1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
use strict; |
use strict; |
4 |
use OpenIsis; |
use Biblio::Isis; |
5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
22 |
my $config; |
my $config; |
23 |
|
|
24 |
#use index_DBI; # default DBI module for index |
#use index_DBI; # default DBI module for index |
25 |
use index_DBI_cache; # faster DBI module using memory cache |
#use index_DBI_cache; # faster DBI module using memory cache |
26 |
|
use index_DBI_filter; # filter support for indexes |
27 |
my $index; |
my $index; |
28 |
|
|
29 |
my %opts; |
my %opts; |
258 |
# init vars so that we go into while... |
# init vars so that we go into while... |
259 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
260 |
|
|
|
# placeholder for all repeatable entries for index |
|
|
|
|
261 |
sub mkformat($$) { |
sub mkformat($$) { |
262 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
263 |
my $data = shift || return; |
my $data = shift || return; |
633 |
|
|
634 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
635 |
|
|
636 |
|
# save database name in global variable path for later |
637 |
|
# (need for index filter creation) |
638 |
|
$path = $database; |
639 |
|
|
640 |
my $type = lc($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"; |
641 |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
642 |
|
|
690 |
my $current = shift; |
my $current = shift; |
691 |
my $total = shift || 1; |
my $total = shift || 1; |
692 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
693 |
if ($p != $last_p) { |
if ($p < $last_p || $current == 1) { |
694 |
|
$start_t = time(); |
695 |
|
$last_p = 0; |
696 |
|
} elsif ($p != $last_p) { |
697 |
my $rate = ($current / (time() - $start_t || 1)); |
my $rate = ($current / (time() - $start_t || 1)); |
698 |
my $eta = ($total-$current) / ($rate || 1); |
my $eta = ($total-$current) / ($rate || 1); |
699 |
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
702 |
} |
} |
703 |
|
|
704 |
my $fake_dir = 1; |
my $fake_dir = 1; |
705 |
|
my $fake_pos = 0; |
706 |
|
my $last_fake_t = time(); |
707 |
sub fakeprogress { |
sub fakeprogress { |
708 |
return if (! $show_progress); |
return if (! $show_progress); |
709 |
my $current = shift @_; |
my $current = shift @_; |
710 |
|
|
711 |
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
my @ind = ('-','\\','|','/','-','\\','|','/'); |
712 |
|
|
713 |
|
if ($current < $fake_pos) { |
714 |
|
$start_t = time(); |
715 |
|
$last_fake_t = 0; |
716 |
|
$fake_dir = 1; |
717 |
|
$fake_pos = 0; |
718 |
|
} |
719 |
|
|
720 |
$last_p += $fake_dir; |
if (time()-$last_fake_t >= 1) { |
721 |
$fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); |
$last_fake_t = time(); |
722 |
if ($last_p % 10 == 0) { |
$fake_pos += $fake_dir; |
723 |
printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); |
$fake_dir = -$fake_dir if ($fake_pos > 38); |
724 |
|
} |
725 |
|
|
726 |
|
if ($current % 10 == 0) { |
727 |
|
my $rate = ($current / (time() - $start_t || 1)); |
728 |
|
printf STDERR ("%5d [%-38s] %0.1f/s\r",$current, " " x $fake_pos .$ind[($current / 10) % 8], $rate); |
729 |
} |
} |
730 |
} |
} |
731 |
|
|
740 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
741 |
|
|
742 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
743 |
my $db = OpenIsis::open( $isis_db ); |
my $db = new Biblio::Isis( isisdb => $isis_db ); |
744 |
|
|
745 |
# check if .txt database for OpenIsis is zero length, |
my $max_rowid = $db->count if ($db); |
|
# 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; |
|
|
} |
|
|
} |
|
746 |
|
|
747 |
# OpenIsis::ERR_BADF |
if (! $max_rowid) { |
748 |
if ($db == -4) { |
print STDERR "FATAL: can't read ISIS database: $isis_db, skipping...\n"; |
|
print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; |
|
749 |
next; |
next; |
|
# OpenIsis::ERR_IO |
|
|
} elsif ($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 ); |
|
750 |
} |
} |
751 |
|
|
752 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
753 |
|
|
|
my $path = $database; |
|
|
|
|
754 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
755 |
my $row = OpenIsis::read( $db, $row_id ); |
my $row = $db->to_hash( $row_id ); |
756 |
if ($row && $row->{mfn}) { |
if ($row) { |
757 |
|
|
758 |
|
$row->{mfn} = $row_id; |
759 |
|
$row->{record} = $db->{record}; |
760 |
|
|
761 |
progress($row->{mfn}, $max_rowid); |
progress($row->{mfn}, $max_rowid); |
762 |
|
|
763 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
771 |
} |
} |
772 |
} |
} |
773 |
} |
} |
|
# 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($db); |
|
774 |
print STDERR "\n"; |
print STDERR "\n"; |
775 |
|
|
776 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
832 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
833 |
} |
} |
834 |
} |
} |
835 |
|
|
836 |
|
print STDERR "\n"; |
837 |
|
|
838 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
839 |
|
|
840 |
require MARC::File::USMARC; |
require MARC::File::USMARC; |
846 |
warn "marc_format is no longer used!" if ($config->{marc_format}); |
warn "marc_format is no longer used!" if ($config->{marc_format}); |
847 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
848 |
|
|
849 |
my $marc = MARC::File::USMARC->in( $marc_file ) |
my $marc = MARC::File::USMARC->in( $marc_file ); |
850 |
|| die "Can't open MARC file '$marc_file': ".$MARC::File::ERROR; |
|
851 |
|
if (! $marc) { |
852 |
|
print STDERR "FATAL: can't read MARC file: $marc_file, skipping...\n"; |
853 |
|
next; |
854 |
|
} |
855 |
|
|
856 |
# count records in MARC file |
# count records in MARC file |
857 |
sub marc_count { |
sub marc_count { |
858 |
my $filename = shift || die; |
my $filename = shift || die; |
859 |
my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; |
my $file = MARC::File::USMARC->in($filename) || return; |
860 |
my $count = 0; |
my $count = 0; |
861 |
while ($file->skip()) { |
while ($file->skip()) { |
862 |
$count++; |
$count++; |
866 |
|
|
867 |
my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; |
my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; |
868 |
|
|
869 |
my $i = 0; |
my $i = 1; |
870 |
|
|
871 |
while( my $rec = $marc->next() ) { |
while( my $rec = $marc->next() ) { |
872 |
|
|
873 |
progress($i++,$count); |
progress($i,$count); |
874 |
|
|
875 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
876 |
|
|
881 |
print "Content-Length: ".(length($xml)+1)."\n"; |
print "Content-Length: ".(length($xml)+1)."\n"; |
882 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
883 |
} |
} |
884 |
|
|
885 |
|
$i++; |
886 |
} |
} |
887 |
|
|
888 |
print STDERR "\n"; |
print STDERR "\n"; |
947 |
|
|
948 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
949 |
|
|
950 |
This command will read ISIS data file using OpenIsis perl module, MARC |
This command will read ISIS data file using Biblio::Isis perl module, MARC |
951 |
records using MARC module and optionally Micro$oft Excel files to |
records using MARC::File module and optionally Micro$oft Excel files to |
952 |
create one XML file for usage with I<SWISH-E> indexer. Dispite it's name, |
create one XML file for usage with I<SWISH-E> indexer. Dispite it's name, |
953 |
this script B<isn't general xml generator> from isis files (isis allready |
this script B<isn't general xml generator> from isis files (isis allready |
954 |
has something like that). Output of this script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |