update workflow

.github/workflows/docker-publish.yml

@@ -1,59 +1,77 @@
-name: Publish image
+name: Build and Publish Docker Image
 
-# Build and push the feeder image to the GitHub Container Registry (ghcr.io):
-#   * every push to main  -> the moving `main` tag
-#   * every `v*` tag       -> semver tags (e.g. v1.0.0 -> 1.0.0, 1.0, 1)
-# The build context is the repo root (the workspace) so the vendored
-# `meta-feeder-sdk` path dependency resolves; the Dockerfile lives under
-# feeder-plugin/torrent-feeder/.
 on:
   push:
-    branches: [main]
+    branches:
-    tags: ['v*']
+      - main
+    tags:
+      - 'v*'
+  pull_request:
+    branches:
+      - main
+  workflow_dispatch:
 
 env:
   REGISTRY: ghcr.io
   IMAGE_NAME: ${{ github.repository }}
 
 jobs:
-  publish:
+  build-and-push:
     runs-on: ubuntu-latest
     permissions:
       contents: read
       packages: write
+
     steps:
-      - name: Checkout
+      - name: Checkout repository
         uses: actions/checkout@v4
 
+      - name: Free disk space
+        uses: jlumbroso/free-disk-space@main
+        with:
+          tool-cache: false
+          android: true
+          dotnet: true
+          haskell: true
+          large-packages: true
+          docker-images: true
+          swap-storage: true
+
+      - name: Set up QEMU
+        uses: docker/setup-qemu-action@v3
+
       - name: Set up Docker Buildx
         uses: docker/setup-buildx-action@v3
 
-      - name: Log in to GitHub Container Registry
+      - name: Log in to Container Registry
         uses: docker/login-action@v3
         with:
           registry: ${{ env.REGISTRY }}
           username: ${{ github.actor }}
           password: ${{ secrets.GITHUB_TOKEN }}
 
-      - name: Extract image metadata
+      - name: Extract metadata (tags, labels)
         id: meta
         uses: docker/metadata-action@v5
         with:
           images: ${{ env.REGISTRY }}/${{ env.IMAGE_NAME }}
           tags: |
             type=ref,event=branch
+            type=ref,event=pr
+            type=sha
             type=semver,pattern={{version}}
             type=semver,pattern={{major}}.{{minor}}
             type=semver,pattern={{major}}
-            type=sha,format=short
+            type=raw,value=latest,enable={{is_default_branch}}
 
-      - name: Build and push
+      - name: Build and push Docker image
         uses: docker/build-push-action@v6
         with:
           context: .
-          file: ./feeder-plugin/torrent-feeder/Dockerfile
+          platforms: linux/amd64,linux/arm64
           push: true
           tags: ${{ steps.meta.outputs.tags }}
           labels: ${{ steps.meta.outputs.labels }}
+          provenance: false
           cache-from: type=gha
           cache-to: type=gha,mode=max

README.md

@@ -1 +1,128 @@
 # meta-feeder-tribler
+
+**Tribler feeder sidecar** for
+[MetaMesh](https://github.com/worph/meta-gateway) — bridges the
+**decentralized [Tribler](https://www.tribler.org/) network** (its IPv8
+peer-to-peer overlay search) into a meta-gateway as a search source.
+
+Unlike a Torznab/Prowlarr feeder that talks to centralized indexer sites,
+Tribler has **no central index**: it runs its own anonymous overlay and each
+peer contributes the torrents it has seen. This feeder does not reimplement
+that overlay — it runs a **headless Tribler core as a sidecar container**
+(`tribler-instance`) and queries its REST API, exactly the way a Torznab feeder
+treats an external indexer. It bridges Tribler's *content corpus* into the
+mesh so those torrents become discoverable and playable via meta-share.
+
+This feeder was **split out of** [`meta-feeder-torrent`](../meta-feeder-torrent)
+so that Tribler is a **single-plugin** feeder. A feeder exposes one `/config`
+surface; a multi-plugin feeder (Torznab + Tribler in one binary) can only route
+that surface to one plugin, so its dashboard config panel could not resolve to
+the Tribler schema. Splitting gives Tribler its own clean config card.
+
+## Metadata-only
+
+This is the **light** feeder. It resolves a Tribler search hit's infohash
+directly into a `btih-v1-file` CID and **advertises** the record — it does
+**not** download or seed the bytes. There is no `librqbit` here: file lists
+come from the Tribler core's own `torrentinfo` metainfo response, not from a
+BitTorrent fetch. So the cold Docker build is much lighter than the torrent
+feeder's, and full-file fetch/seed is left to meta-share's on-demand fetcher
+(the feeder synthesizes magnets with a set of well-known public trackers so
+that fetcher can find peers quickly, since Tribler torrents are often DHT-only).
+
+It carries a **copy** of the source-agnostic "torrent-core" TMDB
+catalog-discovery + enrichment stack (`tmdb`, `tmdb_budget`, `discovery`,
+`enrich`, `title`, `consts`, `filename_meta`). That stack cannot live in a
+container plugin — a container plugin enriches a *known* CID; it cannot answer
+"what's popular" for the home-row browse — so the browse/catalog primitive is
+inherently in-process and shared verbatim by every torrent source. Like every
+feeder it stays **meta-core-free at the transport layer and blockstore-free**:
+it finds records and enriches them; publishing resolved records + TMDB posters
+goes to the meta-core configured in the dashboard.
+
+## Role in MetaMesh
+
+A feeder is a stateless HTTP sidecar. The gateway registers it as a remote
+feeder plugin pointing at its `/` and drives the SDK contract
+([`meta_feeder_sdk::serve_feeders`](crates/meta-feeder-sdk/src/serve.rs)):
+
+| Endpoint | Purpose |
+|----------|---------|
+| `GET /manifest` | feeder identity + served types (`*` / `*` — Tribler is free-text, any query can return anything) |
+| `GET /health` | liveness (`Degraded` until `configure()` runs) |
+| `POST /query`, `POST /query_stream` | structured / streaming search over the Tribler local + remote (IPv8) overlay |
+| `POST /compute` | outcome compute + TMDB enrichment |
+| `GET /fetch/:upstream_id/:record_id`, `GET /blob/:upstream_id/:cid` | SDK byte routes (metadata-only here — no local bytes to serve) |
+| `POST /enrich/callback` | async enrichment callback sink |
+| `GET /config`, `GET /config/schema`, `GET\|PUT /config/values` | schema-driven config UI + API |
+
+## Configuration
+
+All Tribler config is **dashboard config only** — it lives in the persisted
+`config.json` (written via the gateway dashboard) and has **no env var**. This
+is the MetaMesh rule: a field in the web-UI config schema must not also be an
+env seed (config from JSON / web UI is the single source of truth).
+
+Configured on the plugin's card in the gateway dashboard:
+
+| Field | Notes |
+|-------|-------|
+| `meta_core_url` | meta-core this feeder publishes resolved records + TMDB posters to (e.g. `http://metacore-app:9000`). Blank → publish soft-skips. |
+| `sidecar_url` | Tribler core REST base (default `http://tribler-instance:8085`). |
+| `api_key` | Tribler REST `X-Api-Key` (secret); the dev sidecar uses `changeme`. Blank → no auth header. |
+| `tmdb_api_key` | TMDB v3 key / v4 bearer (secret) for poster / overview / tmdbid enrichment. Blank → TMDB enrichment soft-skips. |
+| `tmdb_language` | TMDB metadata language tag (e.g. `en-US`, `fr-FR`, `ja-JP`). Blank → `en-US`. |
+
+Config changes take effect on the next feeder restart (no hot reload). Open the
+Tribler web UI via the "↗ Tribler UI" link on the plugin's card — it opens the
+gated `/ui/` behind single sign-on (put `?key=` **after** the `#`; the UI is a
+HashRouter).
+
+Env vars are **infra only**:
+
+| Env var | Default | Notes |
+|---------|---------|-------|
+| `META_FEEDER_HTTP_LISTEN` | `0.0.0.0:8080` | HTTP listen address |
+| `META_FEEDER_STATE_DIR` | `/data/meta-feeder` | redb midhash cache + state |
+| `RUST_LOG` | `info` | tracing filter |
+
+## Image
+
+```
+ghcr.io/worph/meta-feeder-tribler
+```
+
+Exposes `8080`. Built and pushed by CI on every push to `main` (the moving
+`main` tag) and on `v*` tags (semver tags). Metadata-only (no `librqbit`), so
+the cold build is lighter than `meta-feeder-torrent`.
+
+## Build locally
+
+The build context is the **repo root** (the Cargo workspace) so the vendored
+`meta-feeder-sdk` path dependency resolves:
+
+```bash
+docker build -f feeder-plugin/tribler-feeder/Dockerfile -t ghcr.io/worph/meta-feeder-tribler:dev .
+```
+
+## Repo layout
+
+This repo is a self-contained Cargo workspace vendored out of the
+`meta-gateway` monorepo:
+
+```
+Cargo.toml                      # workspace: members = crates/*, feeder-plugin/*
+                                # (keeps the core2 [patch.crates-io] safety net)
+crates/meta-feeder-sdk/         # vendored shared feeder SDK
+feeder-plugin/tribler-feeder/   # this feeder's crate + Dockerfile
+    src/tribler/                # the Tribler source (IPv8-overlay REST client)
+    src/{tmdb,discovery,enrich,title,filename_meta,...}.rs
+                                # copied source-agnostic "torrent-core" catalog stack
+```
+
+> The workspace keeps the `core2` `[patch.crates-io]` override (a yanked
+> transitive dep) as a safety net, matching the `meta-gateway` workspace.
+
+Upstream source of truth for the SDK and the Tribler source is
+[`worph/meta-gateway`](https://github.com/worph/meta-gateway); changes there are
+vendored back into this repo.