--- trunk/all2xml.pl 2003/07/04 20:11:48 62 +++ trunk/all2xml.pl 2003/07/08 22:13:56 81 @@ -58,6 +58,7 @@ 'isis' => 'isis', 'excel' => 'column', 'marc' => 'marc', + 'feed' => 'feed' ); sub data2xml { @@ -230,7 +231,8 @@ # dump formatted output in if ($html) { - $xml .= xmlify("html",$html); + #$xml .= xmlify("html",$html); + $xml .= ""; } if ($xml) { @@ -283,6 +285,19 @@ } } + my $fake_dir = 1; + sub fakeprogress { + my $current = shift @_; + + my @ind = ('-','\\','|','/','-','\\','|','/', '-'); + + $last_p += $fake_dir; + $fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); + if ($last_p % 10 == 0) { + printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); + } + } + # now read database print STDERR "using: $type...\n"; @@ -373,7 +388,7 @@ } } } elsif ($type_base eq "marc") { - ## XXX + use MARC; $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); @@ -382,8 +397,6 @@ # optional argument is format my $format = x($config->{format}) || 'usmarc'; - my %id_stored; # to aviod duplicates - print STDERR "Reading MARC file '$marc_file'\n"; my $marc = new MARC; @@ -392,17 +405,13 @@ }) || die "Can't open MARC file '$marc_file'"; my $i=0; # record nr. - my $inc=1; - my $max_i=1000; my $rec; while ($marc->nextmarc(1)) { # XXX - progress($i, $max_i); - $i += $inc; - $inc = -$inc if ($i > $max_i || $i < 0); + fakeprogress($i++); my $swishpath = $database."#".$i; @@ -414,6 +423,44 @@ print "Document-Type: XML\n\n$xml\n"; } } + } elsif ($type_base eq "feed") { + + $import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); + my $prog = x($config->{prog}) || die "$database doesn't have 'prog' defined!"; + + print STDERR "Reading feed from program '$prog'\n"; + + open(FEED,"feeds/$prog |") || die "can't start $prog: $!"; + + my $i=1; # record nr. + + my $data; + my $line=1; + + while () { + chomp; + + if (/^$/) { + my $swishpath = $database."#".$i++; + + if (my $xml = data2xml($type_base,$data,$add_xml,$cfg,$database)) { + $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"; + } + $line = 1; + $data = {}; + next; + } + + $line = $1 if (s/^(\d+):\s*//); + $data->{$line++} = $_; + + fakeprogress($i); + + } } } @@ -426,15 +473,21 @@ =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