X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=lib%2FZOOM%2FIRSpy%2FUtils.pm;h=856e60250f41ece095a591499d892a4170763f6e;hb=0cf41562dd3be395b42934a4a1a99404ce2a4e68;hp=c31d188ef6f757ad7f8e88501cdb0a0eacd21e6b;hpb=b2d2832eaa18ae8cfef70651c05bdc57069e4dde;p=irspy-moved-to-github.git
diff --git a/lib/ZOOM/IRSpy/Utils.pm b/lib/ZOOM/IRSpy/Utils.pm
index c31d188..856e602 100644
--- a/lib/ZOOM/IRSpy/Utils.pm
+++ b/lib/ZOOM/IRSpy/Utils.pm
@@ -1,4 +1,3 @@
-# $Id: Utils.pm,v 1.22 2007-03-01 13:51:18 mike Exp $
package ZOOM::IRSpy::Utils;
@@ -6,21 +5,99 @@ use 5.008;
use strict;
use warnings;
+use Scalar::Util;
+
use Exporter 'import';
-our @EXPORT_OK = qw(isodate
+our @EXPORT_OK = qw(utf8param
+ trimField
+ utf8paramTrim
+ isodate
xml_encode
cql_quote
+ cql_target
irspy_xpath_context
+ irspy_make_identifier
+ irspy_record2identifier
+ irspy_identifier2target
modify_xml_document
- bib1_access_point);
+ bib1_access_point
+ render_record
+ validate_record
+ calc_reliability_string
+ calc_reliability_stats);
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) = @_;
@@ -29,6 +106,26 @@ sub isodate {
$year+1900, $mon+1, $mday, $hour, $min, $sec);
}
+# strips whitespaces at start and ends of a field
+sub trimField {
+ my $field = shift;
+
+ $field =~ s/^\s+//;
+ $field =~ s/\s+$//;
+
+ return $field;
+}
+
+# utf8param() with trim
+sub utf8paramTrim {
+ my $result = utf8param(@_);
+
+ if (defined $result) {
+ $result = trimField($result);
+ }
+
+ return $result;
+}
# I can't -- just can't, can't, can't -- believe that this function
# isn't provided by one of the core XML modules. But the evidence all
@@ -65,12 +162,28 @@ sub xml_encode {
sub cql_quote {
my($term) = @_;
- $term =~ s/([""\\])/\\$1/g;
- $term = qq["$term"] if $term =~ /\s/;
+ $term =~ s/([""\\*?])/\\$1/g;
+ $term = qq["$term"] if $term =~ /[\s""\/]/;
return $term;
}
+# Makes a CQL query that finds a specified target. Arguments may be
+# either an ID alone, or a (host, port, db) triple.
+sub cql_target {
+ my($protocol, $host, $port, $db) = @_;
+
+ my $id;
+ if (defined $host) {
+ $id = irspy_make_identifier($protocol, $host, $port, $db);
+ } else {
+ $id = $protocol;
+ }
+
+ return "rec.id=" . cql_quote($id);
+}
+
+
# PRIVATE to irspy_namespace() and irspy_xpath_context()
my %_namespaces = (
e => 'http://explain.z3950.org/dtd/2.0/',
@@ -94,10 +207,19 @@ sub irspy_namespace {
sub irspy_xpath_context {
my($record) = @_;
- my $xml = ref $record ? $record->render() : $record;
- my $parser = new XML::LibXML();
- my $doc = $parser->parse_string($xml);
- my $root = $doc->getDocumentElement();
+ if (ref $record && $record->isa("ZOOM::Record")) {
+ $record = $record->render();
+ }
+
+ my $root;
+ if (ref $record) {
+ $root = $record;
+ } else {
+ my $parser = new XML::LibXML();
+ my $doc = $parser->parse_string($record);
+ $root = $doc->getDocumentElement();
+ }
+
my $xc = XML::LibXML::XPathContext->new($root);
foreach my $prefix (keys %_namespaces) {
$xc->registerNs($prefix, $_namespaces{$prefix});
@@ -106,6 +228,81 @@ 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) = @_;
+
+ confess "_irspy_identifier2target(): id is undefined"
+ if !defined $id;
+
+ my($protocol, $target) = ($id =~ /(.*?):(.*)/);
+ if (uc($protocol) eq "Z39.50" || uc($protocol) eq "TCP") {
+ 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;
+}
+
+
+# Modifies the XML document for which $xc is an XPath context by
+# inserting or replacing the values specified in the hash %$data. The
+# keys are fieldnames, which are looked up in the register
+# $fieldsByKey to determine, among other things, what their XPath is.
+
sub modify_xml_document {
my($xc, $fieldsByKey, $data) = @_;
@@ -135,27 +332,28 @@ 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)
\n";
+ #print STDERR "Elem $key ($xpath): ", _renderchars($old), " -> '", _renderchars($value), "\n";
} else {
warn "unexpected node type $node";
}
} else {
- next if !$value; # No need to create a new empty node
+ next if !defined $value; # No need to create a new empty node
my($ppath, $selector) = $xpath =~ /(.*)\/(.*)/;
dom_add_node($xc, $ppath, $selector, $value, @addAfter);
#print "New $key ($xpath) = '$value'
\n";
@@ -167,6 +365,13 @@ sub modify_xml_document {
}
+sub _renderchars {
+ my($text) = @_;
+
+ return "'" . $text . "'", " (", join(" ", map {ord($_)} split //, $text), "), is_utf8=" , is_utf8($text);
+}
+
+
sub dom_add_node {
my($xc, $ppath, $selector, $value, @addAfter) = @_;
@@ -278,7 +483,7 @@ sub inheritance_tree {
# This function is made available in xslt using the register_function call
sub xslt_strcmp {
my ($arg1, $arg2) = @_;
- return ($arg1->to_literal()) cmp ($arg2->to_literal());
+ return "$arg1" cmp "$arg2";
}
@@ -581,4 +786,90 @@ sub bib1_access_point {
}
+sub render_record {
+ my($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 $rec->render();
+}
+
+
+sub calc_reliability_string {
+ my($xc) = @_;
+
+ my($nok, $nall, $percent) = calc_reliability_stats($xc);
+ return "[untested]" if $nall == 0;
+ return "$nok/$nall = " . $percent . "%";
+}
+
+
+sub calc_reliability_stats {
+ my($xc) = @_;
+
+ my @allpings = $xc->findnodes("i:status/i:probe");
+ my $nall = @allpings;
+ return (0, 0, 0) if $nall == 0;
+ my @okpings = $xc->findnodes('i:status/i:probe[@ok = "1"]');
+ my $nok = @okpings;
+ my $percent = int(100*$nok/$nall + 0.5);
+ return ($nok, $nall, $percent);
+}
+
+#
+# validate_record( record, ( "port" => 1, "database" => 1, "country" => 0, ... ))
+#
+sub validate_record {
+ my $rec = shift;
+ my %args = @_;
+
+ my %required = map { $_ => 1 } qw/port host database protocol/;
+ my %optional = map { $_ => 1 } qw/country type hosturl contact language/;
+ my %tests = ( %required, %args );
+
+ my $xc = irspy_xpath_context($rec);
+
+ my $protocol = $xc->findnodes("e:serverInfo/\@protocol") || "";
+ my $port = $xc->findnodes("e:serverInfo/e:port") || "";
+ my $host = $xc->findnodes("e:serverInfo/e:host") || "";
+ my $dbname = $xc->findnodes("e:serverInfo/e:database") || "";
+
+ my $id = irspy_make_identifier($protocol, $host, $port, $dbname);
+
+ if ($protocol =~ /\s+$/ || $dbname =~ /\s+$/) {
+ warn "xxx: $protocol:$host:$port:$dbname: whitespaces\n";
+ }
+
+ my @errors = $id;
+
+ if ($tests{'protocol'}) {
+ push(@errors, 'protocol number is not valid') if $protocol !~ /^(z39\.50|sru|srw|tcp)$/;
+ }
+
+ if ($tests{'port'}) {
+ push(@errors, 'port number is not valid') if $port !~ /^\d+$/;
+ }
+
+ if ($tests{'host'}) {
+ push(@errors, 'host name is not valid') if $host !~ /^[0-9a-z]+[0-9a-z\.\-]*\.[0-9a-z]+$/i;
+ }
+
+ if ($tests{'database'}) {
+ push(@errors, 'database name is not valid') if $dbname =~ m,/,i;
+ push(@errors, 'database has trailing spaces') if $dbname ne trimField($dbname);
+ }
+
+ if ($tests{'hosturl'}) {
+ my $hosturl = $xc->findnodes("i:status/i:hostURL") || "";
+ push(@errors, 'This hosturl name is not valid') if $hosturl !~ /^\w+$/i;
+ }
+
+ return ( !$#errors, \@errors );
+}
+
1;