| 1 |
#!/usr/bin/perl |
| 2 |
|
| 3 |
#package common; |
| 4 |
|
| 5 |
use strict; |
| 6 |
|
| 7 |
$common::install_dir="/local/index"; |
| 8 |
$common::mpsindex="/local/mps-5.3/bin/mpsindex -l 9 -b"; |
| 9 |
#$common::isis_data="/var/autofs/misc/isis_data/"; |
| 10 |
$common::isis_data="/data/isis_data/"; |
| 11 |
#$common::isis_data="/mnt/20020606/Isis/Data/"; # doma |
| 12 |
|
| 13 |
$common::mps_header="V 5 3 |
| 14 |
L hr-HR |
| 15 |
F 700+ 1 Autor |
| 16 |
F 200+ 2 Naslov |
| 17 |
F 210+ 3 Izdavanje |
| 18 |
F 225 4 Nakladnièka cjelina |
| 19 |
F 300+ 5 Napomene |
| 20 |
F 330 6 Sadr¾aj |
| 21 |
F 464 7 Analitièki radovi |
| 22 |
F 610 8 Kljuène rijeèi |
| 23 |
F 675 9 UDK |
| 24 |
F 686 10 CC |
| 25 |
F 990 11 Signatura |
| 26 |
F 991 12 Inventarni broj |
| 27 |
F ISN 13 ISBN, ISSN |
| 28 |
F MFN 14 MFN |
| 29 |
F 215+ 15 Materijalni opis |
| 30 |
F 101 16 Jezik |
| 31 |
F tip 17 Tip graðe |
| 32 |
"; |
| 33 |
|
| 34 |
$common::database_cf="/local/search/databases.cf"; |
| 35 |
|
| 36 |
#-------------------------------------------------------------------- |
| 37 |
# |
| 38 |
# mps_expand(nr,"space separated string"); |
| 39 |
# |
| 40 |
|
| 41 |
sub mps_expand { |
| 42 |
my $nr = shift @_; |
| 43 |
my $out = ""; |
| 44 |
while (my $fld = shift @_) { |
| 45 |
if ($fld =~ s/\s*[,;\.!?'"<>\[\]]*\s+/ /g) { |
| 46 |
$fld =~ s/></ /g; |
| 47 |
foreach my $w (split(/\s+/,$fld)) { |
| 48 |
# FIX: this should be replaced by stemmer! |
| 49 |
# remove chars from beginning of word |
| 50 |
$w =~ s/^['"<(\[]//g; |
| 51 |
# remove chars from end of word |
| 52 |
$w =~ s/[,;\.!?'">)\]]$//g; |
| 53 |
# preglasi öÖäÄüÜ -> |
| 54 |
if ($w =~ m/[]/) { |
| 55 |
$out .= "W $w $nr\n"; |
| 56 |
$w =~ s/[]/oe/g; |
| 57 |
$w =~ s/[]/ae/g; |
| 58 |
$w =~ s/[]/ue/g; |
| 59 |
$out .= "W $w $nr\n"; |
| 60 |
# Irish Last names |
| 61 |
} elsif (($nr == 1 || $nr == 2 || $nr == 3 || $nr == 6 || $nr == 7 || $nr == 8) && $w =~ m/^Ma?c([A-Z].+)/) { |
| 62 |
$out .= "W Mc$1 $nr\n"; |
| 63 |
$out .= "W Mac$1 $nr\n"; |
| 64 |
} else { |
| 65 |
$out .= "W $w $nr\n"; |
| 66 |
} |
| 67 |
} |
| 68 |
} else { |
| 69 |
$out .= "W $fld $nr\n"; |
| 70 |
} |
| 71 |
} |
| 72 |
return c_852_czs($out); |
| 73 |
} |
| 74 |
|
| 75 |
1; |
| 76 |
|
Parent Directory
| 
Revision Log