Language translation with Hugging Face

Head to the Appwrite Console then click on Functions in the left sidebar and then click on the Create Function button.

1. In the Appwrite Console's sidebar, click Functions.

2. Click Create function.

3. Under Connect Git repository, select your provider.

4. After connecting to GitHub, under Quick start, select the Node.js starter template.

5. In the Variables step, add the HUGGINGFACE_ACCESS_TOKEN, generate it here.

6. Follow the step-by-step wizard and create the function.

Once the function is created, clone the function and open it in your development environment.

Once you have the repository open, you can install the Hugging Face inference SDK by running the following command in your terminal:

npm install @huggingface/inference

After installing the SDK, write the code that will accept a validate the request body.

Open up your src/main.js file and replace the function body with the following code:

export default async ({ req, res }) => {
  if (!req.body.source || typeof req.body.source !== 'string') {
    return res.json({
        ok: false,
        error: 'Missing requrired field `source`',
    }, 400);
  }
}

Add the following import at the top of your src/main.js file:

import { HfInference } from '@huggingface/inference';

In your function body, add the following code after the parameter checks:

export default async ({ req, res }) => {
    // ... existing parameter checks
    
    const hf = new HfInference(process.env.HUGGINGFACE_ACCESS_TOKEN);

    try {
        const translation = await hf.translation({
            model: 'facebook/mbart-large-50-many-to-many-mmt',
            inputs: req.body.source,
            parameters: {
                src_lang: 'en_XX', // English locale
                tgt_lang: 'fr_XX', // French locale
            }
        });
        return res.json({
            ok: true,
            output: translation.translation_text
        });
    } catch (err) {
        return res.json({
            ok: false,
            error: 'Failed to query Hugging Face'
        }, 500);
    }
}

First, ensure the function is called with method POST. Then, make a request to the Hugging Face API to translate the source text from English to French. You can change the src_lang and tgt_lang parameters to any language supported by the model you choose.

Test our function by sending a POST request to the function's endpoint with a JSON body containing the source parameter.

Navigate to your function in the Appwrite Console and click on Execute now. In the modal that appears, enter the following JSON body:

{
    "source": "Hello, how are you?"
}

Click Execute and you should see a response similar to the following:

{
    "ok": true,
    "output": "Bonjour, comment ça va?"
}