X-Git-Url: http://git.indexdata.com/?p=irspy-moved-to-github.git;a=blobdiff_plain;f=lib%2FZOOM%2FIRSpy%2FUtils.pm;h=2dbbe131accc966ede26ee8c141fd0b96f8bec9e;hp=5db0e331c7e4dbad71fde3c70649f83b6084e908;hb=68379b31e731a6cdd942e99bb339bafe1a4c8dc1;hpb=5cc7c7423877cc60242402f6a33553d5447d8457

diff --git a/lib/ZOOM/IRSpy/Utils.pm b/lib/ZOOM/IRSpy/Utils.pm
index 5db0e33..2dbbe13 100644
--- a/lib/ZOOM/IRSpy/Utils.pm
+++ b/lib/ZOOM/IRSpy/Utils.pm
@@ -1,4 +1,4 @@
-# $Id: Utils.pm,v 1.26 2007-03-19 18:51:03 mike Exp $
+# $Id: Utils.pm,v 1.33 2007-06-27 10:44:57 mike Exp $
 
 package ZOOM::IRSpy::Utils;
 
@@ -7,22 +7,41 @@ use strict;
 use warnings;
 
 use Exporter 'import';
-our @EXPORT_OK = qw(isodate
+our @EXPORT_OK = qw(utf8param
+		    isodate
 		    xml_encode 
 		    cql_quote
 		    cql_target
 		    irspy_xpath_context
+		    irspy_make_identifier
+		    irspy_record2identifier
+		    irspy_identifier2target
 		    modify_xml_document
 		    bib1_access_point
 		    render_record);
 
 use XML::LibXML;
 use XML::LibXML::XPathContext;
+use Encode;
+use Encode qw(is_utf8);
+
 
 our $IRSPY_NS = 'http://indexdata.com/irspy/1.0';
 
 
 # Utility functions follow, exported for use of web UI
+sub utf8param {
+    my($r, $key, $value) = @_;
+    die "utf8param() called with value '$value'" if defined $value;
+
+    my $raw = $r->param($key);
+    return undef if !defined $raw;
+    my $cooked = decode_utf8($raw);
+    warn "converted '$raw' to '", $cooked, "'\n" if $cooked ne $raw;
+    return $cooked;
+}
+
+
 sub isodate {
     my($time) = @_;
 
@@ -67,7 +86,7 @@ sub xml_encode {
 sub cql_quote {
     my($term) = @_;
 
-    $term =~ s/([""\\])/\\$1/g;
+    $term =~ s/([""\\*?])/\\$1/g;
     $term = qq["$term"] if $term =~ /[\s""\/]/;
     return $term;
 }
@@ -76,10 +95,16 @@ sub cql_quote {
 # Makes a CQL query that finds a specified target.  Arguments may be
 # either an ID alone, or a (host, port, db) triple.
 sub cql_target {
-    my($host, $port, $db) = @_;
+    my($protocol, $host, $port, $db) = @_;
 
-    $host .= ":$port/$db" if defined $port;
-    return "rec.id=" . cql_quote($host);
+    my $id;
+    if (defined $host) {
+	$id = irspy_make_identifier($protocol, $host, $port, $db);
+    } else {
+	$id = $protocol;
+    }
+
+    return "rec.id=" . cql_quote($id);
 }
 
 
@@ -127,6 +152,73 @@ sub irspy_xpath_context {
 }
 
 
+# Construct an opaque identifier from its components.  Although it's
+# trivial, this is needed in so many places that it really needs to be
+# factored out.
+#
+# This is the converse of _parse_target_string() in IRSpy.pm, which
+# should be renamed and moved into this package.
+#
+sub irspy_make_identifier {
+    my($protocol, $host, $port, $dbname) = @_;
+
+    die "irspy_make_identifier(" . join(", ", map { "'$_'" } @_).
+	"): wrong number of arguments" if @_ != 4;
+
+    die "irspy_make_identifier(): protocol undefined" if !defined $protocol;
+    die "irspy_make_identifier(): host undefined" if !defined $host;
+    die "irspy_make_identifier(): port undefined" if !defined $port;
+    die "irspy_make_identifier(): dbname undefined" if !defined $dbname;
+
+    return "$protocol:$host:$port/$dbname";
+}
+
+
+# Returns the opaque identifier of an IRSpy record based on the
+# XPathContext'ed DOM object, as returned by irspy_xpath_context().
+# This is doing the same thing as irspy_make_identifier() but from a
+# record rather than a set of parameters.
+#
+sub irspy_record2identifier {
+    my($xc) = @_;
+
+    ### Must be kept the same as is used in ../../../zebra/*.xsl
+    return $xc->find("concat(e:serverInfo/\@protocol, ':',
+			     e:serverInfo/e:host, ':',
+			     e:serverInfo/e:port, '/',
+			     e:serverInfo/e:database)");
+}
+
+
+# Transforms an IRSpy opqaue identifier, as returned from
+# irspy_make_identifier() or irspy_record2identifier(), into a YAZ
+# target-string suitable for feeding to ZOOM.  Before we introduced
+# the protocol element at the start of the identifier string, this was
+# a null transform; now we have to be a bit cleverer.
+#
+sub irspy_identifier2target {
+    my $res = _irspy_identifier2target(@_);
+    #carp "converted ID '@_' to target '$res'";
+    return $res;
+}
+
+sub _irspy_identifier2target {
+    my($id) = @_;
+
+    my($protocol, $target) = ($id =~ /(.*?):(.*)/);
+    if (uc($protocol) eq "Z39.50") {
+	return "tcp:$target";
+    } elsif (uc($protocol) eq "SRU") {
+	return "sru=get,http:$target";
+    } elsif (uc($protocol) eq "SRW") {
+	return "sru=srw,http:$target";
+    }
+
+    warn "unrecognised protocol '$protocol' in ID $id";
+    return $target;
+}
+
+
 sub modify_xml_document {
     my($xc, $fieldsByKey, $data) = @_;
 
@@ -156,21 +248,23 @@ sub modify_xml_document {
 		# we'll check whether the element is already
 		# canonical, to determine whether our change is a
 		# no-op.
-		my $old = "???";
+		my $old = "";
 		my @children = $node->childNodes();
 		if (@children == 1) {
 		    my $child = $node->firstChild();
 		    if (ref $child && ref $child eq "XML::LibXML::Text") {
 			$old = $child->getData();
-			next if $value eq $old;
+			print STDERR "child='$child', old=", _renderchars($old), "\n"
+			    if $key eq "title";
 		    }
 		}
+		next if $value eq $old;
 
 		$node->removeChildNodes();
 		my $child = new XML::LibXML::Text($value);
 		$node->appendChild($child);
 		push @changes, $ref;
-		#print "Elem $key: '$old' -> '$value' ($xpath)<br/>\n";
+		print STDERR "Elem $key ($xpath): ", _renderchars($old), " -> '", _renderchars($value), "\n";
 	    } else {
 		warn "unexpected node type $node";
 	    }
@@ -188,6 +282,13 @@ sub modify_xml_document {
 }
 
 
+sub _renderchars {
+    my($text) = @_;
+
+    return "'" . $text . "'", " (", join(" ", map {ord($_)} split //, $text), "), is_utf8=" , is_utf8($text);
+}
+
+
 sub dom_add_node {
     my($xc, $ppath, $selector, $value, @addAfter) = @_;