X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=lib%2FZOOM%2FIRSpy%2FRecord.pm;h=c5b6c1bab63ac6965c9a5fb33b2f5d95dcd154e6;hb=1e23eeb797f1bbe48e3d3977d864a3d148b71597;hp=8a09c541cdad02919a526bf5035ab30acd19f3ca;hpb=65645b766a450a2ba6ab35ffde844ae250ac15c2;p=irspy-moved-to-github.git
diff --git a/lib/ZOOM/IRSpy/Record.pm b/lib/ZOOM/IRSpy/Record.pm
index 8a09c54..c5b6c1b 100644
--- a/lib/ZOOM/IRSpy/Record.pm
+++ b/lib/ZOOM/IRSpy/Record.pm
@@ -1,14 +1,16 @@
-# $Id: Record.pm,v 1.6 2006-07-21 16:50:20 mike Exp $
package ZOOM::IRSpy::Record;
+### I don't think there's any reason for this to be separate from
+# ZOOM::IRSpy::Connection, now that the correspondence is always 1:1
use 5.008;
use strict;
use warnings;
+use Scalar::Util;
use XML::LibXML;
use XML::LibXML::XPathContext;
-
+use ZOOM::IRSpy::Utils qw(xml_encode isodate irspy_xpath_context);
=head1 NAME
@@ -26,32 +28,46 @@ I<## To follow>
sub new {
my $class = shift();
- my($target, $zeerex) = @_;
+ my($irspy, $target, $zeerex) = @_;
if (!defined $zeerex) {
$zeerex = _empty_zeerex_record($target);
}
+ ### Parser should be in the IRSpy object
my $parser = new XML::LibXML();
- return bless {
+ my $this = bless {
+ irspy => $irspy,
target => $target,
+ parser => $parser,
zeerex => $parser->parse_string($zeerex)->documentElement(),
+ zoom_error => { TIMEOUT => 0 },
}, $class;
+
+ #Scalar::Util::weaken($this->{irspy});
+ #Scalar::Util::weaken($this->{parser});
+
+ return $this;
}
+sub zoom_error { return shift->{'zoom_error'} }
sub _empty_zeerex_record {
my($target) = @_;
- ### Doesn't recognise SRU/SRW URLs
- my($host, $port, $db) = ZOOM::IRSpy::_parse_target_string($target);
+ my($protocol, $host, $port, $db) =
+ ZOOM::IRSpy::_parse_target_string($target);
+ my $xprotocol = xml_encode($protocol);
+ my $xhost = xml_encode($host);
+ my $xport = xml_encode($port);
+ my $xdb = xml_encode($db);
return <<__EOT__;
-
- $host
- $port
- $db
+
+ $xhost
+ $xport
+ $xdb
__EOT__
@@ -62,23 +78,104 @@ sub append_entry {
my $this = shift();
my($xpath, $frag) = @_;
- print STDERR "this=$this, xpath='$xpath', frag='$frag'\n";
- my $root = $this->{zeerex}; # XML::LibXML::Element ISA XML::LibXML::Node
- print "Record='", $root->toString(), "'\n";
- my $xc = XML::LibXML::XPathContext->new($root);
+ #print STDERR "this=$this, xpath='$xpath', frag='$frag'\n";
+ my $xc = $this->xpath_context();
$xc->registerNs(zeerex => "http://explain.z3950.org/dtd/2.0/");
- $xc->registerNs(irspy => "http://indexdata.com/irspy/1.0");
+ $xc->registerNs(irspy => $ZOOM::IRSpy::Utils::IRSPY_NS);
my @nodes = $xc->findnodes($xpath);
if (@nodes == 0) {
- ZOOM::Log::log("irspy", "no matches for '$xpath': can't append");
+ # Make the node that we're inserting into, if possible. A
+ # fully general version would work its way through each
+ # component of the XPath, but for now we just treat it as a
+ # single chunk to go inside the top-level node.
+ $this->_half_decent_appendWellBalancedChunk($xc->getContextNode(),
+ "<$xpath>$xpath>");
+ @nodes = $xc->findnodes($xpath);
+ die("still no matches for '$xpath' after creating: can't append")
+ if @nodes == 0;
+ }
+
+ $this->{irspy}->log("warn",
+ scalar(@nodes), " matches for '$xpath': using first")
+ if @nodes > 1;
+
+ $this->_half_decent_appendWellBalancedChunk($nodes[0], $frag);
+}
+
+sub xpath_context {
+ my $this = shift();
+
+ return irspy_xpath_context($this->{zeerex});
+}
+
+sub store_result {
+ my ($this, $type, %info) = @_;
+ my $xml = "\n";
+
+ $this->append_entry('irspy:status', $xml);
+}
+
+
+# *sigh*
+#
+# _Clearly_ the right way to append a well-balanced chunk of XML to
+# a node's children is to call appendWellBalancedChunk() from the
+# XML::LibXML::Element class. However, this fails in the common case
+# where the ZeeRex record we're working with doesn't declare the
+# "irspy" namespace that the inserted fragments use.
+#
+# To my utter astonishment it seems that XML::LibXML (as of version
+# 1.58, 31st March 2004) doesn't provide ANY way to register a
+# namespace for parsing, which makes the parse_balanced_chunk()
+# function that appendWellBalancedChunk() uses effectively useless.
+# It _is_ possible to use setNamespace() on a node, to register a new
+# namespace mapping for that node -- but that only affects pre-parsed
+# trees, and is no use for parsing. Hence the following pair of lines
+# DOES NOT WORK:
+# $node->setNamespace($ZOOM::IRSpy::Utils::IRSPY_NS, "irspy", 0);
+# $node->appendWellBalancedChunk($frag);
+#
+# Instead I have to go the long way round, hence this method. I have
+# two candidate re-implementations, of which the former is marginally
+# less loathsome, but does require that the excess namespace
+# declarations be factored out later -- as least, if you want neat
+# output.
+#
+sub _half_decent_appendWellBalancedChunk {
+ my $this = shift();
+ my($node, $frag) = @_;
+
+ if (1) {
+ $frag =~ s,>, xmlns:irspy="$ZOOM::IRSpy::Utils::IRSPY_NS">,;
+ eval {
+ $node->appendWellBalancedChunk($frag);
+ }; if ($@) {
+ print STDERR "died while trying to appendWellBalancedChunk(), probably due to bad XML:\n$frag";
+ die $@;
+ }
return;
- } elsif (@nodes > 1) {
- ZOOM::Log::log("irspy", scalar(@nodes),
- " matches for '$xpath': using first");
}
- print STDERR "zeerex='$root'\n";
+ # Instead -- and to call this brain-damaged would be an insult
+ # to all those fine people out there with actual brain damage
+ # -- I have to "parse" the XML fragment myself and insert the
+ # resulting hand-build DOM tree. Someone shoot me now.
+ my($open, $content, $close) = $frag =~ /^<(.*?)>(.*)<\/(.*?)>$/;
+ die "can't 'parse' XML fragment '$frag'"
+ if !defined $open;
+ my($tag, $attrs) = $open =~ /(.*?)\s(.*)/;
+ $tag = $open if !defined $tag;
+ die "mismatched XML start/end <$open>...<$close>"
+ if $close ne $tag;
+ print STDERR "tag='$tag', attrs=[$attrs], content='$content'\n";
+ die "## no code yet to make DOM node";
}