38 lines
1.9 KiB
Markdown
38 lines
1.9 KiB
Markdown
# Station Announcer
|
|
|
|
Station Announcer is a small Flask app that keeps your Immich library and Mastodon account in sync. It helps you compose rich Mastodon posts (with GPT-polished copy and cached alt text) and doubles as an Immich alt-text helper.
|
|
|
|
## Getting started
|
|
|
|
1. Create and populate a `.env` file (or reuse the existing `api_keys` file) with:
|
|
|
|
```
|
|
IMMICH_API_URL=https://photos.4angle.com/
|
|
IMMICH_API_KEY=your-immich-api-key
|
|
OPENAI_API_KEY=your-openai-api-key
|
|
OPENAI_MODEL=gpt-4o-mini
|
|
MASTODON_BASE_URL=http://localhost:3000
|
|
MASTODON_ACCESS_TOKEN=your-mastodon-token
|
|
```
|
|
|
|
`OPENAI_MODEL` is optional; set it to override the default model used for both
|
|
alt-text generation and copy refinement.
|
|
|
|
2. Install dependencies and run the development server:
|
|
|
|
```bash
|
|
pip install -r requirements.txt
|
|
flask --app app run --debug
|
|
```
|
|
|
|
3. Visit <http://127.0.0.1:5000> to start composing Mastodon posts from your recent Immich uploads, or hop into the alt-text helper view.
|
|
|
|
## Mastodon workflow
|
|
|
|
* The home page launches the composer. Use **Select photos** there to pick up to four recent Immich photos (200 entries are shown for convenience).
|
|
* The draft view auto-fills cached alt text (or generates it if missing), lets you edit each description, enter post text, and provide optional ChatGPT instructions. It also shows your most recent Mastodon post so you can tick a box to continue a thread.
|
|
* Hit **Improve text with ChatGPT** as many times as you like; the model polishes grammar, tone, and UK English spellings while respecting your extra notes.
|
|
* When ready, choose **Post to Mastodon** to upload the selected media (with their alt text) and publish the refined caption (optionally as a reply) via the configured access token.
|
|
|
|
The app stores cached data in `instance/station_announcer.db` (or wherever `STATION_DB` points). Remove this file to clear the cache.
|