5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
8 |
use Text::Unaccent; |
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
9 |
require Unicode::Map8; |
require Unicode::Map8; |
10 |
|
use DBI; |
11 |
|
|
12 |
my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
my $config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
13 |
|
my $dbh = DBI->connect("DBI:Pg:dbname=webpac","","") || die $DBI::errstr; # FIX |
14 |
|
# FIX; select relname from pg_class where relname like 'index_%' ; |
15 |
|
$dbh->begin_work || die $dbh->errstr(); |
16 |
|
|
17 |
|
$dbh->do("delete from index_author") || die $dbh->errstr(); |
18 |
|
$dbh->do("delete from index_title") || die $dbh->errstr(); |
19 |
|
|
20 |
my %opts; |
my %opts; |
21 |
|
|
22 |
getopts('d:m:q', \%opts); |
# usage: |
23 |
|
# -d directory name |
24 |
|
# -m multiple directories |
25 |
|
# -q quiet |
26 |
|
# -s run swish |
27 |
|
|
28 |
|
getopts('d:m:qs', \%opts); |
29 |
|
|
30 |
my $db_dir = $opts{d} || "ps"; # FIX |
my $db_dir = $opts{d} || "ps"; # FIX |
31 |
|
|
42 |
my $row = shift @_; |
my $row = shift @_; |
43 |
|
|
44 |
my $xml; |
my $xml; |
45 |
|
$xml->{db_dir} = [ $db_dir ]; # FIX remove? |
46 |
|
|
47 |
sub isis_sf { |
sub isis_sf { |
48 |
my $row = shift @_; |
my $row = shift @_; |
64 |
foreach my $field (keys %{$config->{indexer}}) { |
foreach my $field (keys %{$config->{indexer}}) { |
65 |
|
|
66 |
my $display_data = ""; |
my $display_data = ""; |
67 |
|
my $swish_data = ""; |
68 |
my $index_data = ""; |
my $index_data = ""; |
69 |
|
|
70 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
71 |
|
|
72 |
my $display_tmp = ""; |
my $display_tmp = ""; |
73 |
|
my $swish_tmp = ""; |
74 |
my $index_tmp = ""; |
my $index_tmp = ""; |
75 |
|
|
76 |
my $format = $x->{content}; |
my $format = $x->{content}; |
77 |
my $i = 1; # index only |
my $s = 1; # swish only |
78 |
my $d = 1; # display only |
my $d = 1; # display only |
79 |
$i = 0 if (lc($x->{type}) eq "display"); |
my $i = 0; # index only |
80 |
$d = 0 if (lc($x->{type}) eq "index"); |
$s = 0 if (lc($x->{type}) eq "display"); |
81 |
#print "## i: $i d: $d ## $format ##"; |
$d = 0 if (lc($x->{type}) eq "swish"); |
82 |
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
83 |
|
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
84 |
# parse format |
# parse format |
85 |
my $prefix = ""; |
my $prefix = ""; |
86 |
if ($format =~ s/^([^\d]+)//) { |
if ($format =~ s/^([^\d]+)//) { |
92 |
if ($isis_tmp) { |
if ($isis_tmp) { |
93 |
# $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d); |
# $display_tmp .= $prefix . "/$1/$2/".$isis_tmp if ($d); |
94 |
$display_tmp .= $prefix . $isis_tmp if ($d); |
$display_tmp .= $prefix . $isis_tmp if ($d); |
95 |
$index_tmp .= $isis_tmp." " if ($i); |
$swish_tmp .= $isis_tmp." " if ($s); |
96 |
#print " $isis_tmp <--\n"; |
$index_tmp .= $prefix . $isis_tmp if ($i); |
97 |
|
#print STDERR " $isis_tmp <--\n"; |
98 |
} |
} |
99 |
$prefix = ""; |
$prefix = ""; |
100 |
} elsif ($format =~ s/^([^\d]+)//) { |
} elsif ($format =~ s/^([^\d]+)//) { |
106 |
} |
} |
107 |
# add suffix |
# add suffix |
108 |
$display_tmp .= $prefix if ($display_tmp); |
$display_tmp .= $prefix if ($display_tmp); |
109 |
|
$index_tmp .= $prefix if ($index_tmp); |
110 |
|
|
111 |
# $display_data .= $display_tmp if ($display_tmp ne ""); |
# $display_data .= $display_tmp if ($display_tmp ne ""); |
112 |
# $index_data .= $index_tmp if ($index_tmp ne ""); |
# $swish_data .= $swish_tmp if ($swish_tmp ne ""); |
113 |
$display_data .= $display_tmp; |
$display_data .= $display_tmp; |
114 |
|
$swish_data .= $swish_tmp; |
115 |
$index_data .= $index_tmp; |
$index_data .= $index_tmp; |
116 |
|
|
117 |
} |
} |
118 |
#print "--display:$display_data\n--index:$index_data\n"; |
#print "--display:$display_data\n--swish:$swish_data\n"; |
119 |
$xml->{$field}->{display} .= $isis_map->tou($display_data)->utf8 if ($display_data); |
#$xml->{$field."_display"} = $isis_map->tou($display_data)->utf8 if ($display_data); |
120 |
$xml->{$field}->{index} .= unac_string($config->{isis_codepage},$index_data) if ($index_data); |
#$xml->{$field."_swish"} = unac_string($config->{isis_codepage},$swish_data) if ($swish_data); |
121 |
|
$xml->{$field."_display" } = [ $isis_map->tou($display_data)->utf8 ] if ($display_data); |
122 |
|
$xml->{$field."_swish"} = [ unac_string($config->{isis_codepage},$swish_data) ] if ($swish_data); |
123 |
|
|
124 |
|
# index |
125 |
|
if ($index_data && $index_data ne "") { |
126 |
|
my $sql = "select $field from index_$field where upper($field)=upper(?)"; |
127 |
|
my $sth = $dbh->prepare($sql) || die $dbh->errstr(); |
128 |
|
$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr(); |
129 |
|
#print STDERR "--->$index_data<---\n"; |
130 |
|
if (! $sth->fetchrow_hashref) { |
131 |
|
my $sql = "insert into index_$field values (?)"; |
132 |
|
my $sth = $dbh->prepare($sql) || die $dbh->errstr(); |
133 |
|
#print STDERR "$sql: $index_data<!----\n"; |
134 |
|
$sth->execute($index_data) || die "SQL: $sql; ".$dbh->errstr(); |
135 |
|
} |
136 |
|
} |
137 |
|
|
138 |
} |
} |
139 |
if ($xml) { |
if ($xml) { |
140 |
return XMLout($xml, rootname => 'xml', noattr => 1 ); |
return XMLout($xml, rootname => 'xml', keeproot => 0, noattr => 0 ); |
141 |
} else { |
} else { |
142 |
return; |
return; |
143 |
} |
} |
205 |
} |
} |
206 |
|
|
207 |
if (my $xml = isis2xml($row)) { |
if (my $xml = isis2xml($row)) { |
208 |
print "Path-Name: $isis_db#".$row->{mfn}."\n"; |
my $path = $isis_db; |
209 |
print "Content-Length: ".(length($xml)+1)."\n"; |
$path =~ s#$config->{isis_data}/*##g; |
210 |
print "Document-Type: XML\n\n$xml\n"; |
my $out = "Path-Name: $path#".$row->{mfn}."\n"; |
211 |
|
$out .= "Content-Length: ".(length($xml)+1)."\n"; |
212 |
|
$out .= "Document-Type: XML\n\n$xml\n"; |
213 |
|
print $out; |
214 |
} |
} |
215 |
} |
} |
216 |
} |
} |
217 |
print STDERR "\n"; |
print STDERR "\n"; |
218 |
} |
} |
219 |
|
|
220 |
|
$dbh->commit || die $dbh->errstr(); |
221 |
|
|
222 |
1; |
1; |
223 |
__END__ |
__END__ |