1 # $Id: Stats.pm,v 1.4 2006-12-15 17:24:59 mike Exp $
3 package ZOOM::IRSpy::Stats;
8 use ZOOM::IRSpy::Utils qw(irspy_xpath_context);
12 ZOOM::IRSpy::Stats - statistics generated for IRSpy about its targets
16 $stats = new ZOOM::IRSpy::Stats($dbname);
21 Provides a simple API to obtaining statistics about targets registered
22 in IRSpy. This is done just by creating a Stats object. Once this
23 object is made, it can be crudely printed using the built-in debugging
24 C<print()> method, or the application can walk the structure to
31 $stats = new ZOOM::IRSpy::Stats($dbname, "dc.creator=wedel");
33 $stats = new ZOOM::IRSpy::Stats($dbname,
34 new ZOOM::Query::PQF('@attr 1=1003 wedel');
36 $spy = new ZOOM::Connection("target/string/for/irspy/database");
37 $stats = new ZOOM::IRSpy::Stats($spy, $query);
39 Creates a new C<ZOOM::IRSpy::Stats> object and populates it with
40 statistics for the targets in the nominated database. This process
41 involves analysing the nominated IRSpy database at some length, and
42 which therefore takes some time
44 Either one or two arguments are required:
48 =item $conn (mandatory)
50 An indication of the IRSpy database that statistics are required for.
51 This may be in the form of a C<ZOOM::Connection> object or a
52 database-name string such as C<localhost:3313/IR-Explain---1>.
54 =item $query (optional)
56 The query with which to select a subset of the database to be
57 analysed. This may be in the form of a C<ZOOM::Query> object (using
58 any of the supported subclasses) or a CQL string. If this is omitted,
59 then all records in the database are included in the generated
68 my($conn, $query) = @_;
70 $query ||= "cql.allRecords=1",
71 $conn = new ZOOM::Connection($conn) if !ref $conn;
72 $query = new ZOOM::Query::CQL($query) if !ref $query;
74 my $oldSyntax = $conn->option("preferredRecordSyntax");
75 my $oldESN = $conn->option("elementSetName");
76 my $oldPC = $conn->option("presentChunk");
77 $conn->option(preferredRecordSyntax => "xml");
78 $conn->option(elementSetName => "zeerex");
79 # $conn->option(presentChunk => 10);
81 my $rs = $conn->search($query);
91 $this->_gather_stats();
92 $conn->option(preferredRecordSyntax => $oldSyntax);
93 $conn->option(elementSetName => $oldESN);
94 $conn->option(presentChunk => $oldPC);
103 foreach my $i (0 .. $this->{n}-1) {
104 my $rec = $this->{rs}->record($i);
105 my $xc = irspy_xpath_context($rec);
107 # The ten most commonly supported Bib-1 Use attributes
108 foreach my $node ($xc->findnodes('e:indexInfo/e:index[@search="true"]/e:map/e:attr[@type=1 and @set="bib-1"]')) {
109 $this->{bib1AccessPoints}->{$node->findvalue(".")}++;
112 # Record syntax support by database
113 foreach my $node ($xc->findnodes('e:recordInfo/e:recordSyntax/@name')) {
114 $this->{recordSyntaxes}->{$node->findvalue(".")}++;
118 foreach my $node ($xc->findnodes('i:status/i:explain[@ok="1"]/@category')) {
120 $this->{explain}->{$node->findvalue(".")}++;
123 # Z39.50 Protocol Services Support
124 ### Requires XSLT fix
126 # Z39.50 Server Atlas
127 ### TODO -- awkward, should be considered an enhancement
130 my $host = $xc->findvalue('e:serverInfo/e:host');
132 $this->{domains}->{$host}++;
135 ### Requires XSLT fix
144 Prints an ugly but human-readable summary of the statistics on
152 print "database = '", $this->{conn}->option("host"), "'\n";
153 print "query = '", $this->{query}, "'\n";
154 print "result set = '", $this->{rs}, "'\n";
155 print "count = '", $this->{n}, "'\n";
158 print "\nTOP 10 BIB-1 ATTRIBUTES\n";
159 $hr = $this->{bib1AccessPoints};
160 foreach my $key ((sort { $hr->{$b} <=> $hr->{$a}
161 || $a <=> $b } keys %$hr)[0..9]) {
162 print sprintf("%6d%20s%5d (%d%%)\n",
163 $key, "", $hr->{$key}, 100*$hr->{$key}/$this->{n});
166 print "\nRECORD SYNTAXES\n";
167 $hr = $this->{recordSyntaxes};
168 foreach my $key (sort { $hr->{$b} <=> $hr->{$a}
169 || $a cmp $b } keys %$hr) {
170 print sprintf("%-26s%5d (%d%%)\n",
171 $key, $hr->{$key}, 100*$hr->{$key}/$this->{n});
174 print "\nEXPLAIN SUPPORT\n";
175 $hr = $this->{explain};
176 foreach my $key (sort { $hr->{$b} <=> $hr->{$a}
177 || $a cmp $b } keys %$hr) {
178 print sprintf("%-26s%5d (%d%%)\n",
179 $key, $hr->{$key}, 100*$hr->{$key}/$this->{n});
182 print "\nTOP-LEVEL DOMAINS\n";
183 $hr = $this->{domains};
184 foreach my $key (sort { $hr->{$b} <=> $hr->{$a}
185 || $a cmp $b } keys %$hr) {
186 print sprintf("%-26s%5d (%d%%)\n",
187 $key, $hr->{$key}, 100*$hr->{$key}/$this->{n});
198 Mike Taylor, E<lt>mike@indexdata.comE<gt>
200 =head1 COPYRIGHT AND LICENSE
202 Copyright (C) 2006 by Index Data ApS.
204 This library is free software; you can redistribute it and/or modify
205 it under the same terms as Perl itself, either Perl version 5.8.7 or,
206 at your option, any later version of Perl 5 you may have available.