app.py

import gradio as gr
import os
import PyPDF2
import pandas as pd
import openai
import docx
from docx import Document
from langchain_community.embeddings import OpenAIEmbeddings
from langchain_community.vectorstores import FAISS
from langchain_community.llms import OpenAI
from langchain.text_splitter import RecursiveCharacterTextSplitter

def detect_language(text):
    """Detects the language of the input text using OpenAI."""
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "Detect the language of this text."},
            {"role": "user", "content": text}
        ]
    )
    return response["choices"][0]["message"]["content"].strip()

# Set up OpenAI API key (replace with your key)
openai.api_key = "YOUR_OPENAI_API_KEY"

def extract_files_from_folder(folder_path):
    """Scans a folder and its subfolders for PDF, TXT, CSV, and DOCX files."""
    extracted_files = {"pdf": [], "txt": [], "csv": [], "docx": []}
    
    print(f"Scanning folder: {folder_path}")
    for root, subdirs, files in os.walk(folder_path):
        print(f"Checking folder: {root}")  # Debugging log for subfolders
        for file_name in files:
            file_path = os.path.join(root, file_name)
            if file_name.endswith(".pdf"):
                extracted_files["pdf"].append(file_path)
            elif file_name.endswith(".txt"):
                extracted_files["txt"].append(file_path)
            elif file_name.endswith(".csv"):
                extracted_files["csv"].append(file_path)
            elif file_name.endswith(".docx"):
                extracted_files["docx"].append(file_path)
    
    print("Files found:", extracted_files)  # Debugging log
    return extracted_files

def read_text_from_files(file_paths):
    """Reads text content from a list of files."""
    text = ""
    for file_path in file_paths:
        print(f"Reading text file: {file_path}")  # Debugging log
        with open(file_path, "r", encoding="utf-8", errors="ignore") as file:
            text += file.read() + "\n"
    return text

def get_text_from_pdf(pdf_files):
    text = ""
    for pdf_path in pdf_files:
        print(f"Reading PDF file: {pdf_path}")  # Debugging log
        with open(pdf_path, "rb") as pdf_file:
            reader = PyPDF2.PdfReader(pdf_file)
            for page in reader.pages:
                page_text = page.extract_text()
                if page_text:
                    text += page_text + "\n"
                else:
                    text += "[Could not extract text from this page]\n"
    return text

def get_text_from_csv(csv_files):
    text = ""
    for csv_path in csv_files:
        print(f"Reading CSV file: {csv_path}")  # Debugging log
        df = pd.read_csv(csv_path)
        text += df.to_string() + "\n"
    return text

def get_text_from_docx(docx_files):
    text = ""
    for docx_path in docx_files:
        print(f"Reading DOCX file: {docx_path}")  # Debugging log
        doc = Document(docx_path)
        for para in doc.paragraphs:
            text += para.text + "\n"
    return text

def create_vector_database(text):
    splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=100)
    texts = splitter.split_text(text)
    embeddings = OpenAIEmbeddings()
    vector_db = FAISS.from_texts(texts, embeddings)
    return vector_db

def correct_exercises(text):
    """Uses OpenAI to correct and complete exercises found in the documents."""
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": "Analyze the text and complete or correct any incomplete exercises."},
            {"role": "user", "content": text}
        ]
    )
    return response["choices"][0]["message"]["content"].strip()

def get_answer(question, vector_db, corrected_exercises):
    retriever = vector_db.as_retriever()
    docs = retriever.get_relevant_documents(question)
    
    if not docs:
        return "I could not find the answer in the documents. Do you want me to search external sources?"
    
    context = "\n".join([doc.page_content for doc in docs])
    language = detect_language(question)
    response = openai.ChatCompletion.create(
        model="gpt-3.5-turbo",
        messages=[
            {"role": "system", "content": f"You are a Data Analytics assistant. Answer in {language}. Use the documents to answer questions. Also, use the corrected exercises if relevant."},
            {"role": "user", "content": question + "\n\nBased on the following document content:\n" + context + "\n\nCorrected Exercises:\n" + corrected_exercises}
        ]
    )
    return response["choices"][0]["message"]["content"]

def chatbot_interface(question):
    folder_path = "/mnt/data/Data Analitics/"
    extracted_files = extract_files_from_folder(folder_path)
    
    text = get_text_from_pdf(extracted_files["pdf"]) + read_text_from_files(extracted_files["txt"]) + get_text_from_csv(extracted_files["csv"]) + get_text_from_docx(extracted_files["docx"])
    
    if not text:
        return "The folder does not contain valid PDF, TXT, CSV, or DOCX files. Please upload supported file types."
    
    corrected_exercises = correct_exercises(text)
    vector_db = create_vector_database(text)
    return get_answer(question, vector_db, corrected_exercises)

# Gradio interface
demo = gr.Interface(
    fn=chatbot_interface,
    inputs=gr.Textbox(label="Ask a question", placeholder="Type your question here..."),
    outputs=gr.Textbox(label="Answer")
)

demo.launch()