--- trunk/all2xml.pl 2004/12/31 04:22:49 619 +++ trunk/all2xml.pl 2005/01/23 02:02:10 641 @@ -1,7 +1,7 @@ #!/usr/bin/perl -w use strict; -use IsisDB; +use Biblio::Isis; use Getopt::Std; use Data::Dumper; use XML::Simple; @@ -22,7 +22,8 @@ my $config; #use index_DBI; # default DBI module for index -use index_DBI_cache; # faster DBI module using memory cache +#use index_DBI_cache; # faster DBI module using memory cache +use index_DBI_tag; # tag support for indexes my $index; my %opts; @@ -257,8 +258,6 @@ # init vars so that we go into while... ($swish,$display) = (1,1); - # placeholder for all repeatable entries for index - sub mkformat($$) { my $x = shift || die "mkformat needs tag reference"; my $data = shift || return; @@ -684,10 +683,13 @@ my $start_t = time(); sub progress { return if (! $show_progress); - my $current = shift || 1; + my $current = shift; my $total = shift || 1; my $p = int($current * 100 / $total); - if ($p != $last_p) { + if ($p < $last_p || $current == 1) { + $start_t = time(); + $last_p = 0; + } elsif ($p != $last_p) { my $rate = ($current / (time() - $start_t || 1)); my $eta = ($total-$current) / ($rate || 1); printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); @@ -696,16 +698,30 @@ } my $fake_dir = 1; + my $fake_pos = 0; + my $last_fake_t = time(); sub fakeprogress { return if (! $show_progress); my $current = shift @_; - my @ind = ('-','\\','|','/','-','\\','|','/', '-'); + my @ind = ('-','\\','|','/','-','\\','|','/'); + + if ($current < $fake_pos) { + $start_t = time(); + $last_fake_t = 0; + $fake_dir = 1; + $fake_pos = 0; + } - $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) % $#ind]); + if (time()-$last_fake_t >= 1) { + $last_fake_t = time(); + $fake_pos += $fake_dir; + $fake_dir = -$fake_dir if ($fake_pos > 38); + } + + if ($current % 10 == 0) { + my $rate = ($current / (time() - $start_t || 1)); + printf STDERR ("%5d [%-38s] %0.1f/s\r",$current, " " x $fake_pos .$ind[($current / 10) % 8], $rate); } } @@ -720,13 +736,13 @@ my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; $import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); - my $db = new IsisDB( isisdb => $isis_db ); + my $db = new Biblio::Isis( isisdb => $isis_db ); - my $max_rowid = $db->{'maxmfn'} || die "can't find maxmfn"; + my $max_rowid = $db->count || die "can't find maxmfn"; print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; - my $path = $database; + $path = $database; for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { my $row = $db->to_hash( $row_id ); @@ -811,36 +827,40 @@ } } elsif ($type_base eq "marc") { - require MARC; + require MARC::File::USMARC; $import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; # optional argument is format - my $format = x($config->{marc_format}) || 'usmarc'; - + warn "marc_format is no longer used!" if ($config->{marc_format}); print STDERR "Reading MARC file '$marc_file'\n"; - my $marc = new MARC; - my $nr = $marc->openmarc({ - file=>$marc_file, format=>$format - }) || die "Can't open MARC file '$marc_file' with format '$format'"; + my $marc = MARC::File::USMARC->in( $marc_file ) + || die "Can't open MARC file '$marc_file': ".$MARC::File::ERROR; - # read MARC file in memory - $marc->nextmarc(-1); + # count records in MARC file + sub marc_count { + my $filename = shift || die; + my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; + my $count = 0; + while ($file->skip()) { + $count++; + } + return $count; + } - my $max_rec = $marc->marc_count(); + my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; - for(my $i=1; $i<=$max_rec; $i++) { + my $i = 0; - progress($i,$max_rec); + while( my $rec = $marc->next() ) { - # store value for marc_sf.pm - $main::cache->{marc_record} = $i; + progress($i++,$count); my $swishpath = $database."#".$i; - if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { + if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) { $xml = $cp2utf->convert($xml); use bytes; # as opposed to chars print "Path-Name: $swishpath\n"; @@ -911,7 +931,7 @@ =head1 DESCRIPTION -This command will read ISIS data file using OpenIsis perl module, MARC +This command will read ISIS data file using IsisDB perl module, MARC records using MARC module and optionally Micro$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