app.py

import gradio as gr
import numpy as np
import pdfplumber
from sklearn.metrics.pairwise import cosine_similarity
from transformers import AutoTokenizer, AutoModel
import torch
from groq import Groq
import os
from fpdf import FPDF
import PyPDF2
from dotenv import load_dotenv
import pickle
from langchain_community.vectorstores import FAISS
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.prompts import PromptTemplate
from langchain_together import Together

load_dotenv()

def extract_text_from_pdf(pdf_path):
    try:
        with pdfplumber.open(pdf_path) as pdf:
            text = "\n".join(page.extract_text() for page in pdf.pages if page.extract_text())
        return text
    except Exception as e:
        return f"Error extracting text: {str(e)}"

def get_huggingface_embeddings(text_chunks):
    model_name = "sentence-transformers/all-MiniLM-L6-v2"
    tokenizer = AutoTokenizer.from_pretrained(model_name)
    model = AutoModel.from_pretrained(model_name)

    inputs = tokenizer(text_chunks, return_tensors="pt", padding=True, truncation=True)
    with torch.no_grad():
        outputs = model(**inputs)

    embeddings = outputs.last_hidden_state.mean(dim=1).numpy()
    return embeddings

def compute_similarity(embeddings1, embeddings2):
    return cosine_similarity(embeddings1, embeddings2)

def generate_comparison_summary(similarity_score, similar_terms):
    summary = f"Similarity Score: {similarity_score:.2f}%\n"
    summary += "Important terms/phrases:\n"
    summary += "\n".join(similar_terms)
    summary += "\nThis comparison highlights the key related points between the documents."
    return summary

def generate_pdf_report(similarity_score, similar_terms):
    pdf = FPDF()
    pdf.set_auto_page_break(auto=True, margin=15)
    pdf.add_page()
    pdf.set_font("Arial", style='B', size=16)
    pdf.cell(200, 10, "PDF Similarity Report", ln=True, align='C')
    pdf.ln(10)

    pdf.set_font("Arial", size=12)
    pdf.multi_cell(0, 10, f"Similarity Score: {similarity_score:.2f}%\n\n")
    pdf.multi_cell(0, 10, "Important terms/phrases:")

    for term in similar_terms:
        pdf.multi_cell(0, 8, f"- {term}")

    pdf.multi_cell(0, 10, "\nThis comparison highlights the key related points between the documents.")
    pdf_path = "similarity_report.pdf"
    pdf.output(pdf_path)
    return pdf_path

def compare_documents(file1, file2):
    try:
        file1_path = file1.name
        file2_path = file2.name

        text1 = extract_text_from_pdf(file1_path)
        text2 = extract_text_from_pdf(file2_path)

        chunks1 = text1.split(". ")
        chunks2 = text2.split(". ")

        embeddings1 = get_huggingface_embeddings(chunks1)
        embeddings2 = get_huggingface_embeddings(chunks2)

        similarity_scores = compute_similarity(embeddings1, embeddings2)
        overall_similarity = np.mean(similarity_scores) * 100

        groq_api_key = os.getenv("API_KEY")
        if groq_api_key:
            client = Groq(api_key=groq_api_key)
            response = client.chat.completions.create(
                model="llama-3.3-70b-versatile",
                messages=[
                    {"role": "system",
                     "content": "You are a helpful assistant. Only give required and should not give outside context and dont give groq context or anything like that. Stick to work."},
                    {"role": "user",
                     "content": f"Here is the text of two documents. Extract only the important terms or phrases from both, filtering out common terms like 'court', 'judge', etc.\n\nDocument 1:\n{text1}\n\nDocument 2:\n{text2}"}
                ]
            )
            groq_response = response.choices[0].message.content
            similar_terms = groq_response.split('\n')
        else:
            groq_response = "API key not found. Skipping further analysis."
            similar_terms = ["Error retrieving important terms."]

        comparison_summary = generate_comparison_summary(overall_similarity, similar_terms)
        pdf_report_path = generate_pdf_report(overall_similarity, similar_terms)

        return comparison_summary, pdf_report_path

    except Exception as e:
        return f"Error: {str(e)}", None

def extract_text_from_pdf_translator(file):
    reader = PyPDF2.PdfReader(file)
    text = ""
    for page in reader.pages:
        text += page.extract_text()
    return text

def split_text_into_chunks(text, chunk_size=1000):
    chunks = [text[i:i + chunk_size] for i in range(0, len(text), chunk_size)]
    return chunks

def translate_text_with_groq(text, target_language):
    api_key = os.getenv("API_KEY")
    if not api_key:
        raise ValueError("API key is missing. Please set the 'API_KEY' environment variable.")

    client = Groq(api_key=api_key)
    prompt_text = f"Translate the following text to {target_language}:\n{text}\n\nConvert to {target_language}"

    try:
        completion = client.chat.completions.create(
            model="llama-3.3-70b-versatile",
            messages=[{
                "role": "user",
                "content": prompt_text
            }],
            temperature=1,
            max_tokens=1024,
            top_p=1,
            stream=True,
            stop=None,
        )

        translated_text = ""
        for chunk in completion:
            translated_text += chunk.choices[0].delta.content or ""

        return translated_text
    except Exception as e:
        return f"Translation error: {str(e)}. Please try again later."

def get_font_for_language(language):
    font_map = {
        "Tamil": "NotoSansTamil-Regular.ttf",
        "Telugu": "NotoSansTelugu-Regular.ttf",
        "Malayalam": "NotoSansMalayalam-Regular.ttf",
        "Hindi": "NotoSansDevanagari-Regular.ttf"
    }
    return font_map.get(language, "DejaVuSans.ttf")

def save_translated_text_as_pdf(translated_text, target_language, output_filename="translated_text.pdf"):
    pdf = FPDF()
    pdf.add_page()

    font_path = get_font_for_language(target_language)

    try:
        pdf.add_font("CustomFont", "", font_path, uni=True)
        pdf.set_font("CustomFont", size=12)
    except Exception as e:
        return f"Error loading font: {e}"

    pdf.set_auto_page_break(auto=True, margin=15)
    pdf.set_left_margin(15)
    pdf.set_right_margin(15)

    width = pdf.w - 2 * pdf.l_margin
    pdf.multi_cell(width, 10, translated_text, align='L')

    pdf.output(output_filename)
    return output_filename

def process_pdf_and_translate(file, target_language):
    text = extract_text_from_pdf_translator(file)
    chunks = split_text_into_chunks(text)

    translated_text = ""
    for chunk in chunks:
        translated_text += translate_text_with_groq(chunk, target_language)

    pdf_file = save_translated_text_as_pdf(translated_text, target_language)
    return translated_text, pdf_file

def extract_text_from_pdf_summarizer(file):
    reader = PyPDF2.PdfReader(file)
    text = ""
    for page in reader.pages:
        text += page.extract_text()
    return text

def trim_text_for_groq(text, limit=3000):
    if len(text) > 2 * limit:
        return text[:limit] + '\n...\n' + text[-limit:]
    return text

def summarize_text_with_groq(text):
    api_key = os.getenv("API_KEY")
    if not api_key:
        raise ValueError("API key is missing. Please set the 'API_KEY' environment variable.")

    client = Groq(api_key=api_key)
    completion = client.chat.completions.create(
        model="llama-3.3-70b-versatile",
        messages=[
            {
                "role": "user",
                "content": text
            }
        ],
        temperature=1,
        max_tokens=1024,
        top_p=1,
        stream=True,
        stop=None,
    )

    summary = ""
    for chunk in completion:
        summary += chunk.choices[0].delta.content or ""
    return summary

def save_summary_as_pdf(summary, output_filename="summary.pdf"):
    pdf = FPDF()
    pdf.add_page()
    pdf.set_auto_page_break(auto=True, margin=15)
    pdf.set_font("Arial", size=12)
    pdf.multi_cell(0, 10, summary)
    pdf.output(output_filename)
    return output_filename

def process_pdf_and_summarize(file):
    text = extract_text_from_pdf_summarizer(file)
    trimmed_text = trim_text_for_groq(text)
    summary = summarize_text_with_groq(trimmed_text)
    pdf_file = save_summary_as_pdf(summary)
    return summary, pdf_file

def load_embeddings():
    try:
        embeddings = HuggingFaceEmbeddings(
            model_name="nomic-ai/nomic-embed-text-v1",
            model_kwargs={"trust_remote_code": True, "revision": "289f532e14dbbbd5a04753fa58739e9ba766f3c7"}
        )
        print("Embeddings loaded successfully.")
        return embeddings
    except Exception as e:
        raise RuntimeError(f"Error loading embeddings: {e}")

embeddings = load_embeddings()

def load_db():
    try:
        db = FAISS.load_local("law_vector_db", embeddings, allow_dangerous_deserialization=True)
        print(f"FAISS index loaded successfully.")

        with open('law_vector_db/index.pkl', 'rb') as pkl_file:
            metadata = pickle.load(pkl_file)
        print("Pickle file loaded successfully.")

        return db, metadata
    except Exception as e:
        raise RuntimeError(f"Error loading FAISS index or pickle file: {e}")

db, metadata = load_db()
db_retriever = db.as_retriever(search_type="similarity", search_kwargs={"k": 4})

prompt_template = """
<s>[INST]This is a chat template and As a legal chatbot specializing in Indian Penal Code queries, your primary objective is to provide accurate and concise information based on the user's questions. 
Do not generate your own questions and answers. You will adhere strictly to the instructions provided, offering relevant context from the knowledge base while avoiding unnecessary details. 
Your responses will be brief, to the point, and in compliance with the established format. 
If a question falls outside the given context, you will refrain from utilizing the chat history and instead rely on your own knowledge base to generate an appropriate response. 
You will prioritize the user's query and refrain from posing additional questions. 
The aim is to deliver professional, precise, and contextually relevant information pertaining to the Indian Penal Code.
CONTEXT: {context}
CHAT HISTORY: {chat_history}
QUESTION: {question}
ANSWER:</s>[INST]
"""

prompt = PromptTemplate(template=prompt_template, input_variables=['context', 'question', 'chat_history'])

TOGETHER_AI_API = os.getenv("T_API_KEY")

llm = Together(
    model="mistralai/Mistral-7B-Instruct-v0.2",
    temperature=0.5,
    max_tokens=1024,
    together_api_key=TOGETHER_AI_API
)

def ask_question(user_question, chat_history=[]):
    try:
        context_docs = db_retriever.get_relevant_documents(user_question)
        context = "\n".join(
            [doc.page_content for doc in context_docs]) if context_docs else "No relevant context found."

        input_data = {
            "context": context,
            "question": user_question,
            "chat_history": "\n".join(chat_history)
        }

        response = llm(prompt.format(**input_data))
        return response
    except Exception as e:
        return f"Error: {e}"

def chat_bot_interface(user_message, chat_history=[]):
    if not user_message:
        return chat_history, chat_history

    chat_history.append(("User", user_message))
    response = ask_question(user_message, [msg[1] for msg in chat_history if msg[0] == "User"])
    chat_history.append(("Assistant", response))

    return chat_history, chat_history

with gr.Blocks() as demo:
    with gr.Tabs():
        with gr.Tab("Summarizer"):
            gr.Markdown("## PDF Summarizer")
            with gr.Row():
                with gr.Column():
                    pdf_input_summary = gr.File(label="Upload PDF", file_types=[".pdf"])
                    summarize_button = gr.Button("Summarize")
                with gr.Column():
                    summary_output = gr.Textbox(label="Summary", lines=10, interactive=False)
                    download_link_summary = gr.File(label="Download Summary as PDF", interactive=False)
            summarize_button.click(process_pdf_and_summarize, inputs=[pdf_input_summary], outputs=[summary_output, download_link_summary])

        with gr.Tab("Translator"):
            gr.Markdown("## Document Translation")
            with gr.Row():
                with gr.Column():
                    pdf_input = gr.File(label="Upload PDF", file_types=[".pdf"])
                    language_input = gr.Dropdown(
                        label="Select Target Language",
                        choices=["Tamil", "Malayalam", "Telugu", "Hindi"],
                        value="Tamil"
                    )
                    translate_button = gr.Button("Translate")
                with gr.Column():
                    translated_output = gr.Textbox(label="Translated Text", lines=10, interactive=False)
                    download_link = gr.File(label="Download Translated PDF", interactive=False)
            translate_button.click(process_pdf_and_translate, inputs=[pdf_input, language_input], outputs=[translated_output, download_link])

        with gr.Tab("PDF Similarity Checker"):
            gr.Markdown("## PDF Similarity Checker")
            with gr.Row():
                with gr.Column():
                    file1 = gr.File(label="Upload PDF 1")
                    file2 = gr.File(label="Upload PDF 2")
                    compare_button = gr.Button("Compare")
                with gr.Column():
                    result = gr.Textbox(label="Results")
                    report = gr.File(label="Download Report")
            compare_button.click(compare_documents, inputs=[file1, file2], outputs=[result, report])

        with gr.Tab("Law Chatbot"):
            gr.Markdown("<h1 style='text-align: center;'>Legal Chatbot</h1>")
            chatbot = gr.Chatbot(label="Chatbot Interface")
            user_input = gr.Textbox(label="Ask a Question", placeholder="Type your question here...", lines=1)
            clear_button = gr.Button("Clear")
            chat_history = gr.State([])

            def clear_chat():
                return [], []

            user_input.submit(chat_bot_interface, inputs=[user_input, chat_history], outputs=[chatbot, chat_history])
            clear_button.click(clear_chat, outputs=[chatbot, chat_history])

# Launch the app
if __name__ == "__main__":
    demo.launch()