--- trunk/all2xml.pl 2002/11/24 20:52:11 1
+++ trunk/all2xml.pl 2003/01/16 17:35:54 10
@@ -5,20 +5,116 @@
use Getopt::Std;
use Data::Dumper;
use XML::Simple;
+use Text::Unaccent 1.02; # 1.01 won't compile on my platform,
+use Text::Iconv;
-my $config=XMLin();
-print Dumper($config);
+$|=1;
+
+my $config;
+
+$config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1);
+
+use index_DBI; # there is no other, right now ;-)
+
+my $index = new index_DBI(); # open index
my %opts;
-getopts('d:m:q', \%opts);
+# usage:
+# -d directory name
+# -m multiple directories
+# -q quiet
+# -s run swish
+
+getopts('d:m:qs', \%opts);
+
+my $db_dir = $opts{d} || "ps"; # FIX
+
+#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+
+#print Dumper($config->{indexer});
+#print "-" x 70,"\n";
+
+Text::Iconv->raise_error(1); # Conversion errors raise exceptions
+
+my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8');
+my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage});
+my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1');
+
+sub isis2xml {
+
+ use xmlify;
+
+ my $row = shift @_;
-my $db_dir = $opts{d};
+ my $xml;
+ $xml .= xmlify('db_dir',$db_dir); # FIX remove?
-die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts);
+ use parse_format;
-#--------------------------------------------------------------------
+
+ foreach my $field (keys %{$config->{indexer}}) {
+
+ my $swish_data = "";
+ my $display_data = "";
+ my $index_data = "";
+
+ foreach my $x (@{$config->{indexer}->{$field}->{isis}}) {
+
+ my $format = $x->{content};
+ my ($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");
+#print STDERR "## s: $s d: $d i: $i ## $format ##\n";
+
+ $format = $cludge_codepage->convert($format);
+ my ($swish,$display) = parse_format($format,$row);
+#print STDERR "s: $swish\nd: $display\n" if ($swish);
+
+#print STDERR "swish: $swish<-- display: $display<--\n";
+ # FIX: this is ugly, UGLY, cludge: OpenIsis return
+ # UTF8 encoding of strings, but as if source charset
+ # is ISO8859-1 and not some other. This breaks our
+ # isis character encoding, so we convert it first
+ # back to ISO8859-1 (which can actually be different
+ # encoding in isis)
+
+ $swish_data .= $swish if ($s && $swish);
+ $display_data .= $display if ($d && $display);
+ $index_data .= $display if ($i && $display);
+ }
+
+#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data);
+ if ($display_data) {
+ $display_data = $isis_codepage->convert($display_data)."##" || $display_data;
+ $xml .= xmlify($field."_display", $display_data);
+ }
+ if ($swish_data) {
+ my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2');
+ $swish_data = $i->convert($swish_data);
+ $xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data));
+ #$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data;
+ #$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data));
+ }
+
+ # index
+ if ($index_data && $index_data ne "") {
+ $index_data = $index_codepage->convert($index_data) || $index_data;
+ $index->insert($field, $index_data, $db_dir);
+ }
+
+ }
+ if ($xml) {
+#print STDERR "x: $xml\n";
+ return "\n$xml\n";
+ } else {
+ return;
+ }
+}
+
+##########################################################################
my $last_tell=0;
@@ -31,36 +127,97 @@
my @isis_dbs;
foreach (@isis_dirs) {
- if (-e "$common::isis_data/$db_dir/$_/LIBRI") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/LIBRI/LIBRI";
+ if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") {
+ push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI";
}
- if (-e "$common::isis_data/$db_dir/$_/PERI") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/PERI/PERI";
+ if (-e $config->{isis_data}."/$db_dir/$_/PERI") {
+ push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI";
}
- if (-e "$common::isis_data/$db_dir/$_/AMS") {
- push @isis_dbs,"$common::isis_data/$db_dir/$_/AMS/AMS";
+ if (-e $config->{isis_data}."/$db_dir/$_/AMS") {
+ push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS";
}
- if (-e "$common::isis_data/$db_dir/$_/ARTI") {
-# push @isis_dbs,"$common::isis_data/$db_dir/$_/ARTI/ARTI";
+ if (-e $config->{isis_data}."/$db_dir/$_/ARTI") {
+# push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI";
}
}
+print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs);
+
+my $db;
+
foreach my $isis_db (@isis_dbs) {
- my $db = OpenIsis::open( "$isis_db" ) || warn "can't open '$isis_db'";
+
+ my $db = OpenIsis::open( $isis_db );
+ if (0) {
+# # FIX
+# if (! $db ) {
+ print STDERR "WARNING: can't open '$isis_db'\n";
+ next ;
+ }
my $max_rowid = OpenIsis::maxRowid( $db );
- my $last_pcnt = 0;
+ print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
+
+ my $path = $isis_db;
+ $path =~ s#$config->{isis_data}/*##g;
+ my $last_p = 0;
+
+# { my $row_id = 4514;
+# FIX
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) {
my $row = OpenIsis::read( $db, $row_id );
-
- # output current process indicator
- my $pcnt = int($row->{mfn} * 100 / $max_rowid);
- if ($pcnt != $last_pcnt) {
- printf STDERR ("%5d / %5d -- %-2d %%\n",$row->{mfn},$max_rowid,$pcnt) if (! $opts{q});
- $last_pcnt = $pcnt;
+ if ($row && $row->{mfn}) {
+#print STDERR "mfn: ",$row->{mfn},"\n";
+ # output current process indicator
+ my $p = int($row->{mfn} * 100 / $max_rowid);
+ if ($p != $last_p) {
+ printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q});
+ $last_p = $p;
+ }
+
+ if (my $xml = isis2xml($row)) {
+#print STDERR "--ret-->$xml\n";
+ print "Path-Name: $path#".int($row->{mfn})."\n";
+ print "Content-Length: ".(length($xml)+1)."\n";
+ print "Document-Type: XML\n\n$xml\n";
+ }
}
}
+ print STDERR "\n";
}
+
+# call this to commit index
+$index->close;
+
+1;
+__END__
+##########################################################################
+
+=head1 NAME
+
+isis2xml.pl - read isis file and dump XML
+
+=head1 DESCRIPTION
+
+This command will read ISIS data file using OpenIsis perl module and
+create XML file for usage with I
+indexer. Dispite it's name, this script B
+from isis files (isis allready has something like that). Output of this
+script is tailor-made for SWISH-E.
+
+=head1 AUTHOR
+
+Dobrica Pavlinusic
+
+=head1 COPYRIGHT
+
+GNU Public License (GPL) v2 or later
+
+=head1 SEE ALSO
+
+SWISH-E web site at http://www.swish-e.org
+
+=cut