Create app.py

app.py

@@ -0,0 +1,127 @@
+import os
+import tempfile
+import json
+import re
+import gradio as gr
+import google.generativeai as genai
+
+# API Keys
+API_KEYS = [
+    "AIzaSyAtHYSw1iWJYyGLt_hC5mKO3627wkHtN-s",
+    "AIzaSyBfJeK_IAkfLpmnBsMNe6xwjcielrloSFY", 
+    "AIzaSyCprm_rLChQ7Rv7YkHKI_6tcbS213PiPto",
+    "AIzaSyAPFCgH8uSjANmPRF9iHYIYcneTOod8Qi0"
+]
+
+key_index = 0
+
+def get_next_key():
+    global key_index
+    key = API_KEYS[key_index % len(API_KEYS)]
+    key_index += 1
+    return key
+
+def extract_json(text):
+    # Tìm JSON trong ```json blocks
+    match = re.search(r"```json\s*(.*?)\s*```", text, re.IGNORECASE | re.DOTALL)
+    if match:
+        json_text = match.group(1).strip()
+    else:
+        json_text = text.strip()
+    
+    try:
+        return json.loads(json_text)
+    except:
+        # Tìm {...} ngoài cùng
+        first = json_text.find("{")
+        last = json_text.rfind("}")
+        if first != -1 and last != -1 and last > first:
+            try:
+                return json.loads(json_text[first:last+1])
+            except:
+                pass
+        return {"raw_response": text}
+
+def process_image(image, prompt):
+    try:
+        print(f"Received image: {type(image)}")
+        print(f"Received prompt: {prompt}")
+        
+        # Lấy API key
+        api_key = get_next_key()
+        genai.configure(api_key=api_key)
+        
+        # Tạo model
+        model = genai.GenerativeModel("gemini-2.5-flash")
+        
+        # Upload ảnh
+        temp_file = None
+        uploaded_file = None
+        
+        try:
+            # Tạo file tạm
+            with tempfile.NamedTemporaryFile(delete=False, suffix='.png') as tmp:
+                if hasattr(image, 'save'):
+                    # PIL Image object
+                    image.save(tmp.name)
+                elif isinstance(image, str):
+                    # File path
+                    with open(image, 'rb') as f:
+                        tmp.write(f.read())
+                else:
+                    raise ValueError(f"Unsupported image type: {type(image)}")
+                temp_file = tmp.name
+            
+            print(f"Created temp file: {temp_file}")
+            
+            # Upload lên Gemini
+            uploaded_file = genai.upload_file(temp_file)
+            print(f"Uploaded to Gemini: {uploaded_file.name}")
+            
+            # Gọi API
+            response = model.generate_content([prompt, uploaded_file])
+            print(f"Got response: {response.text[:100]}...")
+            
+            # Extract JSON
+            result = extract_json(response.text)
+            return result
+            
+        finally:
+            # Cleanup
+            if temp_file and os.path.exists(temp_file):
+                os.remove(temp_file)
+                print(f"Cleaned temp file: {temp_file}")
+            if uploaded_file:
+                genai.delete_file(uploaded_file.name)
+                print(f"Deleted from Gemini: {uploaded_file.name}")
+                
+    except Exception as e:
+        print(f"Error: {e}")
+        import traceback
+        traceback.print_exc()
+        return {"error": str(e)}
+
+# Gradio Interface
+demo = gr.Interface(
+    fn=process_image,
+    inputs=[
+        gr.File(label="Upload Image", file_types=["image"]),
+        gr.Textbox(lines=5, placeholder="Enter your prompt here...", label="Prompt"),
+    ],
+    outputs=gr.JSON(label="Response"),
+    title="Gemini OCR API",
+    description="Upload image + prompt to get JSON response",
+    flagging_mode="never",
+)
+
+demo.api_name = "/predict"
+
+if __name__ == "__main__":
+    demo.launch(
+        server_name="0.0.0.0",
+        server_port=int(os.getenv("PORT", "7860")),
+        show_error=True,
+        debug=True,
+        # share=True,        
+        # inbrowser=True,    
+    )