services.py

import schemas as _schemas
from PIL import Image
from io import BytesIO
import base64
import utils
import uuid
import cv2
import wikipedia_downloader


"""
Note: make sure .env exist and contains your token
"""
df = utils.load_metadata('wiki.mat')


async def find_look_alike(query: _schemas._QueryBase) -> Image:
    temp_id = str(uuid.uuid4())
    utils.create_temp()
    
    init_image = Image.open(BytesIO(base64.b64decode(query.encoded_base_img[0])))
    init_image = init_image.save(utils.TEMP_PATH + '/' + temp_id + '.jpg')
    
    results = utils.find_similar_face(utils.TEMP_PATH + '/' + temp_id + '.jpg', df, query.gender)
    return_images = []
    return_names = []
    for result in results:
        upscaled_image = wikipedia_downloader.get_image(result['name'], temp_id)
        upscaled_image = utils.crop_face(upscaled_image)
        _, im_png = cv2.imencode(".jpg", upscaled_image)
        return_names.append(result['name'])
        return_images.append(im_png)
    utils.remove_temp_image(temp_id)
    
    return return_images, return_names