12 |
|
|
13 |
use lib './lib'; |
use lib './lib'; |
14 |
use WebPAC; |
use WebPAC; |
15 |
|
use WebPAC::jsFind; |
16 |
|
use WebPAC::Index; |
17 |
|
|
18 |
my $webpac = new WebPAC( |
my $webpac = new WebPAC( |
19 |
code_page => 'ISO-8859-2', |
code_page => 'ISO-8859-2', |
20 |
limit_mfn => 500, |
# limit_mfn => 500, |
21 |
debug => 1, |
# debug => 1, |
22 |
) || die; |
) || die; |
23 |
|
|
24 |
|
my $log = $webpac->_get_logger(); |
25 |
|
|
26 |
|
my $index = new WebPAC::jsFind( |
27 |
|
index_path => './out/index', |
28 |
|
keys => 10, |
29 |
|
) || die; |
30 |
|
|
31 |
|
my $thes; |
32 |
|
|
33 |
$|=1; |
$|=1; |
34 |
|
|
35 |
my $maxmfn = $webpac->open_isis( |
my $maxmfn = $webpac->open_isis( |
40 |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval '=> '"v901^a" eq "Mikrotezaurus"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
41 |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
# { 'eval' => '"v901^a" eq "Deskriptor"', 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
42 |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
{ 'key' => 'a:v561^4:v562^4:v461^1', 'val' => 'v900' }, |
43 |
|
{ 'key' => '900_mfn:v900', 'val' => 'v000' }, |
44 |
], |
], |
45 |
); |
); |
46 |
|
|
47 |
print "rows: $maxmfn\n\n"; |
$log->info("rows: $maxmfn"); |
48 |
|
|
49 |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
$webpac->open_import_xml(type => 'isis_hidra_ths'); |
50 |
|
|
51 |
while (my $rec = $webpac->fetch_rec) { |
while (my $rec = $webpac->fetch_rec) { |
52 |
|
|
53 |
print "-- ",$webpac->parse($rec,'v250^a / [v562^4] v562^a'),"\n"; |
my @ds = $webpac->data_structure($rec); |
54 |
|
|
55 |
print " ", |
if (0 && $log->is_debug) { |
56 |
$webpac->parse($rec,'eval{v901^a eq "Deskriptor"}v250^a / [v562^4] v562^a'), |
$log->debug("rec = ",Dumper($rec)); |
57 |
$webpac->parse($rec,'eval{v901^a ne "Deskriptor"}250a [251]'), |
$log->debug("ds = ",Dumper(\@ds)); |
58 |
"\n"; |
} |
59 |
|
|
60 |
|
next if (! @ds); |
61 |
|
|
62 |
|
my $filename = $webpac->{'current_filename'}; |
63 |
|
|
64 |
|
if ($filename) { |
65 |
|
$webpac->output_file( |
66 |
|
file => $filename, |
67 |
|
template => 'html.tt', |
68 |
|
data => \@ds, |
69 |
|
headline => $webpac->{'headline'}, |
70 |
|
); |
71 |
|
} else { |
72 |
|
print $webpac->output( |
73 |
|
template => 'text.tt', |
74 |
|
data => \@ds, |
75 |
|
headline => $webpac->{'headline'}, |
76 |
|
); |
77 |
|
} |
78 |
|
|
79 |
|
my $headline = $webpac->{'headline'}; |
80 |
|
|
81 |
|
my $f = $filename; |
82 |
|
$f =~ s!out/!!; |
83 |
|
|
84 |
|
# save into index |
85 |
|
foreach my $ds (@ds) { |
86 |
|
next if (! $ds->{'swish'}); |
87 |
|
|
88 |
|
$index->insert( |
89 |
|
index_name => $ds->{'tag'}, |
90 |
|
path => $f, |
91 |
|
headline => $headline, |
92 |
|
words => join(" ",@{$ds->{'swish'}}) |
93 |
|
); |
94 |
|
} |
95 |
|
|
96 |
|
# save into sorted index (thesaurus) |
97 |
|
foreach my $ds (@ds) { |
98 |
|
next if (! $ds->{'index'}); |
99 |
|
|
100 |
|
$thes->{$ds->{'tag'}} ||= new WebPAC::Index; |
101 |
|
|
102 |
|
$thes->{$ds->{'tag'}}->insert( |
103 |
|
path => $f, |
104 |
|
headline => join(" ",@{$ds->{'index'}}), |
105 |
|
); |
106 |
|
} |
107 |
|
|
108 |
my @t = $webpac->fill_in($rec,'v553^1;;v553^a'); |
# print Dumper(\@ds); |
|
print " Uži pojam: ",join("\t\n",@t),"\n" if (@t); |
|
109 |
|
|
110 |
@t = $webpac->fill_in($rec,'eval{v901^a eq "Područje"}[a:v251::];;[d:[a:v251::]]'); |
} |
111 |
print " Područje: ",join("\t\n",@t),"\n" if (@t); |
|
112 |
|
foreach my $t (keys %{$thes}) { |
113 |
|
|
114 |
@t = $webpac->fill_in($rec,'eval{v901^a eq "Mikrotezaurus"}[a:v561^4:v251:];;[d:[a:v561^4:v251:]]'); |
my @e = $thes->{$t}->elements; |
115 |
print " Mikrotezaurus: ",join("\t\n",@t),"\n" if (@t); |
if (! @e) { |
116 |
|
$log->logwarn("no elements in sorted index $t?"); |
117 |
|
next; |
118 |
|
} |
119 |
|
|
120 |
|
$log->info("saving sorted index $t [".scalar(@e)." elements]"); |
121 |
|
|
122 |
|
$webpac->output_file( |
123 |
|
file => "./out/bfilter/$t.txt", |
124 |
|
template => 'index.tt', |
125 |
|
data => \@e, |
126 |
|
index_name => $t, |
127 |
|
); |
128 |
|
} |
129 |
|
|
130 |
@t = $webpac->fill_in($rec,'eval{v901^a eq "Deskriptor"}[a:v561^4:v562^4:v900];;[d:[a:v561^4:v562^4:v900]]'); |
if (0 && $log->is_debug) { |
131 |
print " Deskriptor: ",join("\t\n",@t),"\n" if (@t); |
$log->debug("lookup hash: ",Dumper($webpac->{'lookup'})); |
132 |
|
$log->debug("data hash: ",Dumper($webpac->{'data'})); |
133 |
|
foreach my $t (keys %{$thes}) { |
134 |
|
$log->debug("thesaurus $t hash: ",Dumper($thes->{$t})); |
135 |
|
} |
136 |
} |
} |
137 |
|
|
138 |
print "## lookup ",Dumper($webpac->{'lookup'}); |
$index->close; |
139 |
print "## data ",Dumper($webpac->{'data'}); |
|