guymorlan's picture
Update app.py
0555443
raw
history blame
2.3 kB
from transformers import pipeline
import requests
import json
import gradio as gr
pipe = pipeline("translation", "guymorlan/TokenizerLabeller")
# download json and open
# from https://huggingface.co/guymorlan/TokenizerLabeller/raw/main/playaling_words.json
r = requests.get("https://huggingface.co/guymorlan/TokenizerLabeller/raw/main/playaling_words.json")
data = json.loads(r.text)
# build gradio interface
def predict(input):
out = pipe(input)[0]['translation_text']
raw = out
out = [x.strip() for x in out.split(" + ")]
output = f"""
<div style='direction: rtl; text-align: right; font-size: 20px; font-family: sans-serif; line-height: 1.5'>{raw}<br><br>"""
for o in out:
oo = [x.strip() for x in o.split("+")]
newout = []
for ooo in oo:
if ooo in data:
newout.append(f"""
<span style='color: green; font-family: "Courier New", Courier, monospace;'
onmouseover='showCard(event, "{data[ooo]['translation']}", "{data[ooo]['features']}")'
onmouseout='hideCard(event)'>{data[ooo]['word']}</span>
""")
else:
newout.append(ooo)
output += "+".join(newout) + " | "
output += "</div>"
output += """
<div id='hovercard' style='position: absolute; visibility: hidden; background: white; padding: 10px;
border: 1px solid black; border-radius: 5px;'>
<h3 id='card_title'></h3>
<p id='card_content'></p>
</div>
<script>
function showCard(event, title, content) {
document.getElementById('hovercard').style.visibility = 'visible';
document.getElementById('hovercard').style.top = event.pageY + 'px';
document.getElementById('hovercard').style.left = event.pageX + 'px';
document.getElementById('card_title').innerText = title;
document.getElementById('card_content').innerText = content;
}
function hideCard(event) {
document.getElementById('hovercard').style.visibility = 'hidden';
}
</script>
"""
return output
gr.Interface(predict, "textbox", "html", title="Ammiya Tokenizer", description="Tokenize Ammiya text and show Playaling words").launch()