File size: 5,095 Bytes
0f3b6b0
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
import gevent.pywsgi
from gevent import monkey;monkey.patch_all()
from flask import Flask, request, Response, jsonify
import argparse
import requests
import random
import string
import time
import json
import os

app = Flask(__name__)
app.json.sort_keys = False

parser = argparse.ArgumentParser(description="An example of Qwen demo with a similar API to OAI.")
parser.add_argument("--host", type=str, help="Set the ip address.(default: 0.0.0.0)", default='0.0.0.0')
parser.add_argument("--port", type=int, help="Set the port.(default: 7860)", default=7860)
args = parser.parse_args()

base_url = os.getenv('MODEL_BASE_URL')

@app.route('/api/v1/models', methods=["GET", "POST"])
@app.route('/v1/models', methods=["GET", "POST"])
def model_list():
    time_now = int(time.time())
    model_list = {
        "object": "list",
        "data": [
            {
                "id": "qwen",
                "object": "model",
                "created": time_now,
                "owned_by": "tastypear"
            },
            {
                "id": "gpt-3.5-turbo",
                "object": "model",
                "created": time_now,
                "owned_by": "tastypear"
            }
        ]
    }
    return jsonify(model_list)

@app.route("/", methods=["GET"])
def index():
    return Response(f'QW1_5 OpenAI Compatible API<br><br>'+
        f'Set "{os.getenv("SPACE_URL")}/api" as proxy (or API Domain) in your Chatbot.<br><br>'+
        f'The complete API is: {os.getenv("SPACE_URL")}/api/v1/chat/completions')

@app.route("/api/v1/chat/completions", methods=["POST", "OPTIONS"])
@app.route("/v1/chat/completions", methods=["POST", "OPTIONS"])
def chat_completions():

    if request.method == "OPTIONS":
        return Response(
            headers={
                "Access-Control-Allow-Origin": "*",
                "Access-Control-Allow-Headers": "*",
            }
        )

    data = request.get_json()

    # reorganize data
    system = "You are a helpful assistant."
    chat_history = []
    prompt = ""
    
    if "messages" in data:
        messages = data["messages"]
        message_size = len(messages)

        prompt = messages[-1].get("content")
        for i in range(message_size - 1):
            role_this = messages[i].get("role")
            role_next = messages[i + 1].get("role")
            if role_this == "system":
                system = messages[i].get("content")
            elif role_this == "user":
                if role_next == "assistant":
                    chat_history.append(
                        [messages[i].get("content"), messages[i + 1].get("content")]
                    )
                else:
                    chat_history.append([messages[i].get("content"), " "])

        # print(f'{system = }')
        # print(f'{chat_history = }')
        # print(f'{prompt = }')

        fn_index = 0

        # gen a random char(11) hash
        chars = string.ascii_lowercase + string.digits
        session_hash = "".join(random.choice(chars) for _ in range(11))

        json_prompt = {
            "data": [prompt, chat_history, system],
            "fn_index": fn_index,
            "session_hash": session_hash,
        }

    def generate():
        response = requests.post(f"{base_url}/queue/join", json=json_prompt)
        url = f"{base_url}/queue/data?session_hash={session_hash}"
        data = requests.get(url, stream=True)

        time_now = int(time.time())

        for line in data.iter_lines():
            if line:
                decoded_line = line.decode("utf-8")
                json_line = json.loads(decoded_line[6:])
                if json_line["msg"] == "process_starts":
                    res_data = gen_res_data({}, time_now=time_now, start=True)
                    yield f"data: {json.dumps(res_data)}\n\n"
                elif json_line["msg"] == "process_generating":
                    res_data = gen_res_data(json_line, time_now=time_now)
                    yield f"data: {json.dumps(res_data)}\n\n"
                elif json_line["msg"] == "process_completed":
                    yield "data: [DONE]"

    return Response(
        generate(),
        mimetype="text/event-stream",
        headers={
            "Access-Control-Allow-Origin": "*",
            "Access-Control-Allow-Headers": "*",
        },
    )


def gen_res_data(data, time_now=0, start=False):
    res_data = {
        "id": "chatcmpl",
        "object": "chat.completion.chunk",
        "created": time_now,
        "model": "qwen1_5",
        "choices": [{"index": 0, "finish_reason": None}],
    }

    if start:
        res_data["choices"][0]["delta"] = {"role": "assistant", "content": ""}
    else:
        chat_pair = data["output"]["data"][1]
        if chat_pair == []:
            res_data["choices"][0]["finish_reason"] = "stop"
        else:
            res_data["choices"][0]["delta"] = {"content": chat_pair[-1][-1]}
    return res_data


if __name__ == "__main__":
    # app.run(host=args.host, port=args.port, debug=True)
    gevent.pywsgi.WSGIServer((args.host, args.port), app).serve_forever()