X-getdata: fix regex matching (for real)

fmarotta · Mar 25, 2019 · e0f461f · e0f461f
1 parent ba1f0e9
commit e0f461f
Showing 1 changed file with 16 additions and 7 deletions.
diff --git a/bioinfoconda/local/src/X-getdata.pl b/bioinfoconda/local/src/X-getdata.pl
@@ -19,6 +19,8 @@
 
 # TODO: check if new dest is overriding something.
 
+# TODO: add --info=progress2 to rsync
+
 # TODO: pass @ARGV as @params for rsync, e.g. to specify port or
 # something else in the command line.
 
@@ -183,6 +185,10 @@
 		@output = rsync_simple($v_source, $data_root . $dest);
 	}
 }
+else
+{
+	die("ERROR: unrecognised source. Please specify the protocol or the username\@host.\n");
+}
 
 # Add an entry to the index file
 if (@output)
@@ -191,6 +197,7 @@
 	update_index(\@output);
 }
 
+print "Done.";
 exit 0;
 
 # Subroutines
@@ -410,11 +417,12 @@ sub find_dest_host
 	{
 		if ($dirs[$i] =~ m/[\*\?\[\]\{\}]/)
 		{
-			$localdirs .= '';
+			$localdirs .= '/';
+			last;
 		}
 		else
 		{
-			$localdirs .= $dirs[$i] . '/';
+			$localdirs .= '/' . $dirs[$i];
 		}
 	}
 
@@ -497,13 +505,13 @@ sub wget_simple
                 {
                         print $line;
                         # extract the information to be logged
-                        if ($line =~ m/ saved \[\d+\/\d+\]/)
+                        if ($line =~ m/ saved \[.*\]/)
                         {
                                 # We want @output to contain: date, 
                                 # hour, url, path. But wget simple 
                                 # returns only date, hour, path, so we 
                                 # add the url.
-                                @output = ($line =~ m/(.*) (.*) \(.*\) - '(.*)' saved \[\d+\/\d+\]/);
+                                @output = ($line =~ m/(.*) (.*) \(.*\) - '(.*)' saved \[.*\]/);
                                 splice(@output, 2, 0, $url);
                         }
                 }
@@ -516,7 +524,8 @@ sub wget_recursive
 {
         my $url = $_[0];
         my $dest = $_[1];
-        my $dirname = dirname($dest);
+		# my $dirname = dirname($dest);
+		my $dirname = $dest;
         my @output = ();
         my @urldirs = split("/", $url);
         my $cutdirs;
@@ -526,7 +535,7 @@ sub wget_recursive
                 die("ERROR: cannot make path $dirname.\n");
         }
 
-        $cutdirs = $#urldirs-2;
+		$cutdirs = $#urldirs-2;
         # we subtract the false dirs generated by ://
         open(WGETOUTPUT, "wget -P $dirname \\
                 -N --no-verbose --show-progress \\
@@ -536,7 +545,7 @@ sub wget_recursive
         while (my $line = <WGETOUTPUT>)
         {
                 # update the progress line...
-                if ($line =~ m/^\s+\d/ and $line =~ m/%/)
+                if ($line =~ m/^\s*\d/ and $line =~ m/%/)
                 {
                         chomp $line;
                         print "\r$line";