Is there a way to merge two XML files via Python?

Dave Angel d at
Thu Jan 12 12:02:41 EST 2012

On 01/12/2012 11:39 AM, Stefan Behnel wrote:
> J, 12.01.2012 17:04:
>> This is more a theory exercise and something I'm trying to figure out,
>> and this is NOT a homework assignment...
>> I'm trying to make a tool I use at work more efficient :)
>> So this is at test tool that generates an XML file as it's output that
>> is eventually used by a web service to display test results and system
>> information.
>> The problem is that the testing is broken down into to different runs:
>> Functional and Automated where the Functional tests are all manual,
>> then the automated tests are run separately, usually overnight.
>> Each of those test runs generates essentially an identical XML file.
>> What I want to learn is a way to merge them.
> Ok - how large are these files? (i.e., do they easily fit into memory?)
>> In abstract terms, the idea is essentially to diff the two files
>> creating a patch and then use that patch to merge the two files into a
>> single XML file.
> I wouldn't go through patch. If they fit into memory, just load both, merge
> one into the other eliminating duplicates, and save that.
> Or rather, load just one and process the other one incrementally using
> ElementTree's iterparse().
>> SO what I was hoping I could get pointers on from those of you who are
>> experienced in using Python with XML is what python libraries or means
>> are there for working with XML files specifically, and how easy or
>> difficult would this be?
> Depends on how easy it is to recognise duplicates in your specific data
> format. Once you've managed to do that, the rest is trivial.
>> I'm also doing research on my own in my spare time on this, but I also
>> wanted to ask here to get the opinion of developers who are more
>> experienced in working with XML than I am.
> I recommend looking at the stdlib xml.etree.ElementTree module or the
> external lxml package (which contains the ElementTree compatible lxml.etree
> module). The latter will (likely) make things easier due to full XPath
> support and some other goodies, but ElementTree is also quite quick and
> easy to use by itself.
> Stefan
Question for jeff:

Have you tried doing it by hand?  Do you know when a duplicate should be 
ignored, when it should be replicated, when it should be represented by 
incrementing a count?

xml is very flexible, but the final reader of your file may not be so 
flexible.  (e.g. if it has to match a wsdl)  If two runs differ only by 
some timing field, then you might need to sum those times, and produce 
an average in the final run.  Or a max value, or both.



More information about the Python-list mailing list