Using XSL to make a hash of XML file

Question

I am trying to find a way to \"hash\" the contents of an XML file. At the root of this is a need to compare some text nodes that are passed in to text nodes that I am expec

Answer 1

Sounds like you need a position-dependent checksum. Are you asking for an XSLT implementation, or just the algorithm?

Here is an implementation of Fletcher's checksum in C, which should not be very hard to port to XSLT.

Update: Below is an XSLT 2.0 adaptation of Fletcher's checksum. Whether it's fast enough, depends on the size of your data and the amount of time you have. I'd be interested to hear how your tests go. To optimize, I would attempt to change xs:integer to xs:int.

Note that I have substituted plain addition for the bitwise OR (|) of the implementation I linked to above. I'm not really qualified to analyze the ramifications of this change in regard to uniformity or non-invertibility, but it seems OK as long as you don't have a smart hacker trying to maliciously bypass your checksum checks.

Do note that because of the above change, this implementation will not give the same results as true implementations of Fletcher's checksum (@MDBiker). So you can't compare the output of this function with that of Java's Fletcher16, for example. However it will always return the same result for the same input (it's deterministic), so you can compare the output of this function on two text strings.

    The quick brown fox jumps over the lazy dog.
    The quick frown box jumps over the hazy frog.

    
        Checksum 1:     
        Checksum 2:     
    

    
        
        
        
    

    
    
        
        
        
        
        
        
            
                
            
            
                
                
            
        
    

The output:

    Checksum 1: 65256    
    Checksum 2: 25689

A note on usage: You said you needed to run checksum on "the contents of an XML file. At the root of this is a need to compare some text nodes". If you pass a text node to foo:checksum(), it will work fine: its string value will be extracted.

FYI, I ran a performance test, to calculate the checksum of text nodes in a 535KB XML input file. Here was the initial template I used:

    Checksum of input:     

It finished in 0.8s, using Saxon PE.

Alternatively:

If the amount of text is not very large, it would probably be faster and more accurate to simply compare the strings themselves (instead of checksums) to each other. But maybe you can't get access to both text nodes at the same time, due to your architecture restrictions... I'm not clear on that from your description.