Fixed regex to handle lines that begin with a dot and something other than a capital...

[migration-tools.git] / unicorn / unicorn_to_tsv.pl

diff --git a/unicorn/unicorn_to_tsv.pl b/unicorn/unicorn_to_tsv.pl
index e864a31..3af493f 100755 (executable)
--- a/unicorn/unicorn_to_tsv.pl
+++ b/unicorn/unicorn_to_tsv.pl
@@ -25,8 +25,9 @@ while (<>) {
                next;
        }
 
-       # Is this a FORM= line (which can be ignored)?
-       if ( /^FORM=/ ) {
+       # Is this a FORM= line?
+       if ( /^FORM=(.*)/ ) {
+               $records[$serial]{'l_form'} = $1;
                next;
        }
 
@@ -52,7 +53,7 @@ while (<>) {
 
        # Looks like we've got some actual data!  Let's store it.
        # FIXME: For large batches of data, we may run out of memory and should store this on disk.
-       if ( /^\.(.*?).\s+(\|a)?(.*)$/ ) {
+       if ( /^\.([A-Z]*?).\s+(\|a)?(.*)$/ ) {
 
                # Build the name of this field (taking note of whether we're in a named section of data)
                $field = '';