Update app.py

app.py

@@ -1,161 +1,178 @@
 import os
-import requests
+from typing import List, Optional, Tuple, Any
+
 import gradio as gr
+from groq import Groq
 from pdf2image import convert_from_path
 from PIL import Image
-import base64
-import io
 
-# -----------------------------
-# ENVIRONMENT VARIABLES
-# -----------------------------
-GROQ_API_KEY = os.getenv("GROQ_API_KEY")
-HF_TOKEN = os.getenv("HF_TOKEN")
+from llava13b_runtime import analyze_image
 
+GROQ_API_KEY = os.getenv("GROQ_API_KEY")
 if not GROQ_API_KEY:
-    raise ValueError("GROQ_API_KEY is not set.")
-if not HF_TOKEN:
-    raise ValueError("HF_TOKEN is not set.")
-
-# -----------------------------
-# API ENDPOINTS
-# -----------------------------
-GROQ_URL = "https://api.groq.com/openai/v1/chat/completions"
-HF_ROUTER_URL = "https://router.huggingface.co"
+    raise ValueError("GROQ_API_KEY environment variable is not set.")
 
-VISION_MODEL = "LiquidAI/LFM2.5-VL-1.6B-GGUF"
+groq_client = Groq(api_key=GROQ_API_KEY)
 TEXT_MODEL = "llama-3.1-8b-instant"
 
-# -----------------------------
-# TEXT (GROQ)
-# -----------------------------
-def groq_chat(user_msg, history, mode):
-    system_prompt = {
-        "Normal": "You are NeoHelper, a helpful assistant.",
-        "School": "You explain concepts simply for students.",
-        "Shopping": "You help compare products and give buying advice.",
-        "Gaming": "You speak like a gamer and give gaming tips.",
-    }.get(mode, "You are NeoHelper.")
-
-    messages = [{"role": "system", "content": system_prompt}]
-    if history:
-        for pair in history:
-            if isinstance(pair, list) and len(pair) == 2:
-                user, bot = pair
-                if user:
-                    messages.append({"role": "user", "content": user})
-                if bot:
-                    messages.append({"role": "assistant", "content": bot})
-    messages.append({"role": "user", "content": user_msg})
-
-    payload = {
-        "model": TEXT_MODEL,
-        "messages": messages,
-        "max_tokens": 300,
-        "temperature": 0.4
-    }
-
-    headers = {
-        "Authorization": f"Bearer {GROQ_API_KEY}",
-        "Content-Type": "application/json"
-    }
-
-    response = requests.post(GROQ_URL, json=payload, headers=headers, timeout=30)
-    response.raise_for_status()
-    data = response.json()
-    return data["choices"][0]["message"]["content"]
-
-# -----------------------------
-# VISION (LIQUIDAI via HF Router)
-# -----------------------------
-def analyze_with_liquid_ai(image, prompt):
-    buf = io.BytesIO()
-    image.save(buf, format="PNG")
-    img_b64 = base64.b64encode(buf.getvalue()).decode()
-
-    payload = {
-        "model": VISION_MODEL,
-        "inputs": {
-            "prompt": prompt,
-            "image": img_b64
-        },
-        "parameters": {
-            "max_new_tokens": 256,
-            "temperature": 0.2
-        }
-    }
-
-    headers = {
-        "Authorization": f"Bearer {HF_TOKEN}",
-        "Content-Type": "application/json"
-    }
-
-    response = requests.post(HF_ROUTER_URL, json=payload, headers=headers, timeout=60)
-    response.raise_for_status()
-    data = response.json()
-
-    if isinstance(data, list) and "generated_text" in data[0]:
-        return data[0]["generated_text"]
-    return str(data)
-
-# -----------------------------
-# FILE HANDLER
-# -----------------------------
-def handle_file(file, prompt):
+MODE_PROMPTS = {
+    "Normal Chat": (
+        "You are NeoHelper, Eyad’s branded assistant. "
+        "Be concise, friendly, and helpful."
+    ),
+    "School Helper": (
+        "You are NeoHelper, a school helper for a Grade 5 student. "
+        "Explain clearly, step by step, using simple language."
+    ),
+    "Shopping Assistant": (
+        "You are NeoHelper, a shopping assistant in Saudi Arabia. "
+        "Compare specs, prices, and value clearly and briefly."
+    ),
+    "Gaming Help": (
+        "You are NeoHelper, an energetic gaming helper. "
+        "Give practical tips, fixes, and short, clear steps."
+    ),
+}
+
+def extract_images_from_pdf(pdf_path: str):
+    try:
+        return convert_from_path(pdf_path)
+    except:
+        return None
+def call_groq_text(message, system_prompt):
+    try:
+        resp = groq_client.chat.completions.create(
+            model=TEXT_MODEL,
+            messages=[
+                {"role": "system", "content": system_prompt},
+                {"role": "user", "content": message},
+            ],
+            max_tokens=900,
+        )
+        return resp.choices[0].message.content
+    except Exception as e:
+        return f"⚠️ Text model error: {str(e)}"
+
+def summarize_pdf_analyses(analyses, system_prompt):
+    joined = "\n\n--- PAGE BREAK ---\n\n".join(analyses)
+    prompt = (
+        "Summarize these PDF page analyses into one clear explanation for a student. "
+        "Be structured, simple, and concise:\n\n"
+        f"{joined}"
+    )
+    return call_groq_text(prompt, system_prompt)
+    
+def resolve_gradio_file(file):
     if file is None:
-        return "No file uploaded."
+        return None, None
+
+    if isinstance(file, dict):
+        path = file.get("path") or file.get("name")
+        name = file.get("orig_name") or path
+        return path, name
+
+    path = getattr(file, "name", None)
+    name = getattr(file, "orig_name", path)
+    return path, name
+
+def neohelper_chat(message, history, file, mode):
+    if history is None:
+        history = []
+
+    system_prompt = MODE_PROMPTS.get(mode, MODE_PROMPTS["Normal Chat"])
+
+    if message is None:
+        message = ""
+    message = message.strip()
+
+    if isinstance(file, list):
+        file = file[0] if file else None
+
+    file_path, orig_name = resolve_gradio_file(file)
+
+    # TEXT ONLY
+    if not file_path:
+        reply = call_groq_text(message, system_prompt) if message else "Please type a question or upload a file."
+        history.append({"role": "user", "content": message})
+        history.append({"role": "assistant", "content": reply})
+        return history, ""
+
+    ext = (orig_name or file_path).lower()
+
+    # IMAGE
+    if ext.endswith((".png", ".jpg", ".jpeg", ".webp", ".bmp")):
+        try:
+            prompt = message or "Explain this image in a simple way for a student."
+            img = Image.open(file_path).convert("RGB")
+            reply = analyze_image(img, prompt)
+        except Exception as e:
+            reply = f"⚠️ Error analyzing image: {str(e)}"
+
+        history.append({"role": "user", "content": message or f"[Image: {orig_name}]"})
+        history.append({"role": "assistant", "content": reply})
+        return history, ""
+
+    # PDF
+    if ext.endswith(".pdf"):
+        try:
+            pages = extract_images_from_pdf(file_path)
+            if not pages:
+                reply = "This PDF contains no pages I can analyze."
+            else:
+                analyses = []
+                for idx, img in enumerate(pages, start=1):
+                    prompt = message or f"Explain page {idx} simply."
+                    analyses.append(f"Page {idx}:\n{analyze_image(img, prompt)}")
+                reply = summarize_pdf_analyses(analyses, system_prompt)
+        except Exception as e:
+            reply = f"⚠️ Error processing PDF: {str(e)}"
+
+        history.append({"role": "user", "content": message or f"[PDF: {orig_name}]"})
+        history.append({"role": "assistant", "content": reply})
+        return history, ""
+
+    reply = "Unsupported file type. Please upload an image or a PDF."
+    history.append({"role": "user", "content": message})
+    history.append({"role": "assistant", "content": reply})
+    return history, ""
 
-    path = file.name
-    if path.lower().endswith(".pdf"):
-        pages = convert_from_path(path)
-        if not pages:
-            return "PDF conversion failed."
-        return analyze_with_liquid_ai(pages[0], prompt)
-
-    img = Image.open(path).convert("RGB")
-    return analyze_with_liquid_ai(img, prompt)
-
-# -----------------------------
-# UI
-# -----------------------------
 def build_ui():
-    with gr.Blocks() as ui:
-        gr.Markdown("# **NeoHelper — Groq Text + LiquidAI Vision (Remote)**")
+    with gr.Blocks(title="NeoHelper") as demo:
+        gr.Markdown("# 🧠 NeoHelper")
 
         with gr.Row():
-            with gr.Column(scale=3):
-                chatbot = gr.Chatbot(height=520)
-                with gr.Row():
-                    user_input = gr.Textbox(placeholder="Type your message…", scale=4)
-                    send_btn = gr.Button("Send", scale=1)
-                file_input = gr.File(label="Upload PNG / JPG / PDF")
-
             with gr.Column(scale=1):
-                mode = gr.Radio(
-                    ["Normal", "School", "Shopping", "Gaming"],
-                    value="Normal",
-                    label="Mode"
+                mode_dd = gr.Dropdown(
+                    choices=list(MODE_PROMPTS.keys()),
+                    value="Normal Chat",
+                    label="Mode",
+                )
+                file_input = gr.File(
+                    label="Upload image or PDF",
+                    file_types=["image", ".pdf"],
+                    file_count="single",
                 )
 
-        def process(user_msg, history, mode):
-            reply = groq_chat(user_msg, history, mode)
-            history = history or []
-            history.append([user_msg, reply])
-            return history, ""
-
-        send_btn.click(process, inputs=[user_input, chatbot, mode], outputs=[chatbot, user_input])
-
-        def on_file(f, h, m):
-            result = handle_file(f, "Please describe the content and extract any text.")
-            h = h or []
-            h.append([f.name, result])
-            return h, None
+            with gr.Column(scale=3):
+                chatbot = gr.Chatbot(
+                    label="NeoHelper",
+                    show_label=False,
+                    height=500,
+                )
+                msg = gr.Textbox(label="Message", lines=3)
+                send_btn = gr.Button("Send")
+                clear_btn = gr.Button("Clear")
 
-        file_input.change(on_file, inputs=[file_input, chatbot, mode], outputs=[chatbot, file_input])
+        send_btn.click(
+            neohelper_chat,
+            inputs=[msg, chatbot, file_input, mode_dd],
+            outputs=[chatbot, msg],
+        )
 
-    return ui
+        clear_btn.click(lambda: ([], ""), None, [chatbot, msg])
 
-app = build_ui()
+    return demo
 
 if __name__ == "__main__":
-    app.launch(server_name="0.0.0.0", server_port=7860, theme="gradio/soft")
+    app = build_ui()
+    app.launch(share=False, ssr_mode=False)