5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
|
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
|
8 |
use Text::Iconv; |
use Text::Iconv; |
9 |
use Config::IniFiles; |
use Config::IniFiles; |
10 |
use Encode; |
use Encode; |
16 |
|
|
17 |
my $config_file = $0; |
my $config_file = $0; |
18 |
$config_file =~ s/\.pl$/.conf/; |
$config_file =~ s/\.pl$/.conf/; |
19 |
|
$config_file = $ARGV[0] if ($ARGV[0] && -f $ARGV[0]); |
20 |
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
21 |
|
|
22 |
my $config; |
my $config; |
72 |
# if you are tight on memory, turn this off |
# if you are tight on memory, turn this off |
73 |
my $use_lhash_cache = 1; |
my $use_lhash_cache = 1; |
74 |
|
|
75 |
|
my $last_field_name; # cache to prevent repeated fields |
76 |
|
|
77 |
sub data2xml { |
sub data2xml { |
78 |
|
|
79 |
use xmlify; |
use xmlify; |
119 |
delete $cache->{swish_data}; |
delete $cache->{swish_data}; |
120 |
delete $cache->{swish_exact_data}; |
delete $cache->{swish_exact_data}; |
121 |
delete $cache->{index_data}; |
delete $cache->{index_data}; |
122 |
|
delete $cache->{index_delimiter}; |
123 |
my @page_fields; # names of fields |
my @page_fields; # names of fields |
124 |
|
|
125 |
|
|
140 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
141 |
} |
} |
142 |
if ($field_name) { |
if ($field_name) { |
143 |
return x($field_name); |
if (! $last_field_name) { |
144 |
|
$last_field_name = x($field_name); |
145 |
|
return $last_field_name; |
146 |
|
} elsif ($field_name ne $last_field_name) { |
147 |
|
$last_field_name = x($field_name); |
148 |
|
return $last_field_name; |
149 |
|
} |
150 |
} |
} |
151 |
} |
} |
152 |
|
|
153 |
|
|
154 |
|
# init variables for different types |
155 |
|
sub init_visible_type($) { |
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,$il) = (0,1,0,0,1); |
171 |
|
} elsif ($type) { |
172 |
|
print STDERR "WARNING: unknown type: $type\n"; |
173 |
|
} |
174 |
|
return ($s,$se,$d,$i,$il); |
175 |
|
} |
176 |
|
|
177 |
|
|
178 |
|
# convert |
179 |
|
# |
180 |
|
# <tag> |
181 |
|
# <delimiter>, </delimiter> |
182 |
|
# <value>200a</value> |
183 |
|
# </tag> |
184 |
|
# |
185 |
|
# to |
186 |
|
# |
187 |
|
# <tag delimiter=", ">200a</tag> |
188 |
|
# |
189 |
|
# but without loosing spaces in delimiter (becasue |
190 |
|
# new XML::Simple strips spaces in attribute values |
191 |
|
# as defined in XML specification) |
192 |
|
# |
193 |
|
sub unroll_x($) { |
194 |
|
my $x = shift; |
195 |
|
|
196 |
|
if (defined $x->{value}) { |
197 |
|
my ($v,$d) = ($x->{value}->{content}, $x->{delimiter}->{content}); |
198 |
|
delete $x->{value}; |
199 |
|
delete $x->{delimiter}; |
200 |
|
$x->{content} = $v; |
201 |
|
$x->{delimiter} = $d; |
202 |
|
} |
203 |
|
return $x; |
204 |
|
} |
205 |
|
|
206 |
# begin real work: go field by field |
# begin real work: go field by field |
207 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
208 |
|
|
225 |
my %page_max = (); |
my %page_max = (); |
226 |
# default line_delimiter if using |
# default line_delimiter if using |
227 |
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
228 |
|
$cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter}; |
229 |
|
|
230 |
|
my $format_name = $config->{indexer}->{$field}->{format_name}; |
231 |
|
my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter}; |
232 |
|
if ($format_name && $format_delimiter) { |
233 |
|
$cache->{format}->{$field}->{format_name} = $format_name; |
234 |
|
$cache->{format}->{$field}->{format_delimiter} = $format_delimiter; |
235 |
|
} |
236 |
|
|
237 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
238 |
|
|
239 |
|
$x = unroll_x($x); |
240 |
|
|
241 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
242 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
243 |
|
|
244 |
my $repeat_off = 0; # init repeatable offset |
my $repeat_off = 0; # init repeatable offset |
245 |
|
|
246 |
# swish, swish_exact, display, index, index_lookup |
my ($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) = (0,0,0,1) 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/); |
|
247 |
|
|
248 |
# what will separate last line from this one? |
# what will separate last line from this one? |
249 |
if ($display_data && $x->{append}) { |
if ($display_data && $x->{append}) { |
250 |
$line_delimiter = ' '; |
$line_delimiter = $delimiter; |
251 |
} elsif ($display_data) { |
} elsif ($display_data) { |
252 |
$line_delimiter = '<br/>'; |
$line_delimiter = '<br/>'; |
253 |
} |
} |
257 |
|
|
258 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
259 |
|
|
260 |
sub mkformat { |
sub mkformat($$) { |
261 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
262 |
my $data = shift || return; |
my $data = shift || return; |
263 |
my $format_name = x($x->{format_name}) || return $data; |
my $format_name = x($x->{format_name}) || return $data; |
275 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
276 |
return sprintf($fmt,@data); |
return sprintf($fmt,@data); |
277 |
} else { |
} else { |
278 |
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"; |
279 |
return $data; |
return $data; |
280 |
} |
} |
281 |
} else { |
} else { |
304 |
$display = $new_display; |
$display = $new_display; |
305 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
306 |
} else { |
} else { |
307 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
# print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
308 |
$display = ""; |
$display = ""; |
309 |
$cache->{lhash}->{$display} = $null; |
$cache->{lhash}->{$display} = $null; |
310 |
} |
} |
325 |
} |
} |
326 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
327 |
if ($swish) { |
if ($swish) { |
328 |
|
my $tmp = $swish; |
329 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
330 |
no strict 'refs'; |
no strict 'refs'; |
331 |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
$tmp = join(" ",&$filter($tmp)) if ($s || $se); |
|
$swish_data .= $tmp if ($s); |
|
|
$swish_exact_data .= $tmp if ($se); |
|
|
|
|
|
} else { |
|
|
$swish_data .= $swish if ($s); |
|
|
$swish_exact_data .= $swish if ($se); |
|
332 |
} |
} |
333 |
|
|
334 |
|
$swish_data .= $tmp if ($s && $tmp); |
335 |
|
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
336 |
} |
} |
337 |
|
|
338 |
# type="display" ; field for display |
# type="display" ; field for display |
339 |
if ($d && $display) { |
if ($d && $display) { |
340 |
|
my $ldel = $delimiter; |
341 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
342 |
$display_data .= $line_delimiter; |
$ldel = $line_delimiter; |
343 |
} |
} |
344 |
if ($filter) { |
if ($filter) { |
345 |
no strict 'refs'; |
no strict 'refs'; |
346 |
if ($display_data) { |
my @arr; |
347 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
foreach my $tmp (&$filter($display)) { |
348 |
} else { |
my $tmp2 = mkformat($x,$tmp); |
349 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
push @arr,$tmp2 if ($tmp2); |
350 |
} |
} |
351 |
|
$display_data .= $ldel if ($display_data && @arr); |
352 |
|
$display_data .= join($delimiter,@arr); |
353 |
} else { |
} else { |
354 |
if ($display_data) { |
$display_data .= $ldel if ($display_data); |
355 |
$display_data .= $delimiter.mkformat($x,$display); |
my $tmp = mkformat($x,$display); |
356 |
} else { |
$display_data .= $tmp if ($tmp); |
|
$display_data = mkformat($x,$display); |
|
|
} |
|
357 |
} |
} |
358 |
} |
} |
359 |
|
|
365 |
no strict 'refs'; |
no strict 'refs'; |
366 |
$idisplay = &$filter($idisplay); |
$idisplay = &$filter($idisplay); |
367 |
} |
} |
368 |
push @index_data, $idisplay if (! $iterate_by_page); |
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
369 |
} |
} |
370 |
|
|
371 |
# store fields in lookup |
# store fields in lookup |
374 |
if ($lookup_key) { |
if ($lookup_key) { |
375 |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
376 |
} else { |
} else { |
377 |
$lookup_key = $display; |
if ($filter) { |
378 |
|
no strict 'refs'; |
379 |
|
$lookup_key = &$filter($display); |
380 |
|
} else { |
381 |
|
$lookup_key = $display; |
382 |
|
} |
383 |
} |
} |
384 |
} elsif (lc($x->{type}) eq "lookup_val") { |
} elsif (lc($x->{type}) eq "lookup_val") { |
385 |
if ($lookup_key) { |
if ($lookup_key) { |
386 |
$lhash{$lookup_key} = $display; |
if ($filter) { |
387 |
|
no strict 'refs'; |
388 |
|
$lhash{$lookup_key} = &$filter($display); |
389 |
|
} else { |
390 |
|
$lhash{$lookup_key} = $display; |
391 |
|
} |
392 |
} else { |
} else { |
393 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
394 |
} |
} |
433 |
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
434 |
} |
} |
435 |
if ($x->{append}) { |
if ($x->{append}) { |
436 |
$index_data[$#index_data] .= $idisplay; |
if (@index_data) { |
437 |
|
$index_data[$#index_data] .= $idisplay; |
438 |
|
} else { |
439 |
|
push @index_data, $idisplay; |
440 |
|
} |
441 |
} else { |
} else { |
442 |
push @index_data, $idisplay; |
push @index_data, $idisplay; |
443 |
} |
} |
448 |
} |
} |
449 |
|
|
450 |
if (! $iterate_by_page) { |
if (! $iterate_by_page) { |
451 |
|
my $idel = $x->{index_delimiter}; |
452 |
# fill data in index |
# fill data in index |
453 |
foreach my $d (@index_data) { |
foreach my $tmp (@index_data) { |
454 |
$index->insert($field, $d, $path); |
my $i = $d = $tmp; |
455 |
|
if ($idel && $tmp =~ m/$idel/) { |
456 |
|
($i,$d) = split(/$idel/,$tmp); |
457 |
|
} |
458 |
|
$index->insert($field, $i, $d, $path); |
459 |
} |
} |
460 |
@index_data = (); |
@index_data = (); |
461 |
} |
} |
464 |
# now try to parse variables from configuration file |
# now try to parse variables from configuration file |
465 |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{'config'}}) { |
466 |
|
|
467 |
|
$x = unroll_x($x); |
468 |
|
|
469 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
470 |
my $val = $cfg->val($database, x($x->{content})); |
my $val = $cfg->val($database, x($x->{content})); |
471 |
|
|
472 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
# FIXME index_lookup is not supported! |
473 |
$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"); |
|
474 |
|
|
475 |
if ($val) { |
if ($val) { |
476 |
$display_data .= $delimiter.$val if ($d); |
$display_data .= $delimiter.$val if ($d); |
477 |
$swish_data .= $val if ($s); |
$swish_data .= " ".$val if ($s); |
478 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
479 |
} |
} |
480 |
|
|
503 |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
504 |
#print STDERR Dumper($cache->{display_data}); |
#print STDERR Dumper($cache->{display_data}); |
505 |
for (my $page=0; $page <= $nr_pages; $page++) { |
for (my $page=0; $page <= $nr_pages; $page++) { |
506 |
|
my $display_data; |
507 |
my $display_data = $cache->{display_data}->{$field}->[$page]; |
if ($cache->{format}->{$field}) { |
508 |
|
my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
509 |
|
$display_data=$tmp if ($tmp); |
510 |
|
} else { |
511 |
|
$display_data = $cache->{display_data}->{$field}->[$page]; |
512 |
|
} |
513 |
if ($display_data) { # default |
if ($display_data) { # default |
514 |
if ($field eq "headline") { |
if ($field eq "headline") { |
515 |
$xml .= xmlify("headline", $display_data); |
$xml .= xmlify("headline", $display_data); |
527 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
528 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
529 |
|
|
530 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
531 |
} |
} |
532 |
|
|
533 |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
537 |
|
|
538 |
# add delimiters before and after word. |
# add delimiters before and after word. |
539 |
# That is required to produce exact match |
# That is required to produce exact match |
540 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
541 |
} |
} |
542 |
|
|
543 |
foreach my $d (@{$cache->{index_data}->{$field}->[$page]}) { |
my $idel = $cache->{index_delimiter}->{$field}; |
544 |
$index->insert($field, $d, $path); |
foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { |
545 |
#print STDERR "index $field: $d [$path]\n"; |
my $i = $tmp; |
546 |
|
my $d = $tmp; |
547 |
|
if ($idel && $tmp =~ m/$idel/) { |
548 |
|
($i,$d) = split(/$idel/,$tmp); |
549 |
|
} |
550 |
|
$index->insert($field, $i, $d, $path); |
551 |
|
#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; |
552 |
} |
} |
553 |
} |
} |
554 |
|
|
570 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
571 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
572 |
|
|
573 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
574 |
} |
} |
575 |
|
|
576 |
if ($swish_exact_data) { |
if ($swish_exact_data) { |
579 |
|
|
580 |
# add delimiters before and after word. |
# add delimiters before and after word. |
581 |
# That is required to produce exact match |
# That is required to produce exact match |
582 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
583 |
} |
} |
584 |
} |
} |
585 |
} |
} |
616 |
|
|
617 |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
618 |
|
|
619 |
my $unac_filter = $cfg_global->val('global', 'unac_filter'); |
my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter'); |
620 |
if ($unac_filter) { |
if ($my_unac_filter) { |
621 |
require $unac_filter; |
print STDERR "using $my_unac_filter to filter characters for search\n"; |
622 |
|
require $my_unac_filter; |
623 |
|
} else { |
624 |
|
print STDERR "### fallback to default my_unac_string!\n"; |
625 |
|
eval q{ |
626 |
|
sub main::my_unac_string($$) { |
627 |
|
my ($charset, $string) = (@_); |
628 |
|
return $string; |
629 |
|
} |
630 |
|
}; |
631 |
} |
} |
632 |
|
|
633 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
639 |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
640 |
if ($lookup_file) { |
if ($lookup_file) { |
641 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
642 |
|
if (! -e $lookup_file) { |
643 |
|
open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!"; |
644 |
|
close(LOOKUP); |
645 |
|
} |
646 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
647 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
648 |
|
# delete memory cache for lookup file |
649 |
|
delete $cache->{lhash}; |
650 |
} |
} |
651 |
|
|
652 |
# open existing lookup file |
# open existing lookup file |
663 |
my $type_base = $type; |
my $type_base = $type; |
664 |
$type_base =~ s/_.+$//g; |
$type_base =~ s/_.+$//g; |
665 |
|
|
666 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
667 |
|
|
668 |
# output current progress indicator |
# output current progress indicator |
669 |
my $last_p = 0; |
my $last_p = 0; |
783 |
print STDERR "\n"; |
print STDERR "\n"; |
784 |
|
|
785 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
786 |
use Spreadsheet::ParseExcel; |
require Spreadsheet::ParseExcel; |
787 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
require Spreadsheet::ParseExcel::Utility; |
788 |
|
import Spreadsheet::ParseExcel::Utility qw(int2col); |
789 |
|
|
790 |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
791 |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
813 |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
814 |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
815 |
if ($cell) { |
if ($cell) { |
816 |
$row->{int2col($iC)} = $cell->Value; |
# this conversion is a cludge. |
817 |
|
# Files from Excell could have |
818 |
|
# characters which don't fit into |
819 |
|
# destination encoding. |
820 |
|
$row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value; |
821 |
} |
} |
822 |
} |
} |
823 |
|
|
843 |
} |
} |
844 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
845 |
|
|
846 |
use MARC; |
require MARC; |
847 |
|
|
848 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
849 |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
850 |
|
|
851 |
# optional argument is format |
# optional argument is format |
852 |
my $format = x($config->{format}) || 'usmarc'; |
my $format = x($config->{marc_format}) || 'usmarc'; |
853 |
|
|
854 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
855 |
|
|
856 |
my $marc = new MARC; |
my $marc = new MARC; |
857 |
my $nr = $marc->openmarc({ |
my $nr = $marc->openmarc({ |
858 |
file=>$marc_file, format=>$format |
file=>$marc_file, format=>$format |
859 |
}) || die "Can't open MARC file '$marc_file'"; |
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
860 |
|
|
861 |
|
# read MARC file in memory |
862 |
|
$marc->nextmarc(-1); |
863 |
|
|
864 |
my $i=0; # record nr. |
my $max_rec = $marc->marc_count(); |
865 |
|
|
866 |
my $rec; |
for(my $i=1; $i<=$max_rec; $i++) { |
867 |
|
|
868 |
while ($marc->nextmarc(1)) { |
progress($i,$max_rec); |
869 |
|
|
870 |
# XXX |
# store value for marc_sf.pm |
871 |
fakeprogress($i++); |
$main::cache->{marc_record} = $i; |
872 |
|
|
873 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
874 |
|
|
880 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
881 |
} |
} |
882 |
} |
} |
883 |
|
|
884 |
|
print STDERR "\n"; |
885 |
|
|
886 |
} elsif ($type_base eq "feed") { |
} elsif ($type_base eq "feed") { |
887 |
|
|
888 |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
937 |
|
|
938 |
all2xml.pl - read various file formats and dump XML for SWISH-E |
all2xml.pl - read various file formats and dump XML for SWISH-E |
939 |
|
|
940 |
|
=head1 SYNOPSYS |
941 |
|
|
942 |
|
$ all2xml.pl [test.conf] |
943 |
|
|
944 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
945 |
|
|
946 |
This command will read ISIS data file using OpenIsis perl module, MARC |
This command will read ISIS data file using OpenIsis perl module, MARC |
949 |
this script B<isn't general xml generator> from isis files (isis allready |
this script B<isn't general xml generator> from isis files (isis allready |
950 |
has something like that). Output of this script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |
951 |
|
|
952 |
|
If no configuration file is specified, it will use default one called |
953 |
|
C<all2xml.conf>. |
954 |
|
|
955 |
=head1 BUGS |
=head1 BUGS |
956 |
|
|
957 |
Documentation is really lacking. However, in true Open Source spirit, source |
Documentation is really lacking. However, in true Open Source spirit, source |