text2sql工具使用-Langchain-chatchat-0.3.1版本部署流程以及对接Oracle数 - ToB企服应用市场:ToB评测及商务社交产业平台

git clone https://github.com/chatchat-space/Langchain-Chatchat.git

复制代码

conda create -n llm_tl python==3.11

复制代码

source ~/.bashrc

复制代码

conda init

复制代码

conda activate llm_tl

复制代码

pip install "langchain-chatchat[xinference]" -U

复制代码

pip install "xinference[all]" -i https://pypi.tuna.tsinghua.edu.cn/simple

复制代码

pip install "xinference[transformers]" -i https://pypi.tuna.tsinghua.edu.cn/simple

复制代码

pip install "xinference[vllm]"

复制代码

pip install xinference -i https://pypi.tuna.tsinghua.edu.cn/simple

复制代码

CMAKE_ARGS="-DLLAMA_CUBLAS=on" pip install llama-cpp-python -i https://pypi.tuna.tsinghua.edu.cn/simple

复制代码

pip install 'xinference[sglang]' -i https://pypi.tuna.tsinghua.edu.cn/simple

复制代码

XINFERENCE_HOME=/root/autodl-tmp/xinference XINFERENCE_MODEL_SRC=modelscope xinference-local --host 0.0.0.0 --port 9997

复制代码

xinference launch --model-engine vllm --model-name glm4-chat --size-in-billions 9 --model-format pytorch --quantization none

复制代码

xinference list

复制代码

XINFERENCE_HOME=/root/autodl-tmp/xinference XINFERENCE_MODEL_SRC=modelscope xinference-local --host 0.0.0.0 --port 9997

复制代码

xinference launch --model-name bge-large-zh-v1.5 --model-type embedding

复制代码

conda create -n chatchat_0.3.1 python==3.11

复制代码

conda activate chatchat_0.3.1

复制代码

pip install langchain-chatchat -U

复制代码

# on linux or macos
export CHATCHAT_ROOT=/path/to/chatchat_data

复制代码

chatchat init

复制代码

# 默认选用的 LLM 名称
DEFAULT_LLM_MODEL: qwen1.5-chat
# 默认选用的 Embedding 名称
DEFAULT_EMBEDDING_MODEL: bge-large-zh-v1.5
# 将 `LLM_MODEL_CONFIG` 中 `llm_model, action_model` 的键改成对应的 LLM 模型
# 在 `MODEL_PLATFORMS` 中修改对应模型平台信息

复制代码

chatchat kb -r

复制代码

Traceback (most recent call last):
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/init_database.py", line 156, in main
folder2db(kb_names=args.kb_name, mode="recreate_vs", embed_model=args.embed_model)
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/server/knowledge_base/migrate.py", line 130, in folder2db
kb.create_kb()
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/server/knowledge_base/kb_service/base.py", line 80, in create_kb
self.do_create_kb()
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/server/knowledge_base/kb_service/faiss_kb_service.py", line 51, in do_create_kb
self.load_vector_store()
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/server/knowledge_base/kb_service/faiss_kb_service.py", line 28, in load_vector_store
return kb_faiss_pool.load_vector_store(kb_name=self.kb_name,
File "/Users/hkk/Amap/ai/Langchain-Chatchat/myenv/lib/python3.8/site-packages/chatchat/server/knowledge_base/kb_cache/faiss_cache.py", line 132, in load_vector_store
raise RuntimeError(f"向量库 {kb_name} 加载失败。")
RuntimeError: 向量库 samples 加载失败。

复制代码

pip install faiss-cpu==1.7.4

复制代码

pip install rank_bm25 -i https://mirrors.aliyun.com/pypi/simple

复制代码

----------------------------------------------------------------------------------------------------
知识库名称：samples
知识库类型：faiss
向量模型：：bge-large-zh-v1.5
知识库路径：/root/anaconda3/envs/chatchat/lib/python3.11/site-packages/chatchat/data/knowledge_base/samples
文件总数量：47
入库文件数：42
知识条目数：740
用时：0:02:29.701002
----------------------------------------------------------------------------------------------------
总计用时：0:02:33.414425

复制代码

chatchat start -a

复制代码

instantclient-basic-linux.x64-11.2.0.4.0.zip
instantclient-sqlplus-linux.x64-11.2.0.4.0.zip

复制代码

unzip instantclient-basic-linux.x64-11.2.0.4.0.zip -d /opt/oracle
unzip instantclient-sqlplus-linux.x64-11.2.0.4.0.zip -d /opt/oracle

复制代码

vi ~/.bashrc

复制代码

在文件末尾添加以下内容：
export ORACLE_HOME=/opt/oracle/instantclient_11_2
export LD_LIBRARY_PATH=$ORACLE_HOME:/usr/local/nvidia/lib:/usr/local/nvidia/lib64
export PATH=$ORACLE_HOME:$PATH

复制代码

source ~/.bashrc

复制代码

ln -s /opt/oracle/instantclient_11_2/libclntsh.so.11.1 /opt/oracle/instantclient_11_2/libclntsh.so

复制代码

sudo apt-get install libaio1 # Debian/Ubuntu
sudo yum install libaio # Red Hat/CentOS

复制代码

chmod +r /opt/oracle/instantclient_11_2/*.so

复制代码

sqlplus -v

复制代码

import cx_Oracle
print(cx_Oracle.clientversion())

复制代码

/opt/oracle/instantclient_11_2。你可以使用以下命令检查和修复路径：
echo $LD_LIBRARY_PATH
export LD_LIBRARY_PATH=/opt/oracle/instantclient_11_2:/usr/local/nvidia/lib:/usr/local/nvidia/lib64

复制代码

def query_database(query: str, config: dict):
model_name = config["model_name"]
top_k = config["top_k"]
return_intermediate_steps = config["return_intermediate_steps"]
sqlalchemy_connect_str = config["sqlalchemy_connect_str"]
db = SQLDatabase.from_uri(sqlalchemy_connect_str)
from chatchat.server.utils import get_ChatOpenAI
llm = get_ChatOpenAI(
model_name=model_name,
temperature=0.1,
streaming=True,
local_wrap=True,
verbose=True,
)
table_names = config["table_names"]
result = None
if len(table_names) > 0:
db_chain = SQLDatabaseChain.from_llm(
llm,
db,
verbose=True,
top_k=top_k,
return_intermediate_steps=return_intermediate_steps,
)
result = db_chain.invoke({"query": query, "table_names_to_use": table_names})
# 检查 result['intermediate_steps'] 是否为列表，并包含字典
if isinstance(result['intermediate_steps'], list):
last_step = result['intermediate_steps'][-1]
if isinstance(last_step, dict) and 'sql_cmd' in last_step:
# 提取生成的 SQL 语句
sql_cmd = last_step['sql_cmd']
elif isinstance(last_step, str):
# 如果是字符串，尝试解析并提取SQL语句
print(f"警告：最后一步是字符串，尝试解析: {last_step}")
sql_cmd = last_step.split("SQLQuery:")[-1].strip()
# 移除可能出现的反引号和其他多余字符
sql_cmd = sql_cmd.replace('```sql', '').replace('```', '').strip()
# 移除可能多余的 "Query:" 前缀
sql_cmd = sql_cmd.replace('Query:', '').strip()
else:
raise TypeError(
f"Unexpected format in last intermediate step. Expected a dictionary with 'sql_cmd' key or a parsable string, got: {last_step}")
else:
raise TypeError(
f"Unexpected format in intermediate_steps. Expected a list, got: {type(result['intermediate_steps'])}")
# 打印原始 SQL 语句
print("原始 SQL 语句:", sql_cmd)
# 修正 Oracle 不支持的 FETCH NEXT 语法
if re.search(r'FETCH (FIRST|NEXT) \d+ ROWS ONLY', sql_cmd, re.IGNORECASE):
match = re.search(r'FETCH (FIRST|NEXT) (\d+) ROWS ONLY', sql_cmd, re.IGNORECASE)
if match:
limit = match.group(2)
# 将 SQL 查询封装到一个子查询中，然后用 ROWNUM 限制行数
sql_cmd = re.sub(r'SELECT', f'SELECT * FROM (SELECT', sql_cmd, flags=re.IGNORECASE)
sql_cmd = re.sub(r'FETCH (FIRST|NEXT) \d+ ROWS ONLY', f') WHERE ROWNUM <= {limit}', sql_cmd,
flags=re.IGNORECASE)
print("已修正 SQL 语法:", sql_cmd)
else:
print("无需修正的 SQL 语法:", sql_cmd)
# 移除表名和列名中的双引号
sql_cmd = re.sub(r'"(\w+)"', r'\1', sql_cmd)
# 移除可能多余的分号以避免 ORA-00911 错误
sql_cmd = sql_cmd.rstrip(';')
try:
# 执行调整后的 SQL 查询
final_result = db.run(sql_cmd)
print("执行的最终 SQL 语句:", sql_cmd)
except Exception as e:
print(f"执行 SQL 查询时出错: {e}")
raise e
context = f"""查询结果: {final_result}\n\n"""
return context

复制代码

XINFERENCE_HOME=/root/autodl-tmp/xinference XINFERENCE_MODEL_SRC=modelscope xinference-local --host 0.0.0.0 --port 9997

复制代码

export CHATCHAT_ROOT=/root/autodl-tmp/Langchain-Chatchat/to/chatchat_data

复制代码

chatchat init

复制代码

export CHATCHAT_ROOT=/root/autodl-tmp/Langchain-Chatchat/to/chatchat_data

复制代码

chatchat start -a

复制代码