Create main.py
Browse files
main.py
ADDED
@@ -0,0 +1,83 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import json
|
2 |
+
import os
|
3 |
+
import uuid
|
4 |
+
from typing import AsyncGenerator, NoReturn
|
5 |
+
|
6 |
+
import google.generativeai as genai
|
7 |
+
import uvicorn
|
8 |
+
from dotenv import load_dotenv
|
9 |
+
from fastapi import FastAPI, WebSocket
|
10 |
+
|
11 |
+
load_dotenv()
|
12 |
+
|
13 |
+
genai.configure(api_key=os.getenv("GOOGLE_API_KEY"))
|
14 |
+
model = genai.GenerativeModel("gemini-pro")
|
15 |
+
|
16 |
+
app = FastAPI()
|
17 |
+
|
18 |
+
PROMPT = """
|
19 |
+
You are a helpful assistant, skilled in explaining complex concepts in simple terms.
|
20 |
+
|
21 |
+
{message}
|
22 |
+
""" # noqa: E501
|
23 |
+
|
24 |
+
IMAGE_PROMPT = """
|
25 |
+
Generate an image based on the following description:
|
26 |
+
|
27 |
+
{description}
|
28 |
+
""" # noqa: E501
|
29 |
+
|
30 |
+
async def get_ai_response(message: str) -> AsyncGenerator[str, None]:
|
31 |
+
"""
|
32 |
+
Gemini Response
|
33 |
+
"""
|
34 |
+
response = await model.generate_content_async(
|
35 |
+
PROMPT.format(message=message), stream=True
|
36 |
+
)
|
37 |
+
|
38 |
+
msg_id = str(uuid.uuid4())
|
39 |
+
all_text = ""
|
40 |
+
async for chunk in response:
|
41 |
+
if chunk.candidates:
|
42 |
+
for part in chunk.candidates[0].content.parts:
|
43 |
+
all_text += part.text
|
44 |
+
yield json.dumps({"id": msg_id, "text": all_text})
|
45 |
+
|
46 |
+
async def get_ai_image(description: str) -> str:
|
47 |
+
"""
|
48 |
+
Gemini Image Generation
|
49 |
+
"""
|
50 |
+
response = await model.generate_image_async(
|
51 |
+
IMAGE_PROMPT.format(description=description)
|
52 |
+
)
|
53 |
+
|
54 |
+
if response.images:
|
55 |
+
# Assuming we take the first generated image
|
56 |
+
return json.dumps({"image_url": response.images[0].url})
|
57 |
+
return json.dumps({"error": "No image generated"})
|
58 |
+
|
59 |
+
@app.websocket("/ws")
|
60 |
+
async def websocket_endpoint(websocket: WebSocket) -> NoReturn:
|
61 |
+
"""
|
62 |
+
Websocket for AI responses
|
63 |
+
"""
|
64 |
+
await websocket.accept()
|
65 |
+
while True:
|
66 |
+
message = await websocket.receive_text()
|
67 |
+
async for text in get_ai_response(message):
|
68 |
+
await websocket.send_text(text)
|
69 |
+
|
70 |
+
@app.post("/generate-image/")
|
71 |
+
async def generate_image_endpoint(description: str):
|
72 |
+
"""
|
73 |
+
Endpoint for AI image generation
|
74 |
+
"""
|
75 |
+
image_url = await get_ai_image(description)
|
76 |
+
return json.loads(image_url)
|
77 |
+
|
78 |
+
if __name__ == "__main__":
|
79 |
+
uvicorn.run(
|
80 |
+
app,
|
81 |
+
host="0.0.0.0",
|
82 |
+
port=7860
|
83 |
+
)
|