1 # $Id: IRSpy.pm,v 1.63 2007-02-22 15:43:13 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::Record;
19 use ZOOM::IRSpy::Stats;
20 use ZOOM::IRSpy::Utils qw(cql_target);
23 our $VERSION = '0.02';
24 our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
27 # Enumeration for callback functions to return
28 package ZOOM::IRSpy::Status;
29 sub OK { 29 } # No problems, task is still progressing
30 sub TASK_DONE { 18 } # Task is complete, next task should begin
31 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
32 sub TEST_BAD { 31 } # Whole test is complete, and failed
33 sub TEST_SKIPPED { 12 } # Test couldn't be run
39 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
44 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
45 $spy->targets(@targets);
47 $res = $spy->check("Main");
51 This module exists to implement the IRspy program, which discovers,
52 analyses and monitors IR servers implementing the Z39.50 and SRU/W
53 protocols. It is a successor to the ZSpy program.
58 ZOOM::Log::mask_str("irspy");
59 ZOOM::Log::mask_str("irspy_debug");
60 ZOOM::Log::mask_str("irspy_event");
61 ZOOM::Log::mask_str("irspy_unhandled");
62 ZOOM::Log::mask_str("irspy_test");
63 ZOOM::Log::mask_str("irspy_task");
68 my($dbname, $user, $password) = @_;
71 push @options, (user => $user, password => $password)
74 my $conn = new ZOOM::Connection($dbname, 0, @options)
75 or die "$0: can't connection to IRSpy database 'dbname'";
77 my $xslt = new XML::LibXSLT;
79 $xslt->register_function($ZOOM::IRSpy::Utils::IRSPY_NS, 'strcmp',
80 \&ZOOM::IRSpy::Utils::xslt_strcmp);
82 my $libxml = new XML::LibXML;
83 my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl);
84 my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc);
88 allrecords => 1, # unless overridden by targets()
89 query => undef, # filled in later
90 targets => undef, # filled in later
91 connections => undef, # filled in later
93 irspy_to_zeerex_style => $irspy_to_zeerex_style,
94 tests => [], # stack of tests currently being executed
96 $this->log("irspy", "starting up with database '$dbname'");
108 # Explicitly nominate a set of targets to check, overriding the
109 # default which is to re-check everything in the database. Each
110 # target already in the database results in the existing record being
111 # updated; each new target causes a new record to be added.
117 $this->log("irspy", "setting explicit list of targets ",
118 join(", ", map { "'$_'" } @targets));
119 $this->{allrecords} = 0;
121 foreach my $target (@targets) {
122 my($host, $port, $db, $newtarget) = _parse_target_string($target);
123 if ($newtarget ne $target) {
124 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
125 $target = $newtarget; # This is written through the ref
127 push @qlist, cql_target($host, $port, $db);
130 $this->{targets} = \@targets;
131 $this->{query} = join(" or ", @qlist);
135 # Also used by ZOOM::IRSpy::Record
136 sub _parse_target_string {
139 my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
140 if (!defined $host) {
142 ($host, $db) = ($target =~ /(.*?)\/(.*)/);
143 $target = "$host:$port/$db";
145 die "$0: invalid target string '$target'"
148 return ($host, $port, $db, $target);
152 # There are two cases.
154 # 1. A specific set of targets is nominated on the command line.
155 # - Records must be fetched for those targets that are in the DB
156 # - New, empty records must be made for those that are not.
157 # - Updated records written to the DB may or may not be new.
159 # 2. All records in the database are to be checked.
160 # - Records must be fetched for all targets in the DB
161 # - Updated records written to the DB may not be new.
163 # That's all -- what could be simpler?
169 if ($this->{allrecords}) {
170 # We need to check on every target in the database, which
171 # means we need to do a "find all". According to the BIB-1
172 # semantics document at
173 # http://www.loc.gov/z3950/agency/bib1.html
175 # @attr 2=103 @attr 1=1035 x
176 # should find all records, but it seems that Zebra doesn't
177 # support this. Furthermore, when using the "alvis" filter
178 # (as we do for IRSpy) it doesn't support the use of any BIB-1
179 # access point -- not even 1035 "everywhere" -- so instead we
180 # hack together a search that we know will find all records.
181 $this->{query} = "port=?*";
183 # Prepopulate the target map with nulls so that after we fill
184 # in what we can from the database query, we know which target
185 # IDs we need new records for.
186 foreach my $target (@{ $this->{targets} }) {
187 $target2record{lc($target)} = undef;
191 $this->log("irspy_debug", "query '", $this->{query}, "'");
192 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
193 delete $this->{query}; # No longer needed at all
194 $this->log("irspy_debug", "found ", $rs->size(), " target records");
195 foreach my $i (1 .. $rs->size()) {
196 my $target = _render_record($rs, $i-1, "id");
197 my $zeerex = _render_record($rs, $i-1, "zeerex");
198 #print STDERR "making '$target' record with '$zeerex'\n";
199 $target2record{lc($target)} =
200 new ZOOM::IRSpy::Record($this, $target, $zeerex);
201 push @{ $this->{targets} }, $target
202 if $this->{allrecords};
205 # Make records for targets not previously in the database
206 foreach my $target (keys %target2record) {
207 my $record = $target2record{$target};
208 if (!defined $record) {
209 $this->log("irspy_debug", "made new record for '$target'");
210 $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
212 $this->log("irspy_debug", "using existing record for '$target'");
217 foreach my $target (@{ $this->{targets} }) {
218 my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
219 $conn->option(host => $target);
220 my $record = delete $target2record{lc($target)};
221 $conn->record($record);
222 push @connections, $conn;
224 die("remaining target2record = { " .
225 join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
226 sort keys %target2record) . " }")
229 $this->{connections} = \@connections;
230 delete $this->{targets}; # The information is now in {connections}
235 my($rs, $which, $elementSetName) = @_;
237 # There is a slight race condition here on the element-set name,
238 # but it shouldn't be a problem as this is (currently) only called
239 # from parts of the program that run single-threaded.
240 my $old = $rs->option(elementSetName => $elementSetName);
241 my $rec = $rs->record($which);
242 $rs->option(elementSetName => $old);
244 return $rec->render();
248 sub _irspy_to_zeerex {
250 my($conn, $save_xml) = @_;
251 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
254 unlink('/tmp/irspy_orig.xml');
255 open FH, '>/tmp/irspy_orig.xml';
256 print FH $irspy_doc->toString();
260 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
262 unlink('/tmp/irspy_transformed.xml');
263 open FH, '>/tmp/irspy_transformed.xml';
264 print FH $result->toString();
268 return $result->documentElement();
272 sub _rewrite_record {
276 $conn->log("irspy", "rewriting XML record");
277 my $rec = $this->_irspy_to_zeerex($conn, $ENV{IRSPY_SAVE_XML});
278 _really_rewrite_record($this->{conn}, $rec);
282 sub _really_rewrite_record {
283 my($conn, $rec) = @_;
285 my $p = $conn->package();
286 $p->option(action => "specialUpdate");
287 my $xml = $rec->toString();
288 $p->option(record => $xml);
292 $p = $conn->package();
299 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
304 # The approach: gather declarative information about test hierarchy,
305 # then go into a loop. In the loop, we ensure that each connection is
306 # running a test, and within that test a task, until its list of tests
307 # is exhausted. No individual test ever calls wait(): tests just queue
308 # up tasks and return immediately. When the tasks are run (one at a
309 # time on each connection) they generate events, and it is these that
310 # are harvested by ZOOM::event(). Since each connection knows what
311 # task it is running, it can invoke the appropriate callbacks.
312 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
313 # loop how to continue.
316 # While a connection is running a task, its current_task()
317 # points at the task structure. When it finishes its task,
318 # next_task() is pointed at the next task to execute (if there
319 # is one), and its current_task() is set to zero. When the next
320 # task is executed, the connection's next_task() is set to zero
321 # and its current_task() pointed to the task structure.
322 # current_task() and next_task() are both zero only when there
323 # are no more queued tasks, which is when a new test is
326 # Each connection's current test is stored in its
327 # "current_test_address" option. The next test to execute is
328 # calculated by walking the declarative tree of tests. This
329 # option begins empty; the "next test" after this is of course
336 $tname = "Main" if !defined $tname;
337 $this->{tree} = $this->_gather_tests($tname)
338 or die "No tests defined for '$tname'";
339 #$this->{tree}->print(0);
342 my @conn = @{ $this->{connections} };
345 ROUND_AND_ROUND_WE_GO:
347 my @copy_conn = @conn; # avoid alias problems after splice()
348 my $nconn = scalar(@copy_conn);
349 foreach my $i0 (0 .. $#copy_conn) {
350 my $conn = $copy_conn[$i0];
351 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
352 if (!$conn->current_task()) {
353 if (!$conn->next_task()) {
354 # Out of tasks: we need a new test
356 my $address = $conn->option("current_test_address");
358 if (!defined $address) {
361 $this->log("irspy_test",
362 "checking for next test after '$address'");
363 $nextaddr = $this->_next_test($address);
365 if (!defined $nextaddr) {
366 $conn->log("irspy", "has no more tests: removing");
367 splice @conn, $i0, 1;
368 $this->_rewrite_record($conn);
369 $conn->option(rewrote_record => 1);
373 my $node = $this->{tree}->select($nextaddr)
374 or die "invalid nextaddr '$nextaddr'";
375 $conn->option(current_test_address => $nextaddr);
376 my $tname = $node->name();
377 $conn->log("irspy_test",
378 "starting test '$nextaddr' = $tname");
379 my $tasks = $conn->tasks();
380 my $oldcount = @$tasks;
381 "ZOOM::IRSpy::Test::$tname"->start($conn);
382 $tasks = $conn->tasks();
383 if (@$tasks > $oldcount) {
384 # Prepare to start the first of the newly added tasks
385 $conn->next_task($tasks->[$oldcount]);
387 $conn->log("irspy_task",
388 "no tasks added by new test $tname");
393 my $task = $conn->next_task();
394 die "no next task queued for $conn" if !defined $task;
395 $conn->log("irspy_task", "preparing task $task");
397 $conn->current_task($task);
401 # Do we need to test $conn->is_idle()? I don't think so!
404 my $i0 = ZOOM::event(\@conn);
405 $this->log("irspy_event",
406 "ZOOM_event(", scalar(@conn), " connections) = $i0");
409 0 => "no events remain",
410 -1 => "ZOOM::event() argument not a reference",
411 -2 => "ZOOM::event() reference not an array",
412 -3 => "no connections remain",
413 -4 => "too many connections for ZOOM::event()",
415 my $message = $messages{$i0} || "ZOOM::event() returned $i0";
416 $this->log("irspy", $message);
420 my $conn = $conn[$i0-1];
421 my $ev = $conn->last_event();
422 my $evstr = ZOOM::event_str($ev);
423 $conn->log("irspy_event", "event $ev ($evstr)");
425 my $task = $conn->current_task();
426 die "$conn has no current task for event $ev ($evstr)" if !$task;
427 eval { $conn->_check() };
429 ($ev == ZOOM::Event::RECV_DATA ||
430 $ev == ZOOM::Event::ZEND)) {
431 # An error in, say, a search response, becomes visible to
432 # ZOOM before the Receive Data event is sent and persists
433 # until after the End, which means that successive events
434 # each report the same error. So we just ignore errors on
435 # "unimportant" events. ### But this doesn't work for,
436 # say, a Connection Refused, as the only event that shows
437 # us this error is the ZEND.
438 $conn->log("irspy_event", "ignoring error ",
439 "on event $ev ($evstr): $@");
445 my $sub = $task->{cb}->{exception};
446 die $@ if !defined $sub;
447 $res = &$sub($conn, $task, $task->udata(), $@);
449 my $sub = $task->{cb}->{$ev};
451 $conn->log("irspy_unhandled", "event $ev ($evstr)");
455 $res = &$sub($conn, $task, $task->udata(), $ev);
458 if ($res == ZOOM::IRSpy::Status::OK) {
459 # Nothing to do -- life continues
461 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
462 my $task = $conn->current_task();
463 die "no task for TASK_DONE on $conn" if !$task;
464 die "next task already defined for $conn" if $conn->next_task();
465 $conn->log("irspy_task", "completed task $task");
466 $conn->next_task($task->{next});
467 $conn->current_task(0);
469 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
470 $res == ZOOM::IRSpy::Status::TEST_BAD) {
471 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
472 $conn->log("irspy_task", "test ended during task $task ($x)");
473 $conn->log("irspy_test", "test completed ($x)");
474 $conn->current_task(0);
476 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
477 my $address = $conn->option('current_test_address');
478 ($address, my $n) = $this->_last_sibling_test($address);
479 if (defined $address) {
480 $conn->log("irspy_test", "skipped $n tests");
481 $conn->option(current_test_address => $address);
486 } elsif ($res == ZOOM::IRSpy::Status::TEST_SKIPPED) {
487 $conn->log("irspy_test", "test skipped during task $task");
488 $conn->current_task(0);
490 # I think that's all we need to do
493 die "unknown callback return-value '$res'";
497 $this->log("irspy", "exiting main loop");
498 # Sanity checks: none of the following should ever happen
500 foreach my $conn (@{ $this->{connections} }) {
501 my $test = $conn->option("current_test_address");
502 my $next = $this->_next_test($test);
504 $this->log("irspy", "$conn (in test '$test') has queued test '$next'");
507 if (my $task = $conn->current_task()) {
508 $this->log("irspy", "$conn still has an active task $task");
511 if (my $task = $conn->next_task()) {
512 $this->log("irspy", "$conn still has a queued task $task");
515 if (!$conn->is_idle()) {
516 $this->log("irspy", "$conn still has ZOOM-C level tasks queued: see below");
519 my $ev = $conn->peek_event();
520 if ($ev != 0 && $ev != ZOOM::Event::ZEND) {
521 my $evstr = ZOOM::event_str($ev);
522 $this->log("irspy", "$conn has event $ev ($evstr) waiting");
525 if (!$conn->option("rewrote_record")) {
526 $this->log("irspy", "$conn did not rewrite its ZeeRex record");
531 # This really shouldn't be necessary, and in practice it rarely
532 # helps, but it's belt and braces. For now, we don't do this
533 # (hence the zero in the $nruns check).
536 $this->log("irspy", "back into main loop, ${nruns}th time");
537 goto ROUND_AND_ROUND_WE_GO;
539 $this->log("irspy", "bailing after $nruns main-loop runs");
543 # This shouldn't happen emit anything either:
544 @conn = @{ $this->{connections} };
545 while ((my $i1 = ZOOM::event(\@conn)) > 0) {
546 my $conn = $conn[$i1-1];
547 my $ev = $conn->last_event();
548 my $evstr = ZOOM::event_str($ev);
550 "$conn still has ZOOM-C level task queued: $ev ($evstr)")
551 if $ev != ZOOM::Event::ZEND;
560 my($tname, @ancestors) = @_;
562 die("$0: test-hierarchy loop detected: " .
563 join(" -> ", @ancestors, $tname))
564 if grep { $_ eq $tname } @ancestors;
566 my $slashSeperatedTname = $tname;
567 $slashSeperatedTname =~ s/::/\//g;
568 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
572 $this->log("irspy", "successfully required '$fullName'");
574 $this->log("irspy", "couldn't require '$fullName': $@");
575 $this->log("warn", "can't load test '$tname': skipping",
576 $@ =~ /^Can.t locate/ ? () : " ($@)");
580 $this->log("irspy", "adding test '$tname'");
582 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
583 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
584 push @subnodes, $subtest if defined $subtest;
587 return new ZOOM::IRSpy::Node($tname, @subnodes);
591 # These next three should arguably be Node methods
594 my($address, $omit_child) = @_;
598 my $maybe = $address eq "" ? "0" : "$address:0";
599 return $maybe if $this->{tree}->select($maybe);
602 # The top-level node has no successor or parent
603 return undef if $address eq "";
605 # Try next sibling child
606 my @components = split /:/, $address;
607 my $last = pop @components;
608 my $maybe = join(":", @components, $last+1);
609 return $maybe if $this->{tree}->select($maybe);
611 # This node is exhausted: try the parent's successor
612 return $this->_next_test(join(":", @components), 1)
616 sub _last_sibling_test {
621 if !defined $this->_next_sibling_test($address);
625 my $maybe = $this->_next_sibling_test($address);
626 last if !defined $maybe;
629 $this->log("irspy", "skipping $nskipped tests to '$address'");
632 return ($address, $nskipped);
636 sub _next_sibling_test {
640 my @components = split /:/, $address;
641 my $last = pop @components;
642 my $maybe = join(":", @components, $last+1);
643 return $maybe if $this->{tree}->select($maybe);
653 ZOOM::IRSpy::Maintenance.
655 The ZOOM-Perl module,
656 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
659 http://indexdata.com/zebra/
663 Mike Taylor, E<lt>mike@indexdata.comE<gt>
665 =head1 COPYRIGHT AND LICENSE
667 Copyright (C) 2006 by Index Data ApS.
669 This library is free software; you can redistribute it and/or modify
670 it under the same terms as Perl itself, either Perl version 5.8.7 or,
671 at your option, any later version of Perl 5 you may have available.