X-Git-Url: http://git.indexdata.com/?p=irspy-moved-to-github.git;a=blobdiff_plain;f=lib%2FZOOM%2FIRSpy.pm;h=0fe969eafded6b1da4992e881f25becdee7cf6f2;hp=b3c0aeb15e8de464b155625bccabe60bd052d75d;hb=393f5d3e8ba6ee9d8e15174914c060d6ff0f0c9d;hpb=09a78b8e8d348d62681278705447c19070cbd6b1 diff --git a/lib/ZOOM/IRSpy.pm b/lib/ZOOM/IRSpy.pm index b3c0aeb..0fe969e 100644 --- a/lib/ZOOM/IRSpy.pm +++ b/lib/ZOOM/IRSpy.pm @@ -1,4 +1,4 @@ -# $Id: IRSpy.pm,v 1.72 2007-02-28 17:36:00 mike Exp $ +# $Id: IRSpy.pm,v 1.85 2007-06-27 11:09:03 mike Exp $ package ZOOM::IRSpy; @@ -15,9 +15,10 @@ 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); +use ZOOM::IRSpy::Utils qw(cql_target render_record + irspy_xpath_context irspy_make_identifier + irspy_record2identifier); our @ISA = qw(); our $VERSION = '0.02'; @@ -43,8 +44,8 @@ ZOOM::IRSpy - Perl extension for discovering and analysing IR services 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 @@ -65,7 +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) @@ -85,13 +86,18 @@ sub new { 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'"); @@ -105,6 +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 @@ -116,15 +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 = $newtarget; # This is written through the ref } - push @qlist, cql_target($host, $port, $db); + push @qlist, cql_target($protocol, $host, $port, $db); } $this->{targets} = \@targets; @@ -132,127 +146,70 @@ sub targets { } -sub find_targets { - my $this = shift(); - my($query) = @_; - - $this->{allrecords} = 0; - $this->{query} = $query; -} - - # Also used by ZOOM::IRSpy::Record sub _parse_target_string { my($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. - $this->{query} = "port=?*"; - } elsif ($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; - } - } - - $this->log("irspy_debug", "query '", $this->{query}, "'"); - my $rs = $this->{conn}->search(new ZOOM::Query::CQL($this->{query})); - $this->log("irspy", "'", $this->{query}, "' found ", - $rs->size(), " target records"); - delete $this->{query}; # No longer needed at all - my $gatherTargets = !$this->{targets}; - 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 $gatherTargets; - } + $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); - # 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 $timeout = "ZOOM::IRSpy::Test::$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"); } } + 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; } @@ -288,12 +245,19 @@ sub _rewrite_record { $conn->log("irspy", "rewriting XML record"); 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"); @@ -302,6 +266,15 @@ 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='$id')"; + _delete_record($conn, $oldid); + } + $p = $conn->package(); $p->send("commit"); $p->destroy(); @@ -314,6 +287,41 @@ sub _really_rewrite_record { } +sub _delete_record { + my($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 @@ -344,15 +352,12 @@ 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'"; - $this->{tree}->resolve(); - #$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; @@ -363,6 +368,7 @@ sub check { 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 @@ -372,16 +378,28 @@ 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"); - ### Does this go wrong if two connections are exhausted? - 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; } @@ -508,9 +526,10 @@ sub check { } $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); @@ -570,6 +589,29 @@ sub check { } +# 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) = @_; @@ -584,7 +626,6 @@ sub _gather_tests { 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",