X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=mig-bin%2Fmig-loadbibs;h=a90ab648845c0d6a0335306833da4d001eca33f4;hp=1d2e4fb5dc2ae3c00cc1bec94897be5675390eb8;hb=d3bf56fcd9351c4ef45b5527d929228b6663ecaf;hpb=1a510ad110ab208ec615a0bf6d8931ad20137159 diff --git a/mig-bin/mig-loadbibs b/mig-bin/mig-loadbibs index 1d2e4fb..a90ab64 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 @@ -58,7 +59,7 @@ pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help'; pod2usage(-verbose => 1) if ! $ARGV[1]; my $next_arg_is_file = 0; -my $append_is_false = 1; +my $append = 0; my $next_arg_is_source = 0; my $source = 'default'; my $file_is_xml = 0; @@ -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; @@ -82,13 +89,13 @@ foreach my $arg (@ARGV) { $next_arg_is_source = 1; next; } - if ($next_arg_is_file) { + if ($next_arg_is_source) { $source = $arg; $next_arg_is_source = 0; next; } if ($arg eq '--append') { - $append_is_false = 0; + $append = 1; next; } if ($arg eq '--xml') { @@ -97,20 +104,13 @@ foreach my $arg (@ARGV) { } } -create_child_table($dbh); +create_child_table($dbh); #and test to see if it exists # normal stage table creation -if ($append_is_false) { create_stage_table($dbh); } - -#sanity check and create stage table if it doesn't exist -my $query = "SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry_stage')"; -my $qsth = $dbh->prepare($query); -$qsth->execute(); -my $f; -while (my @row = $qsth->fetchrow_array) { $f = $row[0]; } -if ($f eq 'f') { create_stage_table($dbh); } - -if ($append_is_false == 0) { create_stage_table($dbh); } +if ($append == 0) { + drop_stage_table($dbh); + create_stage_table($dbh); + } if ($file_is_xml) { $batch = MARC::Batch->new('XML',$infile); @@ -129,7 +129,7 @@ while ( my $record = $batch->next() ) { my @warnings = $batch->warnings(); my $warning_string; if (@warnings) { $warning_string = "'" . join(':',@warnings) . "'"; } else { $warning_string = "'none'"; } - my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry_stage (marc,x_source,x_warnings) VALUES ($xml,$source,$warning_string);"; + my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry_stage (marc,x_source,x_warnings) VALUES ($xml,'$source',$warning_string);"; my $sth = $dbh->prepare($sql); $sth->execute(); report_progress("Records staged", $i) if 0 != $i % 100; @@ -142,10 +142,16 @@ $dbh->do(qq/ print "Finis.\n"; + +sub drop_stage_table { + my $dbh = shift; + $dbh->do("DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;"); + return(); +} + sub create_stage_table { my $dbh = shift; - $dbh->do("DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;"); $dbh->do("CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_stage ( l_bib_id TEXT, x_source TEXT, @@ -174,6 +180,7 @@ 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+