The Code

So, after trying the orthodox approach, I started again. I broke the rule about parsing HTML with regular expressions and wrote the following code:

#!usr/bin/perl -w
use strict; 

my (@s) = m{
      >         # close of previous tag
      ([^<]+)   # text (name of part, e.g., q/BLACK CARTRIDGE/)
      <br>
      ([^<]+)   # part number (e.g., q/HP Part Number: HP C9724A/+)
      (?:<[^>]+>\s*){4} # separated by four tags
      (\d+)       # percent remaining
      |           # --or--
      (?:
            # different text values
            (?:
                Pages\sRemaining
                | Low\sReached
                | Serial\sNumber
                | Pages\sprinted\swith\sthis\ssupply
            ) : (?:\s*<[^>]+>){6}\s* # colon, separated by six tags
        # or just this, within the current element
        | Based\son\shistorical\s\S+\spage\scoverage\sof\s
        )
        (\w+) # and the value we want
    }gx;

A single regular expression (albeit with a /g modifier for global matching) pulls out all I want. Actually, it's not quite perfect, since the resulting array also fills up with a pile of undefs, the unfilled parenthesis on the opposite side of the | alternation to the match. This is easily handled with a simple next unless$index addition to any foreach loop on @s.

Is the code fragile? Not really. The HTML has errors in it, such as <td valign= op">, which can trip up some modules that expect perfectly formed HTML, but HTML::TreeBuilder coped just fine with this too.