8 use Data::Dumper; # For debugging only
12 use XML::LibXML::XPathContext;
14 use Net::Z3950::ZOOM 1.13; # For the ZOOM version-check only
15 use ZOOM::IRSpy::Node;
16 use ZOOM::IRSpy::Connection;
17 use ZOOM::IRSpy::Stats;
18 use ZOOM::IRSpy::Utils qw(cql_target render_record
19 irspy_xpath_context irspy_make_identifier
20 irspy_record2identifier calc_reliability_stats
24 our $VERSION = '1.02';
25 our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
27 our $xslt_max_depth = 250;
28 our $max_timeout_errors = 3;
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,
113 rules => [], # Can be filled by apply_rules()
114 vars => {}, # Will contain name, id, version, etc.
115 # May be written into by tests (mostly
116 # Ping, which gets the Init response)
117 # and read by tests for rules.
119 $this->log("irspy", "starting up with database '$dbname'");
124 # Read from and write to the set of vars
127 my($key, $newval) = @_;
129 if (defined $newval) {
130 $this->{vars}->{$key} = $newval;
131 #warn "$this: set '$key'='$newval'";
133 my $oldval = $this->{vars}->{$key};
134 $oldval = "" if !defined $oldval;
135 #warn "$this: got '$key'='$oldval'";
140 # wrapper to read the IRSpy database name from environment variable / apache config
141 sub connect_to_registry {
144 # XXX: we could also handle her: user, password, elementSetName
146 my $database = $ENV{IRSpyDbName} || "localhost:8018/IR-Explain---1";
161 $this->{query} = $query;
165 # Explicitly nominate a set of targets to check, overriding the
166 # default which is to re-check everything in the database. Each
167 # target already in the database results in the existing record being
168 # updated; each new target causes a new record to be added.
174 $this->log("irspy", "setting explicit list of targets ",
175 join(", ", map { "'$_'" } @targets));
177 foreach my $target (@targets) {
178 my($protocol, $host, $port, $db, $newtarget) =
179 _parse_target_string($target);
180 if ($newtarget ne $target) {
181 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
182 $target = $newtarget; # This is written through the ref
184 push @qlist, cql_target($protocol, $host, $port, $db);
187 $this->{targets} = \@targets;
188 $this->{query} = join(" or ", @qlist);
192 # Also used by ZOOM::IRSpy::Record
193 sub _parse_target_string {
196 my($protocol, $host, $port, $db) = ($target =~ /(.*?):(.*?):(.*?)\/(.*)/);
197 if (!defined $host) {
199 ($protocol, $host, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
200 $target = irspy_make_identifier($protocol, $host, $port, $db);
202 die "$0: invalid target string '$target'"
205 return ($protocol, $host, $port, $db, $target);
209 sub restrict_modulo {
222 my $f = new IO::File("<$fileName")
223 or die "$0: can't open rules file '$fileName': $!";
225 while (my $line = <$f>) {
230 my(@fields) = split /\s+/, $line;
231 my($name, $op, $val, $rule, @params) = @fields;
232 #warn "parsed name='$name', op='$op', val='$val', rule='$rule', params='@params'\n";
233 push @{ $this->{rules} }, [ @fields ];
238 # Records must be fetched for all records satisfying $this->{query} If
239 # $this->{targets} is already set (i.e. a specific list of targets to
240 # check was specified by a call to targets()), then new, empty records
241 # will be made for any targets that are not already in the database.
247 $tname = "Main" if !defined $tname;
248 $this->{test} = $tname;
249 $this->{tree} = $this->_gather_tests($tname)
250 or die "No tests defined for '$tname'";
251 $this->{tree}->resolve();
252 #$this->{tree}->print(0);
254 $this->{timeout} = "ZOOM::IRSpy::Test::$tname"->timeout();
257 my $targets = $this->{targets};
258 if (defined $targets) {
259 @targets = @$targets;
260 delete $this->{targets};
262 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
263 $this->log("irspy", "'", $this->{query}, "' found ",
264 $rs->size(), " target records");
265 delete $this->{query};
267 foreach my $i (1 .. $rs->size()) {
268 push @targets, render_record($rs, $i-1, "id");
272 my $n = $this->{activeSetSize};
273 $n = @targets if $n == 0 || $n > @targets;
275 $this->{queue} = \@targets;
276 $this->{connections} = [];
277 while (@{ $this->{connections} } < $n) {
278 my $conn = $this->_next_connection();
279 last if !defined $conn;
280 push @{ $this->{connections} }, $conn;
285 sub _next_connection {
289 my $n = $this->{modn};
290 my $i = $this->{modi};
292 $target = shift @{ $this->{queue} };
293 return undef if !defined $target;
296 $target = shift @{ $this->{queue} };
297 return undef if !defined $target;
298 my $h = _hash($target);
300 last if $hmodn == $i;
301 #$this->log("irspy", "'$target' hash $h % $n = $hmodn != $i");
305 die "oops -- target is undefined" if !defined $target;
306 return create ZOOM::IRSpy::Connection($this, $target, async => 1,
307 timeout => $this->{timeout});
315 foreach my $s (split //, $target) {
323 sub _irspy_to_zeerex {
327 my $save_xml = $ENV{IRSPY_SAVE_XML};
328 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
331 unlink('/tmp/irspy_orig.xml');
332 open FH, '>/tmp/irspy_orig.xml'
333 or die "can't write irspy_orig.xml: $!";
334 print FH $irspy_doc->toString();
338 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
340 unlink('/tmp/irspy_transformed.xml');
341 open FH, '>/tmp/irspy_transformed.xml'
342 or die "can't write irspy_transformed.xml: $!";
343 print FH $result->toString();
347 return $result->documentElement();
351 sub _rewrite_irspy_record {
355 $conn->log("irspy", "rewriting XML record");
356 my $rec = $this->_irspy_to_zeerex($conn);
358 # Since IRSpy can run for a long time between writes back to the
359 # database, it's quite possible for the server to have closed the
360 # connection as idle. So re-establish it if necessary.
361 $this->{conn}->connect($conn->option("host"));
363 _rewrite_zeerex_record($this->{conn}, $rec);
364 $conn->log("irspy", "rewrote XML record");
368 my $_reliabilityField = {
369 reliability => [ reliability => 0,
370 "Calculated reliability of server",
371 "e:serverInfo/e:reliability" ],
374 sub _rewrite_zeerex_record {
375 my($conn, $rec, $oldid) = @_;
377 # Add reliability score
378 my $xc = irspy_xpath_context($rec);
379 my($nok, $nall, $percent) = calc_reliability_stats($xc);
380 modify_xml_document($xc, $_reliabilityField, { reliability => $percent });
382 my $p = $conn->package();
383 $p->option(action => "specialUpdate");
384 my $xml = $rec->toString();
385 $p->option(record => $xml);
389 # This is the expression in the ID-making stylesheet
390 # ../../zebra/zeerex2id.xsl
391 my $id = irspy_record2identifier($xc);
392 if (defined $oldid && $id ne $oldid) {
393 warn "IDs differ (old='$oldid' new='$id')";
394 _delete_record($conn, $oldid);
397 $p = $conn->package();
404 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
412 # We can't delete records using recordIdOpaque, since character
413 # sets are handled differently here in extended services from how
414 # they are used in the Alvis filter's record-parsing, and so
415 # non-ASCII characters come out differently in the two contexts.
416 # Instead, we must send a record whose contents indicate the ID of
417 # that which we wish to delete. There are two ways, both
418 # unsatisfactory: we could either fetch the actual record them
419 # resubmit it in the deletion request (which wastes a search and a
420 # fetch) or we could build a record by hand from the parsed-out
421 # components (which is error-prone and which I am not 100% certain
422 # will work since the other contents of the record will be
423 # different). The former evil seems to be the lesser.
425 warn "$conn deleting record '$id'";
427 my $rs = $conn->search(new ZOOM::Query::CQL(cql_target($id)));
428 die "no such ID '$id'" if $rs->size() == 0;
429 my $rec = $rs->record(0);
430 my $xml = $rec->render();
432 my $p = $conn->package();
433 $p->option(action => "recordDelete");
434 $p->option(record => $xml);
438 $p = $conn->package();
444 # The approach: gather declarative information about test hierarchy,
445 # then go into a loop. In the loop, we ensure that each connection is
446 # running a test, and within that test a task, until its list of tests
447 # is exhausted. No individual test ever calls wait(): tests just queue
448 # up tasks and return immediately. When the tasks are run (one at a
449 # time on each connection) they generate events, and it is these that
450 # are harvested by ZOOM::event(). Since each connection knows what
451 # task it is running, it can invoke the appropriate callbacks.
452 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
453 # loop how to continue.
456 # While a connection is running a task, its current_task()
457 # points at the task structure. When it finishes its task,
458 # next_task() is pointed at the next task to execute (if there
459 # is one), and its current_task() is set to zero. When the next
460 # task is executed, the connection's next_task() is set to zero
461 # and its current_task() pointed to the task structure.
462 # current_task() and next_task() are both zero only when there
463 # are no more queued tasks, which is when a new test is
466 # Each connection's current test is stored in its
467 # "current_test_address" option. The next test to execute is
468 # calculated by walking the declarative tree of tests. This
469 # option begins empty; the "next test" after this is of course
475 my $topname = $this->{tree}->name();
476 my $timeout = $this->{timeout};
477 $this->log("irspy", "beginnning with test '$topname' (timeout $timeout)");
480 my @conn = @{ $this->{connections} };
483 ROUND_AND_ROUND_WE_GO:
485 my @copy_conn = @conn; # avoid alias problems after splice()
486 my $nconn = scalar(@copy_conn);
488 foreach my $i0 (0 .. $#copy_conn) {
489 my $conn = $copy_conn[$i0];
490 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
491 next if !defined $conn;
493 if (!$conn->current_task()) {
494 if (!$conn->next_task()) {
495 # Out of tasks: we need a new test
497 my $address = $conn->option("current_test_address");
499 if (!defined $address) {
502 $conn->log("irspy_test",
503 "checking for next test after '$address'");
504 $nextaddr = $this->_next_test($address);
507 if (ZOOM::IRSpy::Test::zoom_error_timeout_check($conn)) {
508 $conn->log("irspy", "Got to many timeouts, stop testing");
512 if (!defined $nextaddr) {
513 $conn->log("irspy", "has no more tests: removing");
514 $this->_rewrite_irspy_record($conn);
515 $conn->option(rewrote_record => 1);
516 my $newconn = $this->_next_connection();
517 if (!defined $newconn) {
518 # Do not destroy: needed for later sanity checks
519 splice @conn, $i0, 1;
522 $conn[$i0] = $newconn;
523 $conn[$i0]->option(current_test_address => "");
524 $conn[$i0]->log("irspy", "entering active pool - ",
525 scalar(@{ $this->{queue} }),
526 " targets remain in queue");
531 my $node = $this->{tree}->select($nextaddr)
532 or die "invalid nextaddr '$nextaddr'";
533 $conn->option(current_test_address => $nextaddr);
534 my $tname = $node->name();
535 if ($this->should_skip_test($tname)) {
536 warn "skipping test '$tname' due to rule";
540 $conn->log("irspy_test",
541 "starting test '$nextaddr' = $tname");
542 my $tasks = $conn->tasks();
543 my $oldcount = @$tasks;
544 "ZOOM::IRSpy::Test::$tname"->start($conn);
545 $tasks = $conn->tasks();
546 if (@$tasks > $oldcount) {
547 # Prepare to start the first of the newly added tasks
548 $conn->next_task($tasks->[$oldcount]);
550 $conn->log("irspy_task",
551 "no tasks added by new test $tname");
556 my $task = $conn->next_task();
557 die "no next task queued for $conn" if !defined $task;
559 # do not run the next task if we got too many timeouts
560 if (ZOOM::IRSpy::Test::zoom_error_timeout_check($conn)) {
561 $conn->log("irspy_task", "Got too many timeouts for this target, do not start a new task");
565 $conn->log("irspy_task", "preparing task $task");
567 $conn->current_task($task);
573 my $i0 = ZOOM::event(\@conn);
574 $this->log("irspy_event",
575 "ZOOM_event(", scalar(@conn), " connections) = $i0");
578 0 => "no events remain",
579 -1 => "ZOOM::event() argument not a reference",
580 -2 => "ZOOM::event() reference not an array",
581 -3 => "no connections remain",
582 -4 => "too many connections for ZOOM::event()",
584 my $message = $messages{$i0} || "ZOOM::event() returned $i0";
585 $this->log("irspy", $message);
589 my $conn = $conn[$i0-1];
590 my $ev = $conn->last_event();
591 my $evstr = ZOOM::event_str($ev);
592 $conn->log("irspy_event", "event $ev ($evstr)");
593 goto NEXT_EVENT if $ev != ZOOM::Event::ZEND;
595 my $task = $conn->current_task();
596 die "$conn has no current task for event $ev ($evstr)" if !$task;
599 eval { $conn->check() };
600 if ($@ && ref $@ && $@->isa("ZOOM::Exception")) {
601 my $sub = $task->{cb}->{exception};
602 die $@ if !defined $sub;
603 $res = &$sub($conn, $task, $task->udata(), $@);
605 die "Unexpected non-ZOOM exception: " . ref($@) . " ($@)";
607 my $sub = $task->{cb}->{$ev};
609 $conn->log("irspy_unhandled", "event $ev ($evstr)");
613 $res = &$sub($conn, $task, $task->udata(), $ev);
616 if ($res == ZOOM::IRSpy::Status::OK) {
617 # Nothing to do -- life continues
619 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
620 my $task = $conn->current_task();
621 die "no task for TASK_DONE on $conn" if !$task;
622 die "next task already defined for $conn" if $conn->next_task();
623 $conn->log("irspy_task", "completed task $task");
624 $conn->next_task($task->{next});
625 $conn->current_task(0);
627 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
628 $res == ZOOM::IRSpy::Status::TEST_BAD) {
629 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
630 $conn->log("irspy_task", "test ended during task $task ($x)");
631 $conn->log("irspy_test", "test completed ($x)");
632 $conn->current_task(0);
634 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
635 my $address = $conn->option('current_test_address');
636 $conn->log("irspy", "top-level test failed!")
638 my $node = $this->{tree}->select($address);
640 while (defined $node->next() &&
641 length($node->next()->address()) >= length($address)) {
642 $conn->log("irspy_debug", "skipping from '",
643 $node->address(), "' to '",
644 $node->next()->address(), "'");
645 $node = $node->next();
649 $conn->option(current_test_address => $node->address());
650 $conn->log("irspy_test", "skipped $skipcount tests");
651 $nskipped += $skipcount;
654 } elsif ($res == ZOOM::IRSpy::Status::TEST_SKIPPED) {
655 $conn->log("irspy_test", "test skipped during task $task");
656 $conn->current_task(0);
661 die "unknown callback return-value '$res'";
665 $this->log("irspy", "exiting main loop");
667 # Sanity checks: none of the following should ever happen
669 $this->log("irspy", "performing end-of-run sanity-checks");
670 foreach my $conn (@conn) {
671 my $test = $conn->option("current_test_address");
672 my $next = $this->_next_test($test);
675 "$conn (in test '$test') has queued test '$next'");
678 if (my $task = $conn->current_task()) {
679 $this->log("irspy", "$conn still has an active task $task");
682 if (my $task = $conn->next_task()) {
683 $this->log("irspy", "$conn still has a queued task $task");
686 if (!$conn->is_idle()) {
688 "$conn still has ZOOM-C level tasks queued: see below");
691 my $ev = $conn->peek_event();
692 if ($ev != 0 && $ev != ZOOM::Event::ZEND) {
693 my $evstr = ZOOM::event_str($ev);
694 $this->log("irspy", "$conn has event $ev ($evstr) waiting");
697 if (!$conn->option("rewrote_record")) {
698 $this->log("irspy", "$conn did not rewrite its ZeeRex record");
703 # This really shouldn't be necessary, and in practice it rarely
704 # helps, but it's belt and braces. (For now, we don't do this
705 # hence the zero in the $nruns check).
708 $this->log("irspy", "back into main loop, ${nruns}th time");
709 goto ROUND_AND_ROUND_WE_GO;
711 $this->log("irspy", "bailing after $nruns main-loop runs");
715 # This shouldn't happen emit anything either:
716 while ((my $i1 = ZOOM::event(\@conn)) > 0) {
717 my $conn = $conn[$i1-1];
718 my $ev = $conn->last_event();
719 my $evstr = ZOOM::event_str($ev);
721 "$conn still has ZOOM-C level task queued: $ev ($evstr)")
722 if $ev != ZOOM::Event::ZEND;
729 ### This could of course call a much more general looping evaluator
730 sub should_skip_test {
734 foreach my $rule (@{ $this->{rules} }) {
735 my($name, $op, $val, $rule, @params) = @$rule;
736 #warn "using name='$name', op='$op', val='$val', rule='$rule', params='@params'\n";
739 $params[0] eq $tname) {
741 $re =~ s/^\/(.*)\/$/$1/;
742 if ($this->var($name) =~ $re) {
743 #warn "$name '", $this->var($name), "' matches /$re/ -- skipping '$tname'";
753 # Exactly equivalent to ZOOM::event() except that it is tolerant to
754 # undefined values in the array being passed in.
756 sub __UNUSED_tolerant_ZOOM_event {
760 foreach my $i (0 .. @$connref-1) {
761 my $conn = $connref->[$i];
768 my $res = ZOOM::event(\@conn);
769 return $res if $res <= 0;
770 my $res2 = $map[$res-1] + 1;
771 print STDERR "*** tolerant_ZOOM_event() returns $res->$res2\n";
778 my($tname, @ancestors) = @_;
780 die("$0: test-hierarchy loop detected: " .
781 join(" -> ", @ancestors, $tname))
782 if grep { $_ eq $tname } @ancestors;
784 my $slashSeperatedTname = $tname;
785 $slashSeperatedTname =~ s/::/\//g;
786 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
791 $this->log("irspy", "couldn't require '$fullName': $@");
792 $this->log("warn", "can't load test '$tname': skipping",
793 $@ =~ /^Can.t locate/ ? () : " ($@)");
797 $this->log("irspy", "adding test '$tname'");
799 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
800 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
801 push @subnodes, $subtest if defined $subtest;
804 return new ZOOM::IRSpy::Node($tname, @subnodes);
808 # These next three should arguably be Node methods
811 my($address, $omit_child) = @_;
815 my $maybe = $address eq "" ? "0" : "$address:0";
816 return $maybe if $this->{tree}->select($maybe);
819 # The top-level node has no successor or parent
820 return undef if $address eq "";
822 # Try next sibling child
823 # The use of a regexp in split() is to avoid confusing Emacs's
824 # Perl mode; the third colon is to avoid looking like a POSIX
825 # named character class ... *sigh*
826 my @components = split /[:::]/, $address;
827 my $last = pop @components;
828 my $maybe = join(":", @components, $last+1);
829 return $maybe if $this->{tree}->select($maybe);
831 # This node is exhausted: try the parent's successor
832 return $this->_next_test(join(":", @components), 1)
836 sub _last_sibling_test {
837 die "_last_sibling_test() called -- I thought that never happened?";
842 if !defined $this->_next_sibling_test($address);
846 my $maybe = $this->_next_sibling_test($address);
847 last if !defined $maybe;
850 $this->log("irspy", "skipping $nskipped tests to '$address'");
853 return ($address, $nskipped);
857 sub _next_sibling_test {
858 die "_next_sibling_test() called -- I thought that never happened?";
862 my @components = split /[:::]/, $address;
863 my $last = pop @components;
864 my $maybe = join(":", @components, $last+1);
865 return $maybe if $this->{tree}->select($maybe);
875 ZOOM::IRSpy::Maintenance.
877 The ZOOM-Perl module,
878 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
881 http://indexdata.com/zebra/
885 Mike Taylor, E<lt>mike@indexdata.comE<gt>
887 =head1 COPYRIGHT AND LICENSE
889 Copyright (C) 2006 by Index Data ApS.
891 This library is free software; you can redistribute it and/or modify
892 it under the same terms as Perl itself, either Perl version 5.8.7 or,
893 at your option, any later version of Perl 5 you may have available.