8 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
11 my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
15 getopts('d:m:q', \%opts);
17 my $db_dir = $opts{d} || "ps"; # FIX
19 #die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
21 #print Dumper($config->{indexer});
24 # how to convert isis code page to UTF8?
25 my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die;
35 my $isis_id = shift @_;
36 my $subfield = shift @_;
37 if ($row->{$isis_id}->[0]) {
38 my $sf = OpenIsis::subfields($row->{$isis_id}->[0]);
39 if (! defined $subfield || length($subfield) == 0) {
40 # subfield list undef, empty or no defined subfields for this record
41 my $all_sf = $row->{$isis_id}->[0];
42 $all_sf =~ s/\^./ /g; nuke definirions
44 } elsif ($sf->{$subfield}) {
45 return $sf->{$subfield};
50 foreach my $field (keys %{$config->{indexer}}) {
52 my $display_data = "";
55 foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
60 my $format = $x->{content};
61 my $i = 1; # index only
62 my $d = 1; # display only
63 $i = 0 if (lc($x->{type}) eq "display");
64 $d = 0 if (lc($x->{type}) eq "index");
65 #print "## i: $i d: $d ## $format ##";
68 if ($format =~ s/^([^\d]+)//) {
72 if ($format =~ s/^(\d\d\d)(\w?)//) {
73 my $isis_tmp = isis_sf($row,$1,$2);
75 # $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d);
76 $display_tmp .= $prefix . $isis_tmp if ($d);
77 $index_tmp .= $isis_tmp." " if ($i);
78 #print " $isis_tmp <--\n";
81 } elsif ($format =~ s/^([^\d]+)//) {
84 print STDERR "WARNING: unparsed format '$format'\n";
89 $display_tmp .= $prefix if ($display_tmp);
91 # $display_data .= $display_tmp if ($display_tmp ne "");
92 # $index_data .= $index_tmp if ($index_tmp ne "");
93 $display_data .= $display_tmp;
94 $index_data .= $index_tmp;
97 #print "--display:$display_data\n--index:$index_data\n";
98 $xml->{$field}->{display} .= $isis_map->tou($display_data)->utf8 if ($display_data);
99 $xml->{$field}->{index} .= unac_string($config->{isis_codepage},$index_data) if ($index_data);
103 return XMLout($xml, rootname => 'xml', noattr => 1 );
109 ##########################################################################
113 my @isis_dirs = ( '.' ); # use dirname as database name
116 @isis_dirs = split(/,/,$opts{m});
121 foreach (@isis_dirs) {
122 if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") {
123 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI";
125 if (-e $config->{isis_data}."/$db_dir/$_/PERI") {
126 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI";
128 if (-e $config->{isis_data}."/$db_dir/$_/AMS") {
129 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS";
131 if (-e $config->{isis_data}."/$db_dir/$_/ARTI") {
132 # push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI";
136 print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs);
140 foreach my $isis_db (@isis_dbs) {
143 my $db = OpenIsis::open( $isis_db );
147 print STDERR "WARNING: can't open '$isis_db'\n";
151 my $max_rowid = OpenIsis::maxRowid( $db );
153 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
159 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
160 my $row = OpenIsis::read( $db, $row_id );
161 if ($row && $row->{mfn}) {
163 # output current process indicator
164 my $p = int($row->{mfn} * 100 / $max_rowid);
166 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
170 if (my $xml = isis2xml($row)) {
171 print "Path-Name: $isis_db#".$row->{mfn}."\n";
172 print "Content-Length: ".(length($xml)+1)."\n";
173 print "Document-Type: XML\n\n$xml\n";
183 ##########################################################################
187 isis2xml.pl - read isis file and dump XML
191 This command will read ISIS data file using OpenIsis perl module and
192 create XML file for usage with I<SWISH-E>
193 indexer. Dispite it's name, this script B<isn't general xml generator>
194 from isis files (isis allready has something like that). Output of this
195 script is tailor-made for SWISH-E.
199 Dobrica Pavlinusic <dpavlin@rot13.org>
203 GNU Public License (GPL) v2 or later
207 SWISH-E web site at http://www.swish-e.org