spaCy for Lambda

lambda_function.py

import os
import urllib.request
import spacy
import tarfile
from pathlib import Path


def maybe_download(model, dest):
    save_path = Path(dest) / model
    if not os.path.exists(save_path):
        print('Downloading...')
        url = f'https://github.com/explosion/spacy-models/releases/download/{model}/{model}.tar.gz'
        filename = Path(dest) / f'{model}.tar.gz'
        res = urllib.request.urlretrieve(url, filename)
        with tarfile.open(filename) as f:
            f.extractall(path=dest)
    dirname = model.split('-')[0]
    return save_path / dirname / model


def lambda_handler(event, context):
    doc = nlp(event['text'])
    response = [
        {
            'text': ent.text,
            'label': ent.label_,
            'start': ent.start_char,
            'end': ent.end_char
        }
        for ent in doc.ents
        
    ]
    return response


model = 'en_core_web_sm-2.3.0'
mnt_path = '/mnt/models'
model_path = maybe_download(model, mnt_path)
nlp = spacy.load(model_path)