How to auto pick pictures from wikimedia representing an item from arbitrary words?

I want my site to automatically load pictures from wikimedia (or any other no-license site) to represent food items (on my nutrition site*)
I haven't done anything like this before and was wondering if anyone has any experience with this.

For example, the string "lentils, pink, cooked, with salt" would generate a picture of cooked pink lentils. And if that not exist just some pink lentils. And if that not exist just any lentils.

Do I just try any combination for "pink" "lentils" and "cooked" or any synonyms from it? Then how? Do I use an API? Is there a better site than wikimedia? Is there already some sort algorithm for this? Or any tips at all?

Of course I am gonna let users know the picture is a guess and let them even approve if this is the right one. It doesn't have to be perfect and can grow. I am doing this with js and store results with PHP/MySQL, but I think the question is not language dependent.

Solution

use the Wikimedia Commons API for searching and retrieving images based on input string. Follow these steps:

Parse input string: Extract relevant keywords from the input string. In your example, keywords would be "lentils," "pink," "cooked," and "with salt."
Construct API query: Use extracted keywords to construct a query for Wikimedia Commons API. Here's an example using Python:

import requests

def build_api_query(keywords):
    query = " ".join(keywords)
    return f'https://commons.wikimedia.org/w/api.php?action=query&list=search&format=json&srsearch={query}&srnamespace=6&srlimit=10&origin=*'

Fetch and display image: Use the API query to fetch results and display image on your website. Example with Python's requests:

def fetch_image(keywords):
    api_url = build_api_query(keywords)
    response = requests.get(api_url)
    data = response.json()

    if len(data['query']['search']) > 0:
        first_result = data['query']['search'][0]
        image_url = f'https://commons.wikimedia.org/wiki/File:{first_result["title"]}'
        # Display image using image_url
    else:
        # Handle case when no image is found