app.py

import os
import gradio as gr
import json
import torch
from rxn.reaction import Reaction
import re
from getReaction import generate_combined_image


PROMPT_DIR = "prompts/"
ckpt_path = "./rxn/model/model.ckpt"
model = Reaction(ckpt_path, device=torch.device('cpu'))

example_diagram = "examples/exp.png"

PROMPT_NAMES = {
    "2_RxnOCR.txt": "Reaction Image Parsing Workflow",
}


def list_prompt_files_with_names():
    prompt_files = {}
    for f in os.listdir(PROMPT_DIR):
        if f.endswith(".txt"):
            friendly_name = PROMPT_NAMES.get(f, f"Task: {os.path.splitext(f)[0]}")
            prompt_files[friendly_name] = f
    return prompt_files


def format_local_predictions(predictions):
    """
    Convert local model predictions to the same structured format
    that the original RXNIM/GPT-4o pipeline would produce.
    """
    reactions_data = {"reactions": []}
    smiles_list = []

    for i, reaction in enumerate(predictions):
        rxn_entry = {
            "reaction_id": str(i + 1),
            "reactants": [],
            "conditions": [],
            "products": [],
        }

        # Reactants
        reactant_smiles = []
        for r in reaction.get("reactants", []):
            smiles = r.get("smiles", "")
            if smiles:
                rxn_entry["reactants"].append({"smiles": smiles, "label": "None"})
                reactant_smiles.append(smiles)

        # Conditions (text from OCR)
        for c in reaction.get("conditions", []):
            text_parts = c.get("text", [])
            if isinstance(text_parts, list):
                text = " ".join(text_parts)
            else:
                text = str(text_parts)
            if text.strip():
                rxn_entry["conditions"].append({"role": "reagent", "text": text.strip()})

        # Products
        product_smiles = []
        for p in reaction.get("products", []):
            smiles = p.get("smiles", "")
            if smiles:
                rxn_entry["products"].append({"smiles": smiles, "label": "None"})
                product_smiles.append(smiles)

        reactions_data["reactions"].append(rxn_entry)

        if reactant_smiles and product_smiles:
            rxn_smiles = f"{'.'.join(reactant_smiles)}>>{'.'.join(product_smiles)}"
            smiles_list.append(rxn_smiles)

    return reactions_data, smiles_list


def format_reactions_html(reactions_data):
    """Format reactions data as HTML for display."""
    detailed_output = []
    for reaction in reactions_data.get("reactions", []):
        reaction_id = reaction.get("reaction_id", "?")
        reactants = [r.get("smiles", "?") for r in reaction.get("reactants", [])]
        conditions = [
            f"<span style='color:red'>{c.get('text', c.get('smiles', '?'))} [{c.get('role', '?')}]</span>"
            for c in reaction.get("conditions", [])
        ]
        products = [
            f"<span style='color:orange'>{p.get('smiles', '?')}</span>"
            for p in reaction.get("products", [])
        ]

        html = f"<b>Reaction: </b> {reaction_id}<br>"
        html += f"  Reactants: <span style='color:blue'>{', '.join(reactants)}</span><br>"
        html += f"  Conditions: {', '.join(conditions) if conditions else 'N/A'}<br>"
        html += f"  Products: {', '.join(products)}<br><br>"
        detailed_output.append(html)

    return detailed_output


def process_chem_image(image, selected_task):
    import traceback
    try:
        image_path = "temp_image.png"
        if image is None:
            raise ValueError("No image provided")
        image.save(image_path)

        # Run local model (MolScribe + OCR) — no external API needed
        predictions = model.predict_image_file(image_path, molscribe=True, ocr=True)

        # Format predictions into structured data
        reactions_data, smiles_list = format_local_predictions(predictions)

        # Generate HTML output
        detailed_reactions = format_reactions_html(reactions_data)

        # Generate visualization
        try:
            combined_image_path = generate_combined_image(predictions, image_path)
        except Exception:
            combined_image_path = None

        # Save JSON
        json_file_path = "output.json"
        with open(json_file_path, "w") as f:
            json.dump(reactions_data, f, indent=4)

        return (
            "\n\n".join(detailed_reactions),
            "\n".join(smiles_list),
            combined_image_path,
            example_diagram,
            json_file_path,
        )
    except Exception as e:
        tb = traceback.format_exc()
        error_html = f"<b>Error:</b> {str(e)}<br><pre>{tb}</pre>"
        return (error_html, "", None, example_diagram, None)


prompts_with_names = list_prompt_files_with_names()

examples = [
    ["examples/reaction1.png", "Reaction Image Parsing Workflow"],
    ["examples/reaction2.png", "Reaction Image Parsing Workflow"],
    ["examples/reaction3.png", "Reaction Image Parsing Workflow"],
    ["examples/reaction4.png", "Reaction Image Parsing Workflow"],
]

with gr.Blocks() as demo:
    gr.Markdown("""
    <center><h1>RxnIM — Chemical Reaction Image Mining</h1></center>

    Upload a reaction image and extract SMILES, conditions, and structural data.
    This fork uses the local MolScribe + OCR model (no external API required).
    """)

    with gr.Row(equal_height=False):
        with gr.Column(scale=1):
            image_input = gr.Image(type="pil", label="Upload Reaction Image")
            task_radio = gr.Radio(
                choices=list(prompts_with_names.keys()),
                label="Select a predefined task",
            )
            with gr.Row():
                clear_button = gr.Button("Clear")
                process_button = gr.Button("Run", elem_id="submit-btn")

            gr.Markdown("### Reaction Image Parsing Output")
            reaction_output = gr.HTML(label="Reaction outputs")

        with gr.Column(scale=1):
            gr.Markdown("### Reaction Extraction Output")
            visualization_output = gr.Image(label="Visualization Output")
            schematic_diagram = gr.Image(value=example_diagram, label="Schematic Diagram")

        with gr.Column(scale=1):
            gr.Markdown("### Machine-readable Data Output")
            smiles_output = gr.Textbox(
                label="Reaction SMILES",
                show_copy_button=True,
                interactive=False,
            )
            download_json = gr.File(label="Download JSON File")

    gr.Examples(
        examples=examples,
        inputs=[image_input, task_radio],
        outputs=[reaction_output, smiles_output, visualization_output],
    )

    clear_button.click(
        lambda: (None, None, None, None, None),
        inputs=[],
        outputs=[image_input, task_radio, reaction_output, smiles_output, visualization_output],
    )

    process_button.click(
        process_chem_image,
        inputs=[image_input, task_radio],
        outputs=[reaction_output, smiles_output, visualization_output, schematic_diagram, download_json],
    )

demo.css = """
#submit-btn {
    background-color: #FF914D;
    color: white;
    font-weight: bold;
}
"""
demo.launch(server_name="0.0.0.0", server_port=7860, share=True)