Add CI workflow, full env docs, fix Up button navigation

- .gitea/workflows/build.yaml: Gitea Actions — push latest on main
- docker-compose.example.yml: image-based pull, all NAT20_ vars shown
- settings.py: add NAT20_PLAYER_RECAP_STYLE / CUSTOM_PROMPT overrides
- README.md: full env var reference table with accepted values + rationale
- files.py: fix parent_path to avoid up-button bug; add path traversal guard
- Files.tsx: remove copy/paste/autoPlay, scope to campaign

.gitea/workflows/build.yaml

@@ -0,0 +1,29 @@
+name: Build and Push
+on:
+  push:
+    branches: [main]
+jobs:
+  build:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+
+      - uses: docker/setup-buildx-action@v3
+
+      - name: Login to Gitea Container Registry
+        run: echo "${{ secrets.GITEA_TOKEN }}" | docker login gitea.kansaigaijin.com -u ${{ gitea.repository_owner }} --password-stdin
+
+      - name: Build and push backend
+        uses: docker/build-push-action@v5
+        with:
+          context: ./backend
+          target: runtime
+          push: true
+          tags: gitea.kansaigaijin.com/${{ gitea.repository }}/backend:latest
+
+      - name: Build and push frontend
+        uses: docker/build-push-action@v5
+        with:
+          context: ./frontend
+          push: true
+          tags: gitea.kansaigaijin.com/${{ gitea.repository }}/frontend:latest

README.md

@@ -40,39 +40,62 @@ A reference compose file using named volumes only (no host paths) is at
 
 ## Data layout
 
-The app stores everything under `/data` (inside the container), which by default
+The app stores everything under `/data` (inside the container):
-is a [bind mount](./docker-compose.yml) to a host path of your choice:
 
 | Directory / File | Contents |
 |---|---|
-| `audio/` | Uploaded recordings and extracted audio |
+| `campaigns/{id}/audio/` | Uploaded recordings and extracted audio, organised per campaign |
-| `transcriptions/` | Per-session transcript JSON files |
+| `campaigns/{id}/transcriptions/` | Per-session transcript JSON files |
-| `notes/` | Generated notes (GM log + player recap) |
+| `campaigns/{id}/notes/` | Generated notes (GM log + player recap) |
-| `app.db` | SQLite database (sessions, speakers, settings, jobs) |
+| `app.db` | SQLite database (sessions, speakers, campaigns, settings, jobs) |
 
 ## Configuration
 
-### Prefilling the setup wizard
+### Environment variables (`NAT20_*`)
 
-Set `NAT20_*` environment variables under the `backend` service in
+Set these under the `backend` service in `docker-compose.yml` to prefill
-`docker-compose.yml` — the wizard will pick them up as defaults:
+the setup wizard and override defaults. All are optional — the wizard
+and Settings page can set them at runtime.
 
-```yaml
+#### Transcription
-environment:
-  NAT20_HF_TOKEN: "hf_..."
-  NAT20_WHISPER_MODEL: medium
-  NAT20_OLLAMA_HOST: http://localhost:11434
-  NAT20_WORLD_CONTEXT_PATH: /data/campaign-context.txt
-```
 
-See the `environment:` block in `docker-compose.yml` for the full list.
+| Variable | What it does | Values | Default | Notes |
+|---|---|---|---|---|
+| `NAT20_WHISPER_MODEL` | Whisper model size | `tiny` `base` `small` `medium` `large-v3` | `medium` | Larger = more accurate but uses more VRAM. `medium` fits most 6-8 GB GPUs; `large-v3` needs ~10 GB+ |
+| `NAT20_WHISPER_COMPUTE_TYPE` | Compute precision | `int8` `float16` `float32` | `int8` | `int8` = fastest / least VRAM. `float16` = more accurate, more VRAM. `float32` = full precision, slowest |
+| `NAT20_HF_TOKEN` | HuggingFace token for speaker diarization | `hf_...` | (none) | Required for speaker attribution. Must accept pyannote gated-model terms with the same account first |
+
+#### LLM backend
+
+| Variable | What it does | Values | Default | Notes |
+|---|---|---|---|---|
+| `NAT20_OLLAMA_HOST` | Ollama server URL | URL | `http://localhost:11434` | Set to your Ollama host |
+| `NAT20_OLLAMA_MODEL` | Ollama model name | any model on your server | `qwen2.5:7b` | |
+| `NAT20_API_BASE_URL` | OpenAI-compatible API base | URL | `https://api.openai.com/v1` | Uncomment and set to switch from Ollama |
+| `NAT20_API_KEY` | API key | string | (none) | |
+| `NAT20_API_MODEL` | API model name | string | `gpt-4o-mini` | |
+
+#### Summarization
+
+| Variable | What it does | Values | Default | Notes |
+|---|---|---|---|---|
+| `NAT20_CHUNK_WORD_TARGET` | Target words per summarisation chunk | number | `2500` | Long transcripts are split into chunks, each summarised separately, then combined. **Lower** = more LLM calls but finer granularity (good for models with small context windows). **Higher** = more context per chunk but may exceed the model's window. `2500` is safe for most models (8K–128K context) |
+| `NAT20_WORLD_CONTEXT` | Campaign context string | text | (none) | Injected into every summarisation prompt so the LLM recognises NPCs, places, and lore correctly |
+| `NAT20_WORLD_CONTEXT_PATH` | Path to campaign context file inside container | container path | (none) | Alternative to `NAT20_WORLD_CONTEXT` for large campaign bibles you update independently |
+| `NAT20_PLAYER_RECAP_STYLE` | Player recap format | `story` `diary` `bullets` `custom` | `story` | |
+| `NAT20_PLAYER_RECAP_CUSTOM_PROMPT` | Custom prompt (only when style is `custom`) | text | (none) | |
+
+#### Setup wizard
+
+| Variable | What it does | Values | Default | Notes |
+|---|---|---|---|---|
+| `NAT20_ONBOARDING_COMPLETED` | Skip the setup wizard | `true` or `false` | `false` | Set to `"true"` after you finish the wizard once |
 
 ### Campaign context
 
 You can paste context directly in the Settings page, or point to a file
-inside the container using the `world_context_path` setting (or the
+inside the container using `NAT20_WORLD_CONTEXT_PATH`. The file version is
-`NAT20_WORLD_CONTEXT_PATH` env var). The file path version is useful for
+useful for large campaign bibles that you update independently.
-large campaign bibles that you update independently.
 
 ## Notes on hardware
 

backend/app/routers/settings.py

@@ -25,6 +25,8 @@ _ENV_OVERRIDES = {
     "chunk_word_target": "NAT20_CHUNK_WORD_TARGET",
     "world_context": "NAT20_WORLD_CONTEXT",
     "world_context_path": "NAT20_WORLD_CONTEXT_PATH",
+    "player_recap_style": "NAT20_PLAYER_RECAP_STYLE",
+    "player_recap_custom_prompt": "NAT20_PLAYER_RECAP_CUSTOM_PROMPT",
 }
 
 _BOOL_KEYS = {"onboarding_completed"}

docker-compose.example.yml

@@ -1,26 +1,69 @@
-# Example Docker Compose — uses a named volume so data persists across rebuilds
+# Example Docker Compose — copy to docker-compose.yml and customise.
-# without needing a specific host path. Copy to docker-compose.yml and customise.
+# Build locally:  docker compose build
+# Pull pre-built: docker compose pull
+# Run:            docker compose up -d
 services:
-  whisperx-base:
-    build:
-      context: ./backend
-      target: whisperx-base
-    image: nat20-whisperx-base:latest
-    restart: "no"
-
   backend:
-    build:
+    # ── Pre-built image (pull from registry) ──────────────────
-      context: ./backend
+    image: gitea.kansaigaijin.com/Jamie/Nat20-Notes/backend:latest
-      target: runtime
+    # ── Or build locally (uncomment below) ────────────────────
-    image: nat20-backend:latest
+    # build:
+    #   context: ./backend
+    #   target: runtime
+
     restart: unless-stopped
-    depends_on:
-      - whisperx-base
     environment:
-      # Skip setup wizard on container restart (or "false" to force re-run)
+      # ── Setup wizard ──────────────────────────────────────
-      NAT20_ONBOARDING_COMPLETED: "true"
+      # "false" (or omit) to show the wizard on first run.
+      # Set to "true" after onboarding completes.
+      NAT20_ONBOARDING_COMPLETED: "false"
+
+      # ── Transcription ─────────────────────────────────────
+      # Model size:  tiny | base | small | medium | large-v3
+      NAT20_WHISPER_MODEL: medium
+      # Compute precision:  int8 (fastest/least VRAM)
+      #                    | float16 (more accurate)
+      #                    | float32 (full precision, slowest)
+      NAT20_WHISPER_COMPUTE_TYPE: int8
+
+      # Required for speaker diarization (accept HF gated-model terms first)
+      # NAT20_HF_TOKEN: "hf_..."
+
+      # ── LLM backend (Ollama — local) ──────────────────────
+      NAT20_OLLAMA_HOST: http://localhost:11434
+      NAT20_OLLAMA_MODEL: qwen2.5:7b
+
+      # ── LLM backend (API — OpenAI-compatible) ─────────────
+      # Uncomment these to use a hosted API instead of Ollama
+      # NAT20_API_BASE_URL: https://api.openai.com/v1
+      # NAT20_API_KEY: "sk-..."
+      # NAT20_API_MODEL: gpt-4o-mini
+
+      # ── Summarization ─────────────────────────────────────
+      # Target words per chunk. Long transcripts are split into
+      # chunks, each summarized separately.  Lower = more LLM
+      # calls but finer granularity.   Higher = more context per
+      # chunk but may exceed the model's context window.
+      # Default 2500 works for most models (8K–128K context).
+      # NAT20_CHUNK_WORD_TARGET: "2500"
+
+      # Campaign context injected into every summarization prompt.
+      # Inline string or path to a file inside the container.
+      # NAT20_WORLD_CONTEXT: ""
+      # NAT20_WORLD_CONTEXT_PATH: /data/campaign-context.txt
+
+      # Player recap format:  story | diary | bullets | custom
+      # NAT20_PLAYER_RECAP_STYLE: story
+      # NAT20_PLAYER_RECAP_CUSTOM_PROMPT: ""
+
     volumes:
+      # ── Data persistence ──────────────────────────────────
+      # Option A: Named volume (auto-managed, no host path needed)
       - app_data:/data
+      # Option B: Host bind mount (replace with your path)
+      # - /mnt/media/dnd-sessions:/data
+
+      # HuggingFace + Torch model caches (avoid re-downloading)
       - hf_cache:/root/.cache/huggingface
       - torch_cache:/root/.cache/torch
 
@@ -35,7 +78,8 @@ services:
       - "8000:8000"
 
   frontend:
-    build: ./frontend
+    image: gitea.kansaigaijin.com/Jamie/Nat20-Notes/frontend:latest
+    # build: ./frontend
     restart: unless-stopped
     depends_on:
       - backend