Bulk AMP URL validation

bulk-amp-validator.pl

#!/usr/bin/env perl
use strict;
use Modern::Perl;
use Mojo::UserAgent;
use Mojo::JSON qw(decode_json);
use Excel::Writer::XLSX;
use Data::Dumper;

my $in_file = 'bhm.txt';
my $output = 'bhm.xlsx';

open(my $fh, "<", $in_file)
    or die "Failed to open file: $!\n";

my @urls;
while(<$fh>) { 
    chomp; 
    push @urls, $_;
} 
close $fh;

my $workbook  = Excel::Writer::XLSX->new( $output );
my $worksheet = $workbook->add_worksheet();

$worksheet->write( 0, 0, [
	"URL",
    "AMP Status",
    "Canonical URL",
    "AMP URL",
    "AMP Required Markup - Status",
    "AMP Required Markup - Warning Count",
    "AMP Required Markup - Warning Status",
    "AMP Validation - Status",
    "Google AMP Cache - Status",
    "Google AMP Cache - Result",
    "Google AMP Cache - URL",
    "Google AMP Cache - Viewer URL",
    "Robots.txt - Status",
    "Robots.txt Googlebot - Status",
    "Robots.txt Googlebot-Smartphone - Status",
    "Robots Meta - Status",
    "X-Robots Tag Header - Status",
    "Structured Data - Status",
    "Structured Data - Result",
    "Structured Data - Kind",
    "Structured Data - Type",
    "Structured Data Is AMP?",
    "Structured Data Logo - Status",
    "Structured Data Logo - Result",
    "Structured Data Article - Status",
    "Structured Data Article - Result"
    ] 
);

my $ua = Mojo::UserAgent->new;
$ua->transactor->name('Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/48.0.2564.116 Safari/537.36');
my $ampbench_url = "https://ampbench.appspot.com"; # Replace URL if running locally

my $amp_api = $ampbench_url . "/api2"; # Replace with desired API version
my $i = 1;
for my $url (@urls) {
	say "fetching $url";
    my $tx = $ua->get($amp_api . "?url=$url");
    if (my $res = $tx->success) {
    	my $json = decode_json $res->body;
    	say "response: $json->{status}";
    	# say Dumper($json);
    	$worksheet->write( $i, 0, [
    		$url,
    		$json->{'status'},
            $json->{'amp_links'}->{'canonical_url'},
            $json->{'amp_links'}->{'amphtml_url'},
            '',#$json->{'amp_required_markup'}->{'status'},
            $json->{'amp_required_markup'}->{'warning_count'},
            $json->{'amp_required_markup'}->{'warning_status'},
            $json->{'amp_validation'}->{'status'},
            $json->{'google_amp_cache'}->{'status'},
            $json->{'google_amp_cache'}->{'result'},
            $json->{'google_amp_cache'}->{'google_amp_cache_url'},
            $json->{'google_amp_cache'}->{'google_amp_viewer_url'},
            $json->{'robots'}->{'robots_txt_status'},
            $json->{'robots'}->{'robots_txt_googlebot_status'},
            $json->{'robots'}->{'robots_txt_googlebot_smartphone_status'},
            $json->{'robots'}->{'robots_meta_status'},
            $json->{'robots'}->{'x_robots_tag_header_status'},
            $json->{'sd_validation'}->{'status'},
            $json->{'sd_validation'}->{'result'},
            $json->{'sd_validation'}->{'sd_kind'},
            $json->{'sd_validation'}->{'sd_type'},
            $json->{'sd_validation'}->{'sd_type_is_amp'},
            $json->{'sd_validation'}->{'sd_logo_image'}->{'status'},
            $json->{'sd_validation'}->{'sd_logo_image'}->{'result'},
            $json->{'sd_validation'}->{'sd_article'}->{'status'},
            $json->{'sd_validation'}->{'sd_article'}->{'result'},
		]);
		$i++;
    } else {
    	say "error";
    }
}