AI大模型应用开发：从API调用到可交付工程系统的架构与实践

from langchain_community.document_loaders import PyPDFLoader
from langchain_text_splitters import RecursiveCharacterTextSplitter
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain_chroma import Chroma
from langchain.chains import RetrievalQA

loader = PyPDFLoader("./knowledge_base/manual.pdf")
docs = loader.load()

splitter = RecursiveCharacterTextSplitter(
    chunk_size=500, chunk_overlap=50,
    separators=["\n\n", "\n", "。", "！", "？", "；", "，", " ", ""]
)
chunks = splitter.split_documents(docs)

vectorstore = Chroma.from_documents(
    chunks, OpenAIEmbeddings(), persist_directory="./chroma_db"
)
retriever = vectorstore.as_retriever(
    search_type="mmr", search_kwargs={"k": 5, "fetch_k": 20}
)
qa = RetrievalQA.from_chain_type(
    llm=ChatOpenAI(model="gpt-4o-mini", temperature=0.2),
    retriever=retriever, return_source_documents=True
)
result = qa("CPU使用率持续超过90%如何处理？")
print(result["result"], result["source_documents"])

from openai import OpenAI
from pydantic import BaseModel

class QueryOrderInput(BaseModel):
    start_time: str
    end_time: str

client = OpenAI()
tools = [{
    "type": "function",
    "function": {
        "name": "query_order_count",
        "description": "查询指定时间范围内的订单数量",
        "parameters": QueryOrderInput.model_json_schema()
    }
}]

response = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "查一下上个月的订单量"}],
    tools=tools
)
tool_call = response.choices[0].message.tool_calls[0]
args = QueryOrderInput.model_validate_json(tool_call.function.arguments)
print(f"执行 query_order_count({args.start_time}, {args.end_time})")