Question

It’s a small thing, really: I have this function that converts dict objects to xml.

Here’s the function:

def dictToXml(d):     from xml.sax.saxutils import escape      def unicodify(o):         if o is None:             return u'';         return unicode(o)      lines = []     def addDict(node, offset):         for name, value in node.iteritems():             if isinstance(value, dict):                 lines.append(offset + u'<%s>' % name)                 addDict(value, offset + u' ' * 4)                 lines.append(offset + u'</%s>' % name)             elif isinstance(value, list):                 for item in value:                     if isinstance(item, dict):                         lines.append(offset + u'<%s>' % name)                         addDict(item, offset + u' ' * 4)                         lines.append(offset + u'</%s>' % name)                     else:                         lines.append(offset + u'<%s>%s</%s>' % (name, escape(unicodify(item)), name))             else:                 lines.append(offset + u'<%s>%s</%s>' % (name, escape(unicodify(value)), name))      addDict(d, u'')     lines.append(u'')     return u'\n'.join(lines) 

For example, it converts this dictionary

{ 'site': { 'name': 'stackoverflow', 'blogger': [ 'Jeff', 'Joel' ] } } 

to:

<site>     <name>stackoverflow</name>     <blogger>jeff</blogger>     <blogger>joel</blogger> </site> 

It works, but the addDict function looks a little too repetitive. I’m sure there’s a way to refactor it into 3 co-recursive functions named addDict, addList and addElse, but my brain is stuck. Any help?

Also, any way to get rid of the offset + thing in every line would be nice.

NOTE: I chose these semantics because I’m trying to match the behavior of the json-to-xml converter in org.json, which I use in a different part of my project. If you got to this page just looking for a dictionary to xml converter, there are some really good options in some of the answers. (Especially pyfo).

Answer 1

I noticed you have commonality in adding items. Using this commonality I would refactor adding an item to a separate function.

def addItem(item, name, offset):           if isinstance(item, dict):                 lines.append(offset + u'<%s>' % name)                 addDict(item, offset + u' ' * 4)                 lines.append(offset + u'</%s>' % name)           else:                 lines.append(offset + u'<%s>%s</%s>' % (name, escape(unicodify(item)), name))  def addList(value,name, offset):         for item in value:             addItem(item, name, offset)  def addDict(node, offset):         for name, value in node.iteritems():             if isinstance(value, list):                 addList(value, name, offset)             else:                 addItem(value, name, offset) 

Advisory warning: this code is not tested or written by anybody who actually uses Python.