Do not filter out undefined entity errors for standalone XML docs (#7007).

2 files changed, 28 insertions, 8 deletions

diff --git a/htdocs/whatsnew.html b/htdocs/whatsnew.html
index c36f4ef..92ee09b 100755
--- a/htdocs/whatsnew.html
+++ b/htdocs/whatsnew.html
@@ -1,5 +1,5 @@
-<!--#set var="revision" value="\$Id: whatsnew.html,v 1.92 2009-12-12 11:15:12 ville Exp $"
---><!--#set var="date" value="\$Date: 2009-12-12 11:15:12 $"
+<!--#set var="revision" value="\$Id: whatsnew.html,v 1.93 2009-12-14 21:18:33 ville Exp $"
+--><!--#set var="date" value="\$Date: 2009-12-14 21:18:33 $"
 --><!--#set var="title" value="What's New at The W3C Markup Validation Service"
 --><!--#set var="relroot" value="./"
 --><!--#set var="feeds" value="1"
@@ -158,6 +158,11 @@
               lines longer than 4k bytes</a> could not be processed.
           </li>
           <li>
+            Bug fix: an error was not issued for
+            <a href="http://www.w3.org/Bugs/Public/show_bug.cgi?id=7007">standalone
+              XML documents containing external entities</a>.
+          </li>
+          <li>
             Dependency changes: new dependency: JSON &gt;= 2.00; new minimum
             required versions: libwww-perl 5.802, Config-General 2.32,
             SGML-Parser-OpenSP 0.991; String-Approx is no longer required.
diff --git a/httpd/cgi-bin/check b/httpd/cgi-bin/check
index 91adf57..a332c29 100755
--- a/httpd/cgi-bin/check
+++ b/httpd/cgi-bin/check
@@ -14,7 +14,7 @@
 # This source code is available under the license at:
 #     http://www.w3.org/Consortium/Legal/copyright-software
 #
-# $Id: check,v 1.752 2009-12-14 20:44:49 ville Exp $
+# $Id: check,v 1.753 2009-12-14 21:18:33 ville Exp $
 
 #
 # We need Perl 5.8.0+.
@@ -197,7 +197,7 @@ EOF
 
     #
     # Strings
-    $VERSION = q$Revision: 1.752 $;
+    $VERSION = q$Revision: 1.753 $;
     $VERSION =~ s/Revision: ([\d\.]+) /$1/;
 
     # Read friendly error message file
@@ -627,6 +627,15 @@ if (&is_xml($File)) {
                ([^>].*\?>)
            /$1encoding="UTF-8"$4/sx;
 
+        # Is the document standalone?  Need to check with a regex because
+        # the parser may fail to return a document we could use for this.
+        my $standalone = (
+            $xml_string =~ /^<\?xml\b[^>]*[\x20\x09\x0D\x0A]
+                           standalone[\x20\x09\x0D\x0A]*=[\x20\x09\x0D\x0A]*
+                           (["'])yes\1
+                          /sx
+        );
+
         eval { $xmlparser->parse_string($xml_string); };
         $xml_string = undef;
         my $xml_parse_errors_line = undef;
@@ -656,8 +665,11 @@ if (&is_xml($File)) {
                 $err->{msg}  = $err_obj->message();
 
                 # The validator will sometimes fail to dereference entities
-                # files; we're filtering the resulting bogus error
-                if ($err->{msg} =~ /Entity '\w+' not defined/) {
+                # files; we're filtering the resulting bogus error for
+                # non-standalone documents. @@@TODO: is this still needed?
+                if (!$standalone &&
+                    $err->{msg} =~ /Entity '\w+' not defined/)
+                {
                     $err = undef;
                     next;
                 }
@@ -752,8 +764,11 @@ if (&is_xml($File)) {
                     $err->{msg}  = $xmlwf_error_msg;
 
                     # The validator will sometimes fail to dereference entities
-                    # files; we're filtering the resulting bogus error
-                    if ($err->{msg} =~ /Entity '\w+' not defined/) {
+                    # files; we're filtering the resulting bogus error for
+                    # non-standalone documents. @@@TODO: is this still needed?
+                    if (!$standalone &&
+                        $err->{msg} =~ /Entity '\w+' not defined/)
+                    {
                         $xmlwf_error_line = undef;
                         $xmlwf_error_col  = undef;
                         $xmlwf_error_msg  = undef;