--- trunk/all2xml.pl 2003/07/04 23:29:27 67 +++ trunk/all2xml.pl 2003/11/20 21:23:40 163 @@ -82,20 +82,21 @@ # sort subrouting using order="" attribute sub by_order { - return 0 if (! $config->{indexer}->{$a}->{order}); - return 0 if (! $config->{indexer}->{$b}->{order}); + my $va = $config->{indexer}->{$a}->{order} || + $config->{indexer}->{$a}; + my $vb = $config->{indexer}->{$b}->{order} || + $config->{indexer}->{$b}; - return $config->{indexer}->{$a}->{order} <=> - $config->{indexer}->{$b}->{order} ; + return $va <=> $vb; } foreach my $field (sort by_order keys %{$config->{indexer}}) { $field=x($field); - $field_usage{$field}++; my $swish_data = ""; + my $swish_exact_data = ""; my $display_data = ""; my $line_delimiter; @@ -109,9 +110,10 @@ 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? @@ -124,8 +126,44 @@ # init vars so that we go into while... ($swish,$display) = (1,1); - if ($swish || $display) { + # placeholder for all repeatable entries for index + 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 " 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 ( tag)\n"; + } + } + + # while because of repeatable fields + while ($swish || $display) { ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); + if ($repeat_off > 1000) { + print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; + last; + } + # filter="name" ; filter this field through # filter/[name].pm my $filter = $x->{filter}; @@ -133,12 +171,16 @@ require "filter/".$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); } } @@ -150,26 +192,37 @@ } 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,$display); } } } # type="index" ; insert into index if ($i && $display) { - my $index_data = $display; - if ($filter) { - no strict 'refs'; - foreach my $d (&$filter($index_data)) { - $index->insert($field, $d, $path); - } - } else { - $index->insert($field, $index_data, $path); + push @index_data, $display; + $index_filter = $filter if ($filter); + } + } + + # fill data in index + if (@index_data) { + if ($index_filter) { + no strict 'refs'; + foreach my $d (@index_data) { + $index->insert($field, &$index_filter($d), $path); + } + } else { + foreach my $d (@index_data) { + $index->insert($field, $d, $path); } } } @@ -184,6 +237,8 @@ my ($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) { @@ -226,12 +281,22 @@ $xml .= xmlify($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')); + } + } # dump formatted output in if ($html) { - $xml .= xmlify("html",$html); + #$xml .= xmlify("html",$html); + $xml .= ""; } if ($xml) { @@ -258,6 +323,8 @@ $cfg_global->val('global', 'dbi_passwd') || '', ); +my $show_progress = $cfg_global->val('global', 'show_progress'); + foreach my $database ($cfg->Sections) { my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; @@ -269,12 +336,12 @@ 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; sub progress { - #return if (! $opts{q}); # FIXME + return if (! $show_progress); my $current = shift; my $total = shift || 1; my $p = int($current * 100 / $total); @@ -286,6 +353,7 @@ my $fake_dir = 1; sub fakeprogress { + return if (! $show_progress); my $current = shift @_; my @ind = ('-','\\','|','/','-','\\','|','/', '-'); @@ -307,8 +375,56 @@ $import2cp = Text::Iconv->new($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 $isis_db\n"; + next; + # OpenIsis::ERR_IO + } elsif ($db == -5) { + print STDERR "FATAL: OpenIsis can't access file $isis_db\n"; + next; + } elsif ($db < 0) { + print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n"; + next; + } + 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 = $database; @@ -330,6 +446,10 @@ } } } + # for this to work with current version of OpenIsis (0.9.0) + # you might need my patch from + # http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff + OpenIsis::close($db); print STDERR "\n"; } elsif ($type_base eq "excel") { @@ -431,7 +551,7 @@ open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; - my $i=0; # record nr. + my $i=1; # record nr. my $data; my $line=1; @@ -440,7 +560,7 @@ chomp; if (/^$/) { - my $swishpath = $database."#".$i; + my $swishpath = $database."#".$i++; if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { $xml = $cp2utf->convert($xml); @@ -454,9 +574,10 @@ next; } + $line = $1 if (s/^(\d+):\s*//); $data->{$line++} = $_; - fakeprogress($i++); + fakeprogress($i); } } @@ -471,15 +592,21 @@ =head1 NAME -isis2xml.pl - read isis file and dump XML +all2xml.pl - read various file formats and dump XML for SWISH-E =head1 DESCRIPTION -This command will read ISIS data file using OpenIsis perl module and -create XML file for usage with I -indexer. Dispite it's name, this script B -from isis files (isis allready has something like that). Output of this -script is tailor-made for SWISH-E. +This command will read ISIS data file using OpenIsis perl module, MARC +records using MARC module and optionally Micro$oft Excel files to +create one XML file for usage with I 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 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