Because we now properly decode UTF8, in order to strip the BOM, we

need to strip \N{U+FEFF}, not 0xEFBBBF.

diff --git a/scripts/process b/scripts/process
index 17aa15be7b297939a66e8ea97e1eff656fbf4645..d42a1f92b995c8a8c98a77969e62a436a595a348 100755 (executable)
--- a/scripts/process
+++ b/scripts/process
@@ -224,6 +224,7 @@ for my $phline (@bodylines)
     # Remove BOM markers from UTF-8 strings
     # Fixes #488554
     $phline =~ s/\xef\xbb\xbf//g;
+    $phline =~ s/\N{U+FEFF}//g;
     last if $phline !~ m/^([\w-]+):\s*(\S.*)/;
     my ($fn, $fv) = ($1, $2);
     $fv =~ s/\s*$//;