app.py

import streamlit as st
import streamlit.components.v1 as components
import requests
import hashlib
from entity_extraction import extract_entities

st.title('Entity Linking Demo')
st.markdown("""Linking named entities in an article to
wikidata entries (allowing us to pull the images).

*Note: Only trained on entities before May 2020*""")

article = st.text_area('Article to analyze:', value=open("example.txt").read())

if st.button('Submit'):
    with st.spinner(text="Extracting..."):
        good_ents = []

        ents = extract_entities(article)
        for i, ent in enumerate(ents):
            r = requests.get("https://www.wikidata.org/w/api.php?action=wbgetclaims&format=json&property=P18&entity=" + ent._.kb_qid)
            data = r.json()["claims"]
            if "P18" in data.keys():
                data = data["P18"][0]["mainsnak"]
                img_name = data["datavalue"]["value"].replace(" ", "_")
                img_name_hash = hashlib.md5(img_name.encode("utf-8")).hexdigest()
                a = img_name_hash[0]
                b = img_name_hash[1]
                url= f"https://upload.wikimedia.org/wikipedia/commons/{a}/{a}{b}/{img_name}"
                good_ents.append((ent.text, ent.label_, ent._.kb_qid, ent._.url_wikidata, ent._.nerd_score, url))
        cols = st.columns(len(good_ents))
        for i, ent in enumerate(good_ents):
            with cols[i]:
                components.html(f"<image style='border-radius: 50%;object-fit:cover;width:100px;height:100px' src='{ent[-1]}'/>", height=110, width=110)
                st.caption(ent[0])