8 use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
11 my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
17 # -m multiple directories
21 getopts('d:m:qs', \%opts);
23 my $db_dir = $opts{d} || "ps"; # FIX
25 #die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
27 #print Dumper($config->{indexer});
30 # how to convert isis code page to UTF8?
31 my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die;
38 $xml->{db_dir} = [ $db_dir ]; # FIX remove?
42 my $isis_id = shift @_;
43 my $subfield = shift @_;
44 if ($row->{$isis_id}->[0]) {
45 my $sf = OpenIsis::subfields($row->{$isis_id}->[0]);
46 if (! defined $subfield || length($subfield) == 0) {
47 # subfield list undef, empty or no defined subfields for this record
48 my $all_sf = $row->{$isis_id}->[0];
49 $all_sf =~ s/\^./ /g; nuke definirions
51 } elsif ($sf->{$subfield}) {
52 return $sf->{$subfield};
57 foreach my $field (keys %{$config->{indexer}}) {
59 my $display_data = "";
62 foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
67 my $format = $x->{content};
68 my $i = 1; # index only
69 my $d = 1; # display only
70 $i = 0 if (lc($x->{type}) eq "display");
71 $d = 0 if (lc($x->{type}) eq "index");
72 #print "## i: $i d: $d ## $format ##";
75 if ($format =~ s/^([^\d]+)//) {
79 if ($format =~ s/^(\d\d\d)(\w?)//) {
80 my $isis_tmp = isis_sf($row,$1,$2);
82 # $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d);
83 $display_tmp .= $prefix . $isis_tmp if ($d);
84 $index_tmp .= $isis_tmp." " if ($i);
85 #print " $isis_tmp <--\n";
88 } elsif ($format =~ s/^([^\d]+)//) {
91 print STDERR "WARNING: unparsed format '$format'\n";
96 $display_tmp .= $prefix if ($display_tmp);
98 # $display_data .= $display_tmp if ($display_tmp ne "");
99 # $index_data .= $index_tmp if ($index_tmp ne "");
100 $display_data .= $display_tmp;
101 $index_data .= $index_tmp;
104 #print "--display:$display_data\n--index:$index_data\n";
105 #$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data);
106 #$xml->{$field."_index"} = unac_string($config->{isis_codepage},$index_data) if ($index_data);
107 $xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data);
108 $xml->{$field."_index"} = [ unac_string($config->{isis_codepage},$index_data)." jabuka" ] if ($index_data);
112 return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 );
118 ##########################################################################
122 my @isis_dirs = ( '.' ); # use dirname as database name
125 @isis_dirs = split(/,/,$opts{m});
130 foreach (@isis_dirs) {
131 if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") {
132 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI";
134 if (-e $config->{isis_data}."/$db_dir/$_/PERI") {
135 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI";
137 if (-e $config->{isis_data}."/$db_dir/$_/AMS") {
138 push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS";
140 if (-e $config->{isis_data}."/$db_dir/$_/ARTI") {
141 # push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI";
145 print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs);
149 foreach my $isis_db (@isis_dbs) {
152 my $db = OpenIsis::open( $isis_db );
156 print STDERR "WARNING: can't open '$isis_db'\n";
160 my $max_rowid = OpenIsis::maxRowid( $db );
162 print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
168 for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
169 my $row = OpenIsis::read( $db, $row_id );
170 if ($row && $row->{mfn}) {
172 # output current process indicator
173 my $p = int($row->{mfn} * 100 / $max_rowid);
175 printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
179 if (my $xml = isis2xml($row)) {
181 $path =~ s#$config->{isis_data}/*##g;
182 my $out = "Path-Name: $path#".$row->{mfn}."\n";
183 $out .= "Content-Length: ".(length($xml)+1)."\n";
184 $out .= "Document-Type: XML\n\n$xml\n";
195 ##########################################################################
199 isis2xml.pl - read isis file and dump XML
203 This command will read ISIS data file using OpenIsis perl module and
204 create XML file for usage with I<SWISH-E>
205 indexer. Dispite it's name, this script B<isn't general xml generator>
206 from isis files (isis allready has something like that). Output of this
207 script is tailor-made for SWISH-E.
211 Dobrica Pavlinusic <dpavlin@rot13.org>
215 GNU Public License (GPL) v2 or later
219 SWISH-E web site at http://www.swish-e.org