1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
use strict; |
use strict; |
4 |
use OpenIsis; |
use IsisDB; |
5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
139 |
} else { |
} else { |
140 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
141 |
} |
} |
142 |
|
|
143 |
if ($field_name) { |
if ($field_name) { |
144 |
|
$field_name = x($field_name); |
145 |
if (! $last_field_name) { |
if (! $last_field_name) { |
146 |
$last_field_name = x($field_name); |
$last_field_name = $field_name; |
147 |
return $last_field_name; |
return $last_field_name; |
148 |
} elsif ($field_name ne $last_field_name) { |
} elsif ($field_name ne $last_field_name) { |
149 |
$last_field_name = x($field_name); |
$last_field_name = $field_name; |
150 |
return $last_field_name; |
return $last_field_name; |
151 |
} |
} |
152 |
} |
} |
170 |
($s,$se,$d,$i) = (0,1,0,0); |
($s,$se,$d,$i) = (0,1,0,0); |
171 |
} elsif (lc($type) =~ /^lookup/) { |
} elsif (lc($type) =~ /^lookup/) { |
172 |
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
173 |
|
} elsif ($type) { |
174 |
|
print STDERR "WARNING: unknown type: $type\n"; |
175 |
} |
} |
176 |
return ($s,$se,$d,$i,$il); |
return ($s,$se,$d,$i,$il); |
177 |
} |
} |
641 |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
642 |
if ($lookup_file) { |
if ($lookup_file) { |
643 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
644 |
|
if (! -e $lookup_file) { |
645 |
|
open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!"; |
646 |
|
close(LOOKUP); |
647 |
|
} |
648 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
649 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
650 |
# delete memory cache for lookup file |
# delete memory cache for lookup file |
667 |
|
|
668 |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
669 |
|
|
670 |
|
# helper for progress bar |
671 |
|
sub fmt_time { |
672 |
|
my $t = shift || 0; |
673 |
|
my $out = ""; |
674 |
|
|
675 |
|
my ($ss,$mm,$hh) = gmtime($t); |
676 |
|
$out .= "${hh}h" if ($hh); |
677 |
|
$out .= sprintf("%02d:%02d", $mm,$ss); |
678 |
|
$out .= " " if ($hh == 0); |
679 |
|
return $out; |
680 |
|
} |
681 |
|
|
682 |
# output current progress indicator |
# output current progress indicator |
683 |
my $last_p = 0; |
my $last_p = 0; |
684 |
|
my $start_t = time(); |
685 |
sub progress { |
sub progress { |
686 |
return if (! $show_progress); |
return if (! $show_progress); |
687 |
my $current = shift; |
my $current = shift || 1; |
688 |
my $total = shift || 1; |
my $total = shift || 1; |
689 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
690 |
if ($p != $last_p) { |
if ($p != $last_p) { |
691 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
my $rate = ($current / (time() - $start_t || 1)); |
692 |
|
my $eta = ($total-$current) / ($rate || 1); |
693 |
|
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
694 |
$last_p = $p; |
$last_p = $p; |
695 |
} |
} |
696 |
} |
} |
720 |
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!"; |
721 |
|
|
722 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
723 |
my $db = OpenIsis::open( $isis_db ); |
my $db = new IsisDB( isisdb => $isis_db ); |
724 |
|
|
725 |
# check if .txt database for OpenIsis is zero length, |
my $max_rowid = $db->{'maxmfn'} || die "can't find maxmfn"; |
|
# 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 ); |
|
|
} |
|
726 |
|
|
727 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
728 |
|
|
729 |
my $path = $database; |
my $path = $database; |
730 |
|
|
731 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
732 |
my $row = OpenIsis::read( $db, $row_id ); |
my $row = $db->to_hash( $row_id ); |
733 |
if ($row && $row->{mfn}) { |
if ($row) { |
734 |
|
|
735 |
|
$row->{mfn} = $row_id; |
736 |
|
$row->{record} = $db->{record}; |
737 |
|
|
738 |
progress($row->{mfn}, $max_rowid); |
progress($row->{mfn}, $max_rowid); |
739 |
|
|
740 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
748 |
} |
} |
749 |
} |
} |
750 |
} |
} |
|
# for this to work with current version of OpenIsis (0.9.0) |
|
|
# you might need my patch from |
|
|
# https://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff |
|
|
OpenIsis::close($db); |
|
751 |
print STDERR "\n"; |
print STDERR "\n"; |
752 |
|
|
753 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
781 |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
782 |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
783 |
if ($cell) { |
if ($cell) { |
784 |
$row->{int2col($iC)} = $cell->Value; |
# this conversion is a cludge. |
785 |
|
# Files from Excell could have |
786 |
|
# characters which don't fit into |
787 |
|
# destination encoding. |
788 |
|
$row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value; |
789 |
} |
} |
790 |
} |
} |
791 |
|
|