Windows环境下LangChain 0.1.2中HuggingFaceInstructEmbeddings运行报错求助

问题背景 #

Windows系统中使用Python 3.10.0、LangChain 0.1.2及GoogleGenerativeAI时，运行包含HuggingFaceInstructEmbeddings的代码触发pwd模块找不到的错误。原因是langchain_community库的pebblo.py文件导入了仅Linux/Unix系统存在的pwd模块，Windows无此模块。

解决方案 #

方案1：修改pebblo.py适配Windows #

1. 找到文件路径：D:\Program Files\Python310\lib\site-packages\langchain_community\document_loaders\pebblo.py
2. 打开文件，将顶部的import pwd替换为条件导入：

   try:
       import pwd
   except ImportError:
       pwd = None
   

3. 找到文件中使用pwd的代码段（通常是获取当前用户名的部分），替换为跨平台兼容写法：

   if pwd is not None:
       username = pwd.getpwuid(os.getuid())[0]
   else:
       username = os.environ.get('USERNAME', 'unknown')
   

方案2：修改导入语句避开Pebblo模块 #

若不需要使用PebbloSafeLoader，直接从langchain_community导入CSVLoader，避免触发pebblo.py加载：
将代码中的：

from langchain.document_loaders.csv_loader import CSVLoader

修改为：

from langchain_community.document_loaders.csv_loader import CSVLoader

用户原始代码 #

from langchain_google_genai import GoogleGenerativeAI   
from dotenv import load_dotenv 
import os
from langchain.chains import RetrievalQA
from langchain.document_loaders.csv_loader import CSVLoader    
from langchain_community.embeddings import HuggingFaceInstructEmbeddings
from langchain_community.vectorstores import FAISS

load_dotenv()
google_api_key= os.getenv('GOOGLE_API_KEY')
llm = GoogleGenerativeAI(model="models/text-bison-001", google_api_key=google_api_key, temperature = 0.7)

loader = CSVLoader(file_path="codebasics_faqs.csv", source_column="prompt")
docs = loader.load()

instructor_embeddings = HuggingFaceInstructEmbeddings(
    query_instruction="Represent the query for retrieval: "
)
vectordb = FAISS.from_documents(documents = docs, embeddings = instructor_embeddings)
retriever = vectordb.as_retriever()
rdocs = vectordb.get_relevent_documents("do I get a job gurantee?")
print(rdocs)

完整报错信息 #

File "c:\Users\USER\Desktop\project\tut_langchain\tut_googleplam.py", line 5, in <module>
    from langchain.document_loaders.csv_loader import CSVLoader
  File "D:\Program Files\Python310\lib\site-packages\langchain\document_loaders\csv_loader.py", line 1, in <module>
    from langchain_community.document_loaders.csv_loader import (
  File "D:\Program Files\Python310\lib\site-packages\langchain_community\document_loaders\__init__.py", line 163, in <module>    from langchain_community.document_loaders.pebblo import PebbloSafeLoader
  File "D:\Program Files\Python310\lib\site-packages\langchain_community\document_loaders\pebblo.py", line 5, in <module>    
    import pwd
ModuleNotFoundError: No module named 'pwd'

内容的提问来源于stack exchange，提问作者Monidipa Das