如何将Streamlit上传的YAML文件转换为Python字典？

问题场景 #

开发Streamlit应用时，通过streamlit.sidebar.file_uploader()上传YAML配置文件，尝试读取并转换为Python字典时连续遇到报错：

• 直接用open()读取触发TypeError
• 改用.getvalue()后触发OSError

初始代码及错误 #

import streamlit as st
import yaml

config = st.sidebar.file_uploader("Upload config file")

if config is not None:
  with open(config, 'r') as uploaded_config:
    loaded_config = yaml.safe_load(uploaded_config)
  st.write(loaded_config['scoring'])

错误信息：

TypeError: expected str, bytes or os.PathLike object, not UploadedFile
Traceback:
File "/Users/user/opt/anaconda3/envs/npextension_app_env/lib/python3.8/site-packages/streamlit/scriptrunner/script_runner.py", line 557, in _run_script
exec(code, module.__dict__)
File "app.py", line 8, in <module>
with open(config, 'r') as uploaded_config:

修改后代码及新错误 #

import streamlit as st
import yaml

config = st.sidebar.file_uploader("Upload config file")

if config is not None:
  with open(config.getvalue(), 'r') as uploaded_config:
    loaded_config = yaml.safe_load(uploaded_config)
  st.write(loaded_config['scoring'])

新错误信息：

OSError: [Errno 63] File name too long: b'features_list:\n - NO2 Mean\n - 03 Mean\n - SO2 Mean\n - is_weekend\n\nevents_list:\n - is_xmas\n\nds: Date Local\n\ny: CO Mean\n\nscoring: mae\n\nfreq: D\n\ntest_p: 0.06\n\nvalid_p: 0.1\n\nk: 3\n'
Traceback:
File "/Users/user/opt/anaconda3/envs/npextension_app_env/lib/python3.8/site-packages/streamlit/scriptrunner/script_runner.py", line 557, in _run_script
exec(code, module.__dict__)
File "app.py", line 8, in <module>
with open(config.getvalue(), 'r') as uploaded_config:

上传的YAML文件内容 #

features_list:
  - NO2 Mean
  - 03 Mean
  - SO2 Mean
  - is_weekend
events_list:
  - is_xmas
ds: Date Local
y: CO Mean
scoring: mae
freq: D
test_p: 0.06
valid_p: 0.1
k: 3

错误原因 #

• 第一个错误：streamlit.sidebar.file_uploader()返回的是UploadedFile对象，而非文件路径字符串，open()函数需要传入路径，因此类型不匹配报错。
• 第二个错误：UploadedFile.getvalue()返回的是文件的字节内容，把字节内容当作文件名传给open()，自然会出现“文件名过长”的错误。

正确解决方案 #

直接利用UploadedFile对象的read()方法读取字节内容，再传给yaml.safe_load()解析：

import streamlit as st
import yaml

config = st.sidebar.file_uploader("Upload config file")

if config is not None:
    # 读取文件字节内容并解析为YAML
    loaded_config = yaml.safe_load(config.read())
    # 验证读取结果
    st.write(loaded_config['scoring'])

如果需要处理编码问题，可以先将字节内容解码为字符串：

loaded_config = yaml.safe_load(config.read().decode('utf-8'))

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