app.py

# app.py
import json
import traceback
import tempfile
import os
import gradio as gr
from typing import Tuple, Optional, Any # Added Any for file_data_binary

# Import config first to ensure JAVA_HOME is set early
import config 

from main_analyzer import analyze_pdf 
# Import language_tool_python only for the test in __main__ if needed
# import language_tool_python 

def process_upload(file_data_binary: Optional[Any]) -> Tuple[str, Optional[str]]: # Use Optional[Any] for Gradio File type="binary"
    if file_data_binary is None or not hasattr(file_data_binary, 'read'): # Check if it's a file-like object
        # Gradio's binary type for gr.File returns a tempfile._TemporaryFileWrapper object
        # If it's None, no file was uploaded.
        # If it's not None but doesn't have 'read', it's an unexpected type.
        # However, gradio usually passes the bytes directly if type="binary" was used in older versions
        # or a TemporaryFileWrapper which is file-like.
        # For robustness, let's check if it's bytes.
        if isinstance(file_data_binary, bytes):
             pass # Good, it's bytes
        elif file_data_binary is None:
             return json.dumps({"error": "No file uploaded or file data is None"}, indent=2), None
        elif not hasattr(file_data_binary, 'read'): # It's not None, not bytes, not file-like
             return json.dumps({"error": f"Unexpected file data type: {type(file_data_binary)}"}), None
        # If it has 'read', it's a file-like object, proceed.

    # analyze_pdf now handles stream-to-temp-file logic internally via original_pdf_access_path
    # So we can pass the file_data_binary (which is a file-like object from Gradio) directly.
    try:
        print(f"App: Processing uploaded file...")
        # If file_data_binary is bytes, wrap it in BytesIO for file-like interface
        # analyze_pdf expects a path or a file-like object with read() and seek()
        
        # Gradio with type="binary" gives a tempfile._TemporaryFileWrapper.
        # This object is already file-like and can be passed directly.
        # No need to create another temp file here in app.py if main_analyzer handles it.
        
        # analyze_pdf will create its own temp file if it receives a stream.
        results_dict, _ = analyze_pdf(file_data_binary) 
        
        results_json = json.dumps(results_dict, indent=2, ensure_ascii=False)
        return results_json, None 

    except Exception as e:
        print(f"Error in process_upload: {e}\n{traceback.format_exc()}")
        error_message = json.dumps({"error": str(e), "traceback": traceback.format_exc()}, indent=2)
        return error_message, None
    # No finally block needed here for temp file, as analyze_pdf handles its own if it creates one
    # and Gradio handles the temp file wrapper it provides.


def create_interface():
    with gr.Blocks(title="PDF Analyzer") as interface:
        gr.Markdown("# PDF Analyzer")
        gr.Markdown(
            "Upload a PDF document to analyze its structure, references, language, and more. "
            "Language issues are derived from font-filtered text. "
            "Regex issues and general document checks use unfiltered text from the original PDF. "
            "All issue coordinates (if found) are mapped back to the original PDF."
        )
        
        with gr.Row():
            file_input = gr.File(
                label="Upload PDF",
                file_types=[".pdf"],
                # type="binary" # Gradio's File component returns a TemporaryFileWrapper which is file-like
            )
        
        with gr.Row():
            analyze_btn = gr.Button("Analyze PDF")
        
        with gr.Row():
            results_output = gr.JSON(
                label="Analysis Results", # Simplified label
                show_label=True
            )
        
        with gr.Row():
            pdf_output = gr.File( 
                label="Annotated PDF (Placeholder - View Coordinates in JSON)",
                show_label=True,
                interactive=False # Not interactive as it's a placeholder
            )
        
        analyze_btn.click(
            fn=process_upload,
            inputs=[file_input],
            outputs=[results_output, pdf_output]
        )
    return interface

if __name__ == "__main__":
    print("\n--- Launching Gradio Interface ---")
    # config.set_java_home() is called when config.py is imported.

    # Optional: Test LanguageTool initialization
    try:
        import language_tool_python # Import here for the test
        lt_test = language_tool_python.LanguageTool('en-US')
        lt_test.close()
        print("App: LanguageTool initialized successfully for test.")
    except Exception as lt_e:
        print(f"App: Warning: Could not initialize LanguageTool for test. Language checks might fail: {lt_e}")
        print("Please ensure Java is installed and JAVA_HOME is correctly set (see config.py).")
    
    app_interface = create_interface()
    app_interface.launch(
        share=False, # Set to True for public link if ngrok is installed and desired
        # server_port=7860 # Optionally specify a port
    )