6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
8 |
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
9 |
require Unicode::Map8; |
use Text::Iconv; |
|
use DBI; |
|
10 |
|
|
|
my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
|
|
my $dbh = DBI->connect("DBI:Pg:dbname=webpac","","") || die $DBI::errstr; # FIX |
|
|
# FIX; select relname from pg_class where relname like 'index_%' ; |
|
|
$dbh->begin_work || die $dbh->errstr(); |
|
11 |
|
|
12 |
$dbh->do("delete from index_author") || die $dbh->errstr(); |
$|=1; |
13 |
$dbh->do("delete from index_title") || die $dbh->errstr(); |
|
14 |
|
my $config; |
15 |
|
|
16 |
|
$config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
17 |
|
|
18 |
|
use index_DBI; # there is no other, right now ;-) |
19 |
|
|
20 |
|
my $index = new index_DBI(); # open index |
21 |
|
|
22 |
my %opts; |
my %opts; |
23 |
|
|
36 |
#print Dumper($config->{indexer}); |
#print Dumper($config->{indexer}); |
37 |
#print "-" x 70,"\n"; |
#print "-" x 70,"\n"; |
38 |
|
|
39 |
# how to convert isis code page to UTF8? |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
40 |
my $isis_map = Unicode::Map8->new($config->{isis_codepage}) || die; |
|
41 |
|
my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
42 |
|
my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
43 |
|
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
44 |
|
|
45 |
sub isis2xml { |
sub isis2xml { |
46 |
|
|
47 |
|
use xmlify; |
48 |
|
|
49 |
my $row = shift @_; |
my $row = shift @_; |
50 |
|
|
51 |
my $xml; |
my $xml; |
52 |
$xml->{db_dir} = [ $db_dir ]; # FIX remove? |
$xml .= xmlify('db_dir',$db_dir); # FIX remove? |
53 |
|
|
54 |
|
use parse_format; |
55 |
|
|
|
sub isis_sf { |
|
|
my $row = shift @_; |
|
|
my $isis_id = shift @_; |
|
|
my $subfield = shift @_; |
|
|
if ($row->{$isis_id}->[0]) { |
|
|
my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); |
|
|
if (! defined $subfield || length($subfield) == 0) { |
|
|
# subfield list undef, empty or no defined subfields for this record |
|
|
my $all_sf = $row->{$isis_id}->[0]; |
|
|
$all_sf =~ s/\^./ /g; nuke definirions |
|
|
return $all_sf; |
|
|
} elsif ($sf->{$subfield}) { |
|
|
return $sf->{$subfield}; |
|
|
} |
|
|
} |
|
|
} |
|
56 |
|
|
57 |
foreach my $field (keys %{$config->{indexer}}) { |
foreach my $field (keys %{$config->{indexer}}) { |
58 |
|
|
|
my $display_data = ""; |
|
59 |
my $swish_data = ""; |
my $swish_data = ""; |
60 |
|
my $display_data = ""; |
61 |
my $index_data = ""; |
my $index_data = ""; |
62 |
|
|
63 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
64 |
|
|
|
my $display_tmp = ""; |
|
|
my $swish_tmp = ""; |
|
|
my $index_tmp = ""; |
|
|
|
|
65 |
my $format = $x->{content}; |
my $format = $x->{content}; |
66 |
my $s = 1; # swish only |
my ($s,$d,$i) = (1,1,0); # swish, display default |
|
my $d = 1; # display only |
|
|
my $i = 0; # index only |
|
67 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
68 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
69 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
70 |
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
|
# parse format |
|
|
my $prefix = ""; |
|
|
if ($format =~ s/^([^\d]+)//) { |
|
|
$prefix = $1; |
|
|
} |
|
|
while ($format) { |
|
|
if ($format =~ s/^(\d\d\d)(\w?)//) { |
|
|
my $isis_tmp = isis_sf($row,$1,$2); |
|
|
if ($isis_tmp) { |
|
|
# $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d); |
|
|
$display_tmp .= $prefix . $isis_tmp if ($d); |
|
|
$swish_tmp .= $isis_tmp." " if ($s); |
|
|
$index_tmp .= $prefix . $isis_tmp if ($i); |
|
|
#print STDERR " $isis_tmp <--\n"; |
|
|
} |
|
|
$prefix = ""; |
|
|
} elsif ($format =~ s/^([^\d]+)//) { |
|
|
$prefix = $1; |
|
|
} else { |
|
|
print STDERR "WARNING: unparsed format '$format'\n"; |
|
|
last; |
|
|
}; |
|
|
} |
|
|
# add suffix |
|
|
$display_tmp .= $prefix if ($display_tmp); |
|
|
$index_tmp .= $prefix if ($index_tmp); |
|
|
|
|
|
# $display_data .= $display_tmp if ($display_tmp ne ""); |
|
|
# $swish_data .= $swish_tmp if ($swish_tmp ne ""); |
|
|
$display_data .= $display_tmp; |
|
|
$swish_data .= $swish_tmp; |
|
|
$index_data .= $index_tmp; |
|
71 |
|
|
72 |
|
$format = $cludge_codepage->convert($format); |
73 |
|
my ($swish,$display) = parse_format($format,$row); |
74 |
|
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
75 |
|
|
76 |
|
#print STDERR "swish: $swish<-- display: $display<--\n"; |
77 |
|
# FIX: this is ugly, UGLY, cludge: OpenIsis return |
78 |
|
# UTF8 encoding of strings, but as if source charset |
79 |
|
# is ISO8859-1 and not some other. This breaks our |
80 |
|
# isis character encoding, so we convert it first |
81 |
|
# back to ISO8859-1 (which can actually be different |
82 |
|
# encoding in isis) |
83 |
|
|
84 |
|
$swish_data .= $swish if ($s && $swish); |
85 |
|
$display_data .= $display if ($d && $display); |
86 |
|
$index_data .= $display if ($i && $display); |
87 |
|
} |
88 |
|
|
89 |
|
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
90 |
|
if ($display_data) { |
91 |
|
$display_data = $isis_codepage->convert($display_data)."##" || $display_data; |
92 |
|
$xml .= xmlify($field."_display", $display_data); |
93 |
|
} |
94 |
|
if ($swish_data) { |
95 |
|
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
96 |
|
$swish_data = $i->convert($swish_data); |
97 |
|
$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data)); |
98 |
|
#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data; |
99 |
|
#$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data)); |
100 |
} |
} |
|
#print "--display:$display_data\n--swish:$swish_data\n"; |
|
|
#$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data); |
|
|
#$xml->{$field."_swish"} = unac_string($config->{isis_codepage},$swish_data) if ($swish_data); |
|
|
$xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data); |
|
|
$xml->{$field."_swish"} = [ unac_string($config->{isis_codepage},$swish_data) ] if ($swish_data); |
|
101 |
|
|
102 |
# index |
# index |
103 |
if ($index_data && $index_data ne "") { |
if ($index_data && $index_data ne "") { |
104 |
my $sql = "select $field from index_$field where upper($field)=upper(?)"; |
$index_data = $index_codepage->convert($index_data) || $index_data; |
105 |
my $sth = $dbh->prepare($sql) || die $dbh->errstr(); |
$index->insert($field, $index_data, $db_dir); |
|
$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr(); |
|
|
#print STDERR "--->$index_data<---\n"; |
|
|
if (! $sth->fetchrow_hashref) { |
|
|
my $sql = "insert into index_$field values (?)"; |
|
|
my $sth = $dbh->prepare($sql) || die $dbh->errstr(); |
|
|
#print STDERR "$sql: $index_data<!----\n"; |
|
|
$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr(); |
|
|
} |
|
106 |
} |
} |
107 |
|
|
108 |
} |
} |
109 |
if ($xml) { |
if ($xml) { |
110 |
return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 ); |
#print STDERR "x: $xml\n"; |
111 |
|
return "<xml>\n$xml</xml>\n"; |
112 |
} else { |
} else { |
113 |
return; |
return; |
114 |
} |
} |
160 |
|
|
161 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
162 |
|
|
163 |
|
my $path = $isis_db; |
164 |
|
$path =~ s#$config->{isis_data}/*##g; |
165 |
|
|
166 |
my $last_p = 0; |
my $last_p = 0; |
167 |
|
|
168 |
# { my $row_id = 1; |
# { my $row_id = 4514; |
169 |
# FIX |
# FIX |
170 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
171 |
my $row = OpenIsis::read( $db, $row_id ); |
my $row = OpenIsis::read( $db, $row_id ); |
172 |
if ($row && $row->{mfn}) { |
if ($row && $row->{mfn}) { |
173 |
|
#print STDERR "mfn: ",$row->{mfn},"\n"; |
174 |
# output current process indicator |
# output current process indicator |
175 |
my $p = int($row->{mfn} * 100 / $max_rowid); |
my $p = int($row->{mfn} * 100 / $max_rowid); |
176 |
if ($p != $last_p) { |
if ($p != $last_p) { |
179 |
} |
} |
180 |
|
|
181 |
if (my $xml = isis2xml($row)) { |
if (my $xml = isis2xml($row)) { |
182 |
my $path = $isis_db; |
#print STDERR "--ret-->$xml\n"; |
183 |
$path =~ s#$config->{isis_data}/*##g; |
print "Path-Name: $path#".int($row->{mfn})."\n"; |
184 |
my $out = "Path-Name: $path#".$row->{mfn}."\n"; |
print "Content-Length: ".(length($xml)+1)."\n"; |
185 |
$out .= "Content-Length: ".(length($xml)+1)."\n"; |
print "Document-Type: XML\n\n$xml\n"; |
|
$out .= "Document-Type: XML\n\n$xml\n"; |
|
|
print $out; |
|
186 |
} |
} |
187 |
} |
} |
188 |
} |
} |
189 |
print STDERR "\n"; |
print STDERR "\n"; |
190 |
} |
} |
191 |
|
|
192 |
$dbh->commit || die $dbh->errstr(); |
# call this to commit index |
193 |
|
$index->close; |
194 |
|
|
195 |
1; |
1; |
196 |
__END__ |
__END__ |