hitomi-get-new2.pl - handle new picture names Jan 2020

#!/usr/bin/perl

$url=shift;
unless ($url =~ m!https://hitomi.la/\w+/.*-(\d+)\.html$!) { die "ungültige URL"; }
$gallid=$1;
warn "trying to retrieve hitomi gallery $gallid\n";

$subdom = shift;

open(F,"wget -4nv -O - '$url'|");
while (<F>) {
  if (m!<h1><a[^>]+>(.*)<\/a>!) {
    $title = $1;
  }
  elsif (m!href="/artist/[^>]+>(.*)</a>!) {
    push(@artist,$1);
  }
  elsif (m!href="/group/[^>]+>(.*)</a>!) {
    push(@group,$1);
  }
  elsif (m!href="/series/[^>]+>(.*)</a>!) {
    push(@series,$1) unless $1 eq 'original';
  }
  elsif (m!<td>Language</td><td><a href="/index-(\w+)-1.html!) {
    $lang=$1;
  }
  elsif (m!href="/tag/[^>]+>(.*)</a>!) {
    $tag=$1; $tag =~ s/ ♂/_M/; $tag =~ s/ ♀/_F/; $tag =~ s/ /_/g;
    push(@tag,$tag);
  }
}
close(F);
open(F,"wget -4nv -O - https://ltn.hitomi.la/galleries/$gallid.js|");
while (<F>) {
  while (s!,"hash":"(\w{61}(\w\w)(\w))","haswebp":\d,"name":"([^"]+\.(\w+))"!!) {
   # push(@imgurl,"/galleries/$gallid/$1");
    push(@imgurl,"/images/$3/$2/$1.$5");
    $rename{"$1.$5"}=$4;
  }
}
close(F);
unless ($title) { die "no title extracted, wrong URL?"; }
$name=$title;
if (@artist) { $name .= '_'.join(",",@artist); }
if (@group) { $name .= '('.join(",",@group).')'; }
$name =~ s!&#39;!'!g;
$name =~ s!&quot;!"!g;
$name =~ s!&amp;!&!g;
$name =~ s!/!#!g;
if (length($name)>200) {
  warn "shortening overlong name $name\n";
  $name = substr($name,0,200);
}
$name .= '.'.$gallid;
if ( -e $name ) {
    die "$name exists - duplicate download request";
}
warn "mkdir ".$name."\n";
mkdir ($name) && chdir ($name) || die "couldn't mkdir,chdir $name";
warn "tags are ".join(",",@tag)."\n";
warn "will fetch ".scalar(@imgurl)." images.\n";
if ($lang) { warn "language is $lang\n"; }
unless ($subdom) {
  $first = shift @imgurl;
  $subdom = 'ca';
  warn "trying subdom $subdom\n";
  system "wget -4nv --referer=https://hitomi.la/reader/$gallid.html --no-use-server-timestamps https://$subdom.hitomi.la$first";
  if ($?) {
  $subdom = 'aa';
  warn "trying subdom $subdom\n";
  system "wget -4nv --referer=https://hitomi.la/reader/$gallid.html --no-use-server-timestamps https://$subdom.hitomi.la$first";
  if ($?) {
  $subdom = 'ba';
  warn "trying subdom $subdom\n";
  system "wget -4nv --referer=https://hitomi.la/reader/$gallid.html --no-use-server-timestamps https://$subdom.hitomi.la$first";
  if ($?) {
    chdir (".."); rmdir ($name);
    die "unknown subdom";
  }
  }
  }
}
warn "using subdom $subdom\n";
system "wget -4nv --referer=https://hitomi.la/reader/$gallid.html --no-use-server-timestamps ".join(" ",map { "https://$subdom.hitomi.la".$_ } @imgurl);
  foreach $f (keys(%rename)) {
    rename($f,$rename{$f});
  }
if ($lang) { open(F,">00LANG_$lang"); close(F); }
if (@series) { open(F,">00SERIES_".join(",",@series)); close(F); }
while (scalar(@tag)>10) {
  open(F,">00TAGS_".join(",",splice(@tag,0,10))); close(F);
}
open(F,">00TAGS_".join(",",@tag)); close(F);