You need to enable JavaScript to run this app.
优惠活动
大模型
产品
解决方案
定价
更多

Python中llama_index与llama_parse导入无法解析问题求助

问题描述(中文翻译)

我正在开发一个多LLM代理Python程序,参考Tech with Tim的教程视频完成了环境配置:创建并激活了虚拟环境(venv),安装了llama_index、llama_parse等所需依赖。但运行代码时,所有llama_index相关导入及llama_parse导入均提示“Import 'xxx' could not be resolved”,我的Python版本为3.11.8,相关代码如下:

from llama_index.llms.ollama import Ollama
from llama_parse import LlamaParse
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, PromptTemplate
from llama_index.core.embeddings import resolve_embed_model
from llama_index.core.tools import QueryEngineTool, ToolMetadata
from llama_index.core.agent import ReActAgent
from pydantic import BaseModel
from llama_index.core.output_parsers import PydanticOutputParser
from llama_index.core.query_pipeline import QueryPipeline
from prompts import context, code_parser_template
from code_reader import code_reader
from dotenv import load_dotenv
import os
import ast

load_dotenv()

llm = Ollama(model="mistral", request_timeout=30.0)

parser = LlamaParse(result_type="markdown")

file_extractor = {".pdf": parser}
documents = SimpleDirectoryReader("./data", file_extractor=file_extractor).load_data()

embed_model = resolve_embed_model("local:BAAI/bge-m3")
vector_index = VectorStoreIndex.from_documents(documents, embed_model=embed_model)
query_engine = vector_index.as_query_engine(llm=llm)

tools = [
    QueryEngineTool(
        query_engine=query_engine,
        metadata=ToolMetadata(
            name="api_documentation",
            description="this gives documentation about code for an API. Use this for reading docs for the API",
        ),
    ),
    code_reader,
]

code_llm = Ollama(model="codellama")
agent = ReActAgent.from_tools(tools, llm=code_llm, verbose=True, context=context)


class CodeOutput(BaseModel):
    code: str
    description: str
    filename: str


parser = PydanticOutputParser(CodeOutput)
json_prompt_str = parser.format(code_parser_template)
json_prompt_tmpl = PromptTemplate(json_prompt_str)
output_pipeline = QueryPipeline(chain=[json_prompt_tmpl, llm])

while (prompt := input("Enter a prompt (q to quit): ")) != "q":
    retries = 0

    while retries < 3:
        try:
            result = agent.query(prompt)
            next_result = output_pipeline.run(response=result)
            cleaned_json = ast.literal_eval(str(next_result).replace("assistant:", ""))
            break
        except Exception as e:
            retries += 1
            print(f"Error occured, retry #{retries}:", e)

    if retries >= 3:
        print("Unable to process request, try again...")
        continue

    print("Code generated")
    print(cleaned_json["code"])
    print("\n\nDesciption:", cleaned_json["description"])

    filename = cleaned_json["filename"]

    try:
        with open(os.path.join("output", filename), "w") as f:
            f.write(cleaned_json["code"])
        print("Saved file", filename)
    except:
        print("Error saving file...")

请问我哪里操作出错了?


问题排查与解决方法
  • 确认虚拟环境激活状态
    运行代码前必须确保venv已正确激活:

    • Windows系统:执行venv\Scripts\activate
    • macOS/Linux系统:执行source venv/bin/activate
      激活成功后终端会显示(venv)前缀,此时再运行代码。
  • 检查依赖是否安装在虚拟环境中
    激活venv后执行pip list,查看输出是否包含llama-index、llama-parse、pydantic、python-dotenv这些包。如果缺失,重新安装:

    pip install llama-index llama-parse pydantic python-dotenv
    

    注意:PyPI上的包名是llama-index(带连字符),如果之前用llama_index(下划线)安装会失败。

  • 校准IDE的Python解释器
    如果用VS Code或PyCharm,需确保IDE使用的是虚拟环境内的Python解释器:

    • VS Code:按Ctrl+Shift+P(Windows)/Cmd+Shift+P(macOS),输入Python: Select Interpreter,选择venv目录下的Python程序。
    • PyCharm:依次点击File → Settings → Project: [项目名] → Python Interpreter,选择对应的venv解释器。
  • 终端验证导入有效性
    激活venv后打开Python交互环境(输入python),手动执行导入语句:

    from llama_index.llms.ollama import Ollama
    from llama_parse import LlamaParse
    

    如果终端能成功导入,说明是IDE配置问题;如果终端也报错,卸载重装依赖:

    pip uninstall -y llama-index llama-parse
    pip install llama-index llama-parse
    
  • 确认虚拟环境Python版本
    激活venv后执行python --version,确保版本为3.11.8,避免误用到系统全局的其他Python版本。


内容的提问来源于stack exchange,提问作者Joshie

相关产品推荐
方舟 Agent Plan

超全模态模型 × Harness 升级,最新支持 Deepseek-V4.1-Flash、GLM-5.3 系列、Doubao-Seedream-5.0-pro、Kimi-K3 (部分), 限时 9.9 元起

最近更新时间:2026.06.12 12:43:11