--- trunk/all2xml.pl 2003/03/22 22:51:48 44 +++ trunk/all2xml.pl 2003/06/23 20:20:32 54 @@ -19,8 +19,7 @@ my $config; use index_DBI; # there is no other, right now ;-) - -my $index = new index_DBI(); # open index +my $index; my %opts; @@ -46,16 +45,24 @@ return $utf2cp->convert($_[0]); } -# decode isis import codepage -my $isis2cp; +# decode isis/excel or other import codepage +my $import2cp; # outgoing xml must be in UTF-8 my $cp2utf = Text::Iconv->new($codepage,'UTF-8'); -sub isis2xml { +# mapping between data type and tag which specify +# format in XML file +my %type2tag = ( + 'isis' => 'isis', + 'excel' => 'column' +); + +sub data2xml { use xmlify; + my $type = shift @_; my $row = shift @_; my $add_xml = shift @_; @@ -88,12 +95,13 @@ my ($swish,$display); - foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { + my $tag = $type2tag{$type} || die "can't find which tag to use for type $type"; + foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { my $format = x($x->{content}); my $delimiter = x($x->{delimiter}) || ' '; - my $isis_i = 0; # isis repeatable offset + my $repeat_off = 0; # repeatable offset my ($s,$d,$i) = (1,1,0); # swish, display default $s = 0 if (lc($x->{type}) eq "display"); @@ -110,9 +118,8 @@ # init vars so that we go into while... ($swish,$display) = (1,1); - while ($swish || $display) { - ($swish,$display) = parse_format($format,$row,$isis_i++,$isis2cp); - + if ($swish || $display) { + ($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); # filter="name" ; filter this field through # filter/[name].pm my $filter = $x->{filter}; @@ -212,49 +219,123 @@ ########################################################################## +# read configuration for this script my $cfg = new Config::IniFiles( -file => $config_file ); +# read global.conf configuration +my $cfg_global = new Config::IniFiles( -file => 'global.conf' ); + +# open index +$index = new index_DBI( + $cfg_global->val('global', 'dbi_dbd'), + $cfg_global->val('global', 'dbi_dsn'), + $cfg_global->val('global', 'dbi_user'), + $cfg_global->val('global', 'dbi_passwd') || '', + ); + foreach my $database ($cfg->Sections) { - my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; - my $type = $cfg -> val($database, 'type') || die "$database doesn't have 'type' defined"; + my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; my $add_xml = $cfg -> val($database, 'xml'); # optional - $config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); +print STDERR "reading ./import_xml/$type.xml\n"; - $isis2cp = Text::Iconv->new($config->{isis_codepage},$codepage); + $config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type} ], forcecontent => 1); - my $db = OpenIsis::open( $isis_db ); - if (0) { -# # FIX -# if (! $db ) { - print STDERR "WARNING: can't open '$isis_db'\n"; - next ; + # output current progress indicator + my $last_p = 0; + sub progress { + #return if (! $opts{q}); # FIXME + my $current = shift; + my $total = shift; + my $p = int($current * 100 / $total); + if ($p != $last_p) { + printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); + $last_p = $p; + } } - my $max_rowid = OpenIsis::maxRowid( $db ); + # now read database +print STDERR "using: $type...\n"; - print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; + if ($type eq "isis") { + my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; - my $path = $database; # was $isis_db + $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); + my $db = OpenIsis::open( $isis_db ); - my $last_p = 0; + my $max_rowid = OpenIsis::maxRowid( $db ); + + print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; + + my $path = $database; + + for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { + my $row = OpenIsis::read( $db, $row_id ); + if ($row && $row->{mfn}) { + + progress($row->{mfn}, $max_rowid); -# { my $row_id = 4514; -# FIX - for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { - my $row = OpenIsis::read( $db, $row_id ); - if ($row && $row->{mfn}) { - # 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; + my $swishpath = $path."#".int($row->{mfn}); + + if (my $xml = data2xml($type,$row,$add_xml)) { + $xml = $cp2utf->convert($xml); + use bytes; # as opposed to chars + print "Path-Name: $swishpath\n"; + print "Content-Length: ".(length($xml)+1)."\n"; + print "Document-Type: XML\n\n$xml\n"; + } + } + } + print STDERR "\n"; + + } elsif ($type eq "excel") { + use Spreadsheet::ParseExcel; + use Spreadsheet::ParseExcel::Utility qw(int2col); + + $import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); + my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; + + my $sheet = x($config->{sheet}) || die "no sheet in $type.xml"; + my $start_row = x($config->{start_row}) || die "no start_row in $type.xml"; + + my $oBook = Spreadsheet::ParseExcel::Workbook->Parse($excel_file) || die "can't open Excel file '$excel_file'"; + + my $sheet_nr = 0; + foreach my $oWks (@{$oBook->{Worksheet}}) { + #print STDERR "-- SHEET $sheet_nr:", $oWks->{Name}, "\n"; + last if ($oWks->{Name} eq $sheet); + $sheet_nr++; + } + + my $oWorksheet = $oBook->{Worksheet}[$sheet_nr]; + + print STDERR "using sheet: ",$oWorksheet->{Name},"\n"; + defined ($oWorksheet) || die "can't find sheet '$sheet' in $excel_file"; + my $end_row = x($config->{end_row}) || $oWorksheet->{MaxRow}; + + for(my $iR = $oWorksheet->{MinRow} ; defined $end_row && $iR <= $end_row ; $iR++) { + my $row; + for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { + my $cell = $oWorksheet->{Cells}[$iR][$iC]; + if ($cell) { + $row->{int2col($iC)} = $cell->Value; + } } - my $swishpath = $path."#".int($row->{mfn}); + progress($iR, $end_row); + +# print "row[$iR/$end_row] "; +# foreach (keys %{$row}) { +# print "$_: ",$row->{$_},"\t"; +# } +# print "\n"; + + my $swishpath = $database."#".$iR; + + next if (! $row); - if (my $xml = isis2xml($row,$add_xml)) { + if (my $xml = data2xml($type,$row,$add_xml)) { $xml = $cp2utf->convert($xml); use bytes; # as opposed to chars print "Path-Name: $swishpath\n"; @@ -263,7 +344,6 @@ } } } - print STDERR "\n"; } # call this to commit index