lyadmin/perl-script/bookdl.pl

#!/usr/bin/perl

binmode STDOUT, ":utf8";

use warnings;
use strict;

use JSON;

my $THREAD_NO;
my $URL_PREFIX;

my $FN;
my $OUT_DIR;

my $USAGE;
my $ACMD;

my %jh;

my @a1;

$DOMAIN = "lainchan.org";
# $URL_PREFIX = "https://lainchan.org/lit/src/";
$URL_PREFIX = "https://$DOMAIN/lit/src/";
$THREAD_NO = 4619;
# $THREAD_NO = 6105;
# $THREAD_NO = 4345;
# $THREAD_NO = 4953;
$FN = "$THREAD_NO.json";
$OUT_DIR = "./dl/";


$USAGE = "Usage: bookdl.pl [http://someurl/]thread_id";

die "$USAGE" unless scalar @ARGV > 0;
$ACMD = $ARGV[0];

# Read JSON with list of files
open FILE, "<", $FN or die "could not open file";
do{
    my $json_str;

    local $/=undef;
    
    $json_str = <FILE>;
    chomp $json_str;
    
    %jh = %{JSON->new()->decode($json_str)};
};
close FILE;

# anonymous function that returns a list
# of tuples of the below form:
#     (file_name, file_url)
@a1 = sub{
    my @a0;
    my @a2;

    my $f1;

    # filters for file types we
    # dont want to downloads
    sub f1 {
	return $_[0]->{"ext"} && !($_[0]->{"ext"} =~ /jpe?g/);
    }
    sub f2 {
	return !($_[0]->{"ext"} =~ /png/);
    }
    sub f3 {
	return !($_[0]->{"ext"} =~ /gif/);
    }
    sub f4 {
	return !($_[0]->{"ext"} =~ /webm/);
    }
    sub f0 {
	return f1($_[0]) && f2($_[0]) && f3($_[0]) && f4($_[0])	
    }

    # create an array of files
    # that meet our file ext requirement
    @a0 = grep {f0($_)} @{$jh{"posts"}};

    # do the same filter on the
    # extra_files attribute
    @a2 = grep {
	f0($_)
    } map {
	@{$_->{"extra_files"}}
    } grep {
	$_->{"extra_files"}
    } @{$jh{"posts"}};

    # Return our tuple
    return map {
	[
	 sprintf("%s%s", $_->{"filename"}, $_->{"ext"}),          # file_name
	 sprintf("%s%s%s", $URL_PREFIX, $_->{"tim"}, $_->{"ext"}) # file_url
	]
    } (@a0, @a2);
}->();

# Print a list of wget commands from our tuples
for my $i1 (@a1){
    printf("wget -nc %s -O '%s%s'\n", scalar $i1->[1], $OUT_DIR, scalar $i1->[0]);
}