PDF_extract/tester'.py

# kbis_gradio_ui.py — Standalone Gradio UI for K-bis Analysis
# Jonathan Corke © 2025 – ALL RIGHTS RESERVED

import gradio as gr
import os
import tempfile
from datetime import datetime
from weasyprint import HTML  # pip install weasyprint

# Import your core functions
from extract import extract_pdf_text_with_easyocr  # OCR extraction
from parsepers import parse_full_kbis  # Full K-bis parsing

# Gradio Theme & CSS (your dark theme with fixes)
theme = gr.themes.Base(
    primary_hue="violet",
    secondary_hue="orange",
    neutral_hue="gray",
).set(
    body_background_fill="#1a1a1a",
    block_background_fill="#1e1e1e",
    input_background_fill="#2d2d2d",
    button_primary_background_fill="#f97316",
    button_primary_background_fill_hover="#ea580c",
)

css = """
/* Full CSS from your original app — insert here */
.gradio-container { 
    background: #1a1a1a !important; 
    color: #e4e4e7 !important; 
    font-family: 'Inter', system-ui, sans-serif;
    font-size: 15px;
    line-height: 1.6;
}

/* Radio button and input visibility fixes */
.radio label, .radio input[type="radio"] + span {
    color: #ffffff !important;
    background-color: rgb(45, 45, 45) !important;
}
.radio input[type="radio"]:checked + span {
    background-color: #6d28d9 !important;
}
.radio label:hover {
    background-color: #5b21b6 !important;
}

input, textarea {
    color: #ffffff !important;
}
"""

# Export directory
EXPORT_DIR = "exports"
os.makedirs(EXPORT_DIR, exist_ok=True)

with gr.Blocks(theme=theme, css=css, title="K-bis Analyzer") as demo:
    gr.HTML("""
    <div id="chat-header">
      <div class="brand">
        <div class="brand-icon">K-bis</div>
        <div class="brand-text">
          <h1>K-bis Analyzer</h1>
          <p>Automatic analysis of K-bis extracts via OCR and structured parsing</p>
          <p>Jonathan Corke, all rights reserved</p>
        </div>
      </div>
    </div>
    """)

    chatbot = gr.Chatbot(height=620, show_label=False)

    pdf_upload = gr.File(
        label="Upload a K-bis extract (PDF)",
        file_types=[".pdf"],
        container=False
    )

    with gr.Row():
        submit = gr.Button("Analyze", variant="primary", elem_classes="button-primary")
        summarize = gr.Button("Summarize Core Parts", variant="secondary")
        export = gr.Button("Export to PDF", variant="secondary")

    # State to store the last report (full text)
    state_report = gr.State("")


    def analyze(pdf_file, history):
        if pdf_file is None:
            return history + [["User", "Please upload a K-bis PDF."]], ""

        temp_extracted = tempfile.NamedTemporaryFile(mode="w+", suffix="_extracted.txt", delete=False,
                                                     encoding="utf-8").name
        temp_analysis = tempfile.NamedTemporaryFile(mode='w', suffix="_analysis.txt", delete=False,
                                                    encoding="utf-8").name

        try:
            extract_pdf_text_with_easyocr(pdf_file.name, save_to_file=temp_extracted)

            # Added for debugging: Print the extracted text to console
            with open(temp_extracted, "r", encoding="utf-8") as f:
                extracted_text = f.read()
            print("=== Extracted Text from OCR ===")
            print(extracted_text)
            print("=== End of Extracted Text ===")

            parse_full_kbis(temp_extracted, output_file=temp_analysis)

            with open(temp_analysis, "r", encoding="utf-8") as f:
                report = f.read()

            # Display full report in collapsible accordion
            full_report_md = f"""
<details>
<summary>Full Report (click to expand)</summary>
<pre>{report}</pre>
</summary>  
</details>
"""

            history.append([f"Analysis of {os.path.basename(pdf_file.name)}", full_report_md])

        except Exception as e:
            history.append(["User", f"Error: {str(e)}"])
            report = ""

        finally:
            for path in [temp_extracted, temp_analysis]:
                if os.path.exists(path):
                    os.unlink(path)

        return history, report


    def summarize_core(report, history):
        if not report or "Error" in report:
            history.append(["Summarize", "No valid report to summarize."])
            return history

        # Extract core parts from the report (rule-based, no LLM)
        lines = report.split("\n")
        core_keys = ["SIREN", "Dénomination", "Forme juridique", "Adresse", "Gérant(s)", "Associé(s)",
                     "Date de début d'activité"]
        core_summary = "Core Parts Summary\n\n"
        for line in lines:
            for key in core_keys:
                if line.startswith(key + ":"):
                    core_summary += line + "\n"
                    break

        history.append(["Summarize Core Parts", core_summary])
        return history


    def export_pdf(report, history):
        if not report or "Error" in report:
            history.append(["Export", "No valid report to export."])
            return history

        timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
        pdf_path = os.path.join(EXPORT_DIR, f"kbis_report_{timestamp}.pdf")

        html = f"""
        <html>
        <head><meta charset="utf-8"><title>K-bis Report</title></head>
        <body style="font-family: Arial, sans-serif; padding: 40px; line-height: 1.6;">
        <h1 style="text-align: center; color: #6d28d9;">K-bis Analysis</h1>
        <pre style="white-space: pre-wrap; font-size: 14px;">{report}</pre>
        <footer style="margin-top: 50px; text-align: center; color: #666; font-size: 12px;">
        Generated by K-bis Analyzer — Jonathan Corke © 2025
        </footer>
        </body>
        </html>
        """

        try:
            HTML(string=html).write_pdf(pdf_path)
            msg = f"Export successful: {pdf_path}"
        except Exception as e:
            msg = f"Export error: {str(e)}"

        history.append(["Export PDF", msg])
        return history


    submit.click(analyze, inputs=[pdf_upload, chatbot], outputs=[chatbot, state_report])
    summarize.click(summarize_core, inputs=[state_report, chatbot], outputs=chatbot)
    export.click(export_pdf, inputs=[state_report, chatbot], outputs=chatbot)

if __name__ == "__main__":
    demo.launch(server_name="0.0.0.0", server_port=7860, share=False)