Querying a website with Perl LWP::Simple to Process Online Prices

Question

In my free time, I've been trying to improve my perl abilities by working on a script that uses LWP::Simple to poll one specific website's product pages to check the prices of products (I'm somewhat of a perl noob). This script also keeps a very simple backlog of the last price seen for that item (since the prices change frequently).

I was wondering if there was any way I could further automate the script so that I don't have to explicitly add the page's URL to the initial hash (i.e. keep an array of key terms and do a search query amazon to find the page or price?). Is there anyway way I could do this that doesn't involve me just copying Amazon's search URL and parsing in my keywords? (I'm aware that processing HTML with regex is generally bad form, I just used it since I only need one small piece of data).

#!usr/bin/perl
use strict;
use warnings;
use LWP::Simple;

my %oldPrice;
my %nameURL = (
    "Archer Season 1" => "http://www.amazon.com/Archer-Season-H-Jon-Benjamin/dp/B00475B0G2/ref=sr_1_1?ie=UTF8&qid=1297282236&sr=8-1",
    "Code Complete" => "http://www.amazon.com/Code-Complete-Practical-Handbook-Construction/dp/0735619670/ref=sr_1_1?ie=UTF8&qid=1296841986&sr=8-1",
    "Intermediate Perl" => "http://www.amazon.com/Intermediate-Perl-Randal-L-Schwartz/dp/0596102062/ref=sr_1_1?s=books&ie=UTF8&qid=1297283720&sr=1-1",
    "Inglorious Basterds (2-Disc)" => "http://www.amazon.com/Inglourious-Basterds-Two-Disc-Special-Brad/dp/B002T9H2LK/ref=sr_1_3?ie=UTF8&qid=1297283816&sr=8-3"
);

if (-e "backlog.txt"){
    open (LOG, "backlog.txt");
    while(){
        chomp;
        my @temp = split(/:\s/);
        $oldPrice{$temp[0]} = $temp[1];
    }
close(LOG);
}

print "\nChecking Daily Amazon Prices:\n";
open(LOG, ">backlog.txt");
foreach my $key (sort keys %nameURL){
    my $content = get $nameURL{$key} or die;
    $content =~  m{\s*\$(\d+.\d+)} || die;
    if (exists $oldPrice{$key} && $oldPrice{$key} != $1){
        print "$key: \$$1 (Was $oldPrice{$key})\n";
    }
    else{
    print "\n$key: $1\n";
    }
    print LOG "$key: $1\n";
}
close(LOG);

Answer 1

I made simple script to demonstate Amazon search automation. Search url for all departments was changed with escaped search term. The rest of code is simple parsing with HTML::TreeBuilder. Structure of HTML in question can be easily examined with dump method (see commented-out line).

use strict; use warnings;

use LWP::Simple;
use URI::Escape;
use HTML::TreeBuilder;
use Try::Tiny;

my $look_for = "Archer Season 1";

my $contents
  = get "http://www.amazon.com/s/ref=nb_sb_noss?url=search-alias%3Daps&field-keywords="
        . uri_escape($look_for);

my $html = HTML::TreeBuilder->new_from_content($contents);
for my $item ($html->look_down(id => qr/result_\d+/)) {
    # $item->dump;      # find out structure of HTML
    my $title = try { $item->look_down(class => 'productTitle')->as_trimmed_text };
    my $price = try { $item->look_down(class => 'newPrice')->find('span')->as_text };

    print "$title\n$price\n\n";
}
$html->delete;

Answer 2

Yes, the design can be improved. It's probably best to delete everything and start over with an existing full-featured web scraping application or framework, but since you want to learn:

1. The name-to-URL map is configuration data. Retrieve it from outside of the program.
2. Store the historic data in a database.
3. Learn XPath and use it to extract data from HTML, it's easy if you already grok CSS selectors.

---

Other stackers, if you want to amend my post with the rationale for each piece of advice, go ahead and edit it.