Emit bad XML fragment if appendWellBalancedChunk() fails.

[irspy-moved-to-github.git] / lib / ZOOM / IRSpy / Record.pm

diff --git a/lib/ZOOM/IRSpy/Record.pm b/lib/ZOOM/IRSpy/Record.pm
index 2f67dc1..7444d70 100644 (file)
--- a/lib/ZOOM/IRSpy/Record.pm
+++ b/lib/ZOOM/IRSpy/Record.pm
@@ -1,6 +1,8 @@
-# $Id: Record.pm,v 1.8 2006-07-24 16:44:00 mike Exp $
+# $Id: Record.pm,v 1.28 2007-12-12 08:49:58 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;
@@ -8,7 +10,7 @@ use warnings;
 
 use XML::LibXML;
 use XML::LibXML::XPathContext;
-
+use ZOOM::IRSpy::Utils qw(xml_encode isodate irspy_xpath_context);
 
 =head1 NAME
 
@@ -26,14 +28,16 @@ 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 {
+       irspy => $irspy,
        target => $target,
        parser => $parser,
        zeerex => $parser->parse_string($zeerex)->documentElement(),
@@ -44,15 +48,19 @@ sub new {
 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__;
 <explain xmlns="http://explain.z3950.org/dtd/2.0/">
- <serverInfo protocol="Z39.50" version="1995">
-  <host>$host</host>
-  <port>$port</port>
-  <database>$db</database>
+ <serverInfo protocol="$xprotocol">
+  <host>$xhost</host>
+  <port>$xport</port>
+  <database>$xdb</database>
  </serverInfo>
 </explain>
 __EOT__
@@ -63,11 +71,10 @@ 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
-    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) {
@@ -75,21 +82,37 @@ sub append_entry {
        # 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.
-       $this->_half_decent_appendWellBalancedChunk($root,
+       $this->_half_decent_appendWellBalancedChunk($xc->getContextNode(),
                                                    "<$xpath></$xpath>");
        @nodes = $xc->findnodes($xpath);
        die("still no matches for '$xpath' after creating: can't append")
            if @nodes == 0;
     }
 
-    ZOOM::Log::log("irspy",
-                  scalar(@nodes), " matches for '$xpath': using first")
+    $this->{irspy}->log("warn",
+                       scalar(@nodes), " matches for '$xpath': using first")
        if @nodes > 1;
 
-    my $node = $nodes[0];
-    # $node ISA XML::LibXML::ElementXML::LibXML::Element
-    $this->_half_decent_appendWellBalancedChunk($node, $frag);
-    #print STDERR "POST: zeerex='$root' = \n", $root->toString(), "\n";
+    $this->_half_decent_appendWellBalancedChunk($nodes[0], $frag);
+}
+
+sub xpath_context {
+    my $this = shift();
+
+    return irspy_xpath_context($this->{zeerex});
+}
+
+sub store_result {
+    my ($this, $type, %info) = @_;
+    my $xml = "<irspy:$type";
+
+    foreach my $key (keys %info) {
+        $xml .= " $key=\"" . xml_encode($info{$key}) . "\"";
+    }
+
+    $xml .= ">" . isodate(time()) . "</irspy:$type>\n";
+
+    $this->append_entry('irspy:status', $xml);
 }
 
 
@@ -109,7 +132,7 @@ sub append_entry {
 # 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("http://indexdata.com/irspy/1.0", "irspy", 0);
+#      $node->setNamespace($ZOOM::IRSpy::Utils::IRSPY_NS, "irspy", 0);
 #      $node->appendWellBalancedChunk($frag);
 #
 # Instead I have to go the long way round, hence this method.  I have
@@ -123,8 +146,13 @@ sub _half_decent_appendWellBalancedChunk {
     my($node, $frag) = @_;
 
     if (1) {
-       $frag =~ s,>, xmlns:irspy="http://indexdata.com/irspy/1.0">,;
-       $node->appendWellBalancedChunk($frag);
+       $frag =~ s,>, xmlns:irspy="$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;
     }
 
@@ -140,7 +168,7 @@ sub _half_decent_appendWellBalancedChunk {
     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";
+    die "## no code yet to make DOM node";
 }