X-Git-Url: http://git.equinoxoli.org/?p=migration-tools.git;a=blobdiff_plain;f=mig-bin%2Fmig-loadbibs;h=96b12e89da3c084b06ca09a349e33589ea56e64b;hp=d89dfb6aab5bd0d1eedeb476d3dba19f1148f10c;hb=e3f85c63930a230a6e7a5c64dbfc003e76db363c;hpb=d5f19c6b68f57060ece9c04a0f08e685b062dab6

diff --git a/mig-bin/mig-loadbibs b/mig-bin/mig-loadbibs
index d89dfb6..96b12e8 100755
--- a/mig-bin/mig-loadbibs
+++ b/mig-bin/mig-loadbibs
@@ -39,6 +39,8 @@ 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
     MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
@@ -57,7 +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;
@@ -67,6 +69,9 @@ my $i = 0;
 my $batch;
 binmode STDIN, ':utf8';
 
+#MARC::Charset->assume_unicode(1); 
+MARC::Charset->ignore_errors(1);
+
 foreach my $arg (@ARGV) {
     if ($arg eq '--stage_file') {
         $next_arg_is_file = 1;
@@ -81,13 +86,13 @@ 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') {
@@ -96,31 +101,13 @@ foreach my $arg (@ARGV) {
     }
 }
 
-$dbh->do(qq{
-	DO $$
-	DECLARE
-    	t   BOOLEAN;
-	BEGIN
-	SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry') INTO t;
-	IF t = FALSE THEN
-    	PERFORM migration_tools.build_specific_base_staging_table ('$MIGSCHEMA','biblio.record_entry');
-	END IF;
-	END $$;
-});
+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 --append is set but there isn't a stage table
-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 = $sth->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);
 } else {
@@ -129,43 +116,66 @@ if ($file_is_xml) {
 $batch->strict_off();
 
 while ( my $record = $batch->next() ) {
-	my $xml;
-	if ($file_is_xml) { $xml = $record; } 
-		else { $xml = $record->as_xml_record(); } 
+	my $xml = $record->as_xml_record();
+    $xml = marc8_to_utf8($xml);
 	$i++;
 	$xml = clean_marc($xml);
 	$xml = '$_$' . $xml . '$_$';
 	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 $MIGSCHEMA.biblio_record_entry_legacy (marc,x_source,x_warnings) VALUES ($xml,'$source',$warning_string);";
     my $sth = $dbh->prepare($sql);
     $sth->execute();
 	report_progress("Records staged", $i) if 0 != $i % 100;
 }
 
 $dbh->do(qq/
-    CREATE INDEX ${MIGSCHEMA}_biblio_record_entry_stage_idx ON
-        $MIGSCHEMA.biblio_record_entry_stage (id);
+    CREATE INDEX ${MIGSCHEMA}_biblio_record_entry_legacy_idx ON
+        $MIGSCHEMA.biblio_record_entry_legacy (id);
 /);
 
 print "Finis.\n";
 
+
+sub drop_stage_table {
+    my $dbh = shift;
+    $dbh->do("DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_legacy;");
+    return();
+}
+
 sub create_stage_table {
 	my $dbh = shift;
-    $dbh->do(qq{
-        DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;
-        CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_stage (
+
+    $dbh->do("CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_legacy (
             l_bib_id    TEXT,
             x_source    TEXT,
             x_warnings  TEXT,
             x_migrate   BOOLEAN DEFAULT TRUE
-        ) INHERITS ($MIGSCHEMA.biblio_record_entry);
-    });
+        ) INHERITS ($MIGSCHEMA.biblio_record_entry);");
+
+    return();
+}
+
+sub create_child_table {
+    my $dbh = shift;
+
+    $dbh->do("DO \$\$ 
+        DECLARE
+            t   BOOLEAN;
+        BEGIN
+        SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry') INTO t;
+        IF t = FALSE THEN
+            PERFORM migration_tools.build_specific_base_staging_table ('$MIGSCHEMA','biblio.record_entry');
+        END IF;
+        END \$\$;");
+
+    return ();
 }
 
 sub clean_marc {
     my $xml = shift;
+    $xml = marc8_to_utf8($xml);
     $xml =~ s/\n//sog;
     $xml =~ s/^<\?xml.+\?\s*>//go;
     $xml =~ s/>\s+</></go;