1 # $Id: IRSpy.pm,v 1.89 2007-09-20 10:14:37 mike Exp $
9 use Data::Dumper; # For debugging only
13 use XML::LibXML::XPathContext;
15 use Net::Z3950::ZOOM 1.13; # For the ZOOM version-check only
16 use ZOOM::IRSpy::Node;
17 use ZOOM::IRSpy::Connection;
18 use ZOOM::IRSpy::Stats;
19 use ZOOM::IRSpy::Utils qw(cql_target render_record
20 irspy_xpath_context irspy_make_identifier
21 irspy_record2identifier);
24 our $VERSION = '1.01';
25 our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
28 # Enumeration for callback functions to return
29 package ZOOM::IRSpy::Status;
30 sub OK { 29 } # No problems, task is still progressing
31 sub TASK_DONE { 18 } # Task is complete, next task should begin
32 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
33 sub TEST_BAD { 31 } # Whole test is complete, and failed
34 sub TEST_SKIPPED { 12 } # Test couldn't be run
40 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
45 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
46 $spy->targets(@targets);
47 $spy->initialise("Main");
52 This module exists to implement the IRspy program, which discovers,
53 analyses and monitors IR servers implementing the Z39.50 and SRU/W
54 protocols. It is a successor to the ZSpy program.
59 ZOOM::Log::mask_str("irspy");
60 ZOOM::Log::mask_str("irspy_debug");
61 ZOOM::Log::mask_str("irspy_event");
62 ZOOM::Log::mask_str("irspy_unhandled");
63 ZOOM::Log::mask_str("irspy_test");
64 ZOOM::Log::mask_str("irspy_task");
69 my($dbname, $user, $password, $activeSetSize) = @_;
72 push @options, (user => $user, password => $password)
75 my $conn = new ZOOM::Connection($dbname, 0, @options)
76 or die "$0: can't connection to IRSpy database 'dbname'";
78 my $xslt = new XML::LibXSLT;
80 $xslt->register_function($ZOOM::IRSpy::Utils::IRSPY_NS, 'strcmp',
81 \&ZOOM::IRSpy::Utils::xslt_strcmp);
83 my $libxml = new XML::LibXML;
84 my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl);
85 my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc);
89 query => "cql.allRecords=1", # unless overridden
90 modn => undef, # Filled in by restrict_modulo()
91 modi => undef, # Filled in by restrict_modulo()
92 targets => undef, # Filled in later if targets() is
93 # called; used only to keep state from
94 # targets() until initialise() is
96 connections => undef, # Filled in by initialise()
97 queue => undef, # Filled in by initialise()
99 irspy_to_zeerex_style => $irspy_to_zeerex_style,
100 test => undef, # Filled in by initialise()
101 timeout => undef, # Filled in by initialise()
102 tests => undef, # Tree of tests to be executed
103 activeSetSize => defined $activeSetSize ? $activeSetSize : 10,
105 $this->log("irspy", "starting up with database '$dbname'");
121 $this->{query} = $query;
125 # Explicitly nominate a set of targets to check, overriding the
126 # default which is to re-check everything in the database. Each
127 # target already in the database results in the existing record being
128 # updated; each new target causes a new record to be added.
134 $this->log("irspy", "setting explicit list of targets ",
135 join(", ", map { "'$_'" } @targets));
137 foreach my $target (@targets) {
138 my($protocol, $host, $port, $db, $newtarget) =
139 _parse_target_string($target);
140 if ($newtarget ne $target) {
141 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
142 $target = $newtarget; # This is written through the ref
144 push @qlist, cql_target($protocol, $host, $port, $db);
147 $this->{targets} = \@targets;
148 $this->{query} = join(" or ", @qlist);
152 # Also used by ZOOM::IRSpy::Record
153 sub _parse_target_string {
156 my($protocol, $host, $port, $db) = ($target =~ /(.*?):(.*?):(.*?)\/(.*)/);
157 if (!defined $host) {
159 ($protocol, $host, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
160 $target = irspy_make_identifier($protocol, $host, $port, $db);
162 die "$0: invalid target string '$target'"
165 return ($protocol, $host, $port, $db, $target);
169 sub restrict_modulo {
178 # Records must be fetched for all records satisfying $this->{query} If
179 # $this->{targets} is already set (i.e. a specific list of targets to
180 # check was specified by a call to targets()), then new, empty records
181 # will be made for any targets that are not already in the database.
187 $tname = "Main" if !defined $tname;
188 $this->{test} = $tname;
189 $this->{tree} = $this->_gather_tests($tname)
190 or die "No tests defined for '$tname'";
191 $this->{tree}->resolve();
192 #$this->{tree}->print(0);
194 $this->{timeout} = "ZOOM::IRSpy::Test::$tname"->timeout();
197 my $targets = $this->{targets};
198 if (defined $targets) {
199 @targets = @$targets;
200 delete $this->{targets};
202 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
203 $this->log("irspy", "'", $this->{query}, "' found ",
204 $rs->size(), " target records");
205 delete $this->{query};
207 foreach my $i (1 .. $rs->size()) {
208 push @targets, render_record($rs, $i-1, "id");
212 my $n = $this->{activeSetSize};
213 $n = @targets if $n == 0 || $n > @targets;
215 $this->{queue} = \@targets;
216 $this->{connections} = [];
217 while (@{ $this->{connections} } < $n) {
218 my $conn = $this->_next_connection();
219 last if !defined $conn;
220 push @{ $this->{connections} }, $conn;
225 sub _next_connection {
229 my $n = $this->{modn};
230 my $i = $this->{modi};
232 $target = shift @{ $this->{queue} };
233 return undef if !defined $target;
236 $target = shift @{ $this->{queue} };
237 return undef if !defined $target;
238 my $h = _hash($target);
240 last if $hmodn == $i;
241 $this->log("irspy", "'$target' hash $h % $n = $hmodn != $i");
245 die "oops -- target is undefined" if !defined $target;
246 return create ZOOM::IRSpy::Connection($this, $target, async => 1,
247 timeout => $this->{timeout});
255 foreach my $s (split //, $target) {
263 sub _irspy_to_zeerex {
265 my($conn, $save_xml) = @_;
266 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
269 unlink('/tmp/irspy_orig.xml');
270 open FH, '>/tmp/irspy_orig.xml'
271 or die "can't write irspy_orig.xml: $!";
272 print FH $irspy_doc->toString();
276 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
278 unlink('/tmp/irspy_transformed.xml');
279 open FH, '>/tmp/irspy_transformed.xml'
280 or die "can't write irspy_transformed.xml: $!";
281 print FH $result->toString();
285 return $result->documentElement();
289 sub _rewrite_record {
293 $conn->log("irspy", "rewriting XML record");
294 my $rec = $this->_irspy_to_zeerex($conn, $ENV{IRSPY_SAVE_XML});
296 # Since IRSpy can run for a long time between writes back to the
297 # database, it's quite possible for the server to have closed the
298 # connection as idle. So re-establish it if necessary.
299 $this->{conn}->connect($conn->option("host"));
301 _really_rewrite_record($this->{conn}, $rec);
302 $conn->log("irspy", "rewrote XML record");
306 sub _really_rewrite_record {
307 my($conn, $rec, $oldid) = @_;
309 my $p = $conn->package();
310 $p->option(action => "specialUpdate");
311 my $xml = $rec->toString();
312 $p->option(record => $xml);
316 # This is the expression in the ID-making stylesheet
317 # ../../zebra/zeerex2id.xsl
318 my $xc = irspy_xpath_context($rec);
319 my $id = irspy_record2identifier($xc);
320 if (defined $oldid && $id ne $oldid) {
321 warn "IDs differ (old='$oldid' new='$id')";
322 _delete_record($conn, $oldid);
325 $p = $conn->package();
332 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
340 # We can't delete records using recordIdOpaque, since character
341 # sets are handled differently here in extended services from how
342 # they are used in the Alvis filter's record-parsing, and so
343 # non-ASCII characters come out differently in the two contexts.
344 # Instead, we must send a record whose contents indicate the ID of
345 # that which we wish to delete. There are two ways, both
346 # unsatisfactory: we could either fetch the actual record them
347 # resubmit it in the deletion request (which wastes a search and a
348 # fetch) or we could build a record by hand from the parsed-out
349 # components (which is error-prone and which I am not 100% certain
350 # will work since the other contents of the record will be
351 # different). The former evil seems to be the lesser.
353 warn "$conn deleting record '$id'";
355 my $rs = $conn->search(new ZOOM::Query::CQL(cql_target($id)));
356 die "no such ID '$id'" if $rs->size() == 0;
357 my $rec = $rs->record(0);
358 my $xml = $rec->render();
360 my $p = $conn->package();
361 $p->option(action => "recordDelete");
362 $p->option(record => $xml);
366 $p = $conn->package();
372 # The approach: gather declarative information about test hierarchy,
373 # then go into a loop. In the loop, we ensure that each connection is
374 # running a test, and within that test a task, until its list of tests
375 # is exhausted. No individual test ever calls wait(): tests just queue
376 # up tasks and return immediately. When the tasks are run (one at a
377 # time on each connection) they generate events, and it is these that
378 # are harvested by ZOOM::event(). Since each connection knows what
379 # task it is running, it can invoke the appropriate callbacks.
380 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
381 # loop how to continue.
384 # While a connection is running a task, its current_task()
385 # points at the task structure. When it finishes its task,
386 # next_task() is pointed at the next task to execute (if there
387 # is one), and its current_task() is set to zero. When the next
388 # task is executed, the connection's next_task() is set to zero
389 # and its current_task() pointed to the task structure.
390 # current_task() and next_task() are both zero only when there
391 # are no more queued tasks, which is when a new test is
394 # Each connection's current test is stored in its
395 # "current_test_address" option. The next test to execute is
396 # calculated by walking the declarative tree of tests. This
397 # option begins empty; the "next test" after this is of course
403 my $topname = $this->{tree}->name();
404 my $timeout = $this->{timeout};
405 $this->log("irspy", "beginnning with test '$topname' (timeout $timeout)");
408 my @conn = @{ $this->{connections} };
411 ROUND_AND_ROUND_WE_GO:
413 my @copy_conn = @conn; # avoid alias problems after splice()
414 my $nconn = scalar(@copy_conn);
415 foreach my $i0 (0 .. $#copy_conn) {
416 my $conn = $copy_conn[$i0];
417 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
418 next if !defined $conn;
419 if (!$conn->current_task()) {
420 if (!$conn->next_task()) {
421 # Out of tasks: we need a new test
423 my $address = $conn->option("current_test_address");
425 if (!defined $address) {
428 $conn->log("irspy_test",
429 "checking for next test after '$address'");
430 $nextaddr = $this->_next_test($address);
432 if (!defined $nextaddr) {
433 $conn->log("irspy", "has no more tests: removing");
434 $this->_rewrite_record($conn);
435 $conn->option(rewrote_record => 1);
436 my $newconn = $this->_next_connection();
437 if (!defined $newconn) {
438 # Do not destroy: needed for later sanity checks
439 splice @conn, $i0, 1;
442 $conn[$i0] = $newconn;
443 $conn[$i0]->option(current_test_address => "");
444 $conn[$i0]->log("irspy", "entering active pool - ",
445 scalar(@{ $this->{queue} }),
446 " targets remain in queue");
451 my $node = $this->{tree}->select($nextaddr)
452 or die "invalid nextaddr '$nextaddr'";
453 $conn->option(current_test_address => $nextaddr);
454 my $tname = $node->name();
455 $conn->log("irspy_test",
456 "starting test '$nextaddr' = $tname");
457 my $tasks = $conn->tasks();
458 my $oldcount = @$tasks;
459 "ZOOM::IRSpy::Test::$tname"->start($conn);
460 $tasks = $conn->tasks();
461 if (@$tasks > $oldcount) {
462 # Prepare to start the first of the newly added tasks
463 $conn->next_task($tasks->[$oldcount]);
465 $conn->log("irspy_task",
466 "no tasks added by new test $tname");
471 my $task = $conn->next_task();
472 die "no next task queued for $conn" if !defined $task;
473 $conn->log("irspy_task", "preparing task $task");
475 $conn->current_task($task);
481 my $i0 = ZOOM::event(\@conn);
482 $this->log("irspy_event",
483 "ZOOM_event(", scalar(@conn), " connections) = $i0");
486 0 => "no events remain",
487 -1 => "ZOOM::event() argument not a reference",
488 -2 => "ZOOM::event() reference not an array",
489 -3 => "no connections remain",
490 -4 => "too many connections for ZOOM::event()",
492 my $message = $messages{$i0} || "ZOOM::event() returned $i0";
493 $this->log("irspy", $message);
497 my $conn = $conn[$i0-1];
498 my $ev = $conn->last_event();
499 my $evstr = ZOOM::event_str($ev);
500 $conn->log("irspy_event", "event $ev ($evstr)");
501 goto NEXT_EVENT if $ev != ZOOM::Event::ZEND;
503 my $task = $conn->current_task();
504 die "$conn has no current task for event $ev ($evstr)" if !$task;
507 eval { $conn->check() };
508 if ($@ && ref $@ && $@->isa("ZOOM::Exception")) {
509 my $sub = $task->{cb}->{exception};
510 die $@ if !defined $sub;
511 $res = &$sub($conn, $task, $task->udata(), $@);
513 die "Unexpected non-ZOOM exception: " . ref($@) . " ($@)";
515 my $sub = $task->{cb}->{$ev};
517 $conn->log("irspy_unhandled", "event $ev ($evstr)");
521 $res = &$sub($conn, $task, $task->udata(), $ev);
524 if ($res == ZOOM::IRSpy::Status::OK) {
525 # Nothing to do -- life continues
527 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
528 my $task = $conn->current_task();
529 die "no task for TASK_DONE on $conn" if !$task;
530 die "next task already defined for $conn" if $conn->next_task();
531 $conn->log("irspy_task", "completed task $task");
532 $conn->next_task($task->{next});
533 $conn->current_task(0);
535 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
536 $res == ZOOM::IRSpy::Status::TEST_BAD) {
537 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
538 $conn->log("irspy_task", "test ended during task $task ($x)");
539 $conn->log("irspy_test", "test completed ($x)");
540 $conn->current_task(0);
542 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
543 my $address = $conn->option('current_test_address');
544 $conn->log("irspy", "top-level test failed!")
546 my $node = $this->{tree}->select($address);
548 while (defined $node->next() &&
549 length($node->next()->address()) >= length($address)) {
550 $conn->log("irspy_debug", "skipping from '",
551 $node->address(), "' to '",
552 $node->next()->address(), "'");
553 $node = $node->next();
557 $conn->option(current_test_address => $node->address());
558 $conn->log("irspy_test", "skipped $skipcount tests");
559 $nskipped += $skipcount;
562 } elsif ($res == ZOOM::IRSpy::Status::TEST_SKIPPED) {
563 $conn->log("irspy_test", "test skipped during task $task");
564 $conn->current_task(0);
569 die "unknown callback return-value '$res'";
573 $this->log("irspy", "exiting main loop");
575 # Sanity checks: none of the following should ever happen
577 $this->log("irspy", "performing end-of-run sanity-checks");
578 foreach my $conn (@conn) {
579 my $test = $conn->option("current_test_address");
580 my $next = $this->_next_test($test);
583 "$conn (in test '$test') has queued test '$next'");
586 if (my $task = $conn->current_task()) {
587 $this->log("irspy", "$conn still has an active task $task");
590 if (my $task = $conn->next_task()) {
591 $this->log("irspy", "$conn still has a queued task $task");
594 if (!$conn->is_idle()) {
596 "$conn still has ZOOM-C level tasks queued: see below");
599 my $ev = $conn->peek_event();
600 if ($ev != 0 && $ev != ZOOM::Event::ZEND) {
601 my $evstr = ZOOM::event_str($ev);
602 $this->log("irspy", "$conn has event $ev ($evstr) waiting");
605 if (!$conn->option("rewrote_record")) {
606 $this->log("irspy", "$conn did not rewrite its ZeeRex record");
611 # This really shouldn't be necessary, and in practice it rarely
612 # helps, but it's belt and braces. (For now, we don't do this
613 # hence the zero in the $nruns check).
616 $this->log("irspy", "back into main loop, ${nruns}th time");
617 goto ROUND_AND_ROUND_WE_GO;
619 $this->log("irspy", "bailing after $nruns main-loop runs");
623 # This shouldn't happen emit anything either:
624 while ((my $i1 = ZOOM::event(\@conn)) > 0) {
625 my $conn = $conn[$i1-1];
626 my $ev = $conn->last_event();
627 my $evstr = ZOOM::event_str($ev);
629 "$conn still has ZOOM-C level task queued: $ev ($evstr)")
630 if $ev != ZOOM::Event::ZEND;
637 # Exactly equivalent to ZOOM::event() except that it is tolerant to
638 # undefined values in the array being passed in.
640 sub __UNUSED_tolerant_ZOOM_event {
644 foreach my $i (0 .. @$connref-1) {
645 my $conn = $connref->[$i];
652 my $res = ZOOM::event(\@conn);
653 return $res if $res <= 0;
654 my $res2 = $map[$res-1] + 1;
655 print STDERR "*** tolerant_ZOOM_event() returns $res->$res2\n";
662 my($tname, @ancestors) = @_;
664 die("$0: test-hierarchy loop detected: " .
665 join(" -> ", @ancestors, $tname))
666 if grep { $_ eq $tname } @ancestors;
668 my $slashSeperatedTname = $tname;
669 $slashSeperatedTname =~ s/::/\//g;
670 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
675 $this->log("irspy", "couldn't require '$fullName': $@");
676 $this->log("warn", "can't load test '$tname': skipping",
677 $@ =~ /^Can.t locate/ ? () : " ($@)");
681 $this->log("irspy", "adding test '$tname'");
683 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
684 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
685 push @subnodes, $subtest if defined $subtest;
688 return new ZOOM::IRSpy::Node($tname, @subnodes);
692 # These next three should arguably be Node methods
695 my($address, $omit_child) = @_;
699 my $maybe = $address eq "" ? "0" : "$address:0";
700 return $maybe if $this->{tree}->select($maybe);
703 # The top-level node has no successor or parent
704 return undef if $address eq "";
706 # Try next sibling child
707 my @components = split /:/, $address;
708 my $last = pop @components;
709 my $maybe = join(":", @components, $last+1);
710 return $maybe if $this->{tree}->select($maybe);
712 # This node is exhausted: try the parent's successor
713 return $this->_next_test(join(":", @components), 1)
717 sub _last_sibling_test {
722 if !defined $this->_next_sibling_test($address);
726 my $maybe = $this->_next_sibling_test($address);
727 last if !defined $maybe;
730 $this->log("irspy", "skipping $nskipped tests to '$address'");
733 return ($address, $nskipped);
737 sub _next_sibling_test {
741 my @components = split /:/, $address;
742 my $last = pop @components;
743 my $maybe = join(":", @components, $last+1);
744 return $maybe if $this->{tree}->select($maybe);
754 ZOOM::IRSpy::Maintenance.
756 The ZOOM-Perl module,
757 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
760 http://indexdata.com/zebra/
764 Mike Taylor, E<lt>mike@indexdata.comE<gt>
766 =head1 COPYRIGHT AND LICENSE
768 Copyright (C) 2006 by Index Data ApS.
770 This library is free software; you can redistribute it and/or modify
771 it under the same terms as Perl itself, either Perl version 5.8.7 or,
772 at your option, any later version of Perl 5 you may have available.