llm_router_web.anonymizer#

llm_router_web: anonymizer is a lightweight Flask web interface that provides a simple UI for text anonymization.
It forwards the supplied text to an external LLM‑router anonymization service (/api/fast_text_mask) and displays the masked result, with on‑the‑fly highlighting of detected tags ({{…}}).

llm_router_web.anonymizer is the web interface component of the llm-router library.

Features#

• Web form – Paste text and trigger anonymization with a single click.
• HTMX‑powered UI – Asynchronous request/response without a full page reload.
• Result highlighting – Detected placeholders are wrapped in a colored span for easy spotting.
• Spinner indicator – Visual feedback while the request is in progress.
• Error handling – Returns clear messages for missing input or communication problems.
• Configurable backend – Target anonymization service URL is supplied via LLM_ROUTER_HOST environment variable.

Installation#

The project follows the same conventions as the rest of the llm‑router repository and uses Python 3.10.6 with virtualenv.

```shell script

Clone the repository (if not already done)#

git clone https://github.com/radlab-dev-group/llm-router.git cd llm-router/llm_router_web

Create and activate a virtual environment#

python3 -m venv venv source venv/bin/activate

Install dependencies (the base requirements already contain Flask)#

pip install -r requirements.txt

> **Note:** The only additional packages required for the anonymizer are already listed in `requirements.txt` (`flask`,
`requests`, `htmx`, `alpine.js` are loaded from CDN).

## Running the Application

### Development server

```shell script
# Start the Flask development server for the anonymizer module
python -m web.anonymizer

The UI will be reachable at http://localhost:5000/anonymize.
The root path (/) redirects to the form page.

Production (Gunicorn)#

A small launch script is included in the main repository. Example:

```shell script LLM_ROUTER_HOST=http://localhost:8000 \ FLASK_SECRET_KEY=super-secret \ gunicorn -w 4 -b 0.0.0.0:8082 "web.anonymizer:create_anonymize_app()"

- `LLM_ROUTER_HOST` – base URL of the external anonymization service (default: `http://localhost:8000`).
- `FLASK_SECRET_KEY` – secret key for session signing (default: `change-me-anonymizer`).

Adjust the number of workers (`-w`) as needed.

## Configuration

All configuration is performed via environment variables:

| Variable           | Description                               | Default                 |
|--------------------|-------------------------------------------|-------------------------|
| `FLASK_SECRET_KEY` | Secret key for Flask session signing      | `change-me-anonymizer`  |
| `LLM_ROUTER_HOST`  | URL of the external anonymization service | `http://localhost:8000` |

The variables are read in `web/anonymizer/__init__.py` when `create_anonymize_app()` is called.

## Endpoints Overview

| URL              | Methods | Description                                                                                                                        |
|------------------|---------|------------------------------------------------------------------------------------------------------------------------------------|
| `/` (root)       | GET     | Redirects to `/anonymize/`.                                                                                                        |
| `/anonymize/`    | GET     | Renders the anonymization form (`anonymize.html`).                                                                                 |
| `/anonymize/`    | POST    | Accepts `text` form field, forwards it to the external service, and returns the rendered result (`anonymize_result_partial.html`). |
| *Error handlers* | –       | Returns JSON payloads for 400, 404, and 500 errors.                                                                                |

### Request flow (POST `/anonymize/`)

1. The form posts the `text` field via HTMX.
2. The server builds the target URL: `"{LLM_ROUTER_HOST.rstrip('/')}/api/fast_text_mask"`.
3. It sends a JSON payload `{ "text": "<raw text>" }` to the external service.
4. On success the response text (or the `text` field from JSON) is injected back into the page, where JavaScript
   highlights any `{{…}}` tags.

## Development

### Project layout

web/ └─ anonymizer/ ├─ templates/ │ ├─ anonymize.html # Main form page (HTMX enabled) │ ├─ anonymize_result_partial.html # Partial used to render the result │ └─ base_anonymizer.html # Base layout shared by both templates ├─ init.py # Flask app factory (create_anonymize_app) └─ routes.py # Blueprint with view functions ```

Adding new features#

1. New routes – Define them in routes.py and register on anonymize_bp.
2. Templates – Place additional Jinja2 files in templates/ and extend base_anonymizer.html.
3. Static assets – Add CSS/JS to the main static/ directory; they are automatically served because the Flask app points to the parent static folder.
4. Configuration – Extend create_anonymize_app() to read extra environment variables as needed.

License#

llm_router_web.anonymizer is part of the llm-router project and is released under the same license as the parent repository. See the repository’s LICENSE file for details.