Fix PDF metadata extraction when no header block is present

diff --git a/input/regression/no-header.ly b/input/regression/no-header.ly
new file mode 100644 (file)
index 0000000..faf2275
--- /dev/null
+++ b/input/regression/no-header.ly
@@ -0,0 +1,6 @@
+\version "2.12.0"
+
+\markuplines \wordwrap-lines {
+  This regtest does not contain any header and paper blocks. Its purpose is to
+  test whether anything breaks if these blocks are absent.
+}