/[Z3950-HTML-Scraper]/COBISS.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /COBISS.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 10 - (hide annotations)
Fri Mar 26 16:26:59 2010 UTC (14 years ago) by dpavlin
File size: 3717 byte(s)
fix mapping for 210 -> 260
1 dpavlin 1 package COBISS;
2    
3     use warnings;
4     use strict;
5    
6     use WWW::Mechanize;
7 dpavlin 2 use MARC::Record;
8 dpavlin 9 use Data::Dump qw/dump/;
9 dpavlin 1
10 dpavlin 2 binmode STDOUT, ':utf8';
11    
12     my $cobiss_marc21 = {
13     '010' => { a => [ '020', 'a' ] },
14     200 => {
15     a => [ 245 , 'a' ],
16     f => [ 245 , 'f' ],
17     },
18     205 => { a => [ 250 , 'a' ] },
19     210 => {
20 dpavlin 10 a => [ 260 , 'a' ],
21     c => [ 260 , 'b' ],
22     d => [ 260 , 'c' ],
23 dpavlin 2 },
24     215 => {
25     a => [ 300 , 'a' ],
26     c => [ 300 , 'b' ],
27     d => [ 300 , 'c' ],
28     },
29     700 => {
30     a => [ 100 , 'a' ],
31     },
32     };
33    
34 dpavlin 4 our $mech = WWW::Mechanize->new();
35     our $hits;
36    
37     sub diag {
38     print "# ", @_, $/;
39     }
40    
41     # Koha Z39.50 query:
42     #
43     # Bib-1 @and @and @and @and @and @and @and @or
44     # @attr 1=8 isbn-issn
45     # @attr 1=7 isbn-issn
46     # @attr 1=4 title
47     # @attr 1=1003 author
48     # @attr 1=16 dewey
49     # @attr 1=21 subject-holding
50     # @attr 1=12 control-no
51     # @attr 1=1007 standard-id
52     # @attr 1=1016 any
53    
54     our $usemap = {
55     8 => 'BN', # FIXME check
56     7 => 'SN', # FIXME check
57     4 => 'TI',
58     1003 => 'TI',
59     16 => 'CU',
60     21 => 'SU',
61     # 12 => '',
62     # 1007 => '',
63     # 1016 => '',
64    
65     };
66    
67     sub usemap {
68     my $f = shift || die;
69     $usemap->{$f};
70     }
71    
72 dpavlin 1 sub search {
73 dpavlin 4 my ( $self, $query ) = @_;
74 dpavlin 1
75 dpavlin 4 die "need query" unless defined $query;
76    
77 dpavlin 1 my $url = 'http://cobiss.izum.si/scripts/cobiss?ukaz=GETID&lani=en';
78    
79 dpavlin 7 diag "get $url";
80 dpavlin 1
81     $mech->get( $url );
82    
83 dpavlin 7 diag "got session";
84 dpavlin 1
85     $mech->follow_link( text_regex => qr/union/ );
86    
87 dpavlin 7 diag "switch to advanced form (select)";
88 dpavlin 1
89 dpavlin 4 $mech->follow_link( url_regex => qr/mode=3/ );
90    
91 dpavlin 7 diag "submit search $query";
92 dpavlin 4
93 dpavlin 1 $mech->submit_form(
94     fields => {
95 dpavlin 4 'SS1' => $query,
96 dpavlin 1 },
97     );
98    
99 dpavlin 4 $hits = 0;
100 dpavlin 1 if ( $mech->content =~ m{hits:\s*<b>\s*(\d+)\s*</b>}s ) {
101     $hits = $1;
102     } else {
103 dpavlin 4 diag "get't find results in ", $mech->content;
104     return;
105 dpavlin 1 }
106    
107 dpavlin 7 diag "got $hits results, get first one";
108 dpavlin 1
109     $mech->follow_link( url_regex => qr/ukaz=DISP/ );
110    
111 dpavlin 7 diag "in COMARC format";
112 dpavlin 1
113     $mech->follow_link( url_regex => qr/fmt=13/ );
114 dpavlin 4 }
115 dpavlin 1
116 dpavlin 4
117 dpavlin 9 sub fetch_rec {
118     my ($self,$format) = @_;
119 dpavlin 4
120 dpavlin 9 $format ||= 'unimarc';
121    
122     die "unknown format: $format" unless $format =~ m{(uni|us)marc};
123    
124 dpavlin 1 my $comarc;
125    
126 dpavlin 8 if ( $mech->content =~ m{<pre>\s*(.+?(\d+)\.\s+ID=(\d+).+?)\s*</pre>}s ) {
127 dpavlin 4
128 dpavlin 1 my $comarc = $1;
129 dpavlin 4 my $nr = $2;
130     my $id = $3;
131    
132 dpavlin 9 diag "fetch_marc $nr [$id] $format";
133 dpavlin 4
134 dpavlin 1 $comarc =~ s{</?b>}{}gs;
135 dpavlin 2 $comarc =~ s{<font[^>]*>}{<s>}gs;
136     $comarc =~ s{</font>}{<e>}gs;
137 dpavlin 1
138 dpavlin 8 open(my $out, '>:utf8', "comarc/$id");
139     print $out $comarc;
140     close($out);
141 dpavlin 4
142 dpavlin 1 print $comarc;
143    
144 dpavlin 2 my $marc = MARC::Record->new;
145    
146     foreach my $line ( split(/[\r\n]+/, $comarc) ) {
147    
148 dpavlin 4 if ( $line !~ s{^(\d\d\d)([01 ])([01 ])}{} ) {
149     diag "SKIP: $line";
150 dpavlin 2 } else {
151 dpavlin 9 our @f = ( $1, $2, $3 );
152 dpavlin 2 $line .= "<eol>";
153    
154 dpavlin 9 if ( $format eq 'unimarc' ) {
155    
156     diag dump(@f), "line: $line";
157     sub sf_uni {
158     warn "sf ",dump(@_);
159     push @f, @_;
160     }
161     $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf_uni($1, $2)}ges;
162     diag "f:", dump(@f), " left: |$line|";
163     $marc->add_fields( @f );
164    
165     } elsif ( $format eq 'usmarc' ) {
166    
167     my ( $f, $i1, $i2 ) = @f;
168    
169     our $out = {};
170    
171     sub sf_us {
172     my ($f,$sf,$v) = @_;
173     if ( my $m = $cobiss_marc21->{$f}->{$sf} ) {
174     push @{ $out->{ $m->[0] } }, ( $m->[1], $v );
175     }
176     return;
177     }
178     $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf_us($f,$1, $2)}ges;
179    
180     diag "converted marc21 ",dump( $out );
181    
182     foreach my $f ( keys %$out ) {
183     $marc->add_fields( $f, $i1, $i2, @{ $out->{$f} } );
184     }
185     }
186 dpavlin 2 }
187     }
188    
189 dpavlin 9 my $path = "marc/$id.$format";
190    
191     open($out, '>:utf8', $path);
192 dpavlin 2 print $out $marc->as_usmarc;
193     close($out);
194    
195 dpavlin 9 diag "created $path ", -s $path, " bytes";
196    
197 dpavlin 4 diag $marc->as_formatted;
198 dpavlin 2
199 dpavlin 8 $nr++;
200     $mech->follow_link( url_regex => qr/rec=$nr/ );
201    
202 dpavlin 4 return $marc->as_usmarc;
203 dpavlin 1 } else {
204     die "can't fetch COMARC format from ", $mech->content;
205     }
206    
207     }
208    
209     1;

  ViewVC Help
Powered by ViewVC 1.1.26