1 |
#!/usr/bin/perl -w |
2 |
|
3 |
use strict; |
4 |
use OpenIsis; |
5 |
use Getopt::Std; |
6 |
use Data::Dumper; |
7 |
use XML::Simple; |
8 |
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
9 |
use Text::Iconv; |
10 |
use Config::IniFiles; |
11 |
|
12 |
$|=1; |
13 |
|
14 |
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 |
my $config; |
19 |
|
20 |
use index_DBI; # there is no other, right now ;-) |
21 |
|
22 |
my $index = new index_DBI(); # open index |
23 |
|
24 |
my %opts; |
25 |
|
26 |
# usage: |
27 |
# -d directory name |
28 |
# -m multiple directories |
29 |
# -q quiet |
30 |
# -s run swish |
31 |
|
32 |
getopts('d:m:qs', \%opts); |
33 |
|
34 |
my $db_dir; |
35 |
|
36 |
#die "usage: $0 -d [database_dir] -m [database1,database2] " if (! %opts); |
37 |
|
38 |
#print Dumper($config->{indexer}); |
39 |
#print "-" x 70,"\n"; |
40 |
|
41 |
Text::Iconv->raise_error(1); # Conversion errors raise exceptions |
42 |
|
43 |
#my $isis_codepage = Text::Iconv->new($config->{isis_codepage},'UTF8'); |
44 |
#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'); |
48 |
my $xml_codepage; |
49 |
|
50 |
sub isis2xml { |
51 |
|
52 |
use xmlify; |
53 |
|
54 |
my $row = shift @_; |
55 |
my $add_xml = shift @_; |
56 |
|
57 |
my $xml; |
58 |
|
59 |
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}}) { |
66 |
|
67 |
$field_usage{$field}++; |
68 |
|
69 |
my $swish_data = ""; |
70 |
my $display_data = ""; |
71 |
my $index_data = ""; |
72 |
|
73 |
foreach my $x (@{$config->{indexer}->{$field}->{isis}}) { |
74 |
|
75 |
my $format = $x->{content}; |
76 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
77 |
$s = 0 if (lc($x->{type}) eq "display"); |
78 |
$d = 0 if (lc($x->{type}) eq "swish"); |
79 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
80 |
#print STDERR "## s: $s d: $d i: $i ## $format ##\n"; |
81 |
|
82 |
$format = $cludge_codepage->convert($format); |
83 |
my ($swish,$display) = parse_format($format,$row); |
84 |
#print STDERR "s: $swish\nd: $display\n" if ($swish); |
85 |
|
86 |
#print STDERR "swish: $swish<-- display: $display<--\n"; |
87 |
# FIX: this is ugly, UGLY, cludge: OpenIsis return |
88 |
# UTF8 encoding of strings, but as if source charset |
89 |
# is ISO8859-1 and not some other. This breaks our |
90 |
# isis character encoding, so we convert it first |
91 |
# back to ISO8859-1 (which can actually be different |
92 |
# encoding in isis) |
93 |
|
94 |
$swish_data .= $swish if ($s && $swish); |
95 |
$display_data .= $display if ($d && $display); |
96 |
$index_data .= $display if ($i && $display); |
97 |
} |
98 |
|
99 |
|
100 |
#print STDERR "s_d: $swish_data\nd_d: $display_data\n" if ($swish_data); |
101 |
if ($display_data) { |
102 |
$display_data = $isis_codepage->convert($display_data) || die "Can't convert '$display_data' !"; |
103 |
# FIX: this is removed and replaced by html tag. |
104 |
#$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) { |
126 |
my $i = Text::Iconv->new($config->{isis_codepage},'ISO8859-2'); |
127 |
$swish_data = $i->convert($swish_data); |
128 |
$xml .= xmlify($field."_swish",unac_string('ISO8859-2',$swish_data)); |
129 |
#$swish_data = $isis_codepage->convert($swish_data)."##" || $swish_data; |
130 |
#$xml .= xmlify($field."_swish",unac_string($config->{isis_codepage},$swish_data)); |
131 |
} |
132 |
|
133 |
# index |
134 |
if ($index_data && $index_data ne "") { |
135 |
$index_data = $index_codepage->convert($index_data) || $index_data; |
136 |
$index->insert($field, $index_data, $db_dir); |
137 |
} |
138 |
|
139 |
} |
140 |
|
141 |
# dump formatted output in <html> |
142 |
if ($html) { |
143 |
$xml .= xmlify("html",$html); |
144 |
} |
145 |
|
146 |
if ($xml) { |
147 |
#print STDERR "x: $xml\n"; |
148 |
$xml .= $add_xml if ($add_xml); |
149 |
return "<xml>\n$xml</xml>\n"; |
150 |
} else { |
151 |
return; |
152 |
} |
153 |
} |
154 |
|
155 |
########################################################################## |
156 |
|
157 |
my $cfg = new Config::IniFiles( -file => $config_file ); |
158 |
|
159 |
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 ); |
172 |
if (0) { |
173 |
# # FIX |
174 |
# if (! $db ) { |
175 |
print STDERR "WARNING: can't open '$isis_db'\n"; |
176 |
next ; |
177 |
} |
178 |
|
179 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
180 |
|
181 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
182 |
|
183 |
my $path = $database; # was $isis_db |
184 |
|
185 |
my $last_p = 0; |
186 |
|
187 |
# { my $row_id = 4514; |
188 |
# FIX |
189 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
190 |
my $row = OpenIsis::read( $db, $row_id ); |
191 |
if ($row && $row->{mfn}) { |
192 |
#print STDERR "mfn: ",$row->{mfn},"\n"; |
193 |
# output current process indicator |
194 |
my $p = int($row->{mfn} * 100 / $max_rowid); |
195 |
if ($p != $last_p) { |
196 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$row->{mfn},$max_rowid,"=" x ($p/2).">", $p ) if (! $opts{q}); |
197 |
$last_p = $p; |
198 |
} |
199 |
|
200 |
if (my $xml = isis2xml($row,$add_xml)) { |
201 |
#print STDERR "--ret-->$xml\n"; |
202 |
print "Path-Name: $path#".int($row->{mfn})."\n"; |
203 |
print "Content-Length: ".(length($xml)+1)."\n"; |
204 |
print "Document-Type: XML\n\n$xml\n"; |
205 |
} |
206 |
} |
207 |
} |
208 |
print STDERR "\n"; |
209 |
} |
210 |
|
211 |
# call this to commit index |
212 |
$index->close; |
213 |
|
214 |
1; |
215 |
__END__ |
216 |
########################################################################## |
217 |
|
218 |
=head1 NAME |
219 |
|
220 |
isis2xml.pl - read isis file and dump XML |
221 |
|
222 |
=head1 DESCRIPTION |
223 |
|
224 |
This command will read ISIS data file using OpenIsis perl module and |
225 |
create XML file for usage with I<SWISH-E> |
226 |
indexer. Dispite it's name, this script B<isn't general xml generator> |
227 |
from isis files (isis allready has something like that). Output of this |
228 |
script is tailor-made for SWISH-E. |
229 |
|
230 |
=head1 AUTHOR |
231 |
|
232 |
Dobrica Pavlinusic <dpavlin@rot13.org> |
233 |
|
234 |
=head1 COPYRIGHT |
235 |
|
236 |
GNU Public License (GPL) v2 or later |
237 |
|
238 |
=head1 SEE ALSO |
239 |
|
240 |
SWISH-E web site at http://www.swish-e.org |
241 |
|
242 |
=cut |