/[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 5 - (hide annotations)
Sat Jun 20 19:42:00 2009 UTC (14 years, 10 months ago) by dpavlin
File size: 2853 byte(s)
remove debugging content output

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 4 use File::Slurp;
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     a => [ 250 , 'a' ],
21     c => [ 260 , 'b' ],
22     d => [ 260 , 'c' ],
23     },
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 4 diag "# get $url";
80 dpavlin 1
81     $mech->get( $url );
82    
83 dpavlin 4 diag "# got session";
84 dpavlin 1
85     $mech->follow_link( text_regex => qr/union/ );
86    
87 dpavlin 4 diag "# switch to advanced form (select)";
88 dpavlin 1
89 dpavlin 4 $mech->follow_link( url_regex => qr/mode=3/ );
90    
91     diag "# submit search $query";
92    
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 4 diag "# got $hits results, get first one";
108 dpavlin 1
109     $mech->follow_link( url_regex => qr/ukaz=DISP/ );
110    
111 dpavlin 4 diag "# in COMARC format";
112 dpavlin 1
113     $mech->follow_link( url_regex => qr/fmt=13/ );
114 dpavlin 4 }
115 dpavlin 1
116 dpavlin 4
117     sub fetch_marc {
118     my ($self) = @_;
119    
120 dpavlin 1 my $comarc;
121    
122 dpavlin 4 if ( $mech->content =~ m{<pre>\s*(.+?(\d+\.)\s+ID=(\d+).+?)\s*</pre>}s ) {
123    
124 dpavlin 1 my $comarc = $1;
125 dpavlin 4 my $nr = $2;
126     my $id = $3;
127    
128     diag "# fetch_marc $nr [$id]";
129    
130 dpavlin 1 $comarc =~ s{</?b>}{}gs;
131 dpavlin 2 $comarc =~ s{<font[^>]*>}{<s>}gs;
132     $comarc =~ s{</font>}{<e>}gs;
133 dpavlin 1
134 dpavlin 4 write_file "comarc/$id", $comarc;
135    
136 dpavlin 1 print $comarc;
137    
138 dpavlin 2 my $marc = MARC::Record->new;
139    
140     foreach my $line ( split(/[\r\n]+/, $comarc) ) {
141     our @f;
142    
143 dpavlin 4 if ( $line !~ s{^(\d\d\d)([01 ])([01 ])}{} ) {
144     diag "SKIP: $line";
145 dpavlin 2 } else {
146     $line .= "<eol>";
147    
148     @f = ( $1, $2, $3 );
149 dpavlin 4 sub sf { push @f, @_; }
150 dpavlin 2 $line =~ s{<s>(\w)<e>([^<]+)\s*}{sf($1, $2)}ges;
151 dpavlin 4 diag "# f:", join('|', @f), " left: |$line|";
152 dpavlin 2 $marc->add_fields( @f );
153     }
154     }
155    
156 dpavlin 4 open(my $out, '>:utf8', "marc/$id");
157 dpavlin 2 print $out $marc->as_usmarc;
158     close($out);
159    
160 dpavlin 4 diag $marc->as_formatted;
161 dpavlin 2
162 dpavlin 4 return $marc->as_usmarc;
163 dpavlin 1 } else {
164     die "can't fetch COMARC format from ", $mech->content;
165     }
166    
167     }
168    
169     1;

  ViewVC Help
Powered by ViewVC 1.1.26