1 # $Id: IRSpy.pm,v 1.47 2006-11-02 13:16:49 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::Utils;
22 our $VERSION = '0.02';
23 our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
26 # Enumeration for callback functions to return
27 package ZOOM::IRSpy::Status;
28 sub OK { 29 } # No problems, task is still progressing
29 sub TASK_DONE { 18 } # Task is complete, next task should begin
30 sub TEST_GOOD { 8 } # Whole test is complete, and succeeded
31 sub TEST_BAD { 31 } # Whole test is complete, and failed
37 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
42 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
43 print $spy->report_status();
47 This module exists to implement the IRspy program, which discovers,
48 analyses and monitors IR servers implementing the Z39.50 and SRU/W
49 protocols. It is a successor to the ZSpy program.
54 ZOOM::Log::mask_str("irspy");
55 ZOOM::Log::mask_str("irspy_debug");
56 ZOOM::Log::mask_str("irspy_event");
57 ZOOM::Log::mask_str("irspy_unhandled");
58 ZOOM::Log::mask_str("irspy_test");
59 ZOOM::Log::mask_str("irspy_task");
64 my($dbname, $user, $password) = @_;
67 push @options, (user => $user, password => $password)
70 my $conn = new ZOOM::Connection($dbname, 0, @options)
71 or die "$0: can't connection to IRSpy database 'dbname'";
73 my $xslt = new XML::LibXSLT;
74 my $libxml = new XML::LibXML;
75 my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl);
76 my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc);
80 allrecords => 1, # unless overridden by targets()
81 query => undef, # filled in later
82 targets => undef, # filled in later
83 connections => undef, # filled in later
85 irspy_to_zeerex_style => $irspy_to_zeerex_style,
86 tests => [], # stack of tests currently being executed
88 $this->log("irspy", "starting up with database '$dbname'");
100 # Explicitly nominate a set of targets to check, overriding the
101 # default which is to re-check everything in the database. Each
102 # target already in the database results in the existing record being
103 # updated; each new target causes a new record to be added.
109 $this->log("irspy", "setting explicit list of targets ",
110 join(", ", map { "'$_'" } @targets));
111 $this->{allrecords} = 0;
113 foreach my $target (@targets) {
114 my($host, $port, $db, $newtarget) = _parse_target_string($target);
115 if ($newtarget ne $target) {
116 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
117 $target = $newtarget; # This is written through the ref
119 push @qlist, (qq[(host="$host" and port="$port" and path="$db")]);
122 $this->{targets} = \@targets;
123 $this->{query} = join(" or ", @qlist);
127 # Also used by ZOOM::IRSpy::Record
128 sub _parse_target_string {
131 my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
132 if (!defined $host) {
134 ($host, $db) = ($target =~ /(.*?)\/(.*)/);
135 $target = "$host:$port/$db";
137 die "$0: invalid target string '$target'"
140 return ($host, $port, $db, $target);
144 # There are two cases.
146 # 1. A specific set of targets is nominated on the command line.
147 # - Records must be fetched for those targets that are in the DB
148 # - New, empty records must be made for those that are not.
149 # - Updated records written to the DB may or may not be new.
151 # 2. All records in the database are to be checked.
152 # - Records must be fetched for all targets in the DB
153 # - Updated records written to the DB may not be new.
155 # That's all -- what could be simpler?
161 if ($this->{allrecords}) {
162 # We need to check on every target in the database, which
163 # means we need to do a "find all". According to the BIB-1
164 # semantics document at
165 # http://www.loc.gov/z3950/agency/bib1.html
167 # @attr 2=103 @attr 1=1035 x
168 # should find all records, but it seems that Zebra doesn't
169 # support this. Furthermore, when using the "alvis" filter
170 # (as we do for IRSpy) it doesn't support the use of any BIB-1
171 # access point -- not even 1035 "everywhere" -- so instead we
172 # hack together a search that we know will find all records.
173 $this->{query} = "port=?*";
175 # Prepopulate the target map with nulls so that after we fill
176 # in what we can from the database query, we know which target
177 # IDs we need new records for.
178 foreach my $target (@{ $this->{targets} }) {
179 $target2record{lc($target)} = undef;
183 $this->log("irspy_debug", "query '", $this->{query}, "'");
184 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
185 delete $this->{query}; # No longer needed at all
186 $this->log("irspy_debug", "found ", $rs->size(), " target records");
187 foreach my $i (1 .. $rs->size()) {
188 my $target = _render_record($rs, $i-1, "id");
189 my $zeerex = _render_record($rs, $i-1, "zeerex");
190 #print STDERR "making '$target' record with '$zeerex'\n";
191 $target2record{lc($target)} =
192 new ZOOM::IRSpy::Record($this, $target, $zeerex);
193 push @{ $this->{targets} }, $target
194 if $this->{allrecords};
197 # Make records for targets not previously in the database
198 foreach my $target (keys %target2record) {
199 my $record = $target2record{$target};
200 if (!defined $record) {
201 $this->log("irspy_debug", "made new record for '$target'");
202 $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
204 $this->log("irspy_debug", "using existing record for '$target'");
209 foreach my $target (@{ $this->{targets} }) {
210 my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
211 $conn->option(host => $target);
212 my $record = delete $target2record{lc($target)};
213 $conn->record($record);
214 push @connections, $conn;
216 die("remaining target2record = { " .
217 join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
218 sort keys %target2record) . " }")
221 $this->{connections} = \@connections;
222 delete $this->{targets}; # The information is now in {connections}
227 my($rs, $which, $elementSetName) = @_;
229 # There is a slight race condition here on the element-set name,
230 # but it shouldn't be a problem as this is (currently) only called
231 # from parts of the program that run single-threaded.
232 my $old = $rs->option(elementSetName => $elementSetName);
233 my $rec = $rs->record($which);
234 $rs->option(elementSetName => $old);
236 return $rec->render();
240 sub _irspy_to_zeerex {
241 my ($this, $conn) = @_;
242 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
244 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
246 return $result->documentElement();
250 sub _rewrite_record {
254 $conn->log("irspy", "rewriting XML record");
255 my $rec = $this->_irspy_to_zeerex($conn);
256 _really_rewrite_record($this->{conn}, $rec);
260 sub _really_rewrite_record {
261 my($conn, $rec) = @_;
263 my $p = $conn->package();
264 $p->option(action => "specialUpdate");
265 my $xml = $rec->toString();
266 $p->option(record => $xml);
270 $p = $conn->package();
277 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
282 # The approach: gather declarative information about test hierarchy,
283 # then go into a loop. In the loop, we ensure that each connection is
284 # running a test, and within that test a task, until its list of tests
285 # is exhausted. No individual test ever calls wait(): tests just queue
286 # up tasks and return immediately. When the tasks are run (one at a
287 # time on each connection) they generate events, and it is these that
288 # are harvested by ZOOM::event(). Since each connection knows what
289 # task it is running, it can invoke the appropriate callbacks.
290 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
291 # loop how to continue.
294 # While a connection is running a task, its current_task()
295 # points at the task structure. When it finishes its task,
296 # next_task() is pointed at the next task to execute (if there
297 # is one), and its current_task() is set to zero. When the next
298 # task is executed, the connection's next_task() is set to zero
299 # and its current_task() pointed to the task structure.
300 # current_task() and next_task() are both zero only when there
301 # are no more queued tasks, which is when a new test is
304 # Each connection's current test is stored in its
305 # "current_test_address" option. The next test to execute is
306 # calculated by walking the declarative tree of tests. This
307 # option begins empty; the "next test" after this is of course
314 $tname = "Main" if !defined $tname;
315 $this->{tree} = $this->_gather_tests($tname)
316 or die "No tests defined for '$tname'";
317 #$this->{tree}->print(0);
320 my @conn = @{ $this->{connections} };
323 my @copy_conn = @conn; # avoid alias problems after splice()
324 my $nconn = scalar(@copy_conn);
325 foreach my $i0 (0 .. $#copy_conn) {
326 my $conn = $copy_conn[$i0];
327 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
328 if (!$conn->current_task()) {
329 if (!$conn->next_task()) {
330 # Out of tasks: we need a new test
332 my $address = $conn->option("current_test_address");
334 if (!defined $address) {
337 $this->log("irspy_test",
338 "checking for next test after '$address'");
339 $nextaddr = $this->_next_test($address);
341 if (!defined $nextaddr) {
342 $conn->log("irspy", "has no more tests: removing");
343 splice @conn, $i0, 1;
344 $this->_rewrite_record($conn);
345 $conn->option(rewrote_record => 1);
349 my $node = $this->{tree}->select($nextaddr)
350 or die "invalid nextaddr '$nextaddr'";
351 $conn->option(current_test_address => $nextaddr);
352 my $tname = $node->name();
353 $conn->log("irspy_test",
354 "starting test '$nextaddr' = $tname");
355 my $tasks = $conn->tasks();
356 my $oldcount = @$tasks;
357 "ZOOM::IRSpy::Test::$tname"->start($conn);
358 $tasks = $conn->tasks();
359 if (@$tasks > $oldcount) {
360 # Prepare to start the first of the newly added tasks
361 $conn->next_task($tasks->[$oldcount]);
363 $conn->log("irspy_task",
364 "no tasks added by new test $tname");
369 my $task = $conn->next_task();
370 die "no next task queued for $conn" if !defined $task;
371 $conn->log("irspy_task", "preparing task $task");
373 $conn->current_task($task);
377 # Do we need to test $conn->is_idle()? I don't think so!
380 my $i0 = ZOOM::event(\@conn);
381 $this->log("irspy_event",
382 "ZOOM_event(", scalar(@conn), " connections) = $i0");
383 last if $i0 == 0 || $i0 == -3; # no events or no connections
384 my $conn = $conn[$i0-1];
385 my $ev = $conn->last_event();
386 my $evstr = ZOOM::event_str($ev);
387 $conn->log("irspy_event", "event $ev ($evstr)");
389 my $task = $conn->current_task();
390 die "$conn has no current task for event $ev ($evstr)" if !$task;
391 eval { $conn->_check() };
393 ($ev == ZOOM::Event::RECV_DATA ||
394 $ev == ZOOM::Event::RECV_APDU ||
395 $ev == ZOOM::Event::ZEND)) {
396 # An error in, say, a search response, becomes visible to
397 # ZOOM before the Receive Data event is sent and persists
398 # until after the End, which means that successive events
399 # each report the same error. So we just ignore errors on
400 # "unimportant" events. ### But this doesn't work for,
401 # say, a Connection Refused, as the only event that shows
402 # us this error is the End.
403 $conn->log("irspy_event", "ignoring error ",
404 "on event $ev ($evstr): $@");
410 my $sub = $task->{cb}->{exception};
411 die $@ if !defined $sub;
412 $res = &$sub($conn, $task, $task->udata(), $@);
414 my $sub = $task->{cb}->{$ev};
416 $conn->log("irspy_unhandled", "event $ev ($evstr)");
420 $res = &$sub($conn, $task, $task->udata(), $ev);
423 if ($res == ZOOM::IRSpy::Status::OK) {
424 # Nothing to do -- life continues
426 } elsif ($res == ZOOM::IRSpy::Status::TASK_DONE) {
427 my $task = $conn->current_task();
428 die "no task for TASK_DONE on $conn" if !$task;
429 die "next task already defined for $conn" if $conn->next_task();
430 $conn->log("irspy_task", "completed task $task");
431 $conn->next_task($task->{next});
432 $conn->current_task(0);
434 } elsif ($res == ZOOM::IRSpy::Status::TEST_GOOD ||
435 $res == ZOOM::IRSpy::Status::TEST_BAD) {
436 my $x = ($res == ZOOM::IRSpy::Status::TEST_GOOD) ? "good" : "bad";
437 $conn->log("irspy_task", "test ended during task $task ($x)");
438 $conn->log("irspy_test", "test completed ($x)");
439 $conn->current_task(0);
441 if ($res == ZOOM::IRSpy::Status::TEST_BAD) {
442 my $address = $conn->option('current_test_address');
443 ($address, my $n) = $this->_last_sibling_test($address);
444 if (defined $address) {
445 $conn->log("irspy_test", "skipped $n tests");
446 $conn->option(current_test_address => $address);
451 die "unknown callback return-value '$res'";
455 $this->log("irspy", "exiting main loop");
456 # Sanity checks: none of the following should ever happen
457 foreach my $conn (@{ $this->{connections} }) {
458 my $test = $conn->option("current_test_address");
459 my $next = $this->_next_test($test);
461 warn "$conn (in test '$test') has queued test '$next'";
463 if (my $task = $conn->current_task()) {
464 warn "$conn still has an active task $task";
466 if (my $task = $conn->next_task()) {
467 warn "$conn still has a queued task $task";
469 if (!$conn->is_idle()) {
470 warn "$conn is not idle (still has ZOOM-C level tasks queued)";
472 if (!$conn->option("rewrote_record")) {
473 warn "$conn did not rewrite its ZeeRex record";
483 my($tname, @ancestors) = @_;
485 die("$0: test-hierarchy loop detected: " .
486 join(" -> ", @ancestors, $tname))
487 if grep { $_ eq $tname } @ancestors;
489 my $slashSeperatedTname = $tname;
490 $slashSeperatedTname =~ s/::/\//g;
491 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
495 $this->log("irspy", "successfully required '$fullName'");
497 $this->log("irspy", "couldn't require '$fullName': $@");
498 $this->log("warn", "can't load test '$tname': skipping",
499 $@ =~ /^Can.t locate/ ? () : " ($@)");
503 $this->log("irspy", "adding test '$tname'");
505 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
506 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
507 push @subnodes, $subtest if defined $subtest;
510 return new ZOOM::IRSpy::Node($tname, @subnodes);
514 # These next three should arguably be Node methods
517 my($address, $omit_child) = @_;
521 my $maybe = $address eq "" ? "0" : "$address:0";
522 return $maybe if $this->{tree}->select($maybe);
525 # The top-level node has no successor or parent
526 return undef if $address eq "";
528 # Try next sibling child
529 my @components = split /:/, $address;
530 my $last = pop @components;
531 my $maybe = join(":", @components, $last+1);
532 return $maybe if $this->{tree}->select($maybe);
534 # This node is exhausted: try the parent's successor
535 return $this->_next_test(join(":", @components), 1)
539 sub _last_sibling_test {
544 if !defined $this->_next_sibling_test($address);
548 my $maybe = $this->_next_sibling_test($address);
549 last if !defined $maybe;
551 $this->log("irspy", "skipping $nskipped tests to '$address'");
555 return ($address, $nskipped);
559 sub _next_sibling_test {
563 my @components = split /:/, $address;
564 my $last = pop @components;
565 my $maybe = join(":", @components, $last+1);
566 return $maybe if $this->{tree}->select($maybe);
576 ZOOM::IRSpy::Maintenance.
578 The ZOOM-Perl module,
579 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
582 http://indexdata.com/zebra/
586 Mike Taylor, E<lt>mike@indexdata.comE<gt>
588 =head1 COPYRIGHT AND LICENSE
590 Copyright (C) 2006 by Index Data ApS.
592 This library is free software; you can redistribute it and/or modify
593 it under the same terms as Perl itself, either Perl version 5.8.7 or,
594 at your option, any later version of Perl 5 you may have available.