151 |
} |
} |
152 |
|
|
153 |
|
|
154 |
|
# init variables for different types |
155 |
|
sub init_visible_type($) { |
156 |
|
my $type = shift; |
157 |
|
|
158 |
|
# swish, swish_exact, display, index, index_lookup |
159 |
|
# swish and display defaults |
160 |
|
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
161 |
|
if (lc($type) eq "display") { |
162 |
|
$s = 0; |
163 |
|
} elsif (lc($type) eq "swish") { |
164 |
|
$d = 0; |
165 |
|
} elsif (lc($type) eq "index") { |
166 |
|
($s,$se,$d,$i) = (0,1,0,1); |
167 |
|
} elsif (lc($type) eq "swish_exact") { |
168 |
|
($s,$se,$d,$i) = (0,1,0,0); |
169 |
|
} elsif (lc($type) =~ /^lookup/) { |
170 |
|
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
171 |
|
} |
172 |
|
return ($s,$se,$d,$i,$il); |
173 |
|
} |
174 |
|
|
175 |
|
|
176 |
|
# convert |
177 |
|
# |
178 |
|
# <tag> |
179 |
|
# <delimiter>, </delimiter> |
180 |
|
# <value>200a</value> |
181 |
|
# </tag> |
182 |
|
# |
183 |
|
# to |
184 |
|
# |
185 |
|
# <tag delimiter=", ">200a</tag> |
186 |
|
# |
187 |
|
# but without loosing spaces in delimiter (becasue |
188 |
|
# new XML::Simple strips spaces in attribute values |
189 |
|
# as defined in XML specification) |
190 |
|
# |
191 |
|
sub unroll_x($) { |
192 |
|
my $x = shift; |
193 |
|
|
194 |
|
if (defined $x->{value}) { |
195 |
|
my ($v,$d) = ($x->{value}->{content}, $x->{delimiter}->{content}); |
196 |
|
delete $x->{value}; |
197 |
|
delete $x->{delimiter}; |
198 |
|
$x->{content} = $v; |
199 |
|
$x->{delimiter} = $d; |
200 |
|
} |
201 |
|
return $x; |
202 |
|
} |
203 |
|
|
204 |
# begin real work: go field by field |
# begin real work: go field by field |
205 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
206 |
|
|
234 |
|
|
235 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
236 |
|
|
237 |
|
$x = unroll_x($x); |
238 |
|
|
239 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
240 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
241 |
|
|
242 |
my $repeat_off = 0; # init repeatable offset |
my $repeat_off = 0; # init repeatable offset |
243 |
|
|
244 |
# swish, swish_exact, display, index, index_lookup |
my ($s,$se,$d,$i,$il) = init_visible_type($x->{type}); |
|
# swish and display defaults |
|
|
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
|
|
$s = 0 if (lc($x->{type}) eq "display"); |
|
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index"); |
|
|
($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact"); |
|
|
($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/); |
|
245 |
|
|
246 |
# what will separate last line from this one? |
# what will separate last line from this one? |
247 |
if ($display_data && $x->{append}) { |
if ($display_data && $x->{append}) { |
255 |
|
|
256 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
257 |
|
|
|
sub chk_eval($) { |
|
|
my $data = shift; |
|
|
return if (! defined($data)); |
|
|
if ($data && $data =~ s/\s*eval{([^}]+)}\s*//) { |
|
|
if (eval "$1") { |
|
|
return $data; |
|
|
} else { |
|
|
return undef; |
|
|
} |
|
|
} else { |
|
|
return $data; |
|
|
} |
|
|
} |
|
|
|
|
258 |
sub mkformat($$) { |
sub mkformat($$) { |
259 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
260 |
my $data = shift || return; |
my $data = shift || return; |
261 |
my $format_name = x($x->{format_name}) || return chk_eval($data); |
my $format_name = x($x->{format_name}) || return $data; |
262 |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
263 |
my $format_delimiter = x($x->{format_delimiter}); |
my $format_delimiter = x($x->{format_delimiter}); |
264 |
my @data; |
my @data; |
271 |
if ($fmt) { |
if ($fmt) { |
272 |
my $nr = scalar $fmt =~ s/%s/%s/g; |
my $nr = scalar $fmt =~ s/%s/%s/g; |
273 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
274 |
return chk_eval(sprintf($fmt,@data)); |
return sprintf($fmt,@data); |
275 |
} else { |
} else { |
276 |
#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
277 |
return chk_eval($data); |
return $data; |
278 |
} |
} |
279 |
} else { |
} else { |
280 |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
323 |
} |
} |
324 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
325 |
if ($swish) { |
if ($swish) { |
326 |
my $tmp; |
my $tmp = $swish; |
327 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
328 |
no strict 'refs'; |
no strict 'refs'; |
329 |
$tmp = join(" ",&$filter($swish)) if ($s || $se); |
$tmp = join(" ",&$filter($tmp)) if ($s || $se); |
330 |
} |
} |
331 |
|
|
|
$tmp = chk_eval($tmp); |
|
332 |
$swish_data .= $tmp if ($s && $tmp); |
$swish_data .= $tmp if ($s && $tmp); |
333 |
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
334 |
} |
} |
363 |
no strict 'refs'; |
no strict 'refs'; |
364 |
$idisplay = &$filter($idisplay); |
$idisplay = &$filter($idisplay); |
365 |
} |
} |
|
$idisplay = chk_eval($idisplay); |
|
366 |
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
367 |
} |
} |
368 |
|
|
462 |
# now try to parse variables from configuration file |
# now try to parse variables from configuration file |
463 |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
464 |
|
|
465 |
|
$x = unroll_x($x); |
466 |
|
|
467 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
468 |
my $val = $cfg->val($database, x($x->{content})); |
my $val = $cfg->val($database, x($x->{content})); |
469 |
|
|
470 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
# FIXME index_lookup is not supported! |
471 |
$s = 0 if (lc($x->{type}) eq "display"); |
my ($s,$se,$d,$i,$il) = init_visible_type($x->{type}); |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
# no support for swish exact in config. |
|
|
# IMHO, it's useless |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
|
472 |
|
|
473 |
if ($val) { |
if ($val) { |
474 |
$display_data .= $delimiter.$val if ($d); |
$display_data .= $delimiter.$val if ($d); |
475 |
$swish_data .= $val if ($s); |
$swish_data .= " ".$val if ($s); |
476 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
477 |
} |
} |
478 |
|
|
648 |
my $type_base = $type; |
my $type_base = $type; |
649 |
$type_base =~ s/_.+$//g; |
$type_base =~ s/_.+$//g; |
650 |
|
|
651 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
652 |
|
|
653 |
# output current progress indicator |
# output current progress indicator |
654 |
my $last_p = 0; |
my $last_p = 0; |