port complete (Sortie 2/3)**: all ~140 SQLite domain methods
+  across `catalog`/`queue`/`jobs`/`users` are now implemented for the `asyncpg`/Postgres
+  backend (`_pg_catalog_db.py`, `_pg_queue_db.py`, `_pg_jobs_db.py`, `_pg_users_db.py`),
+  wired into `Database.__init__` for `database.backend == "postgres"`. `search()`/
+  `search_count()` use a hybrid `tsvector` + `pg_trgm` two-CTE query in place of SQLite's
+  FTS5. Live-tested end-to-end against a disposable `webqueue_test` database on
+  `chandra-1` (`tests/test_postgres_domain_crud.py`).
+
+### Fixed
+
+- **`Database` facade never dispatched cross-domain orchestration methods (`browse`,
+  `search`, `get_item`, `is_restricted`, etc.) for the Postgres backend** — these all
+  checked `layout == "partitioned"` only, never `backend == "postgres"`, so a Postgres
+  `Database` would silently fall through to the SQLite-monolith code path. Introduced
+  `Database._domain_dispatch` (`layout == "partitioned" or backend == "postgres"`) and
+  updated every dispatch check to use it.
+- **`rotate_playlist_item_to_bottom` silently failed on every real play completion** in
+  both partitioned SQLite and Postgres: it queried a bare `catalog` table from within the
+  `queue`-only connection/schema, which doesn't exist there (`OperationalError`/
+  `UndefinedTableError`), swallowed by a bare `except Exception` in
+  `queue/completion.py`. Added an `is_partitioned` flag to skip the same-connection
+  lookup, and a `Database.rotate_playlist_item_to_bottom` facade override that resolves
+  `friendly_token -> manifest_url` via the `catalog` domain first, mirroring the existing
+  `_get_reserved_tokens()` pattern.
+
+## [0.48.1] - 2026-09-23
+
+### Fixed
+
+- **Production 500 on `/catalog/browse` in partitioned SQLite mode**: `get_tags()` (facet
+  counts) unconditionally ran the monolith reserved-item exclusion subquery referencing
+  `saved_playlist_items`/`saved_playlists`, tables that live in `queue.sqlite3`, not
+  `catalog.sqlite3`. Every browse/search page load crashed with
+  `sqlite3.OperationalError: no such table: saved_playlist_items` once the site was
+  cut over to `layout="partitioned"`. `get_tags` now accepts `exclude_tokens`/
+  `is_partitioned`, and the `Database` facade resolves reserved/blackout exclusions via
+  the `queue` domain before calling it, matching the existing `browse`/`search` pattern.
+- **Reserved-item exclusion leak in partitioned mode**: `queue.get_reserved_media_ids()`
+  returns raw `media_id` values, which are usually CyTube manifest URLs, not bare
+  `friendly_token`s. `browse`, `browse_count`, `search`, and `search_count` were comparing
+  these raw values directly against `catalog.friendly_token`, so immutable/promo-pool
+  items whose playlist entry stored a manifest URL were **not excluded** from public
+  results. Added `catalog.resolve_friendly_tokens()` to resolve either shape back to
+  `friendly_token`, and all four call sites now use it via a shared
+  `Database._get_reserved_tokens()` helper.
+
 ## [0.48.0] - 2026-09-23
 
 ### Added

{kryten_webqueue-0.48.0 → kryten_webqueue-0.49.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.5
 Name: kryten-webqueue
-Version: 0.48.0
+Version: 0.49.0
 Summary: Netflix/Tubi-style catalog browser and pay-to-play queue management for CyTube
 Author: grobertson
 License-Expression: MIT

{kryten_webqueue-0.48.0 → kryten_webqueue-0.49.0}/deploy/nginx-queue.conf

@@ -13,8 +13,13 @@ server {
 
     add_header Strict-Transport-Security "max-age=31536000" always;
 
+    # Upstream application backend:
+    # - Local grindhouse service: http://127.0.0.1:2010
+    # - Cross-network Chandra-1 Podman container: http://chandra-1.local:2010
+
     # WebSocket upgrade
     location /ws {
+        # Use http://chandra-1.local:2010 when deployed to Chandra-1 Podman pod
         proxy_pass http://127.0.0.1:2010;
         proxy_http_version 1.1;
         proxy_set_header Upgrade $http_upgrade;
@@ -28,6 +33,7 @@ server {
 
     # API + pages
     location / {
+        # Use http://chandra-1.local:2010 when deployed to Chandra-1 Podman pod
         proxy_pass http://127.0.0.1:2010;
         proxy_http_version 1.1;
         proxy_set_header Connection "";

{kryten_webqueue-0.48.0 → kryten_webqueue-0.49.0}/docs/DATABASE_ARCHITECTURE_PLAN.md

@@ -133,18 +133,29 @@ Upon completion of Part 1, the partitioned domain model directly maps to Postgre
 - `tmdb.*` — Persistent, queryable local TMDB dump index.
 
 ```
+┌─────────────────────────────────────────┐
+│          grindhouse.local Host          │
+│                                         │
+│   ┌─────────────────────────────────┐   │
+│   │           nginx proxy           │   │
+│   │    queue.dropsugar.co (SSL)     │   │
+│   └────────────────┬────────────────┘   │
+└────────────────────┼────────────────────┘
+                     │ LAN Reverse Proxy
+                     │ http://chandra-1.local:2010
+                     ▼
 ┌────────────────────────────────────────────────────────────────────────────────────────┐
 │                                 chandra-1 Host System                                  │
 │                                                                                        │
 │  ┌──────────────────────────────────────────────────────────────────────────────────┐  │
-│  │                            Podman Pod: kryten-webqueue                           │  │
+│  │                     Podman Container: webqueue-app                               │  │
 │  │                                                                                  │  │
-│  │  ┌─────────────────────────────┐           ┌──────────────────────────────────┐  │  │
-│  │  │    kryten-webqueue-app      │           │           nginx proxy            │  │  │
-│  │  │   (FastAPI + uvicorn)       │◄─────────►│     (Reverse Proxy + SSL)        │  │  │
-│  │  │   SQLAlchemy 2.0 (asyncpg)  │           │                                  │  │  │
-│  │  └──────────────┬──────────────┘           └────────────────┬─────────────────┘  │  │
-│  └─────────────────┼───────────────────────────────────────────┼────────────────────┘  │
+│  │  ┌─────────────────────────────┐                                                 │  │
+│  │  │    kryten-webqueue-app      │                                                 │  │
+│  │  │   (FastAPI + uvicorn:2010)  │                                                 │  │
+│  │  │   SQLAlchemy 2.0 (asyncpg)  │                                                 │  │
+│  │  └──────────────┬──────────────┘                                                 │  │
+│  └─────────────────┼───────────────────────────────────────────┬────────────────────┘  │
 │                    │                                           │                       │
 │                    ▼                                           ▼                       │
 │  ┌─────────────────────────────────────┐     ┌──────────────────────────────────────┐  │
@@ -156,6 +167,11 @@ Upon completion of Part 1, the partitioned domain model directly maps to Postgre
 └────────────────────────────────────────────────────────────────────────────────────────┘
 ```
 
+The public hostname `https://queue.dropsugar.co/` terminates TLS via Let's Encrypt Nginx on
+`grindhouse.local`. During the Chandra-1 cutover, the Nginx reverse proxy configuration on
+`grindhouse.local` (`/etc/nginx/sites-available/queue.conf`) is updated to forward HTTP and
+WebSocket traffic across the local network to `http://chandra-1.local:2010`.
+
 ### 2.2 Core Technical Specifications
 
 1. **ORM & Driver**: **SQLAlchemy 2.0 with `asyncpg`** (`create_async_engine`, `async_sessionmaker`, `AsyncSession`).

{kryten_webqueue-0.48.0 → kryten_webqueue-0.49.0}/docs/postgres-cutover.md

@@ -80,15 +80,73 @@ sudo systemctl start webqueue-app.service
 sudo systemctl status webqueue-app.service --no-pager
 ```
 
-### Step 6: Smoke Testing & Health Verification
+Verify local health inside `chandra-1`:
 ```bash
 curl -I http://127.0.0.1:2010/auth/login
 curl -s http://127.0.0.1:2010/queue/next-schedule
 ```
 
+### Step 6: Update Grindhouse Nginx Reverse Proxy Across Network
+The public website `https://queue.dropsugar.co/` terminates SSL via Let's Encrypt Nginx on `grindhouse.local`. Update its upstream to proxy across the LAN to `chandra-1.local:2010`.
+
+On `grindhouse.local`:
+1. Edit `/etc/nginx/sites-available/queue.conf`:
+```nginx
+    # WebSocket upgrade
+    location /ws {
+        proxy_pass http://chandra-1.local:2010;
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade $http_upgrade;
+        proxy_set_header Connection "upgrade";
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+        proxy_read_timeout 86400;
+    }
+
+    # API + pages
+    location / {
+        proxy_pass http://chandra-1.local:2010;
+        proxy_http_version 1.1;
+        proxy_set_header Connection "";
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+    }
+```
+
+2. Test and reload Nginx:
+```bash
+sudo nginx -t
+sudo systemctl reload nginx
+```
+
+### Step 7: Public Smoke Testing & Verification
+Test public endpoints through the reverse proxy:
+```bash
+curl -I https://queue.dropsugar.co/auth/login
+curl -s https://queue.dropsugar.co/queue/next-schedule
+```
+Monitor logs on `chandra-1`:
+```bash
+sudo journalctl -u webqueue-app.service -f
+```
+
 ---
 
 ## 3. Rollback & Forward Repair Policy
 
-- **Pre-Cutover Failure**: If ETL validation fails or container startup fails before DNS/traffic cutover, immediately restart `kryten-webqueue.service` on `grindhouse.local`. SQLite databases remain untouched and consistent.
-- **Post-Cutover Failure**: Once traffic is switched and PostgreSQL accepts writes (new spend requests, logins, completions), recovery is a **forward repair in PostgreSQL**. The retained SQLite snapshot is preserved read-only for forensics.
+- **Pre-Cutover Failure**: If ETL validation fails, container startup fails on `chandra-1`, or initial smoke tests fail before Nginx is updated:
+  1. Do not update or revert Nginx on `grindhouse.local` (`proxy_pass http://127.0.0.1:2010;`).
+  2. Restart `kryten-webqueue.service` on `grindhouse.local`:
+     ```bash
+     sudo systemctl start kryten-webqueue.service
+     ```
+  3. The SQLite databases remain untouched and consistent.
+- **Traffic Rollback (if Nginx was already switched but app has issues)**:
+  1. Revert `/etc/nginx/sites-available/queue.conf` on `grindhouse.local` to point back to `http://127.0.0.1:2010`.
+  2. Reload Nginx: `sudo nginx -t && sudo systemctl reload nginx`.
+  3. Start `kryten-webqueue.service` on `grindhouse.local`.
+- **Post-Cutover Forward Repair**: Once production traffic is accepted by PostgreSQL (user logins, new spend requests, play completions), recovery is a **forward repair in PostgreSQL**. The retained SQLite snapshot is preserved read-only for forensics.