stefanfeng/huihuiSquare
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 1 Wiki Activity

feat(ai): add dedicated digital avatar model config

Browse Source
This commit is contained in:
stefanfeng
2026-08-25 16:50:33 +08:00
parent 3f7ff9329a
commit dc34a03357
13 changed files with 305 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files
digital-avatar-app/backend/routers/chat.py
+38 -23
View File
@@ -23,13 +23,10 @@ from services.token_billing import (
reserve_avatar_tokens,
settle_reservation,
)
from services.chat_model_config import ChatModelConfig, get_chat_model_config
router = APIRouter(tags=["数字分身聊天"])
CHAT_API_URL = os.getenv("CHAT_API_URL", "https://dashscope.aliyuncs.com/compatible-mode/v1")
CHAT_API_KEY = os.getenv("CHAT_API_KEY", "")
CHAT_MODEL = os.getenv("CHAT_MODEL", "qwen-plus")
CHAT_MAX_OUTPUT_TOKENS = max(128, int(os.getenv("CHAT_MAX_OUTPUT_TOKENS", "1024")))
MAX_MESSAGE_LENGTH = 4000
MAX_HISTORY_MESSAGES = 10
QA_LEXICAL_THRESHOLD = 0.72
@@ -287,22 +284,25 @@ def _search_knowledge(db: Session, avatar_id: str, question: str, top_k: int = 5
return results
def _call_qwen(messages: list[dict], temperature: float) -> dict:
if not CHAT_API_KEY:
def _call_qwen(
messages: list[dict], temperature: float, model_config: ChatModelConfig | None = None
) -> dict:
model_config = model_config or get_chat_model_config()
if not model_config.api_key:
raise RuntimeError("Qwen 模型服务未配置 CHAT_API_KEY")
url = f"{CHAT_API_URL.rstrip('/')}/chat/completions"
url = f"{model_config.api_base_url}/chat/completions"
payload = {
"model": CHAT_MODEL,
"model": model_config.model,
"messages": messages,
"temperature": temperature,
"max_tokens": CHAT_MAX_OUTPUT_TOKENS,
"max_tokens": model_config.max_tokens,
}
try:
response = httpx.post(
url,
headers={"Authorization": f"Bearer {CHAT_API_KEY}"},
headers={"Authorization": f"Bearer {model_config.api_key}"},
json=payload,
timeout=30,
timeout=model_config.timeout_seconds,
)
response.raise_for_status()
data = response.json()
@@ -314,21 +314,30 @@ def _call_qwen(messages: list[dict], temperature: float) -> dict:
return {"answer": answer.strip(), "usage": data.get("usage") or {}}
def _iter_qwen_stream(messages: list[dict], temperature: float):
def _iter_qwen_stream(
messages: list[dict], temperature: float, model_config: ChatModelConfig | None = None
):
"""将 OpenAI 兼容接口的 SSE 分片原样转为文本增量。"""
if not CHAT_API_KEY:
model_config = model_config or get_chat_model_config()
if not model_config.api_key:
raise RuntimeError("模型服务未配置")
url = f"{CHAT_API_URL.rstrip('/')}/chat/completions"
url = f"{model_config.api_base_url}/chat/completions"
payload = {
"model": CHAT_MODEL,
"model": model_config.model,
"messages": messages,
"temperature": temperature,
"max_tokens": CHAT_MAX_OUTPUT_TOKENS,
"max_tokens": model_config.max_tokens,
"stream": True,
"stream_options": {"include_usage": True},
}
try:
with httpx.stream("POST", url, headers={"Authorization": f"Bearer {CHAT_API_KEY}"}, json=payload, timeout=45) as response:
with httpx.stream(
"POST",
url,
headers={"Authorization": f"Bearer {model_config.api_key}"},
json=payload,
timeout=max(45, model_config.timeout_seconds),
) as response:
response.raise_for_status()
for raw_line in response.iter_lines():
line = raw_line.decode() if isinstance(raw_line, bytes) else raw_line
@@ -387,16 +396,21 @@ def _resolve_reply(
if model_client is not None:
answer = model_client(messages=messages, temperature=temperature)
else:
model_config = get_chat_model_config()
reservation = reserve_avatar_tokens(
db,
avatar,
usage_source,
CHAT_MODEL,
model_config.model,
messages,
CHAT_MAX_OUTPUT_TOKENS,
model_config.max_tokens,
)
try:
model_result = _call_qwen(messages=messages, temperature=temperature)
model_result = _call_qwen(
messages=messages,
temperature=temperature,
model_config=model_config,
)
answer = model_result["answer"]
token_usage = settle_reservation(
db,
@@ -436,15 +450,16 @@ def _stream_reply(
config = _config(avatar)
temperature = min(0.45 if references else 0.25, 0.2 + config["creativity"] / 100 * 0.6)
messages = _build_prompt(avatar, history, question, references)
model_config = get_chat_model_config()
reservation = reserve_avatar_tokens(
db,
avatar,
usage_source,
CHAT_MODEL,
model_config.model,
messages,
CHAT_MAX_OUTPUT_TOKENS,
model_config.max_tokens,
)
chunks = _iter_qwen_stream(messages, temperature)
chunks = _iter_qwen_stream(messages, temperature, model_config)
if matched:
messages, reservation = [], None
if public:
digital-avatar-app/backend/services/chat_model_config.py
+93
View File
@@ -0,0 +1,93 @@
import logging
import os
import threading
import time
from dataclasses import dataclass
import httpx
logger = logging.getLogger(__name__)
@dataclass(frozen=True)
class ChatModelConfig:
api_base_url: str
api_key: str
model: str
max_tokens: int
timeout_seconds: float
source: str
_cache_lock = threading.Lock()
_cached_config: ChatModelConfig | None = None
_cache_expires_at = 0.0
def _environment_config() -> ChatModelConfig:
return ChatModelConfig(
api_base_url=os.getenv(
"CHAT_API_URL", "https://dashscope.aliyuncs.com/compatible-mode/v1"
).rstrip("/"),
api_key=os.getenv("CHAT_API_KEY", ""),
model=os.getenv("CHAT_MODEL", "qwen-plus"),
max_tokens=max(128, int(os.getenv("CHAT_MAX_OUTPUT_TOKENS", "1024"))),
timeout_seconds=max(5.0, float(os.getenv("CHAT_TIMEOUT_SECONDS", "30"))),
source="environment",
)
def _fetch_runtime_config() -> ChatModelConfig | None:
url = os.getenv("CHAT_MODEL_CONFIG_URL", "").strip()
token = os.getenv("AVATAR_MODEL_CONFIG_TOKEN", "").strip()
if not url or not token:
return None
response = httpx.get(
url,
headers={"X-Avatar-Config-Token": token},
timeout=max(2.0, float(os.getenv("CHAT_MODEL_CONFIG_TIMEOUT_SECONDS", "5"))),
)
response.raise_for_status()
payload = response.json().get("data") or {}
api_base_url = str(payload.get("api_base_url") or "").rstrip("/")
api_key = str(payload.get("api_key") or "")
model = str(payload.get("model") or "")
if not api_base_url or not api_key or not model:
raise ValueError("数字分身专用模型配置不完整")
return ChatModelConfig(
api_base_url=api_base_url,
api_key=api_key,
model=model,
max_tokens=max(128, int(payload.get("max_tokens") or 1024)),
timeout_seconds=max(5.0, float(payload.get("timeout_seconds") or 30)),
source="admin",
)
def get_chat_model_config(*, force_refresh: bool = False) -> ChatModelConfig:
global _cached_config, _cache_expires_at
now = time.monotonic()
if not force_refresh and _cached_config is not None and now < _cache_expires_at:
return _cached_config
with _cache_lock:
now = time.monotonic()
if not force_refresh and _cached_config is not None and now < _cache_expires_at:
return _cached_config
try:
config = _fetch_runtime_config() or _environment_config()
except (httpx.HTTPError, ValueError, TypeError) as exc:
logger.warning("读取数字分身专用模型配置失败，暂时使用环境变量配置: %s", exc)
config = _environment_config()
_cached_config = config
ttl = max(5, int(os.getenv("CHAT_MODEL_CONFIG_CACHE_SECONDS", "60")))
_cache_expires_at = now + ttl
return config
def clear_chat_model_config_cache() -> None:
global _cached_config, _cache_expires_at
with _cache_lock:
_cached_config = None
_cache_expires_at = 0.0
digital-avatar-app/backend/tests/test_chat_model_config.py
+79
View File
@@ -0,0 +1,79 @@
from unittest.mock import Mock, patch
import httpx
from services.chat_model_config import (
clear_chat_model_config_cache,
get_chat_model_config,
)
def setup_function():
clear_chat_model_config_cache()
def teardown_function():
clear_chat_model_config_cache()
def test_admin_runtime_config_takes_priority(monkeypatch):
monkeypatch.setenv("CHAT_MODEL_CONFIG_URL", "http://config.test/runtime")
monkeypatch.setenv("AVATAR_MODEL_CONFIG_TOKEN", "shared-secret")
response = Mock()
response.raise_for_status.return_value = None
response.json.return_value = {
"data": {
"api_base_url": "https://model.test/v1/",
"api_key": "runtime-key",
"model": "avatar-model",
"max_tokens": 2048,
"timeout_seconds": 42,
}
}
with patch("services.chat_model_config.httpx.get", return_value=response) as request:
config = get_chat_model_config()
assert config.source == "admin"
assert config.api_base_url == "https://model.test/v1"
assert config.model == "avatar-model"
assert config.max_tokens == 2048
request.assert_called_once_with(
"http://config.test/runtime",
headers={"X-Avatar-Config-Token": "shared-secret"},
timeout=5.0,
)
def test_runtime_failure_falls_back_to_environment(monkeypatch):
monkeypatch.setenv("CHAT_MODEL_CONFIG_URL", "http://config.test/runtime")
monkeypatch.setenv("AVATAR_MODEL_CONFIG_TOKEN", "shared-secret")
monkeypatch.setenv("CHAT_API_URL", "https://fallback.test/v1/")
monkeypatch.setenv("CHAT_API_KEY", "fallback-key")
monkeypatch.setenv("CHAT_MODEL", "fallback-model")
monkeypatch.setenv("CHAT_MAX_OUTPUT_TOKENS", "1536")
request = httpx.Request("GET", "http://config.test/runtime")
with patch(
"services.chat_model_config.httpx.get",
side_effect=httpx.ConnectError("offline", request=request),
):
config = get_chat_model_config()
assert config.source == "environment"
assert config.api_base_url == "https://fallback.test/v1"
assert config.api_key == "fallback-key"
assert config.model == "fallback-model"
assert config.max_tokens == 1536
def test_runtime_config_is_cached(monkeypatch):
monkeypatch.setenv("CHAT_MODEL_CONFIG_URL", "")
monkeypatch.setenv("CHAT_MODEL", "first-model")
first = get_chat_model_config()
monkeypatch.setenv("CHAT_MODEL", "second-model")
second = get_chat_model_config()
assert first is second
assert second.model == "first-model"