* add fb photo upload to svn fix

[bin.git] / get_one_manga

diff --git a/get_one_manga b/get_one_manga
index 3fd8118d5bbf84772ce88a9ba7fabcc0033b9ced..6336726bff3ef6c493068899993be5325d991267 100755 (executable)
--- a/get_one_manga
+++ b/get_one_manga
@@ -49,6 +49,7 @@ Display this manual.
 =cut
 
 
+use URI::Escape;
 use WWW::Mechanize;
 use IO::Dir;
 use IO::File;
@@ -92,7 +93,7 @@ my $failure = 0;
 my $m = WWW::Mechanize->new();
 for my $manga (@manga_to_get) {
     # see if the manga exists
-    $m->get($options{onemanga}.'/'.$manga);
+    mm_get($m,$options{onemanga}.'/'.$manga);
     if ($m->status() != 200) {
        print STDERR "Manga $manga doesn't exist\n";
        $failure ||= 1;
@@ -102,24 +103,27 @@ for my $manga (@manga_to_get) {
        mkdir($manga);
     }
     # figure out where to start getting stuff
+    # we need to escape ! apparently; there are probably other characters as well
+    my $manga_escaped = uri_escape($manga,'!');
     my @chapter_links = $m->find_all_links(url_abs_regex => qr{\Q$manga\E\/\d+});
     for my $chapter_link (reverse @chapter_links) {
-       print $chapter_link->url(),qq(\n);
-       my ($chapter) = $chapter_link->url() =~ m/([\d-]+)\/?$/;
-       my $chapter_long = sprintf('%04d',$chapter);
+       my ($chapter) = $chapter_link->url() =~ m/([\d.-]+)\/?$/;
+       my $chapter_long = $chapter =~ /\./ ? join('.',map {sprintf'%04d',$_} split /\./,$chapter) : sprintf('%04d',$chapter);
        if (! -d "$manga/$chapter_long") {
+           print $chapter_link->url(),qq(\n);
            mkdir("$manga/$chapter_long");
-           $m->get($chapter_link->url_abs());
-           $m->follow_link(text_regex => qr{Begin reading});
+           mm_get($m,$chapter_link->url_abs());
+           my $link = $m->find_link(text_regex => qr{Begin reading});
+           mm_get($m,$link->url_abs());
            while ($m->uri() =~ m{\Q$chapter\E/(\d\d[^\/]*)/?$}) {
-               my $image = $m->find_image(alt_regex => qr{Loading\.+\s+media});
-               my $next_link = $m->find_link(url_regex => qr{\Q$manga\E/\Q$chapter\E/(\d\d[^\/]*)});
-               $m->get($image->url_abs());
+               my $image = $m->find_image(alt_regex => qr{Loading\.+\s+(media|img)});
+               my $next_link = $m->find_link(url_regex => qr{\Q$manga_escaped\E/\Q$chapter\E/(\d\d[^\/]*)});
+               mm_get($m,$image->url_abs());
                print "getting ".$image->url_abs()."\n";
                my ($page) = $image->url_abs =~ m/([^\/]+)$/;
                $m->save_content("$manga/$chapter_long/$page");
                last if not defined $next_link;
-               $m->get($next_link->url_abs());
+               mm_get($m,$next_link->url_abs());
                print $m->uri()."\n";
                sleep 3;
            }
@@ -127,5 +131,18 @@ for my $manga (@manga_to_get) {
     }
 }
 
+sub mm_get{
+    my ($m,$url) = @_;
+    my $rerun = 8;
+    my $return;
+    do {
+       eval {
+           $return = $m->get($url);
+       };
+    } while ($@ and
+            ($rerun-- > 0) and sleep 5);
+    return $return;
+}
+
 
 __END__