Skip to content

Commit 0802658

Browse files
JarbasAlJarbasAl
authored
Merge pull request #16 from OpenVoiceOS/release-0.4.1a1
Release 0.4.1a1
2 parents dc14232 + 192a14c commit 0802658

File tree

11 files changed

+1544
-271
lines changed

11 files changed

+1544
-271
lines changed

‎CHANGELOG.md‎

Lines changed: 3 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -1,12 +1,12 @@
11
# Changelog
22

3-
## [0.4.0a1](https://github.com/OpenVoiceOS/ovos-persona-server/tree/0.4.0a1) (2025-04-10)
3+
## [0.4.1a1](https://github.com/OpenVoiceOS/ovos-persona-server/tree/0.4.1a1) (2025-10-17)
44

5-
[Full Changelog](https://github.com/OpenVoiceOS/ovos-persona-server/compare/0.3.2...0.4.0a1)
5+
[Full Changelog](https://github.com/OpenVoiceOS/ovos-persona-server/compare/0.4.0...0.4.1a1)
66

77
**Merged pull requests:**
88

9-
- feat:ollama\_api\_support [\#9](https://github.com/OpenVoiceOS/ovos-persona-server/pull/9) ([JarbasAl](https://github.com/JarbasAl))
9+
- refactor: migrate to fastapi [\#14](https://github.com/OpenVoiceOS/ovos-persona-server/pull/14) ([JarbasAl](https://github.com/JarbasAl))
1010

1111

1212

‎ovos_persona_server/__init__.py‎

Lines changed: 48 additions & 255 deletions
Original file line numberDiff line numberDiff line change
@@ -1,274 +1,67 @@
1-
import datetime
1+
"""
2+
Main application entry point for the OVOS Persona Server.
3+
4+
This module initializes the FastAPI application, sets up CORS middleware,
5+
and includes various API routers for chat, embeddings, Ollama, persona status,
6+
and mock OpenAI Vector Stores. It now centrally manages the unified SQLite database
7+
initialization using SQLAlchemy.
8+
"""
29
import json
3-
import os.path
4-
import random
5-
import string
6-
import time
7-
from typing import Any
10+
import os
11+
from typing import Optional
812

9-
from flask import Flask, request
10-
from ovos_bus_client.session import SessionManager
13+
from fastapi import FastAPI
14+
from fastapi.middleware.cors import CORSMiddleware
1115
from ovos_persona import Persona
1216

17+
import ovos_persona_server.persona
1318

14-
def get_app(persona_json):
15-
app = Flask(__name__)
1619

17-
with open(persona_json) as f:
18-
persona = json.load(f)
19-
persona["name"] = persona.get("name") or os.path.basename(persona_json)
20-
21-
persona = Persona(persona["name"], persona)
22-
23-
#######
24-
@app.route("/status", methods=["GET"])
25-
def status():
26-
return {"persona": persona.name,
27-
"solvers": list(persona.solvers.loaded_modules.keys()),
28-
"models": {s: persona.config.get(s, {}).get("model")
29-
for s in persona.solvers.loaded_modules.keys()}}
30-
31-
##############
32-
# OpenAI api compat
33-
@app.route("/chat/completions", methods=["POST"])
34-
def chat_completions():
35-
data = request.get_json()
36-
stream = data.get("stream", False)
37-
messages = data.get("messages")
38-
39-
completion_id = "".join(random.choices(string.ascii_letters + string.digits, k=28))
40-
completion_timestamp = int(time.time())
41-
42-
if not stream:
43-
return {
44-
"id": f"chatcmpl-{completion_id}",
45-
"object": "chat.completion",
46-
"created": completion_timestamp,
47-
"model": persona.name,
48-
"choices": [
49-
{
50-
"index": 0,
51-
"message": {
52-
"role": "assistant",
53-
"content": persona.chat(messages),
54-
},
55-
"finish_reason": "stop",
56-
}
57-
],
58-
"usage": {
59-
"prompt_tokens": None,
60-
"completion_tokens": None,
61-
"total_tokens": None,
62-
},
63-
}
64-
65-
def streaming():
66-
for chunk in persona.stream(messages):
67-
completion_data = {
68-
"id": f"chatcmpl-{completion_id}",
69-
"object": "chat.completion.chunk",
70-
"created": completion_timestamp,
71-
"model": persona.name,
72-
"choices": [
73-
{
74-
"index": 0,
75-
"delta": {
76-
"content": chunk,
77-
},
78-
"finish_reason": None,
79-
}
80-
],
81-
}
82-
83-
content = json.dumps(completion_data, separators=(",", ":"))
84-
yield f"data: {content}\n\n"
85-
time.sleep(0.1)
86-
87-
end_completion_data: dict[str, Any] = {
88-
"id": f"chatcmpl-{completion_id}",
89-
"object": "chat.completion.chunk",
90-
"created": completion_timestamp,
91-
"model": persona.name,
92-
"choices": [
93-
{
94-
"index": 0,
95-
"delta": {},
96-
"finish_reason": "stop",
97-
}
98-
],
99-
}
100-
content = json.dumps(end_completion_data, separators=(",", ":"))
101-
yield f"data: {content}\n\n"
20+
def create_persona_app(persona_path: str) -> FastAPI:
21+
"""
22+
Creates and configures the FastAPI application for the Persona Server.
10223
103-
return app.response_class(streaming(), mimetype="text/event-stream")
24+
Args:
25+
persona_path (Optional[str]): Optional path to a persona JSON file.
26+
If provided, it overrides the default
27+
persona path from settings or environment.
10428
105-
############
106-
# Ollama api compat
107-
@app.route("/api/chat", methods=["POST"])
108-
def chat():
109-
model = request.json.get("model")
110-
messages = request.json.get("messages")
111-
tools = request.json.get("tools")
112-
stream = request.json.get("stream")
29+
Returns:
30+
FastAPI: The configured FastAPI application instance.
31+
"""
11332

114-
# Format timestamp to the desired format
115-
completion_timestamp = (datetime.datetime.now().strftime('%Y-%m-%dT%H:%M:%S')
116-
+ f'.{int(time.time() * 1_000_000) % 1_000_000:06d}Z')
117-
118-
sess = SessionManager().get()
119-
120-
if not stream:
121-
ans = persona.chat(messages, lang=sess.lang, units=sess.system_unit)
122-
data = {
123-
"model": persona.name,
124-
"created_at": completion_timestamp,
125-
"message": {
126-
"role": "assistant",
127-
"content": ans,
128-
},
129-
"done": True
130-
# "context": [1, 2, 3],
131-
# "total_duration": 5043500667,
132-
# "load_duration": 5025959,
133-
# "prompt_eval_count": 26,
134-
# "prompt_eval_duration": 325953000,
135-
# "eval_count": 290,
136-
# "eval_duration": 4709213000
137-
}
138-
return data
139-
140-
def streaming():
141-
for ans in persona.stream(messages, lang=sess.lang, units=sess.system_unit):
142-
data = {
143-
"model": persona.name,
144-
"created_at": completion_timestamp,
145-
"message": {
146-
"role": "assistant",
147-
"content": ans
148-
},
149-
"done": False,
150-
# "context": [1, 2, 3],
151-
# "total_duration": 10706818083,
152-
# "load_duration": 6338219291,
153-
# "prompt_eval_count": 26,
154-
# "prompt_eval_duration": 130079000,
155-
# "eval_count": 259,
156-
# "eval_duration": 4232710000
157-
}
158-
content = json.dumps(data)
159-
yield content + "\n"
160-
161-
end_completion_data = {
162-
"model": persona.name,
163-
"created_at": completion_timestamp,
164-
"message": {
165-
"role": "assistant",
166-
"content": ""
167-
},
168-
"done": True,
169-
# "context": [1, 2, 3],
170-
# "total_duration": 10706818083,
171-
# "load_duration": 6338219291,
172-
# "prompt_eval_count": 26,
173-
# "prompt_eval_duration": 130079000,
174-
# "eval_count": 259,
175-
# "eval_duration": 4232710000
176-
}
177-
content = json.dumps(end_completion_data)
178-
yield content + "\n"
179-
180-
return app.response_class(streaming(), mimetype="application/json")
181-
182-
@app.route("/api/generate", methods=["POST"])
183-
def generate():
184-
model = request.json.get("model")
185-
prompt = request.json.get("prompt")
186-
suffix = request.json.get("suffix")
187-
system = request.json.get("system")
188-
template = request.json.get("template")
189-
stream = request.json.get("stream")
190-
191-
sess = SessionManager().get()
33+
with open(persona_path) as f:
34+
persona = json.load(f)
35+
persona["name"] = persona.get("name") or os.path.basename(persona_path)
19236

193-
messages = [{
194-
"role": "user",
195-
"content": prompt
196-
}]
197-
if system:
198-
messages.insert(0, {"role": "system", "content": system})
37+
# TODO - move to dependency injection
38+
ovos_persona_server.persona.default_persona = persona = Persona(persona["name"], persona)
19939

200-
# Format timestamp to the desired format
201-
completion_timestamp = (datetime.datetime.now().strftime('%Y-%m-%dT%H:%M:%S')
202-
+ f'.{int(time.time() * 1_000_000) % 1_000_000:06d}Z')
40+
from ovos_persona_server.version import VERSION_MAJOR, VERSION_ALPHA, VERSION_BUILD, VERSION_MINOR
20341

204-
sess = SessionManager().get()
42+
version_str = f"{VERSION_MAJOR}.{VERSION_MINOR}.{VERSION_BUILD}"
43+
if VERSION_ALPHA:
44+
version_str += f"a{VERSION_ALPHA}"
20545

206-
if not stream:
207-
ans = persona.chat(messages, lang=sess.lang, units=sess.system_unit)
208-
data = {
209-
"model": persona.name,
210-
"created_at": completion_timestamp,
211-
"message": {
212-
"role": "assistant",
213-
"content": ans,
214-
},
215-
"done": True
216-
# "context": [1, 2, 3],
217-
# "total_duration": 5043500667,
218-
# "load_duration": 5025959,
219-
# "prompt_eval_count": 26,
220-
# "prompt_eval_duration": 325953000,
221-
# "eval_count": 290,
222-
# "eval_duration": 4709213000
223-
}
224-
return data
46+
app = FastAPI(title="OVOS Persona Server",
47+
description="OpenAI/Ollama compatible API for OVOS Personas and Solvers",
48+
version=version_str)
22549

226-
def streaming():
227-
for ans in persona.stream(messages, lang=sess.lang, units=sess.system_unit):
228-
data = {
229-
"model": persona.name,
230-
"created_at": completion_timestamp,
231-
"message": {
232-
"role": "assistant",
233-
"content": ans
234-
},
235-
"done": False,
236-
# "context": [1, 2, 3],
237-
# "total_duration": 10706818083,
238-
# "load_duration": 6338219291,
239-
# "prompt_eval_count": 26,
240-
# "prompt_eval_duration": 130079000,
241-
# "eval_count": 259,
242-
# "eval_duration": 4232710000
243-
}
244-
content = json.dumps(data)
245-
yield content + "\n"
50+
app.add_middleware(
51+
CORSMiddleware,
52+
allow_origins=["*"], # Allows all origins
53+
allow_credentials=True,
54+
allow_methods=["*"], # Allows all methods (GET, POST, etc.)
55+
allow_headers=["*"], # Allows all headers
56+
)
24657

247-
end_completion_data = {
248-
"model": persona.name,
249-
"created_at": completion_timestamp,
250-
"message": {
251-
"role": "assistant",
252-
"content": ""
253-
},
254-
"done": True,
255-
# "context": [1, 2, 3],
256-
# "total_duration": 10706818083,
257-
# "load_duration": 6338219291,
258-
# "prompt_eval_count": 26,
259-
# "prompt_eval_duration": 130079000,
260-
# "eval_count": 259,
261-
# "eval_duration": 4232710000
262-
}
263-
content = json.dumps(end_completion_data)
264-
yield content + "\n"
58+
# Include routers for different API functionalities
59+
# imported here only after the Persona object is loaded
60+
from ovos_persona_server.chat import chat_router
61+
from ovos_persona_server.ollama import ollama_router
26562

266-
return app.response_class(streaming(), mimetype="text/event-stream")
63+
app.include_router(chat_router)
64+
app.include_router(ollama_router)
26765

268-
@app.route("/api/tags", methods=["GET"])
269-
def tags():
270-
return {"models": [
271-
{"name": persona.name, "model": str(persona.solvers.sort_order[0])}
272-
]}
27366

27467
return app

‎ovos_persona_server/__main__.py‎

Lines changed: 23 additions & 10 deletions
Original file line numberDiff line numberDiff line change
@@ -1,19 +1,32 @@
1+
"""
2+
Command-line entry point for running the OVOS Persona Server.
3+
4+
This module parses command-line arguments to configure and start
5+
the FastAPI application using Uvicorn.
6+
"""
7+
18
import argparse
2-
import os.path
9+
from typing import Any
10+
11+
import uvicorn
12+
13+
from ovos_persona_server import create_persona_app
314

4-
from ovos_persona_server import get_app
515

616

7-
def main():
8-
parser = argparse.ArgumentParser()
9-
parser.add_argument("--persona", help="path to persona .json file", required=True)
10-
parser.add_argument("--host", help="host", default="0.0.0.0")
11-
parser.add_argument("--port", help="port to run server", default=8337)
12-
args = parser.parse_args()
17+
def main() -> None:
18+
"""
19+
Main function to parse arguments and start the Persona Server.
20+
"""
21+
parser = argparse.ArgumentParser(description="OVOS Persona Server")
22+
parser.add_argument("--persona", help="Path to persona .json file", default=None, type=str)
23+
parser.add_argument("--host", help="Host address to bind to", default="0.0.0.0", type=str)
24+
parser.add_argument("--port", help="Port to run server on", default=8337, type=int)
25+
args: Any = parser.parse_args() # Using Any for args as argparse.Namespace is dynamic
1326

14-
app = get_app(os.path.expanduser(args.persona))
27+
app = create_persona_app(args.persona)
1528

16-
app.run(host=args.host, port=args.port, debug=False)
29+
uvicorn.run(app, port=args.port, host=args.host, log_level="debug")
1730

1831

1932
if __name__ == "__main__":

0 commit comments

Comments
 (0)