使用Python Open()读取XML(.csproj)文件时出现无效字符致格式错误

我用Python读取.csproj格式的XML文件，执行file_content=f.read().replace('\n','')时，标签前出现了这类无效字符，导致XML格式不合法报错。相关代码如下：

def parse_csproj(repository_root, repository_name, source):

    # If the file is not XML, then a ParseError
    # is raised. Return None to indicate that the
    # file is not a csproj file.
    try:
        p = repository_root / repository_name / source
        with p.open() as f:
            file_content=f.read().replace('\n','') #I put a breakpoint here and I see invalid characters before <xml tag
            root=ET.fromstring(str(file_content))
            #root = ET.parse(f).getroot()
    except Exception as ex:
        return None

    # Extract the list of NuGet packages that are
    # used by the csproj file.
    refs = []
    for element in root.findall('.//{*}PackageReference'): #You need to add {*} otherwise it won't work. Or you 
        #will need to access based on namespace. Because notice in .csproj, the root element contains a namespace.
        ref = PackageReference(
            ecosystem = 'NuGet',
            repository_root = repository_root,
            repository_name = repository_name,
            source = source,
            package_name = element.attrib['Include'],
            version = element.attrib['Version'])
        refs.append(ref)

    return refs 

注：是UTF-8 BOM（字节顺序标记），Windows环境下生成的.csproj文件常带有这个标记，直接读取会残留该字符导致XML解析失败。

解决方案 #

• 方法一：指定编码读取文件
  修改文件打开方式，使用utf-8-sig编码，它会自动识别并去除UTF-8文件开头的BOM标记：

  with p.open('r', encoding='utf-8-sig') as f:
      file_content = f.read().replace('\n','')
  

• 方法二：直接使用ET.parse读取（更推荐）
  无需手动读取文件内容再解析，ET.parse可以直接处理带BOM的文件，同时简化代码：

  def parse_csproj(repository_root, repository_name, source):
      try:
          p = repository_root / repository_name / source
          root = ET.parse(p).getroot()  # 直接传入文件路径解析
      except Exception as ex:
          return None
  
      refs = []
      for element in root.findall('.//{*}PackageReference'):
          ref = PackageReference(
              ecosystem = 'NuGet',
              repository_root = repository_root,
              repository_name = repository_name,
              source = source,
              package_name = element.attrib['Include'],
              version = element.attrib['Version'])
          refs.append(ref)
      return refs 
  

内容的提问来源于stack exchange，提问作者nikhil