96 |
$field_usage{$field}++; |
$field_usage{$field}++; |
97 |
|
|
98 |
my $swish_data = ""; |
my $swish_data = ""; |
99 |
|
my $swish_exact_data = ""; |
100 |
my $display_data = ""; |
my $display_data = ""; |
101 |
my $line_delimiter; |
my $line_delimiter; |
102 |
|
|
110 |
|
|
111 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # repeatable offset |
112 |
|
|
113 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default |
114 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
115 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
116 |
|
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
117 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
118 |
|
|
119 |
# what will separate last line from this one? |
# what will separate last line from this one? |
130 |
my @index_data; |
my @index_data; |
131 |
my $index_filter; |
my $index_filter; |
132 |
|
|
133 |
|
sub mkformat { |
134 |
|
my $x = shift || die "mkformat needs tag reference"; |
135 |
|
my $data = shift || return; |
136 |
|
my $format_name = x($x->{format_name}) || return $data; |
137 |
|
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
138 |
|
my $format_delimiter = x($x->{format_delimiter}); |
139 |
|
my @data; |
140 |
|
if ($format_delimiter) { |
141 |
|
@data = split(/$format_delimiter/,$data); |
142 |
|
} else { |
143 |
|
push @data,$data; |
144 |
|
} |
145 |
|
|
146 |
|
if ($fmt) { |
147 |
|
my $nr = scalar $fmt =~ s/%s/%s/g; |
148 |
|
if (($#data+1) == $nr) { |
149 |
|
return sprintf($fmt,@data); |
150 |
|
} else { |
151 |
|
print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
152 |
|
return $data; |
153 |
|
} |
154 |
|
} else { |
155 |
|
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
156 |
|
} |
157 |
|
} |
158 |
|
|
159 |
# while because of repeatable fields |
# while because of repeatable fields |
160 |
while ($swish || $display) { |
while ($swish || $display) { |
161 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
171 |
require "filter/".$filter.".pm"; |
require "filter/".$filter.".pm"; |
172 |
} |
} |
173 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
174 |
if ($s && $swish) { |
if ($swish) { |
175 |
if ($filter) { |
if ($filter && ($s || $se)) { |
176 |
no strict 'refs'; |
no strict 'refs'; |
177 |
$swish_data .= join(" ",&$filter($swish)); |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
178 |
|
$swish_data .= $tmp if ($s); |
179 |
|
$swish_exact_data .= $tmp if ($se); |
180 |
|
|
181 |
} else { |
} else { |
182 |
$swish_data .= $swish; |
$swish_data .= $swish if ($s); |
183 |
|
$swish_exact_data .= $swish if ($se); |
184 |
} |
} |
185 |
} |
} |
186 |
|
|
192 |
} |
} |
193 |
if ($filter) { |
if ($filter) { |
194 |
no strict 'refs'; |
no strict 'refs'; |
195 |
$display_data .= $delimiter.&$filter($display); |
if ($display_data) { |
196 |
|
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
197 |
|
} else { |
198 |
|
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
199 |
|
} |
200 |
} else { |
} else { |
201 |
$display_data .= $delimiter.$display; |
if ($display_data) { |
202 |
|
$display_data .= $delimiter.mkformat($x,$display); |
203 |
|
} else { |
204 |
|
$display_data = mkformat($x,$display); |
205 |
|
} |
206 |
} |
} |
207 |
} |
} |
208 |
|
|
237 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$d,$i) = (1,1,0); # swish, display default |
238 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
239 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
240 |
|
# no support for swish exact in config. |
241 |
|
# IMHO, it's useless |
242 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
243 |
|
|
244 |
if ($val) { |
if ($val) { |
281 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
282 |
} |
} |
283 |
|
|
284 |
|
if ($swish_exact_data) { |
285 |
|
$swish_exact_data =~ s/ +/ /g; |
286 |
|
$swish_exact_data =~ s/ +$//g; |
287 |
|
|
288 |
|
# add delimiters before and after word. |
289 |
|
# That is required to produce exact match |
290 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
291 |
|
} |
292 |
|
|
293 |
|
|
294 |
} |
} |
295 |
|
|
336 |
my $type_base = $type; |
my $type_base = $type; |
337 |
$type_base =~ s/_.+$//g; |
$type_base =~ s/_.+$//g; |
338 |
|
|
339 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); |
340 |
|
|
341 |
# output current progress indicator |
# output current progress indicator |
342 |
my $last_p = 0; |
my $last_p = 0; |