3 ###############################################################################
6 =item B<loadbibs> --stage_file foo.mrc
8 Takes a load of bibs from a binary marc file and loads them into mig staging table
11 Takes these optional arguments:
15 When used it does not drop the staging table and instead adds onto it.
19 Sets an x_source value on the staging table to the one supplied instead of the
24 By default the program assumes a USMARC file. This flag will identify it as
25 a MARCXML file instead.
31 ###############################################################################
43 HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA
44 MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
51 use Unicode::Normalize;
52 my $mig_bin = "$FindBin::Bin/";
53 use lib "$FindBin::Bin/";
56 pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help';
57 pod2usage(-verbose => 1) if ! $ARGV[1];
59 my $next_arg_is_file = 0;
60 my $append_is_false = 1;
61 my $next_arg_is_source = 0;
62 my $source = 'default';
64 my $dbh = Mig::db_connect();
68 binmode STDIN, ':utf8';
70 foreach my $arg (@ARGV) {
71 if ($arg eq '--stage_file') {
72 $next_arg_is_file = 1;
75 if ($next_arg_is_file) {
77 $next_arg_is_file = 0;
80 if ($arg eq '--source') {
81 $next_arg_is_source = 1;
84 if ($next_arg_is_file) {
86 $next_arg_is_source = 0;
89 if ($arg eq '--append') {
93 if ($arg eq '--xml') {
104 SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry') INTO t;
106 PERFORM migration_tools.build_specific_base_staging_table ('$MIGSCHEMA','biblio.record_entry');
111 # normal stage table creation
112 if ($append_is_false) { create_stage_table($dbh); }
114 #sanity check and create stage table if --append is set but there isn't a stage table
115 my $query = "SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = 'biblio_record_entry_stage')";
116 my $qsth = $dbh->prepare($query);
119 while (my @row = $sth->fetchrow_array) { $f = ($row[0] }
120 if ($f eq 'f') { create_stage_table($dbh); }
122 if ($append_is_false == 0) { create_stage_table($dbh); }
125 $batch = MARC::Batch->new('XML',$infile);
127 $batch = MARC::Batch->new('USMARC',$infile);
129 $batch->strict_off();
131 while ( my $record = $batch->next() ) {
133 if ($file_is_xml) { $xml = $record; }
134 else { $xml = $record->as_xml_record(); }
136 $xml = clean_marc($xml);
137 $xml = '$_$' . $xml . '$_$';
138 my @warnings = $batch->warnings();
140 if (@warnings) { $warning_string = "'" . join(':',@warnings) . "'"; } else { $warning_string = "'none'"; }
141 my $sql = "INSERT INTO $MIGSCHEMA.biblio_record_entry_stage (marc,x_source,x_warnings) VALUES ($xml,$source,$warning_string);";
142 my $sth = $dbh->prepare($sql);
144 report_progress("Records staged", $i) if 0 != $i % 100;
148 CREATE INDEX ${MIGSCHEMA}_biblio_record_entry_stage_idx ON
149 $MIGSCHEMA.biblio_record_entry_stage (id);
154 sub create_stage_table {
157 DROP TABLE IF EXISTS $MIGSCHEMA.biblio_record_entry_stage;
158 CREATE UNLOGGED TABLE $MIGSCHEMA.biblio_record_entry_stage (
162 x_migrate BOOLEAN DEFAULT TRUE
163 ) INHERITS ($MIGSCHEMA.biblio_record_entry);
170 $xml =~ s/^<\?xml.+\?\s*>//go;
171 $xml =~ s/>\s+</></go;
172 $xml =~ s/\p{Cc}//go;
174 $xml =~ s/&(?!\S+;)/&/gso;
175 $xml =~ s/([\x{0080}-\x{fffd}])/sprintf('&#x%X;',ord($1))/sgoe;
176 $xml =~ s/[\x00-\x1f]//go;
183 print STDERR "$0: $msg", "\n";
187 sub report_progress {
188 my ($msg, $counter) = @_;
189 if (defined $counter) {
190 print STDERR "$msg: $counter\n";
192 print STDERR "$msg\n";