3 ###############################################################################
6 =item B<stagebibs> --file foo.xml
8 Takes a load of bibs from a UTF-8 MARC XML file and loads them into mig staging
9 table of bibio_record_entry. This is done with no checking of file validity
10 so records should be checked before hand and cleaned.
12 Takes one optional arguments:
16 Sets an x_source value on the staging table to the one supplied instead of the
23 ###############################################################################
29 #binmode STDIN, ':bytes';
31 HOME PGHOST PGPORT PGUSER PGDATABASE MIGSCHEMA
32 MIGBASEWORKDIR MIGBASEGITDIR MIGGITDIR MIGWORKDIR
40 my $mig_bin = "$FindBin::Bin/";
41 use lib "$FindBin::Bin/";
44 pod2usage(-verbose => 2) if defined $ARGV[0] && $ARGV[0] eq '--help';
45 pod2usage(-verbose => 1) if ! $ARGV[1];
47 my $next_arg_is_file = 0;
49 my $next_arg_is_source = 0;
50 my $next_arg_is_stage = 0;
51 my $next_arg_is_base_table = 0;
52 my $next_arg_is_stage_table = 0;
53 my $base_table = 'biblio_record_entry';
54 my $stage_table = 'biblio_record_entry_legacy';
55 my $source = 'default';
57 my $dbh = Mig::db_connect();
61 binmode STDIN, ':utf8';
63 foreach my $arg (@ARGV) {
64 if ($arg eq '--auth') {
65 $base_table = 'authority_record_entry';
66 $stage_table = 'authority_record_entry_legacy';
68 if ($arg eq '--serial') {
69 $base_table = 'serial_record_entry';
70 $stage_table = 'serial_record_entry_legacy';
72 if ($arg eq '--file') {
73 $next_arg_is_file = 1;
76 if ($next_arg_is_file) {
78 $next_arg_is_file = 0;
81 if ($arg eq '--source') {
82 $next_arg_is_source = 1;
85 if ($next_arg_is_source) {
87 $next_arg_is_source = 0;
90 if ($arg eq '--base-table') {
91 $next_arg_is_base_table = 1;
94 if ($next_arg_is_base_table) {
96 $next_arg_is_base_table = 0;
99 if ($arg eq '--stage-table') {
100 $next_arg_is_stage_table = 1;
103 if ($next_arg_is_stage_table) {
105 $next_arg_is_stage_table = 0;
110 my $bre_test = check_for_table($dbh,$base_table);
111 my $bre_legacy_test = check_for_table($dbh,$stage_table);
112 if ($bre_test == 0 and $bre_legacy_test == 0 ) { create_bre($dbh); create_child_bre($dbh); }
113 if ($bre_test == 1 and $bre_legacy_test == 0 ) { create_child_bre($dbh); }
115 my $xmig_test = check_for_column($dbh,$stage_table,'x_migrate');
116 if ($xmig_test == 0) { add_column($dbh,$stage_table,'x_migrate','BOOLEAN DEFAULT TRUE'); }
118 my $xsource_test = check_for_column($dbh,$stage_table,'x_source');
119 if ($xsource_test == 0) { add_column($dbh,$stage_table,'x_source','TEXT'); }
121 #flatten out MARC XML FILE
122 open my $xml, "<:encoding(utf8)", $infile or abort('could not open MARC XML file');
125 while(my $line = <$xml>) {
126 if ($line =~ /^<\/?collection/) { next; }
128 $record = $record . $line;
129 if ($line =~ /<\/record>$/) {
130 stage_record($dbh,$record,$source);
133 if (($i % 100) == 0) { report_progress('Records stage', $i); }
138 if ($i == 0) { print "No XML was processed, are you sure this is an XML file?\n"; }
141 # beyond here be functions
149 SELECT EXISTS(SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = '$base_table') INTO t;
151 PERFORM migration_tools.build_specific_base_staging_table ('$MIGSCHEMA',REGEXP_REPLACE('$base_table','_','.'));
158 sub create_child_bre {
162 CREATE TABLE $MIGSCHEMA.$stage_table (x_migrate BOOLEAN DEFAULT TRUE, x_source TEXT) INHERITS ($MIGSCHEMA.$base_table);
170 print STDERR "$0: $msg", "\n";
174 sub report_progress {
175 my ($msg, $counter) = @_;
176 if (defined $counter) {
177 print STDERR "$msg: $counter\n";
179 print STDERR "$msg\n";
187 my $last_xact = "'$MIGSCHEMA'";
188 $record = '$_$' . $record . '$_$';
190 if ($source eq 'default') { $sql = "INSERT INTO $MIGSCHEMA.$stage_table (last_xact_id,marc) VALUES ($last_xact,$record);"; }
191 else { $sql = "INSERT INTO $MIGSCHEMA.$stage_table (last_xact_id,marc,x_source) VALUES ($last_xact,$record,'$source');"; }
192 my $sth = $dbh->prepare($sql);
197 sub check_for_table {
200 my $sql = "SELECT 1 FROM information_schema.tables WHERE table_schema = '$MIGSCHEMA' AND table_name = '$table';";
201 my $sth = $dbh->prepare($sql);
203 my @sqlresult = $sth->fetchrow_array;
204 my $r = pop @sqlresult;
205 if ($r) { return $r; } else { return 0; }
208 sub check_for_column {
212 my $sql = "SELECT 1 FROM information_schema.columns WHERE table_schema = '$MIGSCHEMA' AND table_name = '$table' AND column_name = '$column';";
213 my $sth = $dbh->prepare($sql);
215 my @sqlresult = $sth->fetchrow_array;
216 my $r = pop @sqlresult;
217 if ($r) { return $r; } else { return 0; }
224 my $column_type = shift;
225 my $sql = "ALTER TABLE $MIGSCHEMA.$table ADD COLUMN $column $column_type;";
226 my $sth = $dbh->prepare($sql);
228 my @sqlresult = $sth->fetchrow_array;
229 my $r = check_for_column($dbh,$table,$column);
230 if ($r == 0) { abort('failed to create column'); } else { return $r; }