DS_HTML_STEPS_PARSER.pl

#!/usr/bin/perl

# Author: Dawid Mocek
# PP Projekt
# For educational purpose only
# All right reserved

# DS html parser - steps finder

use strict;
use warnings;

use Data::Dumper;
use HTML::TreeBuilder::XPath;
use XML::Writer;
use DBI;
use Config::IniFiles;

sub trim {
 $_[0] =~ s/^\s+|\s+$/g;
}

sub load_ini {

    my $inifile = $_[0];

    unless(-e $inifile) {
         print("File: $inifile does not exists\n");
         exit;
         }

    my $cfg =  Config::IniFiles->new(-file => $inifile, -fallback => "General");
    return $cfg;
}


my $ini_file = './config.ini';
my $cfg = load_ini($ini_file);



my $steps_xpath = '/html/body//div[@class="step-description"]/h4[@class="step-title"]';

### Database ###
my $dbh;
my $sth;
$dbh = DBI->connect('DBI:mysql:database=' . $cfg->val('db', 'name') . ';host=' . $cfg->val('db', 'host'), $cfg->val('db', 'user'), $cfg->val('db', 'pass'), {mysql_auto_reconnect => 1, mysql_enable_utf8 => 1});
$sth = $dbh->prepare('UPDATE `' . $cfg->val('db', 'tb_href') . '` SET steps = ? WHERE ds_id = ?');

$dbh->begin_work();

my $html_file = $cfg->val('prop', 'html_files');

open my $html_fh, $html_file or die "Coulnd not open $html_file: $!";

while(my $line = <$html_fh>) {
    $line =~ s/^\s+|\s+$//g;
    $line =~ /\/share\/przepisy\/(?<ds_id>\d+)\//m;
    my $ds_id = $+{ds_id};

    ### HTML Parser ###
    my $tree = HTML::TreeBuilder::XPath->new(ignore_unknown => 0);
    $tree->parse_file($line);

    my $step = $tree->findvalue($steps_xpath);

    if($step =~ /Krok\s1\s\/\s(?<cnt>\d+)/m ) {
    my $cnt = $+{cnt};
    $sth->execute($cnt, $ds_id);
    }
    $tree->delete;

}

close $html_fh;
$dbh->commit();
$sth->finish();
$dbh->disconnect();