Update app.py

app.py

@@ -1,11 +1,8 @@
-python id="app1"
+python id="fixed1"
 from transformers import AutoModel, AutoTokenizer
 import gradio as gr
 from PIL import Image
 import time
-import pytesseract
-from PyPDF2 import PdfReader
-import speech_recognition as sr
 
 # ======================================================
 # MODEL
@@ -69,22 +66,22 @@ If you cannot do something, explain clearly and suggest alternatives.
 """
 
 # ======================================================
-# MEMORY (simple)
+# MEMORY (SAFE)
 # ======================================================
 
 memory = {}
 
 def get_memory(user_id):
-    return "\n".join(memory.get(user_id, []))
+    return "\n".join(memory.get(user_id, [])[-6:])
 
 def save_memory(user_id, msg):
     if user_id not in memory:
         memory[user_id] = []
-    memory[user_id].append(msg)
+    memory[user_id].append(str(msg)[:300])
     memory[user_id] = memory[user_id][-10:]
 
 # ======================================================
-# FILE HANDLER
+# FILE HANDLER (SAFE ONLY)
 # ======================================================
 
 def process_file(file):
@@ -93,54 +90,38 @@ def process_file(file):
 
     ext = file.name.split(".")[-1].lower()
 
-    # IMAGE + OCR
+    # IMAGE ONLY (safe for MiniCPM-V)
     if ext in ["png", "jpg", "jpeg", "webp"]:
-        img = Image.open(file.name)
         try:
-            text = pytesseract.image_to_string(img)
+            return Image.open(file.name), ""
         except:
-            text = ""
-        return img, text
+            return None, ""
 
-    # PDF
-    if ext == "pdf":
+    # TEXT FILES ONLY
+    if ext in ["txt", "md"]:
         try:
-            reader = PdfReader(file.name)
-            text = ""
-            for page in reader.pages:
-                text += page.extract_text() or ""
-            return None, text[:3000]
+            with open(file.name, "r", encoding="utf-8") as f:
+                return None, f.read()[:2000]
         except:
-            return None, "PDF_ERROR"
+            return None, ""
 
-    # AUDIO
-    if ext in ["mp3", "wav"]:
-        try:
-            r = sr.Recognizer()
-            with sr.AudioFile(file.name) as source:
-                audio = r.record(source)
-            text = r.recognize_google(audio)
-            return None, text
-        except:
-            return None, "AUDIO_ERROR"
-
-    return None, ""
+    return None, f"[FILE RECEIVED: {ext}]"
 
 # ======================================================
-# CHAT FUNCTION
+# CHAT
 # ======================================================
 
 def chat(user_id, message, file):
 
     image, file_text = process_file(file)
 
-    user_mem = get_memory(user_id)
+    history = get_memory(user_id)
 
     prompt = f"""
 {SYSTEM_PROMPT}
 
 Memory:
-{user_mem}
+{history}
 
 File:
 {file_text}
@@ -170,11 +151,11 @@ demo = gr.Interface(
     inputs=[
         gr.Textbox(label="User ID"),
         gr.Textbox(label="Message"),
-        gr.File(label="Upload File (image, pdf, audio)")
+        gr.File(label="Upload File (image or text)")
     ],
     outputs=gr.Textbox(label="Axel AI Response"),
-    title="Axel AI 🚀",
-    description="Multimodal AI (text + image + file + memory)"
+    title="Axel AI 🚀 (Stable Version)",
+    description="Fast multimodal AI for HuggingFace Spaces"
 )
 
 demo.launch()