1 # $Id: IRSpy.pm,v 1.33 2006-10-17 16:22:17 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';
18 our $irspy_ns = 'http://indexdata.com/irspy/1.0';
21 # Enumeration for callback functions to return
22 package ZOOM::IRSpy::Status;
23 sub OK { 29 } # No problems, task is still progressing
24 sub TASK_DONE { 18 } # Task is complete, next task should begin
25 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
26 sub TEST_BAD { 31 } # Whole test is complete, and failed
32 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
37 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
38 print $spy->report_status();
42 This module exists to implement the IRspy program, which discovers,
43 analyses and monitors IR servers implementing the Z39.50 and SRU/W
44 protocols. It is a successor to the ZSpy program.
49 ZOOM::Log::mask_str("irspy");
50 ZOOM::Log::mask_str("irspy_debug");
51 ZOOM::Log::mask_str("irspy_event");
52 ZOOM::Log::mask_str("irspy_unhandled");
53 ZOOM::Log::mask_str("irspy_test");
54 ZOOM::Log::mask_str("irspy_task");
59 my($dbname, $user, $password) = @_;
62 push @options, (user => $user, password => $password)
65 my $conn = new ZOOM::Connection($dbname, 0, @options)
66 or die "$0: can't connection to IRSpy database 'dbname'";
70 allrecords => 1, # unless overridden by targets()
71 query => undef, # filled in later
72 targets => undef, # filled in later
73 connections => undef, # filled in later
74 tests => [], # stack of tests currently being executed
76 $this->log("irspy", "starting up with database '$dbname'");
88 # Explicitly nominate a set of targets to check, overriding the
89 # default which is to re-check everything in the database. Each
90 # target already in the database results in the existing record being
91 # updated; each new target causes a new record to be added.
97 $this->log("irspy", "setting explicit list of targets ",
98 join(", ", map { "'$_'" } @targets));
99 $this->{allrecords} = 0;
101 foreach my $target (@targets) {
102 my($host, $port, $db, $newtarget) = _parse_target_string($target);
103 if ($newtarget ne $target) {
104 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
105 $target = $newtarget; # This is written through the ref
107 push @qlist, (qq[(host="$host" and port="$port" and path="$db")]);
110 $this->{targets} = \@targets;
111 $this->{query} = join(" or ", @qlist);
115 # Also used by ZOOM::IRSpy::Record
116 sub _parse_target_string {
119 my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
120 if (!defined $host) {
122 ($host, $db) = ($target =~ /(.*?)\/(.*)/);
123 $target = "$host:$port/$db";
125 die "$0: invalid target string '$target'"
128 return ($host, $port, $db, $target);
132 # There are two cases.
134 # 1. A specific set of targets is nominated on the command line.
135 # - Records must be fetched for those targets that are in the DB
136 # - New, empty records must be made for those that are not.
137 # - Updated records written to the DB may or may not be new.
139 # 2. All records in the database are to be checked.
140 # - Records must be fetched for all targets in the DB
141 # - Updated records written to the DB may not be new.
143 # That's all -- what could be simpler?
149 if ($this->{allrecords}) {
150 # We need to check on every target in the database, which
151 # means we need to do a "find all". According to the BIB-1
152 # semantics document at
153 # http://www.loc.gov/z3950/agency/bib1.html
155 # @attr 2=103 @attr 1=1035 x
156 # should find all records, but it seems that Zebra doesn't
157 # support this. Furthermore, when using the "alvis" filter
158 # (as we do for IRSpy) it doesn't support the use of any BIB-1
159 # access point -- not even 1035 "everywhere" -- so instead we
160 # hack together a search that we know will find all records.
161 $this->{query} = "port=?*";
163 # Prepopulate the target map with nulls so that after we fill
164 # in what we can from the database query, we know which target
165 # IDs we need new records for.
166 foreach my $target (@{ $this->{targets} }) {
167 $target2record{lc($target)} = undef;
171 $this->log("irspy_debug", "query '", $this->{query}, "'");
172 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
173 delete $this->{query}; # No longer needed at all
174 $this->log("irspy_debug", "found ", $rs->size(), " target records");
175 foreach my $i (1 .. $rs->size()) {
176 my $target = _render_record($rs, $i-1, "id");
177 my $zeerex = _render_record($rs, $i-1, "zeerex");
178 #print STDERR "making '$target' record with '$zeerex'\n";
179 $target2record{lc($target)} =
180 new ZOOM::IRSpy::Record($this, $target, $zeerex);
181 push @{ $this->{targets} }, $target
182 if $this->{allrecords};
185 # Make records for targets not previously in the database
186 foreach my $target (keys %target2record) {
187 my $record = $target2record{$target};
188 if (!defined $record) {
189 $this->log("irspy_debug", "made new record for '$target'");
190 $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
192 $this->log("irspy_debug", "using existing record for '$target'");
197 foreach my $target (@{ $this->{targets} }) {
198 my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
199 $conn->option(host => $target);
200 my $record = delete $target2record{lc($target)};
201 $conn->record($record);
202 push @connections, $conn;
204 die("remaining target2record = { " .
205 join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
206 sort keys %target2record) . " }")
209 $this->{connections} = \@connections;
210 delete $this->{targets}; # The information is now in {connections}
215 my($rs, $which, $elementSetName) = @_;
217 # There is a slight race condition here on the element-set name,
218 # but it shouldn't be a problem as this is (currently) only called
219 # from parts of the program that run single-threaded.
220 my $old = $rs->option(elementSetName => $elementSetName);
221 my $rec = $rs->record($which);
222 $rs->option(elementSetName => $old);
224 return $rec->render();
228 sub _rewrite_record {
232 $conn->log("irspy", "rewriting XML record");
233 my $rec = $conn->record();
234 my $p = $this->{conn}->package();
235 $p->option(action => "specialUpdate");
236 my $xml = $rec->{zeerex}->toString();
237 $p->option(record => $xml);
241 $p = $this->{conn}->package();
248 print "Updated $conn with xml=<br/>\n<pre>$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 my $nconn = scalar(@copy_conn);
296 foreach my $i0 (0 .. $#copy_conn) {
297 my $conn = $copy_conn[$i0];
298 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
299 if (!$conn->current_task()) {
300 if (!$conn->next_task()) {
301 # Out of tasks: we need a new test
303 my $address = $conn->option("current_test_address");
305 if (!defined $address) {
308 $this->log("irspy_test",
309 "checking for next test after '$address'");
310 $nextaddr = $this->_next_test($address);
312 if (!defined $nextaddr) {
313 $conn->log("irspy", "has no more tests: removing");
314 splice @conn, $i0, 1;
315 $this->_rewrite_record($conn);
319 my $node = $this->{tree}->select($nextaddr)
320 or die "invalid nextaddr '$nextaddr'";
321 $conn->option(current_test_address => $nextaddr);
322 my $tname = $node->name();
323 $conn->log("irspy_test",
324 "starting test '$nextaddr' = $tname");
325 my $tasks = $conn->tasks();
326 my $oldcount = @$tasks;
327 "ZOOM::IRSpy::Test::$tname"->start($conn);
328 $tasks = $conn->tasks();
329 if (@$tasks > $oldcount) {
330 # Prepare to start the first of the newly added tasks
331 $conn->next_task($tasks->[$oldcount]);
333 $conn->log("irspy_task",
334 "no tasks added by new test $tname");
339 my $task = $conn->next_task();
340 die "no next task queued for $conn" if !defined $task;
341 $conn->log("irspy_task", "starting task $task");
343 $conn->current_task($task);
347 # Do we need to test $conn->is_idle()? I don't think so!
350 my $i0 = ZOOM::event(\@conn);
351 $this->log("irspy_event",
352 "ZOOM_event(", scalar(@conn), " connections) = $i0");
353 last if $i0 == 0 || $i0 == -3; # no events or no connections
354 my $conn = $conn[$i0-1];
355 my $ev = $conn->last_event();
356 my $evstr = ZOOM::event_str($ev);
357 $conn->log("irspy_event", "event $ev ($evstr)");
359 my $task = $conn->current_task();
360 die "$conn has no current task for event $ev ($evstr)" if !$task;
361 eval { $conn->_check() };
363 ($ev == ZOOM::Event::RECV_DATA ||
364 $ev == ZOOM::Event::RECV_APDU ||
365 $ev == ZOOM::Event::ZEND)) {
366 # An error in, say, a search response, becomes visible to
367 # ZOOM before the Receive Data event is sent and persists
368 # until after the End, which means that successive events
369 # each report the same error. So we just ignore errors on
370 # "unimportant" events. ### But this doesn't work for,
371 # say, a Connection Refused, as the only event that shows
372 # us this error is the End.
373 $conn->log("irspy_event", "ignoring error ",
374 "on event $ev ($evstr): $@");
380 my $sub = $task->{cb}->{exception};
381 die $@ if !defined $sub;
382 $res = &$sub($conn, $task, $task->udata(), $@);
384 my $sub = $task->{cb}->{$ev};
386 $conn->log("irspy_unhandled", "event $ev ($evstr)");
390 $res = &$sub($conn, $task, $task->udata(), $ev);
393 if ($res == ZOOM::IRSpy::Status::OK) {
394 # Nothing to do -- life continues
396 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
397 my $task = $conn->current_task();
398 die "no task for TASK_DONE on $conn" if !$task;
399 die "next task already defined for $conn" if $conn->next_task();
400 $conn->log("irspy_task", "completed task $task");
401 $conn->next_task($task->{next});
402 $conn->current_task(0);
404 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
405 $res == ZOOM::IRSpy::Status::TEST_BAD) {
406 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
407 $conn->log("irspy_task", "test ended during task $task ($x)");
408 $conn->log("irspy_test", "test completed ($x)");
409 $conn->current_task(0);
411 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
412 my $address = $conn->option('current_test_address');
413 ($address, my $n) = $this->_last_sibling_test($address);
414 if (defined $address) {
415 $conn->log("irspy_test", "skipped $n tests");
416 $conn->option(current_test_address => $address);
423 $this->log("irspy", "exiting main loop");
430 my($tname, @ancestors) = @_;
432 die("$0: test-hierarchy loop detected: " .
433 join(" -> ", @ancestors, $tname))
434 if grep { $_ eq $tname } @ancestors;
437 my $slashSeperatedTname = $tname;
438 $slashSeperatedTname =~ s/::/\//g;
439 require "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
441 $this->log("warn", "can't load test '$tname': skipping",
442 $@ =~ /^Can.t locate/ ? () : " ($@)");
446 $this->log("irspy", "adding test '$tname'");
448 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
449 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
450 push @subnodes, $subtest if defined $subtest;
453 return new ZOOM::IRSpy::Node($tname, @subnodes);
457 # These next three should arguably be Node methods
460 my($address, $omit_child) = @_;
464 my $maybe = $address eq "" ? "0" : "$address:0";
465 return $maybe if $this->{tree}->select($maybe);
468 # The top-level node has no successor or parent
469 return undef if $address eq "";
471 # Try next sibling child
472 my @components = split /:/, $address;
473 my $last = pop @components;
474 my $maybe = join(":", @components, $last+1);
475 return $maybe if $this->{tree}->select($maybe);
477 # This node is exhausted: try the parent's successor
478 return $this->_next_test(join(":", @components), 1)
482 sub _last_sibling_test {
487 if !defined $this->_next_sibling_test($address);
491 my $maybe = $this->_next_sibling_test($address);
492 last if !defined $maybe;
494 $this->log("irspy", "skipping $nskipped = '$address'");
498 return ($address, $nskipped);
502 sub _next_sibling_test {
506 my @components = split /:/, $address;
507 my $last = pop @components;
508 my $maybe = join(":", @components, $last+1);
509 return $maybe if $this->{tree}->select($maybe);
519 ZOOM::IRSpy::Maintenance.
521 The ZOOM-Perl module,
522 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
525 http://indexdata.com/zebra/
529 Mike Taylor, E<lt>mike@indexdata.comE<gt>
531 =head1 COPYRIGHT AND LICENSE
533 Copyright (C) 2006 by Index Data ApS.
535 This library is free software; you can redistribute it and/or modify
536 it under the same terms as Perl itself, either Perl version 5.8.7 or,
537 at your option, any later version of Perl 5 you may have available.