From: Rogan Hamby Date: Thu, 19 Apr 2018 20:02:06 +0000 (-0400) Subject: making mig-loadbibs a bit more defensive with marc-8 data X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=commitdiff_plain;h=010f8b06928e9543ec79266d2b9f34e96e6e47cc making mig-loadbibs a bit more defensive with marc-8 data --- diff --git a/mig-bin/mig-loadbibs b/mig-bin/mig-loadbibs index 1d2e4fb..d206a3e 100755 --- a/mig-bin/mig-loadbibs +++ b/mig-bin/mig-loadbibs @@ -39,6 +39,7 @@ use MARC::Record; use MARC::Batch; use MARC::File; use MARC::File::XML; +use MARC::Charset 'marc8_to_utf8'; binmode STDIN, ':bytes'; use Env qw( HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA @@ -68,6 +69,12 @@ my $i = 0; my $batch; binmode STDIN, ':utf8'; +my $ignore = MARC::Charset->ignore_errors(); +MARC::Charset->ignore_errors(1); +my $setting = MARC::Charset->assume_unicode(); +MARC::Charset->assume_unicode(1); +MARC::Charset->ignore_errors(1); + foreach my $arg (@ARGV) { if ($arg eq '--stage_file') { $next_arg_is_file = 1; @@ -174,6 +181,7 @@ sub create_child_table { sub clean_marc { my $xml = shift; + $xml = marc8_to_utf8($xml); $xml =~ s/\n//sog; $xml =~ s/^<\?xml.+\?\s*>//go; $xml =~ s/>\s+