X-Git-Url: http://git.indexdata.com/?p=irspy-moved-to-github.git;a=blobdiff_plain;f=lib%2FZOOM%2FIRSpy%2FUtils.pm;h=7853d20816115e3201b01e0ad747f6c3968d99a8;hp=e24fd6ca569c6478f602f7c74aa98bb87e4c98f8;hb=5be3dea6bacf251661cfa9b3317b3b62e01da295;hpb=7e30dd5f86e43b3e2f19b33d713d413168c7fd6b

diff --git a/lib/ZOOM/IRSpy/Utils.pm b/lib/ZOOM/IRSpy/Utils.pm
index e24fd6c..7853d20 100644
--- a/lib/ZOOM/IRSpy/Utils.pm
+++ b/lib/ZOOM/IRSpy/Utils.pm
@@ -1,4 +1,4 @@
-# $Id: Utils.pm,v 1.27 2007-04-27 14:04:40 mike Exp $
+# $Id: Utils.pm,v 1.38 2009-04-15 18:16:45 wosch Exp $
 
 package ZOOM::IRSpy::Utils;
 
@@ -7,7 +7,8 @@ use strict;
 use warnings;
 
 use Exporter 'import';
-our @EXPORT_OK = qw(isodate
+our @EXPORT_OK = qw(utf8param
+		    isodate
 		    xml_encode 
 		    cql_quote
 		    cql_target
@@ -21,11 +22,77 @@ our @EXPORT_OK = qw(isodate
 
 use XML::LibXML;
 use XML::LibXML::XPathContext;
+use Encode;
+use Encode qw(is_utf8);
+
 
 our $IRSPY_NS = 'http://indexdata.com/irspy/1.0';
 
+# Under Apache 2/mod_perl 2, the ubiquitous $r is no longer and
+# Apache::Request object, nor even an Apache2::Request, but an
+# Apache2::RequestReq ... which, astonishingly, doesn't have the
+# param() method.  So if we're given one of these things, we need to
+# make an Apache::Request out of, which at least isn't too hard.
+# However *sigh* this may not be a cheap operation, so we keep a cache
+# of already-made Request objects.
+#
+my %_apache2request;
+my %_paramsbyrequest;           # Used for Apache2 only
+sub utf8param {
+    my($r, $key, $value) = @_;
+
+    if ($r->isa('Apache2::RequestRec')) {
+        # Running under Apache2
+        if (defined $_apache2request{$r}) {
+            #warn "using existing Apache2::RequestReq for '$r'";
+            $r = $_apache2request{$r};
+        } else {
+            require Apache2::Request;
+            #warn "making new Apache2::RequestReq for '$r'";
+            $r = $_apache2request{$r} = new Apache2::Request($r);
+        }
+    }
+
+    if (!defined $key) {
+        return map { decode_utf8($_) } $r->param();
+    }
+
+    my $raw = undef;
+    $raw = $_paramsbyrequest{$r}->{$key} if $r->isa('Apache2::Request');
+    $raw = $r->param($key) if !defined $raw;
+
+    if (defined $value) {
+        # Argh!  Simply writing through to the underlying method
+        # param() won't work in Apache2, where param() is readonly.
+        # So we have to keep a hash of additional values, which we
+        # consult (above) before the actual parameters.  Ouch ouch.
+        if ($r->isa('Apache2::Request')) {
+            $_paramsbyrequest{$r}->{$key} = encode_utf8($value);
+        } else {
+            $r->param($key, encode_utf8($value));
+        }
+    }
+
+    return undef if !defined $raw;
+    my $cooked = decode_utf8($raw);
+    warn "converted '$raw' to '", $cooked, "'\n" if $cooked ne $raw;
+    return $cooked;
+}
+
 
 # Utility functions follow, exported for use of web UI
+sub utf8param_apache1 {
+    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) = @_;
 
@@ -70,7 +137,7 @@ sub xml_encode {
 sub cql_quote {
     my($term) = @_;
 
-    $term =~ s/([""\\])/\\$1/g;
+    $term =~ s/([""\\*?])/\\$1/g;
     $term = qq["$term"] if $term =~ /[\s""\/]/;
     return $term;
 }
@@ -181,11 +248,27 @@ sub irspy_record2identifier {
 # 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) = @_;
 
+    confess "_irspy_identifier2target(): id is undefined"
+	if !defined $id;
+
     my($protocol, $target) = ($id =~ /(.*?):(.*)/);
-    print STDERR "protocol='$protocol', target='$target'\n";
-    ### This assumes everything is Z39.50
+    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 "_irspy_identifier2target($id): unrecognised protocol '$protocol'";
     return $target;
 }
 
@@ -219,21 +302,22 @@ 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";
 	    }
@@ -251,6 +335,13 @@ 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) = @_;
 
@@ -362,7 +453,7 @@ sub inheritance_tree {
 # This function is made available in xslt using the register_function call
 sub xslt_strcmp {
     my ($arg1, $arg2) = @_;
-    return ($arg1->to_literal()) cmp ($arg2->to_literal());
+    return "$arg1" cmp "$arg2";
 }