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 |
use Text::Iconv; |
use Text::Iconv; |
10 |
|
use Config::IniFiles; |
11 |
|
|
12 |
$|=1; |
$|=1; |
13 |
|
|
14 |
my $config; |
my $config_file = $0; |
15 |
|
$config_file =~ s/\.pl$/.conf/; |
16 |
|
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
17 |
|
|
18 |
$config=XMLin(undef, forcearray => [ 'isis' ], forcecontent => 1); |
my $config; |
19 |
|
|
20 |
use index_DBI; # there is no other, right now ;-) |
use index_DBI; # there is no other, right now ;-) |
21 |
|
|
31 |
|
|
32 |
getopts('d:m:qs', \%opts); |
getopts('d:m:qs', \%opts); |
33 |
|
|
34 |
my $db_dir = $opts{d} || "ps"; # FIX |
my $db_dir; |
35 |
|
|
36 |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
37 |
|
|
40 |
|
|
41 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
42 |
|
|
43 |
my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
#my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
44 |
my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
#my $index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
45 |
|
my $isis_codepage; |
46 |
|
my $index_codepage; |
47 |
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
my $cludge_codepage = Text::Iconv->new('UTF8','ISO8859-1'); |
48 |
|
my $xml_codepage; |
49 |
|
|
50 |
sub isis2xml { |
sub isis2xml { |
51 |
|
|
52 |
use xmlify; |
use xmlify; |
53 |
|
|
54 |
my $row = shift @_; |
my $row = shift @_; |
55 |
|
my $add_xml = shift @_; |
56 |
|
|
57 |
my $xml; |
my $xml; |
|
$xml .= xmlify('db_dir',$db_dir); # FIX remove? |
|
58 |
|
|
59 |
use parse_format; |
use parse_format; |
60 |
|
|
61 |
|
my $html = ""; # html formatted display output |
62 |
|
|
63 |
|
my %field_usage; # counter for usage of each field |
64 |
|
|
65 |
foreach my $field (keys %{$config->{indexer}}) { |
foreach my $field (keys %{$config->{indexer}}) { |
66 |
|
|
67 |
|
$field_usage{$field}++; |
68 |
|
|
69 |
my $swish_data = ""; |
my $swish_data = ""; |
70 |
my $display_data = ""; |
my $display_data = ""; |
71 |
my $index_data = ""; |
my $index_data = ""; |
96 |
$index_data .= $display if ($i && $display); |
$index_data .= $display if ($i && $display); |
97 |
} |
} |
98 |
|
|
99 |
|
|
100 |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
101 |
if ($display_data) { |
if ($display_data) { |
102 |
$display_data = $isis_codepage->convert($display_data)."##" || $display_data; |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
103 |
$xml .= xmlify($field."_display", $display_data); |
# FIX: this is removed and replaced by html tag. |
104 |
|
#$xml .= xmlify($field."_display", $display_data); |
105 |
|
|
106 |
|
if ($field eq "headline") { |
107 |
|
$xml .= xmlify("headline", $display_data); |
108 |
|
} else { |
109 |
|
|
110 |
|
# find field name (signular, plural) |
111 |
|
my $field_name = ""; |
112 |
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) { |
113 |
|
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
114 |
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
115 |
|
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
116 |
|
} else { |
117 |
|
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
118 |
|
} |
119 |
|
if ($field_name) { |
120 |
|
$html .= $xml_codepage->convert($field_name); |
121 |
|
} |
122 |
|
$html .= $display_data."###\n"; |
123 |
|
} |
124 |
} |
} |
125 |
if ($swish_data) { |
if ($swish_data) { |
126 |
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
137 |
} |
} |
138 |
|
|
139 |
} |
} |
140 |
|
|
141 |
|
# dump formatted output in <html> |
142 |
|
if ($html) { |
143 |
|
$xml .= xmlify("html",$html); |
144 |
|
} |
145 |
|
|
146 |
if ($xml) { |
if ($xml) { |
147 |
#print STDERR "x: $xml\n"; |
#print STDERR "x: $xml\n"; |
148 |
|
$xml .= $add_xml if ($add_xml); |
149 |
return "<xml>\n$xml</xml>\n"; |
return "<xml>\n$xml</xml>\n"; |
150 |
} else { |
} else { |
151 |
return; |
return; |
154 |
|
|
155 |
########################################################################## |
########################################################################## |
156 |
|
|
157 |
my $last_tell=0; |
my $cfg = new Config::IniFiles( -file => $config_file ); |
|
|
|
|
my @isis_dirs = ( '.' ); # use dirname as database name |
|
|
|
|
|
if ($opts{m}) { |
|
|
@isis_dirs = split(/,/,$opts{m}); |
|
|
} |
|
|
|
|
|
my @isis_dbs; |
|
|
|
|
|
foreach (@isis_dirs) { |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/LIBRI") { |
|
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/LIBRI/LIBRI"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/PERI") { |
|
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/PERI/PERI"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/AMS") { |
|
|
push @isis_dbs,$config->{isis_data}."/$db_dir/$_/AMS/AMS"; |
|
|
} |
|
|
if (-e $config->{isis_data}."/$db_dir/$_/ARTI") { |
|
|
# push @isis_dbs,$config->{isis_data}."/$db_dir/$_/ARTI/ARTI"; |
|
|
} |
|
|
} |
|
|
|
|
|
print STDERR "FATAL: Can't find isis database.\nPerhaps isis_data (".$config->{isis_data}.") has wrong value?\n" if (! @isis_dbs); |
|
|
|
|
|
my $db; |
|
158 |
|
|
159 |
foreach my $isis_db (@isis_dbs) { |
foreach my $database ($cfg->Sections) { |
160 |
|
|
161 |
|
my $isis_db = $cfg -> val($database, 'isis_db'); |
162 |
|
my $type = $cfg -> val($database, 'type'); |
163 |
|
my $add_xml = $cfg -> val($database, 'xml'); |
164 |
|
|
165 |
|
# read configuration for this type |
166 |
|
$config=XMLin("./import_xml/$type.xml", forcearray => [ 'isis' ], forcecontent => 1); |
167 |
|
$isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
168 |
|
$index_codepage = Text::Iconv->new($config->{isis_codepage},$config->{index_codepage}); |
169 |
|
$xml_codepage = Text::Iconv->new($cfg->val($database,'xml_codepage'),'UTF8'); |
170 |
|
|
171 |
my $db = OpenIsis::open( $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
172 |
if (0) { |
if (0) { |
180 |
|
|
181 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
182 |
|
|
183 |
my $path = $isis_db; |
my $path = $database; # was $isis_db |
|
$path =~ s#$config->{isis_data}/*##g; |
|
184 |
|
|
185 |
my $last_p = 0; |
my $last_p = 0; |
186 |
|
|
197 |
$last_p = $p; |
$last_p = $p; |
198 |
} |
} |
199 |
|
|
200 |
if (my $xml = isis2xml($row)) { |
if (my $xml = isis2xml($row,$add_xml)) { |
201 |
#print STDERR "--ret-->$xml\n"; |
#print STDERR "--ret-->$xml\n"; |
202 |
print "Path-Name: $path#".int($row->{mfn})."\n"; |
print "Path-Name: $path#".int($row->{mfn})."\n"; |
203 |
print "Content-Length: ".(length($xml)+1)."\n"; |
print "Content-Length: ".(length($xml)+1)."\n"; |