$field_usage{$field}++;
my $swish_data = "";
+ my $swish_exact_data = "";
my $display_data = "";
my $line_delimiter;
my $repeat_off = 0; # repeatable offset
- my ($s,$d,$i) = (1,1,0); # swish, display default
+ my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default
$s = 0 if (lc($x->{type}) eq "display");
$d = 0 if (lc($x->{type}) eq "swish");
+ $se = 1 if (lc($x->{type}) eq "swish_exact");
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
# what will separate last line from this one?
my @index_data;
my $index_filter;
+ sub mkformat {
+ my $x = shift || die "mkformat needs tag reference";
+ my $data = shift || return;
+ my $format_name = x($x->{format_name}) || return $data;
+ my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!";
+ my $format_delimiter = x($x->{format_delimiter});
+ my @data;
+ if ($format_delimiter) {
+ @data = split(/$format_delimiter/,$data);
+ } else {
+ push @data,$data;
+ }
+
+ if ($fmt) {
+ my $nr = scalar $fmt =~ s/%s/%s/g;
+ if (($#data+1) == $nr) {
+ return sprintf($fmt,@data);
+ } else {
+ print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n";
+ return $data;
+ }
+ } else {
+ print STDERR "usage of link '$format_name' without defined format (<link> tag)\n";
+ }
+ }
+
# while because of repeatable fields
while ($swish || $display) {
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp);
require "filter/".$filter.".pm";
}
# type="swish" ; field for swish
- if ($s && $swish) {
- if ($filter) {
+ if ($swish) {
+ if ($filter && ($s || $se)) {
no strict 'refs';
- $swish_data .= join(" ",&$filter($swish));
+ my $tmp = join(" ",&$filter($swish)) if ($s || $se);
+ $swish_data .= $tmp if ($s);
+ $swish_exact_data .= $tmp if ($se);
+
} else {
- $swish_data .= $swish;
+ $swish_data .= $swish if ($s);
+ $swish_exact_data .= $swish if ($se);
}
}
}
if ($filter) {
no strict 'refs';
- $display_data .= join($delimiter,&$filter($display));
+ if ($display_data) {
+ $display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display)));
+ } else {
+ $display_data = join($delimiter,mkformat($x,&$filter($display)));
+ }
} else {
if ($display_data) {
- $display_data .= $delimiter.$display;
+ $display_data .= $delimiter.mkformat($x,$display);
} else {
- $display_data .= $display;
+ $display_data = mkformat($x,$display);
}
}
}
if (@index_data) {
if ($index_filter) {
no strict 'refs';
- foreach my $d (&$index_filter(@index_data)) {
- $index->insert($field, $d, $path);
+ foreach my $d (@index_data) {
+ $index->insert($field, &$index_filter($d), $path);
}
} else {
foreach my $d (@index_data) {
my ($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");
+ # no support for swish exact in config.
+ # IMHO, it's useless
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index");
if ($val) {
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data));
}
+ if ($swish_exact_data) {
+ $swish_exact_data =~ s/ +/ /g;
+ $swish_exact_data =~ s/ +$//g;
+
+ # add delimiters before and after word.
+ # That is required to produce exact match
+ $xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx'));
+ }
+
}
my $type_base = $type;
$type_base =~ s/_.+$//g;
- $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1);
+ $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1);
# output current progress indicator
my $last_p = 0;
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage);
my $db = OpenIsis::open( $isis_db );
+ # check if .txt database for OpenIsis is zero length,
+ # if so, erase it and re-open database
+ sub check_txt_db {
+ my $isis_db = shift || die "need isis database name";
+ my $reopen = 0;
+
+ if (-e $isis_db.".TXT") {
+ print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n";
+ unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!";
+ $reopen++;
+ }
+ if (-e $isis_db.".PTR") {
+ print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n";
+ unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!";
+ $reopen++;
+ }
+ return OpenIsis::open( $isis_db ) if ($reopen);
+ }
+
+ # EOF error
+ if ($db == -1) {
+ $db = check_txt_db($isis_db);
+ if ($db == -1) {
+ print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n";
+ next;
+ }
+ }
+
# OpenIsis::ERR_BADF
if ($db == -4) {
print STDERR "FATAL: OpenIsis can't find file $isis_db\n";
my $max_rowid = OpenIsis::maxRowid( $db );
+ # if 0 records, try to rease isis .txt database
+ if ($max_rowid == 0) {
+ # force removal of database
+ $db = check_txt_db($isis_db);
+ $max_rowid = OpenIsis::maxRowid( $db );
+ }
+
print STDERR "Reading database: $isis_db [$max_rowid rows]\n";
my $path = $database;
}
}
}
+ # for this to work with current version of OpenIsis (0.9.0)
+ # you might need my patch from
+ # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff
+ OpenIsis::close($db);
print STDERR "\n";
} elsif ($type_base eq "excel") {