1 # $Id: IRSpy.pm,v 1.45 2006-10-30 16:13:49 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_ns = 'http://indexdata.com/irspy/1.0';
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
38 ZOOM::IRSpy - Perl extension for discovering and analysing IR services
43 $spy = new ZOOM::IRSpy("target/string/for/irspy/database");
44 print $spy->report_status();
48 This module exists to implement the IRspy program, which discovers,
49 analyses and monitors IR servers implementing the Z39.50 and SRU/W
50 protocols. It is a successor to the ZSpy program.
55 ZOOM::Log::mask_str("irspy");
56 ZOOM::Log::mask_str("irspy_debug");
57 ZOOM::Log::mask_str("irspy_event");
58 ZOOM::Log::mask_str("irspy_unhandled");
59 ZOOM::Log::mask_str("irspy_test");
60 ZOOM::Log::mask_str("irspy_task");
65 my($dbname, $user, $password) = @_;
68 push @options, (user => $user, password => $password)
71 my $conn = new ZOOM::Connection($dbname, 0, @options)
72 or die "$0: can't connection to IRSpy database 'dbname'";
74 my $xslt = new XML::LibXSLT;
75 my $libxml = new XML::LibXML;
76 my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl);
77 my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc);
81 allrecords => 1, # unless overridden by targets()
82 query => undef, # filled in later
83 targets => undef, # filled in later
84 connections => undef, # filled in later
86 irspy_to_zeerex_style => $irspy_to_zeerex_style,
87 tests => [], # stack of tests currently being executed
89 $this->log("irspy", "starting up with database '$dbname'");
101 # Explicitly nominate a set of targets to check, overriding the
102 # default which is to re-check everything in the database. Each
103 # target already in the database results in the existing record being
104 # updated; each new target causes a new record to be added.
110 $this->log("irspy", "setting explicit list of targets ",
111 join(", ", map { "'$_'" } @targets));
112 $this->{allrecords} = 0;
114 foreach my $target (@targets) {
115 my($host, $port, $db, $newtarget) = _parse_target_string($target);
116 if ($newtarget ne $target) {
117 $this->log("irspy_debug", "rewriting '$target' to '$newtarget'");
118 $target = $newtarget; # This is written through the ref
120 push @qlist, (qq[(host="$host" and port="$port" and path="$db")]);
123 $this->{targets} = \@targets;
124 $this->{query} = join(" or ", @qlist);
128 # Also used by ZOOM::IRSpy::Record
129 sub _parse_target_string {
132 my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/);
133 if (!defined $host) {
135 ($host, $db) = ($target =~ /(.*?)\/(.*)/);
136 $target = "$host:$port/$db";
138 die "$0: invalid target string '$target'"
141 return ($host, $port, $db, $target);
145 # There are two cases.
147 # 1. A specific set of targets is nominated on the command line.
148 # - Records must be fetched for those targets that are in the DB
149 # - New, empty records must be made for those that are not.
150 # - Updated records written to the DB may or may not be new.
152 # 2. All records in the database are to be checked.
153 # - Records must be fetched for all targets in the DB
154 # - Updated records written to the DB may not be new.
156 # That's all -- what could be simpler?
162 if ($this->{allrecords}) {
163 # We need to check on every target in the database, which
164 # means we need to do a "find all". According to the BIB-1
165 # semantics document at
166 # http://www.loc.gov/z3950/agency/bib1.html
168 # @attr 2=103 @attr 1=1035 x
169 # should find all records, but it seems that Zebra doesn't
170 # support this. Furthermore, when using the "alvis" filter
171 # (as we do for IRSpy) it doesn't support the use of any BIB-1
172 # access point -- not even 1035 "everywhere" -- so instead we
173 # hack together a search that we know will find all records.
174 $this->{query} = "port=?*";
176 # Prepopulate the target map with nulls so that after we fill
177 # in what we can from the database query, we know which target
178 # IDs we need new records for.
179 foreach my $target (@{ $this->{targets} }) {
180 $target2record{lc($target)} = undef;
184 $this->log("irspy_debug", "query '", $this->{query}, "'");
185 my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
186 delete $this->{query}; # No longer needed at all
187 $this->log("irspy_debug", "found ", $rs->size(), " target records");
188 foreach my $i (1 .. $rs->size()) {
189 my $target = _render_record($rs, $i-1, "id");
190 my $zeerex = _render_record($rs, $i-1, "zeerex");
191 #print STDERR "making '$target' record with '$zeerex'\n";
192 $target2record{lc($target)} =
193 new ZOOM::IRSpy::Record($this, $target, $zeerex);
194 push @{ $this->{targets} }, $target
195 if $this->{allrecords};
198 # Make records for targets not previously in the database
199 foreach my $target (keys %target2record) {
200 my $record = $target2record{$target};
201 if (!defined $record) {
202 $this->log("irspy_debug", "made new record for '$target'");
203 $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
205 $this->log("irspy_debug", "using existing record for '$target'");
210 foreach my $target (@{ $this->{targets} }) {
211 my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
212 $conn->option(host => $target);
213 my $record = delete $target2record{lc($target)};
214 $conn->record($record);
215 push @connections, $conn;
217 die("remaining target2record = { " .
218 join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
219 sort keys %target2record) . " }")
222 $this->{connections} = \@connections;
223 delete $this->{targets}; # The information is now in {connections}
228 my($rs, $which, $elementSetName) = @_;
230 # There is a slight race condition here on the element-set name,
231 # but it shouldn't be a problem as this is (currently) only called
232 # from parts of the program that run single-threaded.
233 my $old = $rs->option(elementSetName => $elementSetName);
234 my $rec = $rs->record($which);
235 $rs->option(elementSetName => $old);
237 return $rec->render();
241 sub _irspy_to_zeerex {
242 my ($this, $conn) = @_;
243 my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
245 my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params);
247 return $result->documentElement();
251 sub _rewrite_record {
255 $conn->log("irspy", "rewriting XML record");
256 my $rec = $this->_irspy_to_zeerex($conn);
257 _really_rewrite_record($this->{conn}, $rec);
261 sub _really_rewrite_record {
262 my($conn, $rec) = @_;
264 my $p = $conn->package();
265 $p->option(action => "specialUpdate");
266 my $xml = $rec->toString();
267 $p->option(record => $xml);
271 $p = $conn->package();
278 print "Updated $conn with xml=<br/>\n<pre>$xml</pre>\n";
283 # The approach: gather declarative information about test hierarchy,
284 # then go into a loop. In the loop, we ensure that each connection is
285 # running a test, and within that test a task, until its list of tests
286 # is exhausted. No individual test ever calls wait(): tests just queue
287 # up tasks and return immediately. When the tasks are run (one at a
288 # time on each connection) they generate events, and it is these that
289 # are harvested by ZOOM::event(). Since each connection knows what
290 # task it is running, it can invoke the appropriate callbacks.
291 # Callbacks return a ZOOM::IRSpy::Status value which tells the main
292 # loop how to continue.
295 # While a connection is running a task, its current_task()
296 # points at the task structure. When it finishes its task,
297 # next_task() is pointed at the next task to execute (if there
298 # is one), and its current_task() is set to zero. When the next
299 # task is executed, the connection's next_task() is set to zero
300 # and its current_task() pointed to the task structure.
301 # current_task() and next_task() are both zero only when there
302 # are no more queued tasks, which is when a new test is
305 # Each connection's current test is stored in its
306 # "current_test_address" option. The next test to execute is
307 # calculated by walking the declarative tree of tests. This
308 # option begins empty; the "next test" after this is of course
315 $tname = "Main" if !defined $tname;
316 $this->{tree} = $this->_gather_tests($tname)
317 or die "No tests defined for '$tname'";
318 #$this->{tree}->print(0);
321 my @conn = @{ $this->{connections} };
324 my @copy_conn = @conn; # avoid alias problems after splice()
325 my $nconn = scalar(@copy_conn);
326 foreach my $i0 (0 .. $#copy_conn) {
327 my $conn = $copy_conn[$i0];
328 #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n";
329 if (!$conn->current_task()) {
330 if (!$conn->next_task()) {
331 # Out of tasks: we need a new test
333 my $address = $conn->option("current_test_address");
335 if (!defined $address) {
338 $this->log("irspy_test",
339 "checking for next test after '$address'");
340 $nextaddr = $this->_next_test($address);
342 if (!defined $nextaddr) {
343 $conn->log("irspy", "has no more tests: removing");
344 splice @conn, $i0, 1;
345 $this->_rewrite_record($conn);
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 $this->log("irspy", "exiting main loop");
462 my($tname, @ancestors) = @_;
464 die("$0: test-hierarchy loop detected: " .
465 join(" -> ", @ancestors, $tname))
466 if grep { $_ eq $tname } @ancestors;
468 my $slashSeperatedTname = $tname;
469 $slashSeperatedTname =~ s/::/\//g;
470 my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm";
474 $this->log("irspy", "successfully required '$fullName'");
476 $this->log("irspy", "couldn't require '$fullName': $@");
477 $this->log("warn", "can't load test '$tname': skipping",
478 $@ =~ /^Can.t locate/ ? () : " ($@)");
482 $this->log("irspy", "adding test '$tname'");
484 foreach my $subtname ("ZOOM::IRSpy::Test::$tname"->subtests($this)) {
485 my $subtest = $this->_gather_tests($subtname, @ancestors, $tname);
486 push @subnodes, $subtest if defined $subtest;
489 return new ZOOM::IRSpy::Node($tname, @subnodes);
493 # These next three should arguably be Node methods
496 my($address, $omit_child) = @_;
500 my $maybe = $address eq "" ? "0" : "$address:0";
501 return $maybe if $this->{tree}->select($maybe);
504 # The top-level node has no successor or parent
505 return undef if $address eq "";
507 # Try next sibling child
508 my @components = split /:/, $address;
509 my $last = pop @components;
510 my $maybe = join(":", @components, $last+1);
511 return $maybe if $this->{tree}->select($maybe);
513 # This node is exhausted: try the parent's successor
514 return $this->_next_test(join(":", @components), 1)
518 sub _last_sibling_test {
523 if !defined $this->_next_sibling_test($address);
527 my $maybe = $this->_next_sibling_test($address);
528 last if !defined $maybe;
530 $this->log("irspy", "skipping $nskipped tests to '$address'");
534 return ($address, $nskipped);
538 sub _next_sibling_test {
542 my @components = split /:/, $address;
543 my $last = pop @components;
544 my $maybe = join(":", @components, $last+1);
545 return $maybe if $this->{tree}->select($maybe);
555 ZOOM::IRSpy::Maintenance.
557 The ZOOM-Perl module,
558 http://search.cpan.org/~mirk/Net-Z3950-ZOOM/
561 http://indexdata.com/zebra/
565 Mike Taylor, E<lt>mike@indexdata.comE<gt>
567 =head1 COPYRIGHT AND LICENSE
569 Copyright (C) 2006 by Index Data ApS.
571 This library is free software; you can redistribute it and/or modify
572 it under the same terms as Perl itself, either Perl version 5.8.7 or,
573 at your option, any later version of Perl 5 you may have available.