# format in XML file
my %type2tag = (
'isis' => 'isis',
- 'excel' => 'column'
+ 'excel' => 'column',
+ 'marc' => 'marc',
+ 'feed' => 'feed'
);
sub data2xml {
my $type = shift @_;
my $row = shift @_;
my $add_xml = shift @_;
+ # needed to read values from configuration file
+ my $cfg = shift @_;
+ my $database = shift @_;
my $xml;
}
}
+ # now try to parse variables from configuration file
+ foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) {
+
+ my $delimiter = x($x->{delimiter}) || ' ';
+ my $val = $cfg->val($database, x($x->{content}));
+
+ my ($s,$d,$i) = (1,1,0); # swish, display default
+ $s = 0 if (lc($x->{type}) eq "display");
+ $d = 0 if (lc($x->{type}) eq "swish");
+ ($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
+
+ if ($val) {
+ $display_data .= $delimiter.$val if ($d);
+ $swish_data .= $val if ($s);
+ $index->insert($field, $val, $path) if ($i);
+ }
+
+ }
+
if ($display_data) {
# dump formatted output in <html>
if ($html) {
- $xml .= xmlify("html",$html);
+ #$xml .= xmlify("html",$html);
+ $xml .= "<html><![CDATA[ $html ]]></html>";
}
if ($xml) {
my $type_base = $type;
$type_base =~ s/_.+$//g;
- $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base} ], forcecontent => 1);
+ $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1);
# output current progress indicator
my $last_p = 0;
}
}
+ my $fake_dir = 1;
+ sub fakeprogress {
+ my $current = shift @_;
+
+ my @ind = ('-','\\','|','/','-','\\','|','/', '-');
+
+ $last_p += $fake_dir;
+ $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0);
+ if ($last_p % 10 == 0) {
+ printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]);
+ }
+ }
+
# now read database
print STDERR "using: $type...\n";
my $swishpath = $path."#".int($row->{mfn});
- if (my $xml = data2xml($type_base,$row,$add_xml)) {
+ if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
$xml = $cp2utf->convert($xml);
use bytes; # as opposed to chars
print "Path-Name: $swishpath\n";
next if (! $row);
- if (my $xml = data2xml($type,$row,$add_xml)) {
+ if (my $xml = data2xml($type_base,$row,$add_xml,$cfg,$database)) {
+ $xml = $cp2utf->convert($xml);
+ use bytes; # as opposed to chars
+ print "Path-Name: $swishpath\n";
+ print "Content-Length: ".(length($xml)+1)."\n";
+ print "Document-Type: XML\n\n$xml\n";
+ }
+ }
+ } elsif ($type_base eq "marc") {
+
+ use MARC;
+
+ $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage);
+ my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!";
+
+ # optional argument is format
+ my $format = x($config->{format}) || 'usmarc';
+
+ print STDERR "Reading MARC file '$marc_file'\n";
+
+ my $marc = new MARC;
+ my $nr = $marc->openmarc({
+ file=>$marc_file, format=>$format
+ }) || die "Can't open MARC file '$marc_file'";
+
+ my $i=0; # record nr.
+
+ my $rec;
+
+ while ($marc->nextmarc(1)) {
+
+ # XXX
+ fakeprogress($i++);
+
+ my $swishpath = $database."#".$i;
+
+ if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) {
$xml = $cp2utf->convert($xml);
use bytes; # as opposed to chars
print "Path-Name: $swishpath\n";
print "Document-Type: XML\n\n$xml\n";
}
}
+ } elsif ($type_base eq "feed") {
+
+ $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage);
+ my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!";
+
+ print STDERR "Reading feed from program '$prog'\n";
+
+ open(FEED,"feeds/$prog |") || die "can't start $prog: $!";
+
+ my $i=1; # record nr.
+
+ my $data;
+ my $line=1;
+
+ while (<FEED>) {
+ chomp;
+
+ if (/^$/) {
+ my $swishpath = $database."#".$i++;
+
+ if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) {
+ $xml = $cp2utf->convert($xml);
+ use bytes; # as opposed to chars
+ print "Path-Name: $swishpath\n";
+ print "Content-Length: ".(length($xml)+1)."\n";
+ print "Document-Type: XML\n\n$xml\n";
+ }
+ $line = 1;
+ $data = {};
+ next;
+ }
+
+ $line = $1 if (s/^(\d+):\s*//);
+ $data->{$line++} = $_;
+
+ fakeprogress($i);
+
+ }
}
}
=head1 NAME
-isis2xml.pl - read isis file and dump XML
+all2xml.pl - read various file formats and dump XML for SWISH-E
=head1 DESCRIPTION
-This command will read ISIS data file using OpenIsis perl module and
-create XML file for usage with I<SWISH-E>
-indexer. Dispite it's name, this script B<isn't general xml generator>
-from isis files (isis allready has something like that). Output of this
-script is tailor-made for SWISH-E.
+This command will read ISIS data file using OpenIsis perl module, MARC
+records using MARC module and optionally Micro$oft Excel files to
+create one XML file for usage with I<SWISH-E> indexer. Dispite it's name,
+this script B<isn't general xml generator> from isis files (isis allready
+has something like that). Output of this script is tailor-made for SWISH-E.
+
+=head1 BUGS
+
+Documentation is really lacking. However, in true Open Source spirit, source
+is best documentation. I even made considerable effort to comment parts
+which are not intuitively clear, so...
=head1 AUTHOR
projects / webpac / blobdiff