From: Robert Klemme <bob.news@...>
Date: 2005-06-24T17:55:38+09:00
Subject: Re: REXML Optimization

Robert Klemme wrote:
> Bucco wrote:
>> So, I'm reading this document on XML and using XML as a database.  Of
>> course the author uses this cryptic perl script to parse the xml
>> file:
>>
>> #!/usr/bin/perl
>> use XML::LibXML;
>> my $parser = new XML::LibXML;
>> my $doc = $parser->parse_file( shift @ARGV );
>> my $balance = $doc->findvalue( '/checkbook/@balance-start' );
>> foreach my $record ( $doc->findnodes( '//debit' )) {
>>     $balance -= $record->findvalue( 'amount' );
>> }
>> foreach my $record ( $doc->findnodes( '//deposit' )) {
>>     $balance += $record->findvalue( 'amount' );
>> }
>> print "Current balance: $balance\n";
>>
>> So, since I was trying to figure out how to use xml as a database and
>> how to use REXML I gave the script a whack and tried to write a ruby
>> script to do the same thing.  Below is a sample of the xml file and
>> the ruby script:
>>
>> <?xml version="1.0"?>
>> <checkbook balanceStart="2460.62">
>> <title>expenses: january 2002</title>
>>
>>   <debit category="clothes">
>>     <amount>31.19</amount>
>>     <date><year>2002</year><month>1</month><day>3</day></date>
>>     <payto>Walking Store</payto>
>>     <description>shoes</description>
>>   </debit>
>>
>>   <deposit category="salary">
>>     <amount>1549.58</amount>
>>     <date><year>2002</year><month>1</month><day>7</day></date>
>>     <payor>Bob's Bolts</payor>
>>   </deposit>
>> </checkbook>
>>
>>
>> #!/usr/bin/ruby -w
>> require 'rexml/document'
>>
>> # Read in XML doc
>> doc = REXML::Document.new(File.open('cb.xml'))
>> # Future version need to have entry from command line
>>
>> # Find the balance and assign to float variabl 'balance'
>> balance = doc.root.attributes['balanceStart'].to_f
>>
>> # Calculate debits and balance
>> doc.elements.each("//debit/amount") {|o| balance -= o.text.to_f}
>> # Calculate deposits and balance
>> doc.elements.each("//deposit/amount") {|i| balance += i.text.to_f}
>>
>> #Display final balance:
>> puts balance
>>
>> Of course I was able to complete the same task as teh perl script in
>> ruby with less code. (Not to mention easier to read code)
>>
>> Just to help me complete the learning process, I wish to pose the
>> question to the group: Is there a better way to do this, and is there
>> more optimization I can do to my code?
>
> You could get rid of one traversal by iterating all "amounts" and do
> the calculation based on the parent element's type.

If you want to speed up things even more you can do stream processing with
REXML's SAX like API:
http://www.germane-software.com/software/rexml/docs/tutorial.html#id2248482

Kind regards

    robert