1 # $Id: IRSpy.pm,v 1.90 2008-07-16 11:42:13 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 calc_reliability_stats
25 our $VERSION = '1.02';
26 our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
28 our $xslt_max_depth = 250;
31 # Enumeration for callback functions to return
32 package ZOOM::IRSpy::Status;
33 sub OK { 29 } # No problems, task is still progressing
34 sub TASK_DONE { 18 } # Task is complete, next task should begin
35 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
36 sub TEST_BAD { 31 } # Whole test is complete, and failed
37 sub TEST_SKIPPED { 12 } # Test couldn't be run
43 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
48 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
49 $spy->targets(@targets);
50 $spy->initialise("Main");
55 This module exists to implement the IRspy program, which discovers,
56 analyses and monitors IR servers implementing the Z39.50 and SRU/W
57 protocols. It is a successor to the ZSpy program.
62 ZOOM::Log::mask_str("irspy");
63 ZOOM::Log::mask_str("irspy_debug");
64 ZOOM::Log::mask_str("irspy_event");
65 ZOOM::Log::mask_str("irspy_unhandled");
66 ZOOM::Log::mask_str("irspy_test");
67 ZOOM::Log::mask_str("irspy_task");
72 my($dbname, $user, $password, $activeSetSize) = @_;
76 push @options, (user => $user, password => $password)
79 my $conn = new ZOOM::Connection($dbname, 0, @options)
80 or die "$0: can't connection to IRSpy database 'dbname'";
82 my $xslt = new XML::LibXSLT;
84 # raise the maximum number of nested template calls and variables/params (default 250)
85 warn "raise the maximum number of nested template calls: $xslt_max_depth\n" if $debug;
86 $xslt->max_depth($xslt_max_depth);
88 $xslt->register_function($ZOOM::IRSpy::Utils::IRSPY_NS, 'strcmp',
89 \&ZOOM::IRSpy::Utils::xslt_strcmp);
91 my $libxml = new XML::LibXML;
92 warn "use irspy_to_zeerex_xsl xslt sheet: $irspy_to_zeerex_xsl\n" if $debug;
93 my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl);
94 my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc);
98 query => "cql.allRecords=1", # unless overridden
99 modn => undef, # Filled in by restrict_modulo()
100 modi => undef, # Filled in by restrict_modulo()
101 targets => undef, # Filled in later if targets() is
102 # called; used only to keep state from
103 # targets() until initialise() is
105 connections => undef, # Filled in by initialise()
106 queue => undef, # Filled in by initialise()
108 irspy_to_zeerex_style => $irspy_to_zeerex_style,
109 test => undef, # Filled in by initialise()
110 timeout => undef, # Filled in by initialise()
111 tests => undef, # Tree of tests to be executed
112 activeSetSize => defined $activeSetSize ? $activeSetSize : 10,
114 $this->log("irspy", "starting up with database '$dbname'");
130 $this->{query} = $query;
134 # Explicitly nominate a set of targets to check, overriding the
135 # default which is to re-check everything in the database. Each
136 # target already in the database results in the existing record being
137 # updated; each new target causes a new record to be added.
143 $this->log("irspy", "setting explicit list of targets ",
144 join(", ", map { "'$_'" } @targets));
146 foreach my $target (@targets) {
147 my($protocol, $host, $port, $db, $newtarget) =
148 _parse_target_string($target);
149 if ($newtarget ne $target) {
150 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
151 $target = $newtarget; # This is written through the ref
153 push @qlist, cql_target($protocol, $host, $port, $db);
156 $this->{targets} = \@targets;
157 $this->{query} = join(" or ", @qlist);
161 # Also used by ZOOM::IRSpy::Record
162 sub _parse_target_string {
165 my($protocol, $host, $port, $db) = ($target =~ /(.*?):(.*?):(.*?)\/(.*)/);
166 if (!defined $host) {
168 ($protocol, $host, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
169 $target = irspy_make_identifier($protocol, $host, $port, $db);
171 die "$0: invalid target string '$target'"
174 return ($protocol, $host, $port, $db, $target);
178 sub restrict_modulo {
187 # Records must be fetched for all records satisfying $this->{query} If
188 # $this->{targets} is already set (i.e. a specific list of targets to
189 # check was specified by a call to targets()), then new, empty records
190 # will be made for any targets that are not already in the database.
196 $tname = "Main" if !defined $tname;
197 $this->{test} = $tname;
198 $this->{tree} = $this->_gather_tests($tname)
199 or die "No tests defined for '$tname'";
200 $this->{tree}->resolve();
201 #$this->{tree}->print(0);
203 $this->{timeout} = "ZOOM::IRSpy::Test::$tname"->timeout();
206 my $targets = $this->{targets};
207 if (defined $targets) {
208 @targets = @$targets;
209 delete $this->{targets};
211 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
212 $this->log("irspy", "'", $this->{query}, "' found ",
213 $rs->size(), " target records");
214 delete $this->{query};
216 foreach my $i (1 .. $rs->size()) {
217 push @targets, render_record($rs, $i-1, "id");
221 my $n = $this->{activeSetSize};
222 $n = @targets if $n == 0 || $n > @targets;
224 $this->{queue} = \@targets;
225 $this->{connections} = [];
226 while (@{ $this->{connections} } < $n) {
227 my $conn = $this->_next_connection();
228 last if !defined $conn;
229 push @{ $this->{connections} }, $conn;
234 sub _next_connection {
238 my $n = $this->{modn};
239 my $i = $this->{modi};
241 $target = shift @{ $this->{queue} };
242 return undef if !defined $target;
245 $target = shift @{ $this->{queue} };
246 return undef if !defined $target;
247 my $h = _hash($target);
249 last if $hmodn == $i;
250 #$this->log("irspy", "'$target' hash $h % $n = $hmodn != $i");
254 die "oops -- target is undefined" if !defined $target;
255 return create ZOOM::IRSpy::Connection($this, $target, async => 1,
256 timeout => $this->{timeout});
264 foreach my $s (split //, $target) {
272 sub _irspy_to_zeerex {
276 my $save_xml = $ENV{IRSPY_SAVE_XML};
277 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
280 unlink('/tmp/irspy_orig.xml');
281 open FH, '>/tmp/irspy_orig.xml'
282 or die "can't write irspy_orig.xml: $!";
283 print FH $irspy_doc->toString();
287 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
289 unlink('/tmp/irspy_transformed.xml');
290 open FH, '>/tmp/irspy_transformed.xml'
291 or die "can't write irspy_transformed.xml: $!";
292 print FH $result->toString();
296 return $result->documentElement();
300 sub _rewrite_irspy_record {
304 $conn->log("irspy", "rewriting XML record");
305 my $rec = $this->_irspy_to_zeerex($conn);
307 # Since IRSpy can run for a long time between writes back to the
308 # database, it's quite possible for the server to have closed the
309 # connection as idle. So re-establish it if necessary.
310 $this->{conn}->connect($conn->option("host"));
312 _rewrite_zeerex_record($this->{conn}, $rec);
313 $conn->log("irspy", "rewrote XML record");
317 my $_reliabilityField = {
318 reliability => [ reliability => 0,
319 "Calculated reliability of server",
320 "e:serverInfo/e:reliability" ],
323 sub _rewrite_zeerex_record {
324 my($conn, $rec, $oldid) = @_;
326 # Add reliability score
327 my $xc = irspy_xpath_context($rec);
328 my($nok, $nall, $percent) = calc_reliability_stats($xc);
329 modify_xml_document($xc, $_reliabilityField, { reliability => $percent });
331 my $p = $conn->package();
332 $p->option(action => "specialUpdate");
333 my $xml = $rec->toString();
334 $p->option(record => $xml);
338 # This is the expression in the ID-making stylesheet
339 # ../../zebra/zeerex2id.xsl
340 my $id = irspy_record2identifier($xc);
341 if (defined $oldid && $id ne $oldid) {
342 warn "IDs differ (old='$oldid' new='$id')";
343 _delete_record($conn, $oldid);
346 $p = $conn->package();
353 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
361 # We can't delete records using recordIdOpaque, since character
362 # sets are handled differently here in extended services from how
363 # they are used in the Alvis filter's record-parsing, and so
364 # non-ASCII characters come out differently in the two contexts.
365 # Instead, we must send a record whose contents indicate the ID of
366 # that which we wish to delete. There are two ways, both
367 # unsatisfactory: we could either fetch the actual record them
368 # resubmit it in the deletion request (which wastes a search and a
369 # fetch) or we could build a record by hand from the parsed-out
370 # components (which is error-prone and which I am not 100% certain
371 # will work since the other contents of the record will be
372 # different). The former evil seems to be the lesser.
374 warn "$conn deleting record '$id'";
376 my $rs = $conn->search(new ZOOM::Query::CQL(cql_target($id)));
377 die "no such ID '$id'" if $rs->size() == 0;
378 my $rec = $rs->record(0);
379 my $xml = $rec->render();
381 my $p = $conn->package();
382 $p->option(action => "recordDelete");
383 $p->option(record => $xml);
387 $p = $conn->package();
393 # The approach: gather declarative information about test hierarchy,
394 # then go into a loop. In the loop, we ensure that each connection is
395 # running a test, and within that test a task, until its list of tests
396 # is exhausted. No individual test ever calls wait(): tests just queue
397 # up tasks and return immediately. When the tasks are run (one at a
398 # time on each connection) they generate events, and it is these that
399 # are harvested by ZOOM::event(). Since each connection knows what
400 # task it is running, it can invoke the appropriate callbacks.
401 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
402 # loop how to continue.
405 # While a connection is running a task, its current_task()
406 # points at the task structure. When it finishes its task,
407 # next_task() is pointed at the next task to execute (if there
408 # is one), and its current_task() is set to zero. When the next
409 # task is executed, the connection's next_task() is set to zero
410 # and its current_task() pointed to the task structure.
411 # current_task() and next_task() are both zero only when there
412 # are no more queued tasks, which is when a new test is
415 # Each connection's current test is stored in its
416 # "current_test_address" option. The next test to execute is
417 # calculated by walking the declarative tree of tests. This
418 # option begins empty; the "next test" after this is of course
424 my $topname = $this->{tree}->name();
425 my $timeout = $this->{timeout};
426 $this->log("irspy", "beginnning with test '$topname' (timeout $timeout)");
429 my @conn = @{ $this->{connections} };
432 ROUND_AND_ROUND_WE_GO:
434 my @copy_conn = @conn; # avoid alias problems after splice()
435 my $nconn = scalar(@copy_conn);
436 foreach my $i0 (0 .. $#copy_conn) {
437 my $conn = $copy_conn[$i0];
438 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
439 next if !defined $conn;
440 if (!$conn->current_task()) {
441 if (!$conn->next_task()) {
442 # Out of tasks: we need a new test
444 my $address = $conn->option("current_test_address");
446 if (!defined $address) {
449 $conn->log("irspy_test",
450 "checking for next test after '$address'");
451 $nextaddr = $this->_next_test($address);
453 if (!defined $nextaddr) {
454 $conn->log("irspy", "has no more tests: removing");
455 $this->_rewrite_irspy_record($conn);
456 $conn->option(rewrote_record => 1);
457 my $newconn = $this->_next_connection();
458 if (!defined $newconn) {
459 # Do not destroy: needed for later sanity checks
460 splice @conn, $i0, 1;
463 $conn[$i0] = $newconn;
464 $conn[$i0]->option(current_test_address => "");
465 $conn[$i0]->log("irspy", "entering active pool - ",
466 scalar(@{ $this->{queue} }),
467 " targets remain in queue");
472 my $node = $this->{tree}->select($nextaddr)
473 or die "invalid nextaddr '$nextaddr'";
474 $conn->option(current_test_address => $nextaddr);
475 my $tname = $node->name();
476 $conn->log("irspy_test",
477 "starting test '$nextaddr' = $tname");
478 my $tasks = $conn->tasks();
479 my $oldcount = @$tasks;
480 "ZOOM::IRSpy::Test::$tname"->start($conn);
481 $tasks = $conn->tasks();
482 if (@$tasks > $oldcount) {
483 # Prepare to start the first of the newly added tasks
484 $conn->next_task($tasks->[$oldcount]);
486 $conn->log("irspy_task",
487 "no tasks added by new test $tname");
492 my $task = $conn->next_task();
493 die "no next task queued for $conn" if !defined $task;
494 $conn->log("irspy_task", "preparing task $task");
496 $conn->current_task($task);
502 my $i0 = ZOOM::event(\@conn);
503 $this->log("irspy_event",
504 "ZOOM_event(", scalar(@conn), " connections) = $i0");
507 0 => "no events remain",
508 -1 => "ZOOM::event() argument not a reference",
509 -2 => "ZOOM::event() reference not an array",
510 -3 => "no connections remain",
511 -4 => "too many connections for ZOOM::event()",
513 my $message = $messages{$i0} || "ZOOM::event() returned $i0";
514 $this->log("irspy", $message);
518 my $conn = $conn[$i0-1];
519 my $ev = $conn->last_event();
520 my $evstr = ZOOM::event_str($ev);
521 $conn->log("irspy_event", "event $ev ($evstr)");
522 goto NEXT_EVENT if $ev != ZOOM::Event::ZEND;
524 my $task = $conn->current_task();
525 die "$conn has no current task for event $ev ($evstr)" if !$task;
528 eval { $conn->check() };
529 if ($@ && ref $@ && $@->isa("ZOOM::Exception")) {
530 my $sub = $task->{cb}->{exception};
531 die $@ if !defined $sub;
532 $res = &$sub($conn, $task, $task->udata(), $@);
534 die "Unexpected non-ZOOM exception: " . ref($@) . " ($@)";
536 my $sub = $task->{cb}->{$ev};
538 $conn->log("irspy_unhandled", "event $ev ($evstr)");
542 $res = &$sub($conn, $task, $task->udata(), $ev);
545 if ($res == ZOOM::IRSpy::Status::OK) {
546 # Nothing to do -- life continues
548 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
549 my $task = $conn->current_task();
550 die "no task for TASK_DONE on $conn" if !$task;
551 die "next task already defined for $conn" if $conn->next_task();
552 $conn->log("irspy_task", "completed task $task");
553 $conn->next_task($task->{next});
554 $conn->current_task(0);
556 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
557 $res == ZOOM::IRSpy::Status::TEST_BAD) {
558 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
559 $conn->log("irspy_task", "test ended during task $task ($x)");
560 $conn->log("irspy_test", "test completed ($x)");
561 $conn->current_task(0);
563 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
564 my $address = $conn->option('current_test_address');
565 $conn->log("irspy", "top-level test failed!")
567 my $node = $this->{tree}->select($address);
569 while (defined $node->next() &&
570 length($node->next()->address()) >= length($address)) {
571 $conn->log("irspy_debug", "skipping from '",
572 $node->address(), "' to '",
573 $node->next()->address(), "'");
574 $node = $node->next();
578 $conn->option(current_test_address => $node->address());
579 $conn->log("irspy_test", "skipped $skipcount tests");
580 $nskipped += $skipcount;
583 } elsif ($res == ZOOM::IRSpy::Status::TEST_SKIPPED) {
584 $conn->log("irspy_test", "test skipped during task $task");
585 $conn->current_task(0);
590 die "unknown callback return-value '$res'";
594 $this->log("irspy", "exiting main loop");
596 # Sanity checks: none of the following should ever happen
598 $this->log("irspy", "performing end-of-run sanity-checks");
599 foreach my $conn (@conn) {
600 my $test = $conn->option("current_test_address");
601 my $next = $this->_next_test($test);
604 "$conn (in test '$test') has queued test '$next'");
607 if (my $task = $conn->current_task()) {
608 $this->log("irspy", "$conn still has an active task $task");
611 if (my $task = $conn->next_task()) {
612 $this->log("irspy", "$conn still has a queued task $task");
615 if (!$conn->is_idle()) {
617 "$conn still has ZOOM-C level tasks queued: see below");
620 my $ev = $conn->peek_event();
621 if ($ev != 0 && $ev != ZOOM::Event::ZEND) {
622 my $evstr = ZOOM::event_str($ev);
623 $this->log("irspy", "$conn has event $ev ($evstr) waiting");
626 if (!$conn->option("rewrote_record")) {
627 $this->log("irspy", "$conn did not rewrite its ZeeRex record");
632 # This really shouldn't be necessary, and in practice it rarely
633 # helps, but it's belt and braces. (For now, we don't do this
634 # hence the zero in the $nruns check).
637 $this->log("irspy", "back into main loop, ${nruns}th time");
638 goto ROUND_AND_ROUND_WE_GO;
640 $this->log("irspy", "bailing after $nruns main-loop runs");
644 # This shouldn't happen emit anything either:
645 while ((my $i1 = ZOOM::event(\@conn)) > 0) {
646 my $conn = $conn[$i1-1];
647 my $ev = $conn->last_event();
648 my $evstr = ZOOM::event_str($ev);
650 "$conn still has ZOOM-C level task queued: $ev ($evstr)")
651 if $ev != ZOOM::Event::ZEND;
658 # Exactly equivalent to ZOOM::event() except that it is tolerant to
659 # undefined values in the array being passed in.
661 sub __UNUSED_tolerant_ZOOM_event {
665 foreach my $i (0 .. @$connref-1) {
666 my $conn = $connref->[$i];
673 my $res = ZOOM::event(\@conn);
674 return $res if $res <= 0;
675 my $res2 = $map[$res-1] + 1;
676 print STDERR "*** tolerant_ZOOM_event() returns $res->$res2\n";
683 my($tname, @ancestors) = @_;
685 die("$0: test-hierarchy loop detected: " .
686 join(" -> ", @ancestors, $tname))
687 if grep { $_ eq $tname } @ancestors;
689 my $slashSeperatedTname = $tname;
690 $slashSeperatedTname =~ s/::/\//g;
691 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
696 $this->log("irspy", "couldn't require '$fullName': $@");
697 $this->log("warn", "can't load test '$tname': skipping",
698 $@ =~ /^Can.t locate/ ? () : " ($@)");
702 $this->log("irspy", "adding test '$tname'");
704 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
705 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
706 push @subnodes, $subtest if defined $subtest;
709 return new ZOOM::IRSpy::Node($tname, @subnodes);
713 # These next three should arguably be Node methods
716 my($address, $omit_child) = @_;
720 my $maybe = $address eq "" ? "0" : "$address:0";
721 return $maybe if $this->{tree}->select($maybe);
724 # The top-level node has no successor or parent
725 return undef if $address eq "";
727 # Try next sibling child
728 my @components = split /:/, $address;
729 my $last = pop @components;
730 my $maybe = join(":", @components, $last+1);
731 return $maybe if $this->{tree}->select($maybe);
733 # This node is exhausted: try the parent's successor
734 return $this->_next_test(join(":", @components), 1)
738 sub _last_sibling_test {
743 if !defined $this->_next_sibling_test($address);
747 my $maybe = $this->_next_sibling_test($address);
748 last if !defined $maybe;
751 $this->log("irspy", "skipping $nskipped tests to '$address'");
754 return ($address, $nskipped);
758 sub _next_sibling_test {
762 my @components = split /:/, $address;
763 my $last = pop @components;
764 my $maybe = join(":", @components, $last+1);
765 return $maybe if $this->{tree}->select($maybe);
775 ZOOM::IRSpy::Maintenance.
777 The ZOOM-Perl module,
778 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
781 http://indexdata.com/zebra/
785 Mike Taylor, E<lt>mike@indexdata.comE<gt>
787 =head1 COPYRIGHT AND LICENSE
789 Copyright (C) 2006 by Index Data ApS.
791 This library is free software; you can redistribute it and/or modify
792 it under the same terms as Perl itself, either Perl version 5.8.7 or,
793 at your option, any later version of Perl 5 you may have available.