Python往事：ElementTree的单引号之谜

最近在针对某款设备的界面xml进行更新过程中，被告知回稿的字串放在了一个excel文件中，而我要上传到服务器的界面用语是用xml文件封装的。再经过详细求证了翻译组提供excel文件的原因后，我决定用python来完成界面用语xml的更新，但是在使用ElementTree库的时候，却发现这个库有点小瑕疵。就是会将xml文件的表头<xml/>这段中的双引号换成了单引号，虽然单双引号在解析xml上没有影响。但是如果上提交代码时有强校验的门禁处理等规则的话，就需要额外解释了。为此针对这个问题，查看了下源码并分享一种修改方案。

我遇到的情况如下图所示，原本要替换message的信息，结果执行完替换脚本后，发现xml声明表头也被替换了。这个变更在比较软件中会显得很明显。

发生这种事情的原因在于EelementTree#write()中将表头的格式默认写成了单引号。如下源码展示了write()的实现，可以发现在_get_wirte()的with循环体中直白的执行了一句写入操作：

<?xml version='1.0' encoding='%s'?>

    def write(self, file_or_filename,
              encoding=None,
              xml_declaration=None,
              default_namespace=None,
              method=None, *,
              short_empty_elements=True):
        """Write element tree to a file as XML.

        Arguments:
          *file_or_filename* -- file name or a file object opened for writing

          *encoding* -- the output encoding (default: US-ASCII)

          *xml_declaration* -- bool indicating if an XML declaration should be
                               added to the output. If None, an XML declaration
                               is added if encoding IS NOT either of:
                               US-ASCII, UTF-8, or Unicode

          *default_namespace* -- sets the default XML namespace (for "xmlns")

          *method* -- either "xml" (default), "html, "text", or "c14n"

          *short_empty_elements* -- controls the formatting of elements
                                    that contain no content. If True (default)
                                    they are emitted as a single self-closed
                                    tag, otherwise they are emitted as a pair
                                    of start/end tags

        """
        if not method:
            method = "xml"
        elif method not in _serialize:
            raise ValueError("unknown method %r" % method)
        if not encoding:
            if method == "c14n":
                encoding = "utf-8"
            else:
                encoding = "us-ascii"
        enc_lower = encoding.lower()
        with _get_writer(file_or_filename, enc_lower) as write:
            if method == "xml" and (xml_declaration or
                    (xml_declaration is None and
                     enc_lower not in ("utf-8", "us-ascii", "unicode"))):
                declared_encoding = encoding
                if enc_lower == "unicode":
                    # Retrieve the default encoding for the xml declaration
                    import locale
                    declared_encoding = locale.getpreferredencoding()
                write("<?xml version='1.0' encoding='%s'?>\n" % (
                    declared_encoding,))
            if method == "text":
                _serialize_text(write, self._root)
            else:
                qnames, namespaces = _namespaces(self._root, default_namespace)
                serialize = _serialize[method]
                serialize(write, self._root, qnames, namespaces,
                          short_empty_elements=short_empty_elements)

可能这是ElementTree在设计初为了方便在双引号中引用字串才将version和encoding改为用单引号展示。因为write()中没有复杂的间接依赖，可以直接将该方法复制到自己的工程里。为此，针对该处的修改就是重写ElementTree#write()。重新方案如下，先将源代码中的<?xml version='1.0' encoding='%s'?> 替换成 <?xml version=\"1.0\" encoding=\"%s\"?>。

同时针对提示引用缺失的方法，增加ElementTree前缀来指明调用路径。这样就可以保证整个write()也可以在自己的工程中被执行。修改后的代码如下：

def fix_write(self, file_or_filename,
              encoding=None,
              xml_declaration=None,
              default_namespace=None,
              method=None, *,
              short_empty_elements=True):
    if not method:
        method = "xml"
    elif method not in ElementTree._serialize:
        raise ValueError("unknown method %r" % method)
    if not encoding:
        if method == "c14n":
            encoding = "utf-8"
        else:
            encoding = "us-ascii"
    enc_lower = encoding.lower()
    with ElementTree._get_writer(file_or_filename, enc_lower) as write:
        if method == "xml" and (xml_declaration or
                                    (xml_declaration is None and
                                             enc_lower not in ("utf-8", "us-ascii", "unicode"))):
            declared_encoding = encoding
            if enc_lower == "unicode":
                # Retrieve the default encoding for the xml declaration
                import locale
                declared_encoding = locale.getpreferredencoding()
            write("<?xml version=\"1.0\" encoding=\"%s\"?>\n" % (
                declared_encoding,))
        if method == "text":
            ElementTree._serialize_text(write, self._root)
        else:
            qnames, namespaces = ElementTree._namespaces(self._root, default_namespace)
            serialize = ElementTree._serialize[method]
            serialize(write, self._root, qnames, namespaces,
                      short_empty_elements=short_empty_elements)

修改后，将原来调用ElementTree#write()的地方改成使用fix_write即可，同时不要忘了，将当前工程的elementTree对象作为第一入参穿进去。修改后的运行结果就会发现没有额外的格式变更了。

def update_fix():
    tree = ET.parse('element_test.xml')
    root = tree.getroot()
    messages = root.findall('message')
    messages[0].text = "no, it's so cold,let's take a shower"
    fix_write(tree, 'element_test_update_fix.xml', encoding="utf-8", xml_declaration=True)