XML Parsing and Xpath confusion

naChoZ has asked for the wisdom of the Perl Monks concerning the following question:

I've been trying to write some code to read and write some xml and having no luck. I'm attempting to use XML::XPath, but I'm open to others if anyone has any suggestions. (I tried XML::Simple, but it completely butchers the document in the output.)

Here's some sample xml:

<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE ADI SYSTEM "ADI.DTD">
<ADI>
  <Metadata>
    <AMS Asset_Class="package" Asset_Name="xxxxxxxxxxxxxxxxxxxxxxxx_pa
+ckage" />
    <App_Data App="MOD" Name="Metadata_Spec_Version" Value="xxxxxxxxxx
+xxxxx"/>
    <App_Data App="MOD" Name="Metadata_Tier" Value="xxxxxxxxxxxxxxxxxx
+xxxxx"/>
  </Metadata>
  <Asset>
    <Metadata>
      <AMS Asset_Class="title" Asset_Name="xxxxxxxxxxxxxxxxxxxxxxxx_ti
+tle" />
      <App_Data App="MOD" Name="Type" Value="title" />
      <App_Data App="MOD" Name="Year" Value="1999" />
    </Metadata>
    <Asset>
      <Metadata>
        <AMS Asset_Class="poster" Asset_Name="xxxxxxxxxxxxxxxxxxxxxxxx
+_poster" />
        <App_Data App="MOD" Name="Content_CheckSum" Value="3167abc898d
+ba878794754dc4afca0fd"/>
        <App_Data App="MOD" Name="Content_FileSize" Value="230454"/>
        <App_Data App="MOD" Name="Image_Aspect_Ratio" Value="320x240"/
+>
        <App_Data App="MOD" Name="Type" Value="poster"/>
      </Metadata>
      <Content Value="xxxxxxxxxxxxxxxx.bmp" />
    </Asset>
    <Asset>
      <Metadata>
        <AMS Asset_Class="movie" Asset_Name="xxxxxxxxxxxxxxxxxxxxxxxx_
+movie" />
        <App_Data App="MOD" Name="Setting1" Value="Y"/>
        <App_Data App="MOD" Name="Languages" Value="en"/>
      </Metadata>
      <Content Value="xxxxxxxxxxxxxxxxxxxxxxxxxx.mpg"/>
    </Asset>
  </Asset>
</ADI>
[download]

So, for some example code:

#!/usr/bin/perl

use strict;
use warnings;

use XML::XPath;
use Data::Dumper;

my $xpath = XML::XPath->new(filename => 'TEST.XML');
my $nodeset = $xpath->find('/ADI/Asset/Asset/Metadata[AMS[@Asset_Class
+="poster"]]');
ddump('ref nodeset', __LINE__, ref $nodeset);  # this shows 'XML::XPat
+h::NodeSet'
[download]

After this, I've tried iterating with $nodeset->get_nodelist and then iterating those nodes with the ->getChildNodes and using ->string_value and ->getName methods, but I can never seem to get at what I want. I've tried a variety of Xpath strings, notably .//AMS[@Asset_Class="poster" which with an xpath tester does get me to where I expect, I just can't seem to turn that into something I can code against.

Any help appreciated. Tearing my hair out for way too long on this. Even tried AI to at least look for pointers. No, it didn't work at all. #facepalm

(ddump() is just my little wrapper around Data::Dumper, it's only slightly more advanced than print Dumper().)

--
Andy

Comment on XML Parsing and Xpath confusion Select or Download Code

Replies are listed 'Best First'.
Re: XML Parsing and Xpath confusion by ikegami (Patriarch) on Jul 03, 2024 at 14:09 UTC
Think of a `[]` in an XPath as similar to a `WHERE` clause in an SQL statement, which is to say a filter. `/ADI/Asset/Asset/Metadata` suggests that you want some asset metadata nodes. I'm going to assume that's correct for now. But we only want metadata nodes where `AMS/@Asset_Class` equals `poster`. This is a filter. `/ADI/Asset/Asset/Metadata[ AMS/@Asset_Class = "poster" ]` [download] Since this is a nested collection of assets, maybe you want to look any depth. `//Asset/Metadata[ AMS/@Asset_Class = "poster" ]` [download] With the above, the asset name would be found at relative path `AMS/@Asset_name`. `App_Data` would return rows of metadata. Is the asset metadata really what you want, though? If you wanted the whole asset instead of just its metadata (e.g. if you wanted to access the asset's content), you should use one of the following instead: `/ADI/Asset/Asset[ Metadata/AMS/@Asset_Class = "poster" ]` [download] `//Asset[ Metadata/AMS/@Asset_Class = "poster" ]` [download] With these, the asset name would be found at relative path `Metadata/AMS/@Asset_name`. `Metadata/App_Data` would return rows of metadata. `Content/@Value` would return the poster image. As a side note, an example of multiple `[]` in an XPath, `//Asset[ AMS/@Asset_Class = "package" and AMS/@Asset_Name = "some_pack +age" ]/Asset[ Metadata/AMS/@Asset_Class = "poster" ]` [download] This would return the poster asset for a specific package.	[reply] [d/l] [select]
Re^2: XML Parsing and Xpath confusion by naChoZ (Curate) on Jul 08, 2024 at 14:16 UTC
Thanks ikegami. I was able to use those tips with XML::LibXML to get what I wanted. Now I need to figure out why XML::LibXML changed a completely different part of the xml document. There's a description string elsewhere in the document and it converted an html entity (`'`) to a single quote. I already set `expand_entities` to 0 so I'm not wtf is up with that... ugh... I mean it's technically correct and probably won't break anything, but I'd rather it not touch anything I didn't ask it to touch. -- Andy	[reply] [d/l] [select]
Re: XML Parsing and Xpath confusion by sectokia (Friar) on Jul 04, 2024 at 02:00 UTC
I really like `XML::Twig`. Not sure what you are trying to do exactly but a basic example would be: `use strict; use warnings; use XML::Twig; my $t= XML::Twig->new( twig_handlers => { AMS => sub { print $_->{att}{Asset_Class}." has name ".$_->{att}{Asset_ +Name}."\n"; } })->parsefile('test.xml');` [download] Output: `package has name xxxxxxxxxxxxxxxxxxxxxxxx_package title has name xxxxxxxxxxxxxxxxxxxxxxxx_title poster has name xxxxxxxxxxxxxxxxxxxxxxxx_poster movie has name xxxxxxxxxxxxxxxxxxxxxxxx_movie` [download]	[reply] [d/l] [select]
Re^2: XML Parsing and Xpath confusion by ikegami (Patriarch) on Jul 04, 2024 at 15:18 UTC
The original tried to filter for just the posters, and it attempted to get the Metadata element. You get a different element, and you don't limit to just getting the element relating to the poster. (XML::Twig is also insanely slower.)	[reply]