Update app.py

app.py

@@ -1,77 +1,48 @@
-import gradio as gr
-import google.generativeai as genai
-from transformers import pipeline
-import json
-from ppt_parser import transfer_to_structure
-
-# ✅ Your Google Gemini API Key
-GOOGLE_API_KEY = "your_google_api_key_here"
-genai.configure(api_key=GOOGLE_API_KEY)
-
-# ✅ Load Models
-summarizer = pipeline("summarization", model="facebook/bart-large-cnn")
-gemini_model = genai.GenerativeModel("models/gemini-1.5-flash")
-
-# ✅ Global variable to hold extracted text
-extracted_text = ""
-
-def extract_text_from_pptx_json(parsed_json: dict) -> str:
-    extracted_text = ""
-    for slide_key, slide in parsed_json.items():
-        for shape_key, shape in slide.items():
-            if shape.get('type') == 'group':
-                group = shape.get('group_content', {})
-                for _, group_shape in group.items():
-                    if group_shape.get('type') == 'text':
-                        for para_key, para in group_shape.items():
-                            if para_key.startswith("paragraph_"):
-                                extracted_text += para.get("text", "") + "\n"
-            elif shape.get('type') == 'text':
-                for para_key, para in shape.items():
-                    if para_key.startswith("paragraph_"):
-                        extracted_text += para.get("text", "") + "\n"
-    return extracted_text.strip()
-
-def handle_pptx_upload(pptx_file):
-    global extracted_text
-    tmp_path = pptx_file.name
-    parsed_json_str, _ = transfer_to_structure(tmp_path, "images")
-    parsed_json = json.loads(parsed_json_str)
-    extracted_text = extract_text_from_pptx_json(parsed_json)
-    return extracted_text or "No readable text found in slides."
-
-def summarize_text():
-    global extracted_text
-    if not extracted_text:
-        return "Please upload and extract text from a PPTX file first."
-    summary = summarizer(extracted_text, max_length=200, min_length=50, do_sample=False)[0]['summary_text']
-    return summary
-
-def clarify_concept(question):
-    global extracted_text
-    if not extracted_text:
-        return "Please upload and extract text from a PPTX file first."
-    prompt = f"Context:\n{extracted_text}\n\nQuestion: {question}"
-    response = gemini_model.generate_content(prompt)
-    return response.text if response else "No response from Gemini."
-
-with gr.Blocks() as demo:
-    gr.Markdown("## 🧠 AI-Powered Study Assistant for PowerPoint Lectures")
-
-    pptx_input = gr.File(label="📂 Upload PPTX File", file_types=[".pptx"])
-    extract_btn = gr.Button("📜 Extract & Summarize")
-
-    extracted_output = gr.Textbox(label="📄 Extracted Text", lines=10, interactive=False)
-    summary_output = gr.Textbox(label="📝 Summary", interactive=False)
-
-    extract_btn.click(handle_pptx_upload, inputs=[pptx_input], outputs=[extracted_output])
-    extract_btn.click(summarize_text, outputs=[summary_output])
-
-    question = gr.Textbox(label="❓ Ask a Question")
-    ask_btn = gr.Button("💬 Ask Gemini")
-    ai_answer = gr.Textbox(label="🤖 Gemini Answer", lines=4)
-
-    ask_btn.click(clarify_concept, inputs=[question], outputs=[ai_answer])
-
-if __name__ == "__main__":
-    demo.launch()
+import streamlit as st
+from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
+import PyPDF2
+import torch
+
+st.set_page_config(page_title="Perplexity Clone (Gemma)", layout="wide")
+st.title("📚 Perplexity-Style AI Study Assistant using Gemma")
+
+# Load Gemma model and tokenizer
+@st.cache_resource
+def load_model():
+    tokenizer = AutoTokenizer.from_pretrained("google/gemma-7b-it")
+    model = AutoModelForCausalLM.from_pretrained(
+        "google/gemma-7b-it",
+        torch_dtype=torch.float16,
+        device_map="auto"
+    )
+    pipe = pipeline("text-generation", model=model, tokenizer=tokenizer, max_new_tokens=512)
+    return pipe
+
+textgen = load_model()
+
+# Extract text from uploaded PDF
+def extract_text_from_pdf(file):
+    reader = PyPDF2.PdfReader(file)
+    text = ""
+    for page in reader.pages:
+        text += page.extract_text() + "\n"
+    return text.strip()
+
+# UI Layout
+query = st.text_input("Ask a question or type a query:")
+
+uploaded_file = st.file_uploader("Or upload a PDF to analyze its content:", type=["pdf"])
+
+context = ""
+if uploaded_file:
+    context = extract_text_from_pdf(uploaded_file)
+    st.text_area("Extracted Content", context, height=200)
+
+if st.button("Generate Answer"):
+    with st.spinner("Generating with Gemma..."):
+        prompt = query
+        if context:
+            prompt = f"Context:\n{context}\n\nQuestion: {query}"
+        output = textgen(prompt)[0]["generated_text"]
+        st.success("Answer:")
+        st.write(output)