10 |
use SWISH::API; |
use SWISH::API; |
11 |
use Text::Iconv; |
use Text::Iconv; |
12 |
use Data::Dumper; |
use Data::Dumper; |
13 |
|
use File::Temp qw/ :mktemp /; |
14 |
|
|
15 |
my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8'); |
my $iso2utf = Text::Iconv->new('ISO-8859-2','UTF-8'); |
16 |
my $utf2iso = Text::Iconv->new('UTF-8','ISO-8859-2'); |
my $utf2iso = Text::Iconv->new('UTF-8','ISO-8859-2'); |
22 |
|
|
23 |
if (! $swish) { |
if (! $swish) { |
24 |
|
|
25 |
my $index_file = $self->{index_file} || croak "open_index needs index filename"; |
my $index_file = $self->{index_dir}."/".$self->{config_name}; |
|
$index_file .= "/swish-e"; |
|
26 |
print STDERR "opening index '$index_file'\n"; |
print STDERR "opening index '$index_file'\n"; |
27 |
$swish = SWISH::API->new($index_file); |
$swish = SWISH::API->new($index_file); |
28 |
$swish->AbortLastError if $swish->Error; |
$swish->AbortLastError if $swish->Error; |
119 |
return @res_ids; |
return @res_ids; |
120 |
} |
} |
121 |
|
|
122 |
|
# this function can be null for indexes which doesn't need special |
123 |
|
# setup before add_index is called. however, swish-e support will |
124 |
|
# fork swish binary to create index at this point |
125 |
|
sub create_index { |
126 |
|
my $self = shift; |
127 |
|
|
128 |
|
my $index_prog = $0 || die "can't deduce my own name!"; |
129 |
|
my $config_file = $self->{config_file} || die "no self->config_file"; |
130 |
|
my $index_file = $self->{index_dir} || die "no self->index_dir"; |
131 |
|
$index_file .= "/"; |
132 |
|
$index_file .= $self->{config_name} || die "no self->config_name"; |
133 |
|
|
134 |
|
my ($tmp_fh, $swish_config_file) = mkstemp("/tmp/swishXXXXX"); |
135 |
|
|
136 |
|
print STDERR "creating swish-e configuration file $swish_config_file\n"; |
137 |
|
|
138 |
|
my $swish_config = qq{ |
139 |
|
# swish-e config file |
140 |
|
|
141 |
|
IndexDir $index_prog |
142 |
|
SwishProgParameters --recursive $config_file |
143 |
|
|
144 |
|
# input file definition |
145 |
|
DefaultContents XML2 |
146 |
|
|
147 |
|
# indexed metatags |
148 |
|
MetaNames xml swishdocpath |
149 |
|
|
150 |
|
# stored metatags |
151 |
|
PropertyNames from_phrase from_address |
152 |
|
PropertyNames to_phrase to_address |
153 |
|
PropertyNames cc_phrase cc_address |
154 |
|
PropertyNames subject body |
155 |
|
#PropertyNamesDate date |
156 |
|
PropertyNamesNumeric date_utime |
157 |
|
PropertyNames date |
158 |
|
|
159 |
|
#XMLClassAttributes type |
160 |
|
UndefinedMetaTags auto |
161 |
|
UndefinedXMLAttributes auto |
162 |
|
|
163 |
|
IndexFile $index_file |
164 |
|
|
165 |
|
# Croatian ISO-8859-2 characters to unaccented equivalents |
166 |
|
TranslateCharacters ¹©ðÐèÈæƾ® ssddcccczz |
167 |
|
|
168 |
|
|
169 |
|
# debug |
170 |
|
ParserWarnLevel 3 |
171 |
|
IndexReport 1 |
172 |
|
|
173 |
|
}; |
174 |
|
|
175 |
|
print $tmp_fh $swish_config; |
176 |
|
close($tmp_fh); |
177 |
|
|
178 |
|
exec "swish-e -S prog -c $swish_config_file" || die "can't fork swish with $swish_config_file"; |
179 |
|
exit 0; |
180 |
|
|
181 |
|
} |
182 |
|
|
183 |
sub add_index { |
sub add_index { |
184 |
my $self = shift; |
my $self = shift; |
185 |
|
|