Update app.py

app.py

@@ -1,26 +1,29 @@
-import gradio as gr
+from flask import Flask, request, Response
 from huggingface_hub import InferenceClient
 import os
 from typing import Generator
 
-# Initialize Inference API client
+# Initialize Flask
+app = Flask(__name__)
+
+# Initialize Hugging Face client
 client = InferenceClient(model="Qwen/Qwen2.5-1.5B-Instruct", token=os.getenv("HF_TOKEN"))
 
 def generate_code_and_explanation(prompt: str) -> Generator[str, None, None]:
     """
-    Generates code and explanation using Qwen2.5-1.5B-Instruct via Inference API with streaming.
+    Generates code + explanation with streaming from HF model.
     """
     system_prompt = (
         "You are a coding assistant like Grok. Given a user prompt, generate the requested code "
-        "and provide a clear explanation. Stream the output line by line. Format code in ```python blocks "
-        "and explanations in plain text with bullet points."
+        "and provide a clear explanation. Stream the output line by line. "
+        "Format code in ```python blocks and explanations in plain text with bullet points."
     )
+
     messages = [
         {"role": "system", "content": system_prompt},
         {"role": "user", "content": prompt}
     ]
 
-    # Stream output from Inference API using chat_completion
     for chunk in client.chat_completion(
         messages=messages,
         max_tokens=3000,
@@ -28,33 +31,24 @@ def generate_code_and_explanation(prompt: str) -> Generator[str, None, None]:
         top_p=0.9,
         stream=True
     ):
-        # Extract content from the chunk
         content = chunk.choices[0].delta.content
-        if content:  # Only yield non-empty content
+        if content:
             yield content
 
-# Gradio interface with streaming
-def live_code_generator(prompt: str):
+@app.route("/generate", methods=["POST"])
+def generate():
     """
-    Handles streaming output for the Gradio interface.
+    Flask endpoint to generate code from user prompt.
     """
-    output = ""
-    for chunk in generate_code_and_explanation(prompt):
-        output += chunk
-        yield output
-
-# Gradio app
-with gr.Blocks() as demo:
-    gr.Markdown("# Invescoz AI Studio: Live Code Generator")
-    prompt_input = gr.Textbox(label="Enter your coding prompt", placeholder="e.g., Write a Python function to calculate Fibonacci numbers")
-    output_display = gr.Textbox(label="Generated Code and Explanation", interactive=False, lines=10)
-    submit_button = gr.Button("Generate")
-    
-    submit_button.click(
-        fn=live_code_generator,
-        inputs=prompt_input,
-        outputs=output_display
-    )
+    data = request.json
+    prompt = data.get("prompt", "")
+
+    def event_stream():
+        for chunk in generate_code_and_explanation(prompt):
+            yield chunk
+
+    return Response(event_stream(), mimetype="text/plain")
 
-# Launch the app (handled by Hugging Face Spaces)
-demo.launch()
+if __name__ == "__main__":
+    # Run Flask (Hugging Face Spaces will expose this as API)
+    app.run(host="0.0.0.0", port=7860)