X-Git-Url: http://git.indexdata.com/?p=irspy-moved-to-github.git;a=blobdiff_plain;f=lib%2FZOOM%2FIRSpy.pm;h=bc82ad4ce94fa59d9e4e7127fb86d455c08cbdcc;hp=52c586e8b611db1ca895b5167bff2ed7d478bc6d;hb=2185164008cd6dd14e31e44bb70027de83ce4316;hpb=d1650850febb5bdc6212d60049899c0b53532a4a diff --git a/lib/ZOOM/IRSpy.pm b/lib/ZOOM/IRSpy.pm index 52c586e..bc82ad4 100644 --- a/lib/ZOOM/IRSpy.pm +++ b/lib/ZOOM/IRSpy.pm @@ -1,4 +1,4 @@ -# $Id: IRSpy.pm,v 1.42 2006-10-27 15:36:04 mike Exp $ +# $Id: IRSpy.pm,v 1.84 2007-04-30 11:28:37 mike Exp $ package ZOOM::IRSpy; @@ -6,9 +6,6 @@ use 5.008; use strict; use warnings; -use Exporter 'import'; -our @EXPORT_OK = qw(xml_encode irspy_xpath_context); - use Data::Dumper; # For debugging only use File::Basename; use XML::LibXSLT; @@ -18,11 +15,13 @@ use ZOOM; use Net::Z3950::ZOOM 1.13; # For the ZOOM version-check only use ZOOM::IRSpy::Node; use ZOOM::IRSpy::Connection; -use ZOOM::IRSpy::Record; +use ZOOM::IRSpy::Stats; +use ZOOM::IRSpy::Utils qw(cql_target render_record + irspy_xpath_context irspy_make_identifier + irspy_record2identifier); our @ISA = qw(); our $VERSION = '0.02'; -our $irspy_ns = 'http://indexdata.com/irspy/1.0'; our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl'; @@ -32,6 +31,7 @@ sub OK { 29 } # No problems, task is still progressing sub TASK_DONE { 18 } # Task is complete, next task should begin sub TEST_GOOD { 8 } # Whole test is complete, and succeeded sub TEST_BAD { 31 } # Whole test is complete, and failed +sub TEST_SKIPPED { 12 } # Test couldn't be run package ZOOM::IRSpy; @@ -43,7 +43,9 @@ ZOOM::IRSpy - Perl extension for discovering and analysing IR services use ZOOM::IRSpy; $spy = new ZOOM::IRSpy("target/string/for/irspy/database"); - print $spy->report_status(); + $spy->targets(@targets); + $spy->initialise("Main"); + $res = $spy->check(); =head1 DESCRIPTION @@ -64,7 +66,7 @@ BEGIN { sub new { my $class = shift(); - my($dbname, $user, $password) = @_; + my($dbname, $user, $password, $activeSetSize) = @_; my @options; push @options, (user => $user, password => $password) @@ -74,19 +76,28 @@ sub new { or die "$0: can't connection to IRSpy database 'dbname'"; my $xslt = new XML::LibXSLT; + + $xslt->register_function($ZOOM::IRSpy::Utils::IRSPY_NS, 'strcmp', + \&ZOOM::IRSpy::Utils::xslt_strcmp); + my $libxml = new XML::LibXML; my $xsl_doc = $libxml->parse_file($irspy_to_zeerex_xsl); my $irspy_to_zeerex_style = $xslt->parse_stylesheet($xsl_doc); my $this = bless { conn => $conn, - allrecords => 1, # unless overridden by targets() - query => undef, # filled in later - targets => undef, # filled in later - connections => undef, # filled in later + query => "cql.allRecords=1", # unless overridden + targets => undef, # Filled in later if targets() is + # called; used only to keep state from + # targets() until initialise() is + # called. + connections => undef, # Filled in by initialise() + queue => undef, # Filled in by initialise() libxml => $libxml, irspy_to_zeerex_style => $irspy_to_zeerex_style, - tests => [], # stack of tests currently being executed + test => undef, # Filled in by initialise() + tests => undef, # Tree of tests to be executed + activeSetSize => defined $activeSetSize ? $activeSetSize : 10, }, $class; $this->log("irspy", "starting up with database '$dbname'"); @@ -100,6 +111,14 @@ sub log { } +sub find_targets { + my $this = shift(); + my($query) = @_; + + $this->{query} = $query; +} + + # Explicitly nominate a set of targets to check, overriding the # default which is to re-check everything in the database. Each # target already in the database results in the existing record being @@ -111,15 +130,15 @@ sub targets { $this->log("irspy", "setting explicit list of targets ", join(", ", map { "'$_'" } @targets)); - $this->{allrecords} = 0; my @qlist; foreach my $target (@targets) { - my($host, $port, $db, $newtarget) = _parse_target_string($target); + my($protocol, $host, $port, $db, $newtarget) = + _parse_target_string($target); if ($newtarget ne $target) { $this->log("irspy_debug", "rewriting '$target' to '$newtarget'"); $target = $newtarget; # This is written through the ref } - push @qlist, (qq[(host="$host" and port="$port" and path="$db")]); + push @qlist, cql_target($protocol, $host, $port, $db); } $this->{targets} = \@targets; @@ -131,120 +150,90 @@ sub targets { sub _parse_target_string { my($target) = @_; - my($host, $port, $db) = ($target =~ /(.*?):(.*?)\/(.*)/); + my($protocol, $host, $port, $db) = ($target =~ /(.*?):(.*?):(.*?)\/(.*)/); if (!defined $host) { $port = 210; - ($host, $db) = ($target =~ /(.*?)\/(.*)/); - $target = "$host:$port/$db"; + ($protocol, $host, $db) = ($target =~ /(.*?):(.*?)\/(.*)/); + $target = irspy_make_identifier($protocol, $host, $port, $db); } die "$0: invalid target string '$target'" if !defined $host; - return ($host, $port, $db, $target); + return ($protocol, $host, $port, $db, $target); } -# There are two cases. -# -# 1. A specific set of targets is nominated on the command line. -# - Records must be fetched for those targets that are in the DB -# - New, empty records must be made for those that are not. -# - Updated records written to the DB may or may not be new. -# -# 2. All records in the database are to be checked. -# - Records must be fetched for all targets in the DB -# - Updated records written to the DB may not be new. -# -# That's all -- what could be simpler? +# Records must be fetched for all records satisfying $this->{query} If +# $this->{targets} is already set (i.e. a specific list of targets to +# check was specified by a call to targets()), then new, empty records +# will be made for any targets that are not already in the database. # sub initialise { my $this = shift(); + my($tname) = @_; - my %target2record; - if ($this->{allrecords}) { - # We need to check on every target in the database, which - # means we need to do a "find all". According to the BIB-1 - # semantics document at - # http://www.loc.gov/z3950/agency/bib1.html - # the query - # @attr 2=103 @attr 1=1035 x - # should find all records, but it seems that Zebra doesn't - # support this. Furthermore, when using the "alvis" filter - # (as we do for IRSpy) it doesn't support the use of any BIB-1 - # access point -- not even 1035 "everywhere" -- so instead we - # hack together a search that we know will find all records. - $this->{query} = "port=?*"; - } else { - # Prepopulate the target map with nulls so that after we fill - # in what we can from the database query, we know which target - # IDs we need new records for. - foreach my $target (@{ $this->{targets} }) { - $target2record{lc($target)} = undef; - } - } + $tname = "Main" if !defined $tname; + $this->{test} = $tname; + $this->{tree} = $this->_gather_tests($tname) + or die "No tests defined for '$tname'"; + $this->{tree}->resolve(); + #$this->{tree}->print(0); - $this->log("irspy_debug", "query '", $this->{query}, "'"); - my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query})); - delete $this->{query}; # No longer needed at all - $this->log("irspy_debug", "found ", $rs->size(), " target records"); - foreach my $i (1 .. $rs->size()) { - my $target = _render_record($rs, $i-1, "id"); - my $zeerex = _render_record($rs, $i-1, "zeerex"); - #print STDERR "making '$target' record with '$zeerex'\n"; - $target2record{lc($target)} = - new ZOOM::IRSpy::Record($this, $target, $zeerex); - push @{ $this->{targets} }, $target - if $this->{allrecords}; - } + my $timeout = "ZOOM::IRSpy::Test::$tname"->timeout(); - # Make records for targets not previously in the database - foreach my $target (keys %target2record) { - my $record = $target2record{$target}; - if (!defined $record) { - $this->log("irspy_debug", "made new record for '$target'"); - $target2record{$target} = new ZOOM::IRSpy::Record($this, $target); - } else { - $this->log("irspy_debug", "using existing record for '$target'"); + my @targets; + my $targets = $this->{targets}; + if (defined $targets) { + @targets = @$targets; + delete $this->{targets}; + } else { + my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query})); + $this->log("irspy", "'", $this->{query}, "' found ", + $rs->size(), " target records"); + delete $this->{query}; + + foreach my $i (1 .. $rs->size()) { + push @targets, render_record($rs, $i-1, "id"); } } + my $n = $this->{activeSetSize}; + $n = @targets if $n == 0 || $n > @targets; + my @connections; - foreach my $target (@{ $this->{targets} }) { - my $conn = create ZOOM::IRSpy::Connection($this, async => 1); - $conn->option(host => $target); - my $record = delete $target2record{lc($target)}; - $conn->record($record); - push @connections, $conn; + foreach my $i (1..$n) { + push @connections, create ZOOM::IRSpy::Connection($this, + shift @targets, + async => 1, + timeout => $timeout); } - die("remaining target2record = { " . - join(", ", map { "$_ ->'" . $target2record{$_}. "'" } - sort keys %target2record) . " }") - if %target2record; $this->{connections} = \@connections; - delete $this->{targets}; # The information is now in {connections} -} - - -sub _render_record { - my($rs, $which, $elementSetName) = @_; - - # There is a slight race condition here on the element-set name, - # but it shouldn't be a problem as this is (currently) only called - # from parts of the program that run single-threaded. - my $old = $rs->option(elementSetName => $elementSetName); - my $rec = $rs->record($which); - $rs->option(elementSetName => $old); - - return $rec->render(); + $this->{queue} = \@targets; } sub _irspy_to_zeerex { - my ($this, $conn) = @_; + my $this = shift(); + my($conn, $save_xml) = @_; my $irspy_doc = $conn->record()->{zeerex}->ownerDocument; + + if ($save_xml) { + unlink('/tmp/irspy_orig.xml'); + open FH, '>/tmp/irspy_orig.xml' + or die "can't write irspy_orig.xml: $!"; + print FH $irspy_doc->toString(); + close FH; + } my %params = (); my $result = $this->{irspy_to_zeerex_style}->transform($irspy_doc, %params); + if ($save_xml) { + unlink('/tmp/irspy_transformed.xml'); + open FH, '>/tmp/irspy_transformed.xml' + or die "can't write irspy_transformed.xml: $!"; + print FH $result->toString(); + close FH; + } return $result->documentElement(); } @@ -255,13 +244,20 @@ sub _rewrite_record { my($conn) = @_; $conn->log("irspy", "rewriting XML record"); - my $rec = $this->_irspy_to_zeerex($conn); + my $rec = $this->_irspy_to_zeerex($conn, $ENV{IRSPY_SAVE_XML}); + + # Since IRSpy can run for a long time between writes back to the + # database, it's quite possible for the server to have closed the + # connection as idle. So re-establish it if necessary. + $this->{conn}->connect($conn->option("host")); + _really_rewrite_record($this->{conn}, $rec); + $conn->log("irspy", "rewrote XML record"); } sub _really_rewrite_record { - my($conn, $rec) = @_; + my($conn, $rec, $oldid) = @_; my $p = $conn->package(); $p->option(action => "specialUpdate"); @@ -270,6 +266,22 @@ sub _really_rewrite_record { $p->send("update"); $p->destroy(); + # This is the expression in the ID-making stylesheet + # ../../zebra/zeerex2id.xsl + my $xc = irspy_xpath_context($rec); + my $id = irspy_record2identifier($xc); + if (defined $oldid && $id ne $oldid) { + warn "IDs differ (old='$oldid' new='$id')"; + # Delete old record; + ### Should use same mechanism as delete.mc + my $p = $conn->package(); + $p->option(action => "recordDelete"); + $p->option(recordIdOpaque => $oldid); + $p->option(record => ""); # Work around Zebra bug + $p->send("update"); + $p->destroy(); + } + $p = $conn->package(); $p->send("commit"); $p->destroy(); @@ -312,22 +324,23 @@ sub _really_rewrite_record { # sub check { my $this = shift(); - my($tname) = @_; - $tname = "Main" if !defined $tname; - $this->{tree} = $this->_gather_tests($tname) - or die "No tests defined for '$tname'"; - #$this->{tree}->print(0); - my $nskipped = 0; + my $topname = $this->{tree}->name(); + my $timeout = "ZOOM::IRSpy::Test::$topname"->timeout(); + $this->log("irspy", "beginnning with test '$topname' (timeout $timeout)"); + my $nskipped = 0; my @conn = @{ $this->{connections} }; + my $nruns = 0; + ROUND_AND_ROUND_WE_GO: while (1) { my @copy_conn = @conn; # avoid alias problems after splice() my $nconn = scalar(@copy_conn); foreach my $i0 (0 .. $#copy_conn) { my $conn = $copy_conn[$i0]; #print "connection $i0 of $nconn/", scalar(@conn), " is $conn\n"; + next if !defined $conn; if (!$conn->current_task()) { if (!$conn->next_task()) { # Out of tasks: we need a new test @@ -337,14 +350,28 @@ sub check { if (!defined $address) { $nextaddr = ""; } else { - $this->log("irspy_test", + $conn->log("irspy_test", "checking for next test after '$address'"); $nextaddr = $this->_next_test($address); } if (!defined $nextaddr) { $conn->log("irspy", "has no more tests: removing"); - splice @conn, $i0, 1; $this->_rewrite_record($conn); + $conn->option(rewrote_record => 1); + if (@{ $this->{queue} } == 0) { + # Do not destroy: we need this for later sanity checks + splice @conn, $i0, 1; + } else { + $conn->destroy(); + $conn[$i0] = create + ZOOM::IRSpy::Connection($this, + shift @{ $this->{queue} }, async => 1, + timeout => $timeout); + $conn[$i0]->option(current_test_address => ""); + $conn[$i0]->log("irspy", "entering active pool - ", + scalar(@{ $this->{queue} }), + " targets remain in queue"); + } next; } @@ -375,43 +402,42 @@ sub check { $conn->current_task($task); $task->run(); } - - # Do we need to test $conn->is_idle()? I don't think so! } + NEXT_EVENT: my $i0 = ZOOM::event(\@conn); $this->log("irspy_event", "ZOOM_event(", scalar(@conn), " connections) = $i0"); - last if $i0 == 0 || $i0 == -3; # no events or no connections + if ($i0 < 1) { + my %messages = ( + 0 => "no events remain", + -1 => "ZOOM::event() argument not a reference", + -2 => "ZOOM::event() reference not an array", + -3 => "no connections remain", + -4 => "too many connections for ZOOM::event()", + ); + my $message = $messages{$i0} || "ZOOM::event() returned $i0"; + $this->log("irspy", $message); + last; + } + my $conn = $conn[$i0-1]; my $ev = $conn->last_event(); my $evstr = ZOOM::event_str($ev); $conn->log("irspy_event", "event $ev ($evstr)"); + goto NEXT_EVENT if $ev != ZOOM::Event::ZEND; my $task = $conn->current_task(); die "$conn has no current task for event $ev ($evstr)" if !$task; - eval { $conn->_check() }; - if ($@ && - ($ev == ZOOM::Event::RECV_DATA || - $ev == ZOOM::Event::RECV_APDU || - $ev == ZOOM::Event::ZEND)) { - # An error in, say, a search response, becomes visible to - # ZOOM before the Receive Data event is sent and persists - # until after the End, which means that successive events - # each report the same error. So we just ignore errors on - # "unimportant" events. ### But this doesn't work for, - # say, a Connection Refused, as the only event that shows - # us this error is the End. - $conn->log("irspy_event", "ignoring error ", - "on event $ev ($evstr): $@"); - next; - } my $res; - if ($@) { + eval { $conn->check() }; + if ($@ && ref $@ && $@->isa("ZOOM::Exception")) { my $sub = $task->{cb}->{exception}; die $@ if !defined $sub; $res = &$sub($conn, $task, $task->udata(), $@); + } elsif ($@) { + die "Unexpected non-ZOOM exception: " . ref($@) . " ($@)"; } else { my $sub = $task->{cb}->{$ev}; if (!defined $sub) { @@ -442,23 +468,122 @@ sub check { $conn->next_task(0); if ($res == ZOOM::IRSpy::Status::TEST_BAD) { my $address = $conn->option('current_test_address'); - ($address, my $n) = $this->_last_sibling_test($address); - if (defined $address) { - $conn->log("irspy_test", "skipped $n tests"); - $conn->option(current_test_address => $address); - $nskipped += $n; + $conn->log("irspy", "top-level test failed!") + if $address eq ""; + my $node = $this->{tree}->select($address); + my $skipcount = 0; + while (defined $node->next() && + length($node->next()->address()) >= length($address)) { + $conn->log("irspy_debug", "skipping from '", + $node->address(), "' to '", + $node->next()->address(), "'"); + $node = $node->next(); + $skipcount++; } + + $conn->option(current_test_address => $node->address()); + $conn->log("irspy_test", "skipped $skipcount tests"); + $nskipped += $skipcount; } + + } elsif ($res == ZOOM::IRSpy::Status::TEST_SKIPPED) { + $conn->log("irspy_test", "test skipped during task $task"); + $conn->current_task(0); + $conn->next_task(0); + $nskipped++; + } else { die "unknown callback return-value '$res'"; } } $this->log("irspy", "exiting main loop"); + + # Sanity checks: none of the following should ever happen + my $finished = 1; + $this->log("irspy", "performing end-of-run sanity-checks"); + foreach my $conn (@conn) { + my $test = $conn->option("current_test_address"); + my $next = $this->_next_test($test); + if (defined $next) { + $this->log("irspy", + "$conn (in test '$test') has queued test '$next'"); + $finished = 0; + } + if (my $task = $conn->current_task()) { + $this->log("irspy", "$conn still has an active task $task"); + $finished = 0; + } + if (my $task = $conn->next_task()) { + $this->log("irspy", "$conn still has a queued task $task"); + $finished = 0; + } + if (!$conn->is_idle()) { + $this->log("irspy", + "$conn still has ZOOM-C level tasks queued: see below"); + $finished = 0; + } + my $ev = $conn->peek_event(); + if ($ev != 0 && $ev != ZOOM::Event::ZEND) { + my $evstr = ZOOM::event_str($ev); + $this->log("irspy", "$conn has event $ev ($evstr) waiting"); + $finished = 0; + } + if (!$conn->option("rewrote_record")) { + $this->log("irspy", "$conn did not rewrite its ZeeRex record"); + $finished = 0; + } + } + + # This really shouldn't be necessary, and in practice it rarely + # helps, but it's belt and braces. (For now, we don't do this + # hence the zero in the $nruns check). + if (!$finished) { + if (++$nruns < 0) { + $this->log("irspy", "back into main loop, ${nruns}th time"); + goto ROUND_AND_ROUND_WE_GO; + } else { + $this->log("irspy", "bailing after $nruns main-loop runs"); + } + } + + # This shouldn't happen emit anything either: + while ((my $i1 = ZOOM::event(\@conn)) > 0) { + my $conn = $conn[$i1-1]; + my $ev = $conn->last_event(); + my $evstr = ZOOM::event_str($ev); + $this->log("irspy", + "$conn still has ZOOM-C level task queued: $ev ($evstr)") + if $ev != ZOOM::Event::ZEND; + } + return $nskipped; } +# Exactly equivalent to ZOOM::event() except that it is tolerant to +# undefined values in the array being passed in. +# +sub __UNUSED_tolerant_ZOOM_event { + my($connref) = @_; + + my(@conn, @map); + foreach my $i (0 .. @$connref-1) { + my $conn = $connref->[$i]; + if (defined $conn) { + push @conn, $conn; + push @map, $i; + } + } + + my $res = ZOOM::event(\@conn); + return $res if $res <= 0; + my $res2 = $map[$res-1] + 1; + print STDERR "*** tolerant_ZOOM_event() returns $res->$res2\n"; + return $res2; +} + + sub _gather_tests { my $this = shift(); my($tname, @ancestors) = @_; @@ -472,15 +597,7 @@ sub _gather_tests { my $fullName = "ZOOM/IRSpy/Test/$slashSeperatedTname.pm"; eval { - ### This next line shouldn't be necessary, as we should - # already be running in an environment where the test - # modules are available -- otherwise, how did _this_ module - # get loaded? But it seems that for reasons I don't - # understand we do sometimes (not always!) need this when - # running under Apache. - use lib '/usr/local/src/cvs/irspy/lib'; require $fullName; - $this->log("irspy", "successfully required '$fullName'"); }; if ($@) { $this->log("irspy", "couldn't require '$fullName': $@"); $this->log("warn", "can't load test '$tname': skipping", @@ -536,8 +653,8 @@ sub _last_sibling_test { my $maybe = $this->_next_sibling_test($address); last if !defined $maybe; $nskipped++; - $this->log("irspy", "skipping $nskipped tests to '$address'"); $address = $maybe; + $this->log("irspy", "skipping $nskipped tests to '$address'"); } return ($address, $nskipped); @@ -556,38 +673,6 @@ sub _next_sibling_test { } -# Utility functions follow, exported for use of web UI - -# I can't -- just can't, can't, can't -- believe that this function -# isn't provided by one of the core XML modules. But the evidence all -# says that it's not: among other things, XML::Generator and -# Template::Plugin both roll their own. So I will do likewise. D'oh! -# -sub xml_encode { - my ($text) = @_; - $text =~ s/&/&/g; - $text =~ s//>/g; - $text =~ s/['']/'/g; - $text =~ s/[""]/"/g; - return $text; -} - - -sub irspy_xpath_context { - my($zoom_record) = @_; - - my $xml = $zoom_record->render(); - my $parser = new XML::LibXML(); - my $doc = $parser->parse_string($xml); - my $root = $doc->getDocumentElement(); - my $xc = XML::LibXML::XPathContext->new($root); - $xc->registerNs(e => 'http://explain.z3950.org/dtd/2.0/'); - $xc->registerNs(i => $irspy_ns); - return $xc; -} - - =head1 SEE ALSO ZOOM::IRSpy::Record,