[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

RE: [xsl] Diffing XML


Subject: RE: [xsl] Diffing XML
From: Emma Burrows <Emma.Burrows@xxxxxxxxxxx>
Date: Thu, 25 Oct 2012 16:40:41 +0100

Yes, that's how I thought it would be useful. They don't update every chapter
every month, so this would just narrow things down. Then I can possibly
deep-equal my way through the nested structures to be a bit more precise about
where the changes are. When I'm feeling strong, I'll think about finding out
what is different too!

The only issue I can see looming is that it'll be easy to work out what's new
since I'll be processing the "new" document (no "old" topic with the same id =
new record). I'll have to think about how to find out if something has been
deleted too; probably easy enough for topics by comparing the list of ids for
the two versions, but a bit more tricky for id-less elements.

I seem to have created myself a nice big job to do. ;)

Emma


-----Original Message-----
From: Michael Kay [mailto:mike@xxxxxxxxxxxx]
Sent: 24 October 2012 16:28
To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
Subject: Re: [xsl] Diffing XML

One of the drawbacks of deep-equal is that it doesn't tell you where the
differences are; another is that it doesn't give you any control over how to
do the comparison (Saxon's saxon:deep-equal() variant tries to remedy both
problems). But it can certainly used as part of the solution, by quickly
eliminating subtrees that don't need to be examined any further.

Michael Kay
Saxonica

On 24/10/2012 15:27, Emma Burrows wrote:
> Thinking about it further - I'm wondering whether something like deep-equals
might suffice. What the users apparently really want right now is to know
which parts of the document have changed so they can concentrate on those when
checking the output on a website. In which case, starting with top-level
elements and iterating my way down through the children, I could in theory at
the very least output "Something has changed in <p> number 3 in the topic
entitled 'Topic Title'".
>
> I realise there are many pitfalls ahead and of course the minute they
> see it, they will say "oh, but can't you make it do X?", but if I can
> convince them they don't need to know exactly what has changed (I'm an
> optimist), that might help. Or is there an even better way? (Assuming
> one were daft enough to take on such a project :)
>
> Emma
>
>
> -----Original Message-----
> From: Emma Burrows [mailto:Emma.Burrows@xxxxxxxxxxx]
> Sent: 24 October 2012 14:48
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject: RE: [xsl] Diffing XML
>
> Thanks Michael,
>
> Thanks for the response. Yes, I'm thinking doing it entirely myself might a
bit too ambitious. The data is relatively stable at this point and gets
updated once a month which should theoretically reduce the number of things to
check for each time. But even so, I can tell diffing is an art.
>
> DeltaXML does seem to offer some interesting options and it could probably
be integrated into our CMS (given a chisel and a mallet - the CMS is getting a
bit old), but I don't think we have any budget to buy another tool and it
sounds as if the users have some very specific requirements (like exporting
the list of user-friendly differences to an Excel spreadsheet!). So I was
looking for ideas about how to tackle the problem just in case I do indeed
need to implement it!
>
> Emma
>
>
> -----Original Message-----
> From: Michael Kay [mailto:mike@xxxxxxxxxxxx]
> Sent: 24 October 2012 11:48
> To: xsl-list@xxxxxxxxxxxxxxxxxxxxxx
> Subject: Re: [xsl] Diffing XML
>
> In general differencing well is quite a challenge, e.g. handling an
arbitrary number of inserted elements in either document,  addition or removal
of "div" layers, combining/splitting of paragraphs, reformatted indentation,
etc. Doing it better than a general-purpose product such as DeltaXML could
turn out to be a project that will keep you busy for a while.
>
> Michael Kay
> Saxonica
>
> On 24/10/2012 11:36, Emma Burrows wrote:
>> I have a requirement to produce an end-user-readable "checklist" of all the
places where an XML file has changed since the last version, with custom
explanations of what each difference is. I'm able to run diffs which are fine
for my own purposes, but the end users need the differences spelled out more
precisely in plain language (eg: "there is an extra paragraph here", "the text
'xyz' has changed", "the attribute 'audience' has been changed to 'book'"
etc).
>>
>> Being an XSLT developer, I'm thinking of using an XSLT stylesheet to work
on the "new" version of the file, document() in the "old" version, and then
compare the nodes in the "new" version to those in the "old" version,
generating appropriate messages into an HTML output as I go along.
>>
>> Does that sound like a reasonable approach? Are there existing tools
>> or examples that might do what I'm after? Any recommendations on the
>> best way of comparing individual nodes? I am planning to do this in
>> Oxygen 14 so the world is my oyster as far as XSLT is concerned. :)
>>
>> Just looking for general suggestions to point me in the right direction.
Thanks!
>>
>>
>> _____________________________________________________________________
>> _ This email has been scanned by the Symantec Email Security.cloud
>> service.
>> For more information please visit http://www.symanteccloud.com
>> _____________________________________________________________________
>> _
>
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> For more information please visit http://www.symanteccloud.com
> ______________________________________________________________________
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> For more information please visit http://www.symanteccloud.com
> ______________________________________________________________________
>
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> For more information please visit http://www.symanteccloud.com
> ______________________________________________________________________
>
> ______________________________________________________________________
> This email has been scanned by the Symantec Email Security.cloud service.
> For more information please visit http://www.symanteccloud.com
> ______________________________________________________________________


______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please visit http://www.symanteccloud.com
______________________________________________________________________

______________________________________________________________________
This email has been scanned by the Symantec Email Security.cloud service.
For more information please visit http://www.symanteccloud.com
______________________________________________________________________


Current Thread
Keywords