mirror of
https://github.com/RYDE-WORK/Langchain-Chatchat.git
synced 2026-01-19 21:37:20 +08:00
* 优化configs (#1474) * remove llm_model_dict * optimize configs * fix get_model_path * 更改一些默认参数,添加千帆的默认配置 * Update server_config.py.example * fix merge conflict for #1474 (#1494) * 修复ChatGPT api_base_url错误;用户可以在model_config在线模型配置中覆盖默认的api_base_url (#1496) * 优化LLM模型列表获取、切换的逻辑: (#1497) 1、更准确的获取未运行的可用模型 2、优化WEBUI模型列表显示与切换的控制逻辑 * 更新migrate.py和init_database.py,加强知识库迁移工具: (#1498) 1. 添加--update-in-db参数,按照数据库信息,从本地文件更新向量库 2. 添加--increament参数,根据本地文件增量更新向量库 3. 添加--prune-db参数,删除本地文件后,自动清理相关的向量库 4. 添加--prune-folder参数,根据数据库信息,清理无用的本地文件 5. 取消--update-info-only参数。数据库中存储了向量库信息,该操作意义不大 6. 添加--kb-name参数,所有操作支持指定操作的知识库,不指定则为所有本地知识库 7. 添加知识库迁移的测试用例 8. 删除milvus_kb_service的save_vector_store方法 * feat: support volc fangzhou * 使火山方舟正常工作,添加错误处理和测试用例 * feat: support volc fangzhou (#1501) * feat: support volc fangzhou --------- Co-authored-by: liunux4odoo <41217877+liunux4odoo@users.noreply.github.com> Co-authored-by: liqiankun.1111 <liqiankun.1111@bytedance.com> * 第一版初步agent实现 (#1503) * 第一版初步agent实现 * 增加steaming参数 * 修改了weather.py --------- Co-authored-by: zR <zRzRzRzRzRzRzR> * 添加configs/prompt_config.py,允许用户自定义prompt模板: (#1504) 1、 默认包含2个模板,分别用于LLM对话,知识库和搜索引擎对话 2、 server/utils.py提供函数get_prompt_template,获取指定的prompt模板内容(支持热加载) 3、 api.py中chat/knowledge_base_chat/search_engine_chat接口支持prompt_name参数 * 增加其它模型的参数适配 * 增加传入矢量名称加载 * 1. 搜索引擎问答支持历史记录; 2. 修复知识库问答历史记录传参错误:用户输入被传入history,问题出在webui中重复获取历史消息,api知识库对话接口并无问题。 * langchain日志开关 * move wrap_done & get_ChatOpenAI from server.chat.utils to server.utils (#1506) * 修复faiss_pool知识库缓存key错误 (#1507) * fix ReadMe anchor link (#1500) * fix : Duplicate variable and function name (#1509) Co-authored-by: Jim <zhangpengyi@taijihuabao.com> * Update README.md * fix #1519: streamlit-chatbox旧版BUG,但新版有兼容问题,先在webui中作处理,并限定chatbox版本 (#1525) close #1519 * 【功能新增】在线 LLM 模型支持阿里云通义千问 (#1534) * feat: add qwen-api * 使Qwen API支持temperature参数;添加测试用例 * 将online-api的sdk列为可选依赖 --------- Co-authored-by: liunux4odoo <liunux@qq.com> * 处理序列化至磁盘的逻辑 * remove depends on volcengine * update kb_doc_api: use Form instead of Body when upload file * 将所有httpx请求改为使用Client,提高效率,方便以后设置代理等。 (#1554) 将所有httpx请求改为使用Client,提高效率,方便以后设置代理等。 将本项目相关服务加入无代理列表,避免fastchat的服务器请求错误。(windows下无效) * update QR code * update readme_en,readme,requirements_api,requirements,model_config.py.example:测试baichuan2-7b;更新相关文档 * 新增特性:1.支持vllm推理加速框架;2. 更新支持模型列表 * 更新文件:1. startup,model_config.py.example,serve_config.py.example,FAQ * 1. debug vllm加速框架完毕;2. 修改requirements,requirements_api对vllm的依赖;3.注释掉serve_config中baichuan-7b的device为cpu的配置 * 1. 更新congif中关于vllm后端相关说明;2. 更新requirements,requirements_api; * 增加了仅限GPT4的agent功能,陆续补充,中文版readme已写 (#1611) * Dev (#1613) * 增加了仅限GPT4的agent功能,陆续补充,中文版readme已写 * issue提到的一个bug * 温度最小改成0,但是不应该支持负数 * 修改了最小的温度 * fix: set vllm based on platform to avoid error on windows * fix: langchain warnings for import from root * 修复webui中重建知识库以及对话界面UI错误 (#1615) * 修复bug:webui点重建知识库时,如果存在不支持的文件会导致整个接口错误;migrate中没有导入CHUNK_SIZE * 修复:webui对话界面的expander一直为running状态;简化历史消息获取方法 * 根据官方文档,添加对英文版的bge embedding的指示模板 (#1585) Co-authored-by: zR <2448370773@qq.com> * Dev (#1618) * 增加了仅限GPT4的agent功能,陆续补充,中文版readme已写 * issue提到的一个bug * 温度最小改成0,但是不应该支持负数 * 修改了最小的温度 * 增加了部分Agent支持和修改了启动文件的部分bug * 修改了GPU数量配置文件 * 1 1 * 修复配置文件错误 * 更新readme,稳定测试 * 更改readme 0928 (#1619) * 增加了仅限GPT4的agent功能,陆续补充,中文版readme已写 * issue提到的一个bug * 温度最小改成0,但是不应该支持负数 * 修改了最小的温度 * 增加了部分Agent支持和修改了启动文件的部分bug * 修改了GPU数量配置文件 * 1 1 * 修复配置文件错误 * 更新readme,稳定测试 * 更新readme * fix readme * 处理序列化至磁盘的逻辑 * update version number to v0.2.5 --------- Co-authored-by: qiankunli <qiankun.li@qq.com> Co-authored-by: liqiankun.1111 <liqiankun.1111@bytedance.com> Co-authored-by: zR <2448370773@qq.com> Co-authored-by: glide-the <2533736852@qq.com> Co-authored-by: Water Zheng <1499383852@qq.com> Co-authored-by: Jim Zhang <dividi_z@163.com> Co-authored-by: Jim <zhangpengyi@taijihuabao.com> Co-authored-by: imClumsyPanda <littlepanda0716@gmail.com> Co-authored-by: Leego <leegodev@hotmail.com> Co-authored-by: hzg0601 <hzg0601@163.com> Co-authored-by: WilliamChen-luckbob <58684828+WilliamChen-luckbob@users.noreply.github.com>
207 lines
6.6 KiB
Python
207 lines
6.6 KiB
Python
import requests
|
||
import json
|
||
import sys
|
||
from pathlib import Path
|
||
|
||
root_path = Path(__file__).parent.parent.parent
|
||
sys.path.append(str(root_path))
|
||
from server.utils import api_address
|
||
from configs import VECTOR_SEARCH_TOP_K
|
||
from server.knowledge_base.utils import get_kb_path, get_file_path
|
||
|
||
from pprint import pprint
|
||
|
||
|
||
api_base_url = api_address()
|
||
|
||
|
||
kb = "kb_for_api_test"
|
||
test_files = {
|
||
"FAQ.MD": str(root_path / "docs" / "FAQ.MD"),
|
||
"README.MD": str(root_path / "README.MD"),
|
||
"test.txt": get_file_path("samples", "test.txt"),
|
||
}
|
||
|
||
print("\n\n直接url访问\n")
|
||
|
||
|
||
def test_delete_kb_before(api="/knowledge_base/delete_knowledge_base"):
|
||
if not Path(get_kb_path(kb)).exists():
|
||
return
|
||
|
||
url = api_base_url + api
|
||
print("\n测试知识库存在,需要删除")
|
||
r = requests.post(url, json=kb)
|
||
data = r.json()
|
||
pprint(data)
|
||
|
||
# check kb not exists anymore
|
||
url = api_base_url + "/knowledge_base/list_knowledge_bases"
|
||
print("\n获取知识库列表:")
|
||
r = requests.get(url)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert isinstance(data["data"], list) and len(data["data"]) > 0
|
||
assert kb not in data["data"]
|
||
|
||
|
||
def test_create_kb(api="/knowledge_base/create_knowledge_base"):
|
||
url = api_base_url + api
|
||
|
||
print(f"\n尝试用空名称创建知识库:")
|
||
r = requests.post(url, json={"knowledge_base_name": " "})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 404
|
||
assert data["msg"] == "知识库名称不能为空,请重新填写知识库名称"
|
||
|
||
print(f"\n创建新知识库: {kb}")
|
||
r = requests.post(url, json={"knowledge_base_name": kb})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert data["msg"] == f"已新增知识库 {kb}"
|
||
|
||
print(f"\n尝试创建同名知识库: {kb}")
|
||
r = requests.post(url, json={"knowledge_base_name": kb})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 404
|
||
assert data["msg"] == f"已存在同名知识库 {kb}"
|
||
|
||
|
||
def test_list_kbs(api="/knowledge_base/list_knowledge_bases"):
|
||
url = api_base_url + api
|
||
print("\n获取知识库列表:")
|
||
r = requests.get(url)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert isinstance(data["data"], list) and len(data["data"]) > 0
|
||
assert kb in data["data"]
|
||
|
||
|
||
def test_upload_docs(api="/knowledge_base/upload_docs"):
|
||
url = api_base_url + api
|
||
files = [("files", (name, open(path, "rb"))) for name, path in test_files.items()]
|
||
|
||
print(f"\n上传知识文件")
|
||
data = {"knowledge_base_name": kb, "override": True}
|
||
r = requests.post(url, data=data, files=files)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert len(data["data"]["failed_files"]) == 0
|
||
|
||
print(f"\n尝试重新上传知识文件, 不覆盖")
|
||
data = {"knowledge_base_name": kb, "override": False}
|
||
files = [("files", (name, open(path, "rb"))) for name, path in test_files.items()]
|
||
r = requests.post(url, data=data, files=files)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert len(data["data"]["failed_files"]) == len(test_files)
|
||
|
||
print(f"\n尝试重新上传知识文件, 覆盖,自定义docs")
|
||
docs = {"FAQ.MD": [{"page_content": "custom docs", "metadata": {}}]}
|
||
data = {"knowledge_base_name": kb, "override": True, "docs": json.dumps(docs)}
|
||
files = [("files", (name, open(path, "rb"))) for name, path in test_files.items()]
|
||
r = requests.post(url, data=data, files=files)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert len(data["data"]["failed_files"]) == 0
|
||
|
||
|
||
def test_list_files(api="/knowledge_base/list_files"):
|
||
url = api_base_url + api
|
||
print("\n获取知识库中文件列表:")
|
||
r = requests.get(url, params={"knowledge_base_name": kb})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert isinstance(data["data"], list)
|
||
for name in test_files:
|
||
assert name in data["data"]
|
||
|
||
|
||
def test_search_docs(api="/knowledge_base/search_docs"):
|
||
url = api_base_url + api
|
||
query = "介绍一下langchain-chatchat项目"
|
||
print("\n检索知识库:")
|
||
print(query)
|
||
r = requests.post(url, json={"knowledge_base_name": kb, "query": query})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert isinstance(data, list) and len(data) == VECTOR_SEARCH_TOP_K
|
||
|
||
|
||
def test_update_docs(api="/knowledge_base/update_docs"):
|
||
url = api_base_url + api
|
||
|
||
print(f"\n更新知识文件")
|
||
r = requests.post(url, json={"knowledge_base_name": kb, "file_names": list(test_files)})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert len(data["data"]["failed_files"]) == 0
|
||
|
||
|
||
def test_delete_docs(api="/knowledge_base/delete_docs"):
|
||
url = api_base_url + api
|
||
|
||
print(f"\n删除知识文件")
|
||
r = requests.post(url, json={"knowledge_base_name": kb, "file_names": list(test_files)})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert len(data["data"]["failed_files"]) == 0
|
||
|
||
url = api_base_url + "/knowledge_base/search_docs"
|
||
query = "介绍一下langchain-chatchat项目"
|
||
print("\n尝试检索删除后的检索知识库:")
|
||
print(query)
|
||
r = requests.post(url, json={"knowledge_base_name": kb, "query": query})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert isinstance(data, list) and len(data) == 0
|
||
|
||
|
||
def test_recreate_vs(api="/knowledge_base/recreate_vector_store"):
|
||
url = api_base_url + api
|
||
print("\n重建知识库:")
|
||
r = requests.post(url, json={"knowledge_base_name": kb}, stream=True)
|
||
for chunk in r.iter_content(None):
|
||
data = json.loads(chunk)
|
||
assert isinstance(data, dict)
|
||
assert data["code"] == 200
|
||
print(data["msg"])
|
||
|
||
url = api_base_url + "/knowledge_base/search_docs"
|
||
query = "本项目支持哪些文件格式?"
|
||
print("\n尝试检索重建后的检索知识库:")
|
||
print(query)
|
||
r = requests.post(url, json={"knowledge_base_name": kb, "query": query})
|
||
data = r.json()
|
||
pprint(data)
|
||
assert isinstance(data, list) and len(data) == VECTOR_SEARCH_TOP_K
|
||
|
||
|
||
def test_delete_kb_after(api="/knowledge_base/delete_knowledge_base"):
|
||
url = api_base_url + api
|
||
print("\n删除知识库")
|
||
r = requests.post(url, json=kb)
|
||
data = r.json()
|
||
pprint(data)
|
||
|
||
# check kb not exists anymore
|
||
url = api_base_url + "/knowledge_base/list_knowledge_bases"
|
||
print("\n获取知识库列表:")
|
||
r = requests.get(url)
|
||
data = r.json()
|
||
pprint(data)
|
||
assert data["code"] == 200
|
||
assert isinstance(data["data"], list) and len(data["data"]) > 0
|
||
assert kb not in data["data"]
|