3 |
use strict; |
use strict; |
4 |
use OpenIsis; |
use OpenIsis; |
5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
#use Data::Dumper; |
use Data::Dumper; |
7 |
use common; |
use common; |
8 |
|
|
9 |
my %opts; |
my %opts; |
28 |
print S $common::mps_header; |
print S $common::mps_header; |
29 |
print MPS $common::mps_header; |
print MPS $common::mps_header; |
30 |
|
|
31 |
|
#-------------------------------------------------------------------- |
32 |
|
# init array in_mps_header for config checks later |
33 |
|
|
34 |
|
my %in_mps_header; |
35 |
|
foreach (split(/\n/,$common::mps_header)) { |
36 |
|
if (/^F /) { |
37 |
|
my (undef,$isis,$mps,undef) = split(/ /,$_,4); |
38 |
|
$in_mps_header{$mps}=$isis; |
39 |
|
} |
40 |
|
} |
41 |
|
require "./search/config.pm"; |
42 |
|
|
43 |
|
#-------------------------------------------------------------------- |
44 |
# |
# |
45 |
# expand(nr,"space separated string"); |
# expand(nr,"space separated string"); |
46 |
# |
# |
47 |
|
|
48 |
sub expand { |
sub expand { |
49 |
my $nr = shift @_; |
my $nr = shift @_; |
50 |
|
die "$nr is not in mps_header" if (!$in_mps_header{$nr}); |
51 |
my $out = ""; |
my $out = ""; |
52 |
while (my $fld = c_852_iso(shift @_)) { |
while (my $fld = c_852_iso(shift @_)) { |
53 |
my @words=split(/\s+/,$fld); |
my @words=split(/\s+/,$fld); |
64 |
|
|
65 |
#-------------------------------------------------------------------- |
#-------------------------------------------------------------------- |
66 |
|
|
|
################### ERASE###############3 |
|
|
|
|
|
# expand sub-fileds from ISIS field |
|
|
# (^a.....^b....) |
|
|
my %data; # FIX |
|
|
sub ex_sf { |
|
|
%data = (); |
|
|
my $in = $_[0]; |
|
|
if (my $tmp = $in) { |
|
|
# $tmp =~ tr/џ/ƾ/; # ISIS -> iso-8859-2 |
|
|
$tmp =~ tr//ܫꔼȺ̪㍐슂ٝ/; |
|
|
if ($in =~ m/^\^/) { |
|
|
my @sub = split(/\^/,$in); |
|
|
foreach my $fld (@sub) { |
|
|
$data{$1} = $2 if ($fld =~ m/^(\w+)(.+)$/) |
|
|
} |
|
|
} else { |
|
|
$data{all} = $in."<-- iz polja bez podpolja"; |
|
|
} |
|
|
} |
|
|
} |
|
|
|
|
|
# dump all sub-fields |
|
|
sub all_sf { |
|
|
my $nr = shift @_; |
|
|
my $out=""; |
|
|
foreach my $k (sort keys %data) { |
|
|
$out.=expand($nr,$data{$k}); |
|
|
} |
|
|
return $out; |
|
|
} |
|
|
|
|
|
sub all_sf_r { |
|
|
my $nr = shift @_; |
|
|
my $out=""; |
|
|
foreach my $k (sort {$b cmp $a} keys %data) { |
|
|
$out.=expand($nr,$data{$k}); |
|
|
} |
|
|
return $out; |
|
|
} |
|
|
|
|
|
sub all_sf2bib { |
|
|
my $nr = shift @_; |
|
|
my $max_in_line=shift @_ || 0; |
|
|
my $sep = shift @_ || ' '; |
|
|
my $out; |
|
|
my $i=0; |
|
|
my $bib = ""; |
|
|
foreach my $k (sort keys %data) { |
|
|
if ($out) { |
|
|
$out.= $sep.$data{$k}; |
|
|
} else { |
|
|
$out = $data{$k}; |
|
|
} |
|
|
$i++; |
|
|
if ($i == $max_in_line) { |
|
|
$bib .= $nr." ".$out."\n" if ($out); |
|
|
$i=0; |
|
|
$out=""; |
|
|
} |
|
|
} |
|
|
$bib .= $nr." ".$out."\n" if ($out); |
|
|
return $bib; |
|
|
} |
|
|
|
|
|
sub all_sf2bib_r { |
|
|
my $nr = shift @_; |
|
|
my $max_in_line=shift @_ || 0; |
|
|
my $sep = shift @_ || ' '; |
|
|
my $out; |
|
|
my $i=0; |
|
|
my $bib = ""; |
|
|
foreach my $k (sort {$b cmp $a} keys %data) { |
|
|
if ($out) { |
|
|
$out.= $sep.$data{$k}; |
|
|
} else { |
|
|
$out = $data{$k}; |
|
|
} |
|
|
$i++; |
|
|
if ($i == $max_in_line) { |
|
|
$bib .= $nr." ".$out."\n" if ($out); |
|
|
$i=0; |
|
|
$out=""; |
|
|
} |
|
|
} |
|
|
$bib .= $nr." ".$out."\n" if ($out); |
|
|
return $bib; |
|
|
} |
|
|
|
|
|
#-------------------------------------------------------------------- |
|
|
|
|
67 |
sub c_852_iso { |
sub c_852_iso { |
68 |
my $tmp = $_[0]; |
my $tmp = $_[0]; |
69 |
$tmp =~ tr//ܫꔼȺ̪㍐슂ٝ/ if ($tmp); |
$tmp =~ tr//ܫꔼȺ̪㍐슂ٝ/ if ($tmp); |
80 |
|
|
81 |
#-------------------------------------------------------------------- |
#-------------------------------------------------------------------- |
82 |
|
|
83 |
# $mps .= sf_to_mps($subfiled_hash,"subfields",mps_id); |
# $mps .= sf_to_mps("subfields",$subfiled_hash) |
84 |
# |
# |
85 |
# subfields options: |
# subfields options: |
86 |
# * - all (no sort) |
# * - all (no sort) |
111 |
|
|
112 |
sub sf_to_mps { |
sub sf_to_mps { |
113 |
my ($sf_hash,$subfields,$mps_id) = @_; |
my ($sf_hash,$subfields,$mps_id) = @_; |
114 |
|
die "$mps_id is not in mps_header" if (!$in_mps_header{$mps_id}); |
115 |
my $out=""; |
my $out=""; |
116 |
my @sf_arr = subfields_str_2_arr($subfields,$sf_hash); |
my @sf_arr = subfields_str_2_arr($subfields,$sf_hash); |
117 |
|
|
131 |
my $mps_id = shift @_ || die; |
my $mps_id = shift @_ || die; |
132 |
my $subfields = shift @_; |
my $subfields = shift @_; |
133 |
|
|
134 |
|
die "$mps_id is not in mps_header" if (!$in_mps_header{$mps_id}); |
135 |
|
|
136 |
my $i=0; |
my $i=0; |
137 |
my $out = ""; |
my $out = ""; |
138 |
|
|
176 |
my $group_sort = shift @_ || ''; |
my $group_sort = shift @_ || ''; |
177 |
my $sep = shift @_ || ' '; |
my $sep = shift @_ || ' '; |
178 |
|
|
179 |
|
my $fld = $bib_id; $fld =~ s/^%//; |
180 |
|
die "$fld is not in FieldNames" if (!$default::FieldNames{$fld}); |
181 |
|
|
182 |
my $i=0; |
my $i=0; |
183 |
|
|
184 |
my $bib=""; |
my $bib=""; |
236 |
sub isis_sf { |
sub isis_sf { |
237 |
my $row = shift @_; |
my $row = shift @_; |
238 |
my $isis_id = shift @_; |
my $isis_id = shift @_; |
239 |
my $subfield = shift @_ || 'a'; |
my $subfield = shift @_; |
240 |
my $prefix = shift @_ || ''; |
my $prefix = shift @_ || ''; |
241 |
my $postfix = shift @_ || ''; |
my $postfix = shift @_ || ''; |
242 |
|
|
244 |
|
|
245 |
if ($row->{$isis_id}->[0]) { |
if ($row->{$isis_id}->[0]) { |
246 |
my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); |
my $sf = OpenIsis::subfields($row->{$isis_id}->[0]); |
247 |
if (length($subfield) == 1) { |
if (! defined $subfield || length($subfield) == 0) { |
248 |
|
# subfield list undef, empty or no defined subfields for this record |
249 |
|
return $prefix . $row->{$isis_id}->[0] . $postfix; |
250 |
|
} elsif (length($subfield) == 1) { |
251 |
if ($sf->{$subfield}) { |
if ($sf->{$subfield}) { |
252 |
return $prefix . $sf->{$subfield} . $postfix; |
return $prefix . $sf->{$subfield} . $postfix; |
253 |
} else { |
} else { |
271 |
} |
} |
272 |
|
|
273 |
#-------------------------------------------------------------------- |
#-------------------------------------------------------------------- |
|
#-------------------------------------------------------------------- |
|
|
|
|
274 |
|
|
275 |
my $last_tell=0; |
my $last_tell=0; |
276 |
|
|
306 |
if (my $tmp = $row->{'200'}->[0]) { |
if (my $tmp = $row->{'200'}->[0]) { |
307 |
|
|
308 |
my $bib = "%MFN $row->{mfn}\n"; |
my $bib = "%MFN $row->{mfn}\n"; |
309 |
my $mps; |
my $mps = "W $row->{mfn} 14\n"; |
310 |
|
|
311 |
my $pcnt = int($row->{mfn} * 100 / $max_rowid); |
my $pcnt = int($row->{mfn} * 100 / $max_rowid); |
312 |
if ($pcnt != $last_pcnt) { |
if ($pcnt != $last_pcnt) { |
331 |
$mps .= isis_to_mps($row,'711',1); |
$mps .= isis_to_mps($row,'711',1); |
332 |
$mps .= isis_to_mps($row,'503',1); |
$mps .= isis_to_mps($row,'503',1); |
333 |
$mps .= isis_to_mps($row,'702',1); |
$mps .= isis_to_mps($row,'702',1); |
334 |
|
$mps .= isis_to_mps($row,'200',1,"fg"); |
335 |
|
|
336 |
$bib .= isis_to_bib($row,'205','%205'); |
$bib .= isis_to_bib($row,'205','%205'); |
337 |
|
|
347 |
$book .= ". ".$sf->{c} if ($sf->{c}); |
$book .= ". ".$sf->{c} if ($sf->{c}); |
348 |
$book .= " / ".$sf->{x} if ($sf->{x}); |
$book .= " / ".$sf->{x} if ($sf->{x}); |
349 |
$book .= " ; ".$sf->{y} if ($sf->{y}); |
$book .= " ; ".$sf->{y} if ($sf->{y}); |
350 |
$bib .= "%200 $book\n" if ($book); |
$bib .= "%200+ $book\n" if ($book); |
351 |
|
|
352 |
$mps .= isis_to_mps($row,'200',2,"akcde"); |
$mps .= isis_to_mps($row,'200',2,"akcde"); |
353 |
$mps .= isis_to_mps($row,'532',2); |
$mps .= isis_to_mps($row,'532',2); |
354 |
$mps .= isis_to_mps($row,'424',2); |
$mps .= isis_to_mps($row,'424',2); |
355 |
|
|
356 |
|
$mps .= isis_to_mps($row,'230',2,"ae"); |
357 |
|
$mps .= isis_to_mps($row,'231',2,"ae"); |
358 |
|
$mps .= isis_to_mps($row,'232',2,"ae"); |
359 |
|
$mps .= isis_to_mps($row,'233',2,"ae"); |
360 |
|
|
361 |
|
|
362 |
|
my $tmp; |
363 |
|
$tmp = isis_sf($row,'230','v'). |
364 |
|
isis_sf($row,'230','a',' : '). |
365 |
|
isis_sf($row,'250',undef,'. - '). |
366 |
|
isis_sf($row,'260',undef,'. - '). |
367 |
|
isis_sf($row,'290',undef,'<br>ISBN '); |
368 |
|
$bib .= "%sv $tmp\n" if ($tmp); |
369 |
|
|
370 |
|
$tmp = isis_sf($row,'231','v','<br>'). |
371 |
|
isis_sf($row,'231','a',' : '). |
372 |
|
isis_sf($row,'251',undef,'. - '). |
373 |
|
isis_sf($row,'261',undef,'. - '). |
374 |
|
isis_sf($row,'291',undef,'<br>ISBN '); |
375 |
|
$bib .= "%sv $tmp\n" if ($tmp); |
376 |
|
|
377 |
|
$tmp = isis_sf($row,'232','v','<br>'). |
378 |
|
isis_sf($row,'232','a',' : '). |
379 |
|
isis_sf($row,'252',undef,'. - '). |
380 |
|
isis_sf($row,'262',undef,'. - '). |
381 |
|
isis_sf($row,'292',undef,'<br>ISBN '); |
382 |
|
$bib .= "%sv $tmp\n" if ($tmp); |
383 |
|
|
384 |
|
$tmp = isis_sf($row,'233','v','<br>'). |
385 |
|
isis_sf($row,'233','a',' : '). |
386 |
|
isis_sf($row,'253',undef,'. - '). |
387 |
|
isis_sf($row,'263',undef,'. - '). |
388 |
|
isis_sf($row,'293',undef,'<br>ISBN '); |
389 |
|
|
390 |
|
$mps .= isis_to_mps($row,'270',2); |
391 |
|
$mps .= isis_to_mps($row,'271',2); |
392 |
|
$mps .= isis_to_mps($row,'272',2); |
393 |
|
$mps .= isis_to_mps($row,'273',2); |
394 |
|
|
395 |
$headline .= isis_sf($row,'700','b'," "); |
$headline .= isis_sf($row,'700','b'," "); |
396 |
$headline .= isis_sf($row,'700','a'," "); |
$headline .= isis_sf($row,'700','a'," "); |
397 |
|
|
398 |
# izdavac |
# izdavac |
399 |
$mps .= isis_to_mps($row,'210',3); |
$mps .= isis_to_mps($row,'210',3); |
400 |
|
$mps .= isis_to_mps($row,'250',3); |
401 |
|
$mps .= isis_to_mps($row,'251',3); |
402 |
|
$mps .= isis_to_mps($row,'252',3); |
403 |
|
$mps .= isis_to_mps($row,'253',3); |
404 |
# if (my $sf = OpenIsis::subfields($row->{'210'}->[0])) { |
# if (my $sf = OpenIsis::subfields($row->{'210'}->[0])) { |
405 |
# my $tmp; |
# my $tmp; |
406 |
# $tmp .= $sf->{a} if ($sf->{a}); |
# $tmp .= $sf->{a} if ($sf->{a}); |
408 |
# $tmp .= ", ".$sf->{d} if ($sf->{d}); |
# $tmp .= ", ".$sf->{d} if ($sf->{d}); |
409 |
# $bib .= "%210 $tmp\n" if ($tmp); |
# $bib .= "%210 $tmp\n" if ($tmp); |
410 |
# } |
# } |
411 |
$bib .= "%210 ".isis_sf($row,'210','acd', '','', ('',' : ',', ') )."\n"; |
$bib .= "%210+ ".isis_sf($row,'210','acd', '','', ('',' : ',', ') )."\n"; |
412 |
|
|
413 |
if (my $year = isis_sf($row,'210','d')) { |
if (my $year = isis_sf($row,'210','d')) { |
414 |
$year =~ s/^\s*cop\.*\s*//i; |
$year =~ s/^\s*cop\.*\s*//i; |
417 |
$headline .= " ($year)"; |
$headline .= " ($year)"; |
418 |
} |
} |
419 |
|
|
420 |
$bib .= isis_to_bib($row,'215','%215', '*', undef, undef, ', '); |
$mps .= isis_to_mps($row,'215',15); |
421 |
|
$mps .= isis_to_mps($row,'260',15); |
422 |
|
$mps .= isis_to_mps($row,'261',15); |
423 |
|
$mps .= isis_to_mps($row,'262',15); |
424 |
|
$mps .= isis_to_mps($row,'263',15); |
425 |
|
$bib .= isis_to_bib($row,'215','%215+', '*', undef, undef, ', '); |
426 |
|
|
427 |
# $bib .= isis_to_bib($row,'225','%225', 'aehivw'); |
# $bib .= isis_to_bib($row,'225','%225', 'aehivw'); |
428 |
$bib .= "%225 ".isis_sf($row,'225','aevhiw', '(',')', ('',' : ',' ; ','. ',', ',' ; '))."\n"; |
$bib .= "%225 ".isis_sf($row,'225','aevhiw', '(',')', ('',' : ',' ; ','. ',', ',' ; '))."\n"; |
434 |
$mps .= isis_to_mps($row,'300',5); |
$mps .= isis_to_mps($row,'300',5); |
435 |
$mps .= isis_to_mps($row,'320',5); |
$mps .= isis_to_mps($row,'320',5); |
436 |
$mps .= isis_to_mps($row,'327',5); |
$mps .= isis_to_mps($row,'327',5); |
437 |
|
$mps .= isis_to_mps($row,'280',5); |
438 |
|
$mps .= isis_to_mps($row,'281',5); |
439 |
|
$mps .= isis_to_mps($row,'282',5); |
440 |
|
$mps .= isis_to_mps($row,'283',5); |
441 |
|
|
442 |
$bib .= isis_to_bib($row,'330','%330'); |
$bib .= isis_to_bib($row,'330','%330'); |
443 |
$mps .= isis_to_mps($row,'330',6); |
$mps .= isis_to_mps($row,'330',6); |
459 |
$bib .= isis_to_bib($row,'991','%991'); |
$bib .= isis_to_bib($row,'991','%991'); |
460 |
$mps .= isis_to_mps($row,'991',12); |
$mps .= isis_to_mps($row,'991',12); |
461 |
|
|
462 |
# ISBN |
sub isis_isn_to_mps { |
463 |
if (my $isbn = $row->{10}->[0]) { |
my $row = shift @_ || die; |
464 |
$isbn =~ s/ +//g; # remove spaces |
my $isis_id = shift @_ || die; |
465 |
$mps .= "W $isbn 13\n"; |
my $nr = shift @_ || die; |
466 |
$bib .= "%ISBN $isbn\n"; |
my $i=0; |
467 |
$isbn =~ s/-//g; |
my $mps=''; |
468 |
$mps .= "W $isbn 13\n"; |
while (my $isn=$row->{$isis_id}->[$i]) { |
469 |
|
$isn =~ s/ +//g; # remove spaces |
470 |
|
$mps .= "W $isn $nr\n"; |
471 |
|
if ($isn =~ s/-//g) { |
472 |
|
$mps .= "W $isn $nr\n"; |
473 |
|
} |
474 |
|
$i++; |
475 |
|
} |
476 |
|
return $mps; |
477 |
} |
} |
478 |
$mps .= isis_to_mps($row,'10',12); |
|
479 |
|
# ISBN |
480 |
|
$bib .= isis_to_bib($row,'10','%ISBN'); |
481 |
|
$mps .= isis_isn_to_mps($row,'10',13); |
482 |
|
$mps .= isis_isn_to_mps($row,'290',13); |
483 |
|
$mps .= isis_isn_to_mps($row,'291',13); |
484 |
|
$mps .= isis_isn_to_mps($row,'292',13); |
485 |
|
$mps .= isis_isn_to_mps($row,'293',13); |
486 |
|
|
487 |
|
# ISSN |
488 |
|
#store_isn($row->{11}->[0],14,'%ISSN'); |
489 |
|
|
490 |
$mps .= isis_to_mps($row,'532',1); |
$mps .= isis_to_mps($row,'532',1); |
491 |
|
|
494 |
# headline |
# headline |
495 |
if ($headline) { |
if ($headline) { |
496 |
$headline .= " [".$row->{mfn}."]"; ## debug MFN! |
$headline .= " [".$row->{mfn}."]"; ## debug MFN! |
497 |
|
$headline =~ s/&/∧/g; |
498 |
|
$headline =~ s/</</g; |
499 |
|
$headline =~ s/>/>/g; |
500 |
$mps .= "H ".c_852_iso($headline)."\n"; |
$mps .= "H ".c_852_iso($headline)."\n"; |
501 |
} else { |
} else { |
502 |
$mps .= "H nepoznato\n"; |
$mps .= "H nepoznato\n"; |
512 |
|
|
513 |
print R c_852_iso($bib); |
print R c_852_iso($bib); |
514 |
|
|
515 |
|
# check if all fields are defined |
516 |
|
foreach (split(/\n/,$bib)) { |
517 |
|
if (/^%(\w+)\s/ && !$default::FieldNames{$1}) { |
518 |
|
die "field $1 used but not in FieldNames"; |
519 |
|
} |
520 |
|
} |
521 |
|
|
522 |
|
# print R "%perl ".Dumper($row)."\n"; |
523 |
|
|
524 |
$mps .= "T document text/plain ".(tell(R) - $last_tell)." $dir/bib $last_tell ".tell(R)."\n"; |
$mps .= "T document text/plain ".(tell(R) - $last_tell)." $dir/bib $last_tell ".tell(R)."\n"; |
525 |
$last_tell=tell(R); |
$last_tell=tell(R); |
526 |
|
|