import os
import cv2
import numpy as np
from PIL import Image
import pytesseract
import gradio as gr
from pdf2image import convert_from_path
import PyPDF2
from llama_index.core import VectorStoreIndex, Document
from llama_index.embeddings.openai import OpenAIEmbedding
from llama_index.llms.openai import OpenAI
from llama_index.core import get_response_synthesizer
from sentence_transformers import SentenceTransformer, util
import logging
from openai_tts_tool import generate_audio_and_text
import tempfile

# Set up logging configuration
logging.basicConfig(level=logging.INFO, format='%(asctime)s | %(levelname)s | %(message)s')

# Initialize global variables
vector_index = None
query_log = []
sentence_model = SentenceTransformer('all-MiniLM-L6-v2')

# Get available languages for OCR
try:
    langs = os.popen('tesseract --list-langs').read().split('\n')[1:-1]
except:
    langs = ['eng']  # Fallback to English if tesseract isn't properly configured

def create_temp_dir():
    """Create temporary directory if it doesn't exist"""
    temp_dir = os.path.join(os.getcwd(), 'temp')
    if not os.path.exists(temp_dir):
        os.makedirs(temp_dir)
    return temp_dir

def preprocess_image(image_path):
    img = cv2.imread(image_path)
    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
    gray = cv2.equalizeHist(gray)
    gray = cv2.GaussianBlur(gray, (5, 5), 0)
    processed_image = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_GAUSSIAN_C,
                                          cv2.THRESH_BINARY, 11, 2)
    temp_dir = create_temp_dir()
    temp_filename = os.path.join(temp_dir, "processed_image.png")
    cv2.imwrite(temp_filename, processed_image)
    return temp_filename

def extract_text_from_image(image_path, lang='eng'):
    processed_image_path = preprocess_image(image_path)
    text = pytesseract.image_to_string(Image.open(processed_image_path), lang=lang)
    try:
        os.remove(processed_image_path)
    except:
        pass
    return text

def extract_text_from_pdf(pdf_path, lang='eng'):
    text = ""
    temp_dir = create_temp_dir()
    try:
        with open(pdf_path, 'rb') as file:
            pdf_reader = PyPDF2.PdfReader(file)
            for page_num in range(len(pdf_reader.pages)):
                page = pdf_reader.pages[page_num]
                page_text = page.extract_text()
                if page_text.strip():
                    text += page_text
                else:
                    images = convert_from_path(pdf_path, first_page=page_num + 1, last_page=page_num + 1)
                    for image in images:
                        temp_image_path = os.path.join(temp_dir, f'temp_image_{page_num}.png')
                        image.save(temp_image_path, 'PNG')
                        text += extract_text_from_image(temp_image_path, lang=lang)
                        text += f"\n[OCR applied on page {page_num + 1}]\n"
                        try:
                            os.remove(temp_image_path)
                        except:
                            pass
    except Exception as e:
        return f"Error processing PDF: {str(e)}"
    return text

def extract_text(file_path, lang='eng'):
    file_ext = file_path.lower().split('.')[-1]
    if file_ext in ['pdf']:
        return extract_text_from_pdf(file_path, lang)
    elif file_ext in ['png', 'jpg', 'jpeg']:
        return extract_text_from_image(file_path, lang)
    else:
        return f"Unsupported file type: {file_ext}"

def process_upload(api_key, files, lang):
    global vector_index

    if not api_key:
        return "Please provide a valid OpenAI API Key."

    if not files:
        return "No files uploaded."

    documents = []
    error_messages = []
    image_heavy_docs = []

    for file_path in files:
        try:
            text = extract_text(file_path, lang)
            if "This document consists of" in text and "page(s) of images" in text:
                image_heavy_docs.append(os.path.basename(file_path))
            documents.append(Document(text=text))
        except Exception as e:
            error_message = f"Error processing file {os.path.basename(file_path)}: {str(e)}"
            logging.error(error_message)
            error_messages.append(error_message)

    if documents:
        try:
            embed_model = OpenAIEmbedding(model="text-embedding-3-large", api_key=api_key)
            vector_index = VectorStoreIndex.from_documents(documents, embed_model=embed_model)
            
            success_message = f"Successfully indexed {len(documents)} files."
            if image_heavy_docs:
                success_message += f"\nNote: The following documents consist mainly of images and may require manual review: {', '.join(image_heavy_docs)}"
            if error_messages:
                success_message += f"\nErrors: {'; '.join(error_messages)}"
            
            return success_message
        except Exception as e:
            return f"Error creating index: {str(e)}"
    else:
        return f"No valid documents were indexed. Errors: {'; '.join(error_messages)}"

def query_app(query, model_name, use_similarity_check, api_key):
    global vector_index, query_log

    if vector_index is None:
        return "No documents indexed yet. Please upload documents first."

    if not api_key:
        return "Please provide a valid OpenAI API Key."

    try:
        llm = OpenAI(model=model_name, api_key=api_key)
        response_synthesizer = get_response_synthesizer(llm=llm)
        query_engine = vector_index.as_query_engine(llm=llm, response_synthesizer=response_synthesizer)
        response = query_engine.query(query)
        
        generated_response = response.response
        return generated_response

    except Exception as e:
        logging.error(f"Error during query processing: {e}")
        return f"Error during query processing: {str(e)}"

def create_gradio_interface():
    with gr.Blocks(title="Document Processing and TTS App") as demo:
        gr.Markdown("# 📄 Document Processing, Text & Audio Generation App")
        
        with gr.Tab("📤 Upload Documents"):
            api_key_input = gr.Textbox(
                label="Enter OpenAI API Key",
                placeholder="Paste your OpenAI API Key here",
                type="password"
            )
            file_upload = gr.File(label="Upload Files", file_count="multiple", type="filepath")
            lang_dropdown = gr.Dropdown(choices=langs, label="Select OCR Language", value='eng')
            upload_button = gr.Button("Upload and Index")
            upload_status = gr.Textbox(label="Status", interactive=False)

        with gr.Tab("❓ Ask a Question"):
            query_input = gr.Textbox(label="Enter your question")
            model_dropdown = gr.Dropdown(
                choices=["gpt-4-0125-preview", "gpt-3.5-turbo-0125"],
                label="Select Model",
                value="gpt-3.5-turbo-0125"
            )
            similarity_checkbox = gr.Checkbox(label="Use Similarity Check", value=False)
            query_button = gr.Button("Ask")
            answer_output = gr.Textbox(label="Answer", interactive=False)

        with gr.Tab("🗣️ Generate Audio and Text"):
            text_input = gr.Textbox(label="Enter text for generation")
            voice_type = gr.Dropdown(
                choices=["alloy", "echo", "fable", "onyx", "nova", "shimmer"],
                label="Voice Type",
                value="alloy"
            )
            voice_speed = gr.Slider(
                minimum=0.25,
                maximum=4.0,
                value=1.0,
                label="Voice Speed"
            )
            language = gr.Dropdown(
                choices=["en", "ar", "de", "hi", "es", "fr", "it", "ja", "ko", "pt"],
                label="Language",
                value="en"
            )
            output_option = gr.Radio(
                choices=["audio", "summary_text", "both"],
                label="Output Option",
                value="both"
            )
            summary_length = gr.Slider(
                minimum=50,
                maximum=500,
                value=100,
                step=10,
                label="Summary Length (words)"
            )
            additional_prompt = gr.Textbox(label="Additional Prompt (Optional)")
            generate_button = gr.Button("Generate")
            audio_output = gr.Audio(label="Generated Audio")
            summary_output = gr.Textbox(label="Generated Summary Text")

        # Wire up the components
        upload_button.click(
            fn=process_upload,
            inputs=[api_key_input, file_upload, lang_dropdown],
            outputs=[upload_status]
        )
        
        query_button.click(
            fn=query_app,
            inputs=[query_input, model_dropdown, similarity_checkbox, api_key_input],
            outputs=[answer_output]
        )
        
        generate_button.click(
            fn=generate_audio_and_text,
            inputs=[
                api_key_input, text_input, model_dropdown, voice_type,
                voice_speed, language, output_option, summary_length,
                additional_prompt
            ],
            outputs=[audio_output, summary_output]
        )

    return demo

if __name__ == "__main__":
    demo = create_gradio_interface()
    demo.launch()
else:
    demo = create_gradio_interface()/////////////////////////////////////openai_tts_tool.py// from openai import OpenAI
import tempfile
import os

def generate_audio_and_text(api_key, input_text, model_name, voice_type, voice_speed, language, output_option, summary_length, additional_prompt):
    if not input_text:
        return None, "No input text provided"
    
    try:
        client = OpenAI(api_key=api_key)
        
        # Generate summary if requested
        summary_text = None
        if output_option in ["summary_text", "both"]:
            summary_prompt = f"Summarize the following text in approximately {summary_length} words. {additional_prompt or ''}\n\nText: {input_text}"
            
            summary_response = client.chat.completions.create(
                model=model_name,
                messages=[{"role": "user", "content": summary_prompt}]
            )
            summary_text = summary_response.choices[0].message.content
        
        # Generate audio if requested
        audio_file = None
        if output_option in ["audio", "both"]:
            speech_response = client.audio.speech.create(
                model="tts-1",  # or "tts-1-hd" for higher quality
                voice=voice_type,
                input=input_text,
                speed=float(voice_speed)
            )
            
            # Create temp directory if it doesn't exist
            temp_dir = os.path.join(os.getcwd(), 'temp')
            if not os.path.exists(temp_dir):
                os.makedirs(temp_dir)
            
            # Save the audio to a temporary file
            audio_path = os.path.join(temp_dir, f"output_{hash(input_text)}.mp3")
            with open(audio_path, "wb") as f:
                for chunk in speech_response.iter_bytes():
                    f.write(chunk)
            
            audio_file = audio_path
        
        # Return based on output option
        if output_option == "summary_text":
            return None, summary_text
        elif output_option == "audio":
            return audio_file, None
        elif output_option == "both":
            return audio_file, summary_text
            
    except Exception as e:
        return None, f"Error: {str(e)}"
    
    return None, None