get script now supports urls

1 files changed, 38 insertions, 5 deletions

diff --git a/get.pl b/get.pl
index 4447ab3..8a0d8f9 100755
--- a/get.pl
+++ b/get.pl
@@ -4,11 +4,44 @@ $SIG{TERM} = $SIG{INT} = sub { exit 1 };
 
 chdir('datasets');
 
-foreach (@ARGV){
-  my $s = $_;
-  if ($s !~/\..*$/) { $s .= ".wav"; }
-  system('/usr/bin/wget', 'https://neural:spawn5@asdf.us/neural/' . $s);
-  system('/usr/bin/perl', 'dataset.pl', $s);
+my $fn, $new_fn;
+foreach my $s (@ARGV){
+  if ($s =~ /^http/) {
+    if ($s =~ /(wav|aiff?|flac|mp3|opus)$/i) {
+      my $fn = `basename $s`;
+      system('/usr/bin/wget', $s);
+      system('/usr/bin/perl', 'dataset.pl', $fn);
+    } else {
+      my $yt = `youtube-dl --extract-audio --audio-format flac -o "%(title)s.%(ext)s" $s`;
+      my @partz = split("\n", $yt);
+      foreach $part (@partz) {
+        if ($part =~ /\[ffmpeg\] Destination\: (.*\.flac)$/) {
+          $fn = $1;
+        }
+      }
+      if ($fn) {
+        $new_fn = lc $fn;
+        $new_fn =~ s/\.flac$//g;
+        $new_fn =~ s/\s+/_/g;
+        $new_fn =~ s/\W//g;
+        if (length($new_fn) == 0) {
+          $new_fn = lc $s;
+          $new_fn =~ s/\s+/_/g;
+          $new_fn =~ s/\W//g;
+        }
+	$new_fn =~ s/_+/_/g;
+        $new_fn .= '.flac';
+        system('mv', $fn, $new_fn);
+        print"got fn, $fn => $new_fn\n";
+        system('/usr/bin/perl', 'dataset.pl', $new_fn);
+      }
+    }
+  } else {
+    if ($s !~/\..*$/) { $s .= ".wav"; }
+    system('/usr/bin/wget', 'https://neural:spawn5@asdf.us/neural/' . $s);
+    system('/usr/bin/perl', 'dataset.pl', $s);
+  }
+  $fn = undef;
 }
 
 END {