__all__ = ['learn', 'classify_image', 'categories', 'image', 'label', 'examples', 'intf']

import gradio as gr
from fastai.vision.all import *
from PIL import Image
import rdkit
from rdkit import Chem
from rdkit.Chem import Draw

learn = load_learner('model.pkl')

categories = ("Acid", "Base", "Neutral", "Zwitterion")

def classify_image(SMILES):
    mol = Chem.MolFromSmiles(SMILES)
    Chem.Draw.MolToFile(mol, f'./{SMILES}.png') 
    img = PILImage.create(f'./{SMILES}.png')
    pred,idx,probs = learn.predict(img)
    Probabilities = dict(zip(categories, map(float,probs)))
    Structure = img
    return Probabilities, Structure


image = gr.inputs.Image(shape=(192, 192))
label = gr.outputs.Label()
path = './test/'
examples = [
    "O=C(C)Oc1ccccc1C(=O)O",
    "C1=CNC=C1",
    "CC(O)=O",
    "CCN(CC)CC",
    "C1=CC2=C(C=C1O)C(=CN2)CCN",
    "OC[C@H]1OC(O)[C@H](O)[C@@H](O)[C@@H]1O C([C@@H]1[C@H]([C@@H]([C@H]([C@H](O1)O)O)O)O)O",
    "OC(=O)CC1CNCC2=C1C=CC=C2",
    "CN1CC(CN2C3=CC=C(Cl)C=C3C=NCC2=O)C1",
    "OC(=O)C1CN2CCC1CC2",
    "CS(=O)(=S)NC(=O)C1CC2CCC1CC2",
]

intf = gr.Interface(
    fn=classify_image, 
    inputs=gr.Textbox(lines=1, placeholder="Enter SMILES String Here..."), 
    outputs=[label, image],
    examples=examples)

intf.launch(inline=False)