-# $Id: IRSpy.pm,v 1.73 2007-03-01 14:00:50 mike Exp $
package ZOOM::IRSpy;
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);
+use ZOOM::IRSpy::Utils qw(cql_target render_record
+ irspy_xpath_context irspy_make_identifier
+ irspy_record2identifier calc_reliability_stats
+ modify_xml_document);
our @ISA = qw();
-our $VERSION = '0.02';
+our $VERSION = '1.02';
our $irspy_to_zeerex_xsl = dirname(__FILE__) . '/../../xsl/irspy2zeerex.xsl';
+our $debug = 0;
+our $xslt_max_depth = 250;
+our $max_timeout_errors = 3;
# Enumeration for callback functions to return
use ZOOM::IRSpy;
$spy = new ZOOM::IRSpy("target/string/for/irspy/database");
$spy->targets(@targets);
- $spy->initialise();
- $res = $spy->check("Main");
+ $spy->initialise("Main");
+ $res = $spy->check();
=head1 DESCRIPTION
sub new {
my $class = shift();
- my($dbname, $user, $password) = @_;
+ my($dbname, $user, $password, $activeSetSize) = @_;
+
my @options;
push @options, (user => $user, password => $password)
my $xslt = new XML::LibXSLT;
+ # raise the maximum number of nested template calls and variables/params (default 250)
+ warn "raise the maximum number of nested template calls: $xslt_max_depth\n" if $debug;
+ $xslt->max_depth($xslt_max_depth);
+
$xslt->register_function($ZOOM::IRSpy::Utils::IRSPY_NS, 'strcmp',
\&ZOOM::IRSpy::Utils::xslt_strcmp);
my $libxml = new XML::LibXML;
+ warn "use irspy_to_zeerex_xsl xslt sheet: $irspy_to_zeerex_xsl\n" if $debug;
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,
query => "cql.allRecords=1", # unless overridden
- targets => undef, # filled in later
- connections => undef, # filled in later
+ modn => undef, # Filled in by restrict_modulo()
+ modi => undef, # Filled in by restrict_modulo()
+ 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()
+ timeout => 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'");
return $this;
}
+# wrapper to read the IRSpy database name from environment variable / apache config
+sub connect_to_registry {
+ my %args = @_;
+
+ # XXX: we could also handle her: user, password, elementSetName
+
+ my $database = $ENV{IRSpyDbName} || "localhost:8018/IR-Explain---1";
+
+ return $database;
+}
sub log {
my $this = shift();
join(", ", map { "'$_'" } @targets));
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 = $newtarget; # This is written through the ref
}
- push @qlist, cql_target($host, $port, $db);
+ push @qlist, cql_target($protocol, $host, $port, $db);
}
$this->{targets} = \@targets;
sub _parse_target_string {
my($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);
+}
+
+
+sub restrict_modulo {
+ my $this = shift();
+ my($n, $i) = @_;
+
+ $this->{modn} = $n;
+ $this->{modi} = $i;
}
# 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
-# must be made for any targets that are not already in the database.
+# will be made for any targets that are not already in the database.
#
sub initialise {
my $this = shift();
+ my($tname) = @_;
- my %target2record;
- if ($this->{targets}) {
- # 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);
+
+ $this->{timeout} = "ZOOM::IRSpy::Test::$tname"->timeout();
+
+ 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");
}
}
- delete $this->{targets}; # Information now in keys of %target2record
-
- my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query}));
- $this->log("irspy", "'", $this->{query}, "' found ",
- $rs->size(), " target records");
- delete $this->{query}; # Information now in $rs
-
- foreach my $i (1 .. $rs->size()) {
- my $target = _render_record($rs, $i-1, "id");
- my $zeerex = _render_record($rs, $i-1, "zeerex");
- $target2record{lc($target)} =
- new ZOOM::IRSpy::Record($this, $target, $zeerex);
- }
- # Make records for targets not previously in the database
- foreach my $target (keys %target2record) {
- if (!defined $target2record{$target}) {
- $target2record{$target} = new ZOOM::IRSpy::Record($this, $target);
- $this->log("irspy_debug", "made new record for '$target'");
- } else {
- $this->log("irspy_debug", "using existing record for '$target'");
- }
+ my $n = $this->{activeSetSize};
+ $n = @targets if $n == 0 || $n > @targets;
+
+ $this->{queue} = \@targets;
+ $this->{connections} = [];
+ while (@{ $this->{connections} } < $n) {
+ my $conn = $this->_next_connection();
+ last if !defined $conn;
+ push @{ $this->{connections} }, $conn;
}
+}
- my @connections;
- my @targets = sort keys %target2record;
- foreach my $target (@targets) {
- my $conn = create ZOOM::IRSpy::Connection($this, async => 1);
- $conn->option(host => $target);
- my $record = delete $target2record{lc($target)};
- die "record undefined for '$target'" if !defined $record;
- $conn->record($record);
- push @connections, $conn;
+
+sub _next_connection {
+ my $this = shift();
+
+ my $target;
+ my $n = $this->{modn};
+ my $i = $this->{modi};
+ if (!defined $n) {
+ $target = shift @{ $this->{queue} };
+ return undef if !defined $target;
+ } else {
+ while (1) {
+ $target = shift @{ $this->{queue} };
+ return undef if !defined $target;
+ my $h = _hash($target);
+ my $hmodn = $h % $n;
+ last if $hmodn == $i;
+ #$this->log("irspy", "'$target' hash $h % $n = $hmodn != $i");
+ }
}
- die("remaining target2record = { " .
- join(", ", map { "$_ ->'" . $target2record{$_}. "'" }
- sort keys %target2record) . " }")
- if %target2record;
- $this->{connections} = \@connections;
+ die "oops -- target is undefined" if !defined $target;
+ return create ZOOM::IRSpy::Connection($this, $target, async => 1,
+ timeout => $this->{timeout});
}
-sub _render_record {
- my($rs, $which, $elementSetName) = @_;
+sub _hash {
+ my($target) = @_;
- # 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);
+ my $n = 0;
+ foreach my $s (split //, $target) {
+ $n += ord($s);
+ }
- return $rec->render();
+ return $n;
}
sub _irspy_to_zeerex {
my $this = shift();
- my($conn, $save_xml) = @_;
+ my($conn) = @_;
+
+ my $save_xml = $ENV{IRSPY_SAVE_XML};
my $irspy_doc = $conn->record()->{zeerex}->ownerDocument;
if ($save_xml) {
}
-sub _rewrite_record {
+sub _rewrite_irspy_record {
my $this = shift();
my($conn) = @_;
$conn->log("irspy", "rewriting XML record");
- my $rec = $this->_irspy_to_zeerex($conn, $ENV{IRSPY_SAVE_XML});
- _really_rewrite_record($this->{conn}, $rec);
+ my $rec = $this->_irspy_to_zeerex($conn);
+
+ # 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"));
+
+ _rewrite_zeerex_record($this->{conn}, $rec);
+ $conn->log("irspy", "rewrote XML record");
}
-sub _really_rewrite_record {
- my($conn, $rec) = @_;
+my $_reliabilityField = {
+ reliability => [ reliability => 0,
+ "Calculated reliability of server",
+ "e:serverInfo/e:reliability" ],
+};
+
+sub _rewrite_zeerex_record {
+ my($conn, $rec, $oldid) = @_;
+
+ # Add reliability score
+ my $xc = irspy_xpath_context($rec);
+ my($nok, $nall, $percent) = calc_reliability_stats($xc);
+ modify_xml_document($xc, $_reliabilityField, { reliability => $percent });
my $p = $conn->package();
$p->option(action => "specialUpdate");
$p->send("update");
$p->destroy();
+ # This is the expression in the ID-making stylesheet
+ # ../../zebra/zeerex2id.xsl
+ my $id = irspy_record2identifier($xc);
+ if (defined $oldid && $id ne $oldid) {
+ warn "IDs differ (old='$oldid' new='$id')";
+ _delete_record($conn, $oldid);
+ }
+
$p = $conn->package();
$p->send("commit");
$p->destroy();
}
+sub _delete_record {
+ my($conn, $id) = @_;
+
+ # We can't delete records using recordIdOpaque, since character
+ # sets are handled differently here in extended services from how
+ # they are used in the Alvis filter's record-parsing, and so
+ # non-ASCII characters come out differently in the two contexts.
+ # Instead, we must send a record whose contents indicate the ID of
+ # that which we wish to delete. There are two ways, both
+ # unsatisfactory: we could either fetch the actual record them
+ # resubmit it in the deletion request (which wastes a search and a
+ # fetch) or we could build a record by hand from the parsed-out
+ # components (which is error-prone and which I am not 100% certain
+ # will work since the other contents of the record will be
+ # different). The former evil seems to be the lesser.
+
+ warn "$conn deleting record '$id'";
+
+ my $rs = $conn->search(new ZOOM::Query::CQL(cql_target($id)));
+ die "no such ID '$id'" if $rs->size() == 0;
+ my $rec = $rs->record(0);
+ my $xml = $rec->render();
+
+ my $p = $conn->package();
+ $p->option(action => "recordDelete");
+ $p->option(record => $xml);
+ $p->send("update");
+ $p->destroy();
+
+ $p = $conn->package();
+ $p->send("commit");
+ $p->destroy();
+}
+
+
# The approach: gather declarative information about test hierarchy,
# then go into a loop. In the loop, we ensure that each connection is
# running a test, and within that test a task, until its list of tests
#
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'";
- $this->{tree}->resolve();
- #$this->{tree}->print(0);
- my $nskipped = 0;
+ my $topname = $this->{tree}->name();
+ my $timeout = $this->{timeout};
+ $this->log("irspy", "beginnning with test '$topname' (timeout $timeout)");
+ my $nskipped = 0;
my @conn = @{ $this->{connections} };
my $nruns = 0;
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
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 (ZOOM::IRSpy::Test::zoom_error_timeout_check($conn)) {
+ $conn->log("irspy", "Got to many timeouts, stop testing");
+ undef $nextaddr;
+ }
+
if (!defined $nextaddr) {
$conn->log("irspy", "has no more tests: removing");
- ### Does this go wrong if two connections are exhausted?
- splice @conn, $i0, 1;
- $this->_rewrite_record($conn);
+ $this->_rewrite_irspy_record($conn);
$conn->option(rewrote_record => 1);
+ my $newconn = $this->_next_connection();
+ if (!defined $newconn) {
+ # Do not destroy: needed for later sanity checks
+ splice @conn, $i0, 1;
+ } else {
+ $conn->destroy();
+ $conn[$i0] = $newconn;
+ $conn[$i0]->option(current_test_address => "");
+ $conn[$i0]->log("irspy", "entering active pool - ",
+ scalar(@{ $this->{queue} }),
+ " targets remain in queue");
+ }
next;
}
my $task = $conn->next_task();
die "no next task queued for $conn" if !defined $task;
+
+ # do not run the next task if we got too many timeouts
+ if (ZOOM::IRSpy::Test::zoom_error_timeout_check($conn)) {
+ $conn->log("irspy_task", "Got to many timeouts for this target, do not start a new task");
+ next;
+ }
+
$conn->log("irspy_task", "preparing task $task");
$conn->next_task(0);
$conn->current_task($task);
}
$this->log("irspy", "exiting main loop");
+
# Sanity checks: none of the following should ever happen
my $finished = 1;
- @conn = @{ $this->{connections} };
+ $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);
}
+# 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) = @_;
eval {
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",