1 # $Id: IRSpy.pm,v 1.29 2006-10-12 16:54:13 mike Exp $
9 use Data::Dumper; # For debugging only
11 use Net::Z3950::ZOOM 1.13; # For the ZOOM version-check only
12 use ZOOM::IRSpy::Node;
13 use ZOOM::IRSpy::Connection;
14 use ZOOM::IRSpy::Record;
17 our $VERSION = '0.02';
20 # Enumeration for callback functions to return
21 package ZOOM::IRSpy::Status;
22 sub OK { 29 } # No problems, task is still progressing
23 sub TASK_DONE { 18 } # Task is complete, next task should begin
24 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
25 sub TEST_BAD { 31 } # Whole test is complete, and failed
31 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
36 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
37 print $spy->report_status();
41 This module exists to implement the IRspy program, which discovers,
42 analyses and monitors IR servers implementing the Z39.50 and SRU/W
43 protocols. It is a successor to the ZSpy program.
48 ZOOM::Log::mask_str("irspy");
49 ZOOM::Log::mask_str("irspy_debug");
50 ZOOM::Log::mask_str("irspy_event");
51 ZOOM::Log::mask_str("irspy_unhandled");
52 ZOOM::Log::mask_str("irspy_test");
53 ZOOM::Log::mask_str("irspy_task");
58 my($dbname, $user, $password) = @_;
61 push @options, (user => $user, password => $password)
64 my $conn = new ZOOM::Connection($dbname, 0, @options)
65 or die "$0: can't connection to IRSpy database 'dbname'";
69 allrecords => 1, # unless overridden by targets()
70 query => undef, # filled in later
71 targets => undef, # filled in later
72 connections => undef, # filled in later
73 tests => [], # stack of tests currently being executed
75 $this->log("irspy", "starting up with database '$dbname'");
87 # Explicitly nominate a set of targets to check, overriding the
88 # default which is to re-check everything in the database. Each
89 # target already in the database results in the existing record being
90 # updated; each new target causes a new record to be added.
96 $this->log("irspy", "setting explicit list of targets ",
97 join(", ", map { "'$_'" } @targets));
98 $this->{allrecords} = 0;
100 foreach my $target (@targets) {
101 my($host, $port, $db, $newtarget) = _parse_target_string($target);
102 if ($newtarget ne $target) {
103 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
104 $target = $newtarget; # This is written through the ref
106 push @qlist, (qq[(host="$host" and port="$port" and path="$db")]);
109 $this->{targets} = \@targets;
110 $this->{query} = join(" or ", @qlist);
114 # Also used by ZOOM::IRSpy::Record
115 sub _parse_target_string {
118 my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
119 if (!defined $host) {
121 ($host, $db) = ($target =~ /(.*?)\/(.*)/);
122 $target = "$host:$port/$db";
124 die "$0: invalid target string '$target'"
127 return ($host, $port, $db, $target);
131 # There are two cases.
133 # 1. A specific set of targets is nominated on the command line.
134 # - Records must be fetched for those targets that are in the DB
135 # - New, empty records must be made for those that are not.
136 # - Updated records written to the DB may or may not be new.
138 # 2. All records in the database are to be checked.
139 # - Records must be fetched for all targets in the DB
140 # - Updated records written to the DB may not be new.
142 # That's all -- what could be simpler?
148 if ($this->{allrecords}) {
149 # We need to check on every target in the database, which
150 # means we need to do a "find all". According to the BIB-1
151 # semantics document at
152 # http://www.loc.gov/z3950/agency/bib1.html
154 # @attr 2=103 @attr 1=1035 x
155 # should find all records, but it seems that Zebra doesn't
156 # support this. Furthermore, when using the "alvis" filter
157 # (as we do for IRSpy) it doesn't support the use of any BIB-1
158 # access point -- not even 1035 "everywhere" -- so instead we
159 # hack together a search that we know will find all records.
160 $this->{query} = "port=?*";
162 # Prepopulate the target map with nulls so that after we fill
163 # in what we can from the database query, we know which target
164 # IDs we need new records for.
165 foreach my $target (@{ $this->{targets} }) {
166 $target2record{lc($target)} = undef;
170 $this->log("irspy_debug", "query '", $this->{query}, "'");
171 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
172 delete $this->{query}; # No longer needed at all
173 $this->log("irspy_debug", "found ", $rs->size(), " target records");
174 foreach my $i (1 .. $rs->size()) {
175 my $target = _render_record($rs, $i-1, "id");
176 my $zeerex = _render_record($rs, $i-1, "zeerex");
177 #print STDERR "making '$target' record with '$zeerex'\n";
178 $target2record{lc($target)} =
179 new ZOOM::IRSpy::Record($this, $target, $zeerex);
180 push @{ $this->{targets} }, $target
181 if $this->{allrecords};
184 # Make records for targets not previously in the database
185 foreach my $target (keys %target2record) {
186 my $record = $target2record{$target};
187 if (!defined $record) {
188 $this->log("irspy_debug", "made new record for '$target'");
189 $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
191 $this->log("irspy_debug", "using existing record for '$target'");
196 foreach my $target (@{ $this->{targets} }) {
197 my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
198 $conn->option(host => $target);
199 my $record = delete $target2record{lc($target)};
200 $conn->record($record);
201 push @connections, $conn;
203 die("remaining target2record = { " .
204 join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
205 sort keys %target2record) . " }")
208 $this->{connections} = \@connections;
209 delete $this->{targets}; # The information is now in {connections}
214 my($rs, $which, $elementSetName) = @_;
216 # There is a slight race condition here on the element-set name,
217 # but it shouldn't be a problem as this is (currently) only called
218 # from parts of the program that run single-threaded.
219 my $old = $rs->option(elementSetName => $elementSetName);
220 my $rec = $rs->record($which);
221 $rs->option(elementSetName => $old);
223 return $rec->render();
227 sub _rewrite_records {
230 # Write modified records back to database
231 foreach my $conn (@{ $this->{connections} }) {
232 my $rec = $conn->record();
233 my $p = $this->{conn}->package();
234 $p->option(action => "specialUpdate");
235 my $xml = $rec->{zeerex}->toString();
236 $p->option(record => $xml);
240 $p = $this->{conn}->package();
247 print "Updated with xml=<br/>\n<pre>$xml</pre>\n";
253 # The approach: gather declarative information about test hierarchy,
254 # then go into a loop. In the loop, we ensure that each connection is
255 # running a test, and within that test a task, until its list of tests
256 # is exhausted. No individual test ever calls wait(): tests just queue
257 # up tasks and return immediately. When the tasks are run (one at a
258 # time on each connection) they generate events, and it is these that
259 # are harvested by ZOOM::event(). Since each connection knows what
260 # task it is running, it can invoke the appropriate callbacks.
261 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
262 # loop how to continue.
265 # While a connection is running a task, its current_task()
266 # points at the task structure. When it finishes its task,
267 # next_task() is pointed at the next task to execute (if there
268 # is one), and its current_task() is set to zero. When the next
269 # task is executed, the connection's next_task() is set to zero
270 # and its current_task() pointed to the task structure.
271 # current_task() and next_task() are both zero only when there
272 # are no more queued tasks, which is when a new test is
275 # Each connection's current test is stored in its
276 # "current_test_address" option. The next test to execute is
277 # calculated by walking the declarative tree of tests. This
278 # option begins empty; the "next test" after this is of course
285 $tname = "Main" if !defined $tname;
286 $this->{tree} = $this->_gather_tests($tname)
287 or die "No tests defined";
288 #$this->{tree}->print(0);
291 my @conn = @{ $this->{connections} };
294 my @copy_conn = @conn; # avoid alias problems after splice()
295 foreach my $i0 (0 .. $#copy_conn) {
296 my $conn = $copy_conn[$i0];
297 #print "connection $i0 of ", scalar(@copy_conn), " from ", scalar(@conn), " is $conn\n";
298 if (!$conn->current_task()) {
299 if (!$conn->next_task()) {
300 # Out of tasks: we need a new test
302 my $address = $conn->option("current_test_address");
304 if (!defined $address) {
307 $this->log("irspy_test", "checking for next test after '$address'");
308 $nextaddr = $this->_next_test($address);
310 if (!defined $nextaddr) {
311 $conn->log("irspy", "has no more tests: removing");
312 splice @conn, $i0, 1;
316 my $node = $this->{tree}->select($nextaddr)
317 or die "invalid nextaddr '$nextaddr'";
318 $conn->option(current_test_address => $nextaddr);
319 my $tname = $node->name();
320 $conn->log("irspy_test", "starting test '$nextaddr' = $tname");
321 my $tasks = $conn->tasks();
322 my $oldcount = @$tasks;
323 "ZOOM::IRSpy::Test::$tname"->start($conn);
324 $tasks = $conn->tasks();
325 if (@$tasks > $oldcount) {
326 # Prepare to start the first of the newly added tasks
327 $conn->next_task($tasks->[$oldcount]);
329 $conn->log("irspy_task", "no tasks added by new test $tname");
334 my $task = $conn->next_task();
335 die "no next task queued for $conn" if !defined $task;
336 $conn->log("irspy_task", "starting task $task");
338 $conn->current_task($task);
342 # Do we need to test $conn->is_idle()? I don't think so!
345 my $i0 = ZOOM::event(\@conn);
346 $this->log("irspy_event", "ZOOM_event(", scalar(@conn), " connections) = $i0");
347 last if $i0 == 0 || $i0 == -3; # no events or no connections
348 my $conn = $conn[$i0-1];
349 my $ev = $conn->last_event();
350 my $evstr = ZOOM::event_str($ev);
351 $conn->log("irspy_event", "event $ev ($evstr)");
353 my $task = $conn->current_task();
354 die "$conn has no current task for event $ev ($evstr)" if !$task;
355 eval { $conn->_check() };
357 ($ev == ZOOM::Event::RECV_DATA ||
358 $ev == ZOOM::Event::RECV_APDU ||
359 $ev == ZOOM::Event::ZEND)) {
360 # An error in, say, a search response, becomes visible to
361 # ZOOM before the Receive Data event is sent and persists
362 # until after the End, which means that successive events
363 # each report the same error. So we just ignore errors on
364 # "unimportant" events. ### But this doesn't work for,
365 # say, a Connection Refused, as the only event that shows
366 # us this error is the End.
367 $conn->log("irspy_event", "ignoring error ",
368 "on event $ev ($evstr): $@");
374 my $sub = $task->{cb}->{exception};
375 die $@ if !defined $sub;
376 $res = &$sub($conn, $task, $task->udata(), $@);
378 my $sub = $task->{cb}->{$ev};
380 $conn->log("irspy_unhandled", "event $ev ($evstr)");
384 $res = &$sub($conn, $task, $task->udata(), $ev);
387 if ($res == ZOOM::IRSpy::Status::OK) {
388 # Nothing to do -- life continues
390 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
391 my $task = $conn->current_task();
392 die "no task for TASK_DONE on $conn" if !$task;
393 die "next task already defined for $conn" if $conn->next_task();
394 $conn->log("irspy_task", "completed task $task");
395 $conn->next_task($task->{next});
396 $conn->current_task(0);
398 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
399 $res == ZOOM::IRSpy::Status::TEST_BAD) {
400 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
401 $conn->log("irspy_test", "test completed ($x)");
402 $conn->current_task(0);
404 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
405 ### Should skip over remaining sibling tests if TEST_BAD
406 ### Should count the number of skipped siblings
412 $this->log("irspy", "exiting main loop");
414 #$this->_rewrite_records();
421 my($tname, @ancestors) = @_;
423 die("$0: test-hierarchy loop detected: " .
424 join(" -> ", @ancestors, $tname))
425 if grep { $_ eq $tname } @ancestors;
428 my $slashSeperatedTname = $tname;
429 $slashSeperatedTname =~ s/::/\//g;
430 require "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
432 $this->log("warn", "can't load test '$tname': skipping",
433 $@ =~ /^Can.t locate/ ? () : " ($@)");
437 $this->log("irspy", "adding test '$tname'");
439 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
440 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
441 push @subnodes, $subtest if defined $subtest;
444 return new ZOOM::IRSpy::Node($tname, @subnodes);
450 my($address, $omit_child) = @_;
454 my $maybe = $address eq "" ? "0" : "$address:0";
455 return $maybe if $this->{tree}->select($maybe);
458 # The top-level node has no successor or parent
459 return undef if $address eq "";
461 # Try next sibling child
462 my @components = split /:/, $address;
463 my $last = pop @components;
464 my $maybe = join(":", @components, $last+1);
465 return $maybe if $this->{tree}->select($maybe);
467 # This node is exhausted: try the parent's successor
468 return $this->_next_test(join(":", @components), 1)
477 ZOOM::IRSpy::Maintenance.
479 The ZOOM-Perl module,
480 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
483 http://indexdata.com/zebra/
487 Mike Taylor, E<lt>mike@indexdata.comE<gt>
489 =head1 COPYRIGHT AND LICENSE
491 Copyright (C) 2006 by Index Data ApS.
493 This library is free software; you can redistribute it and/or modify
494 it under the same terms as Perl itself, either Perl version 5.8.7 or,
495 at your option, any later version of Perl 5 you may have available.