fix overrun error reading invalid xhtml + fix loading problem on validation test

org.hl7.fhir.utilities/src/main/java/org/hl7/fhir/utilities/xhtml/XhtmlParser.java

@@ -871,10 +871,20 @@ private boolean elementIsOk(String name) throws FHIRFormatError  {
     return s.toString();
   }
   
+
+  private String readUntil(String sc) throws IOException
+  {
+    StringBuilder s = new StringBuilder();
+    while (peekChar() != 0 && sc.indexOf(peekChar()) == -1)
+      s.append(readChar());
+    readChar();
+    return s.toString();
+  }
+  
   private void parseLiteral(StringBuilder s) throws IOException, FHIRFormatError {
     // UInt16 w;
     readChar();
-    String c = readUntil(';');
+    String c = readUntil(";&'\"><");
     if (c.equals("apos"))
       s.append('\'');
     else if (c.equals("quot"))
@@ -1162,9 +1172,13 @@ private boolean elementIsOk(String name) throws FHIRFormatError  {
       s.append((char) 201D); 
     else if (entities.containsKey(c))
       s.append(entities.get(c));
-    else
+    else if (!mustBeWellFormed) {
+      // we guess that this is an accidentally unescaped &
+      s.append("&"+c);
+    } else {
       throw new FHIRFormatError("unable to parse character reference '" + c + "'' (last text = '" + lastText + "'" + descLoc());
     }
+  }
   
   private boolean isInteger(String s, int base) {
     try {

org.hl7.fhir.utilities/src/test/java/org/hl7/fhir/utilities/tests/XhtmlNodeTest.java

@@ -1,5 +1,6 @@
 package org.hl7.fhir.utilities.tests;
 
+import org.hl7.fhir.exceptions.FHIRException;
 import org.hl7.fhir.exceptions.FHIRFormatError;
 import org.hl7.fhir.utilities.xhtml.XhtmlNode;
 import org.hl7.fhir.utilities.xhtml.XhtmlParser;
@@ -95,4 +96,17 @@ public class XhtmlNodeTest {
   public void testParseBadChars() throws FHIRFormatError, IOException {
     XhtmlNode x = new XhtmlParser().parse(BaseTestingUtilities.loadTestResource("xhtml", "bad-chars.html"), "div");
   }  
+  
+  @Test
+  public void testParseBadLink1() throws FHIRFormatError, IOException {
+    XhtmlNode x = new XhtmlParser().setMustBeWellFormed(false).parse(BaseTestingUtilities.loadTestResource("xhtml", "bad-link.html"), "div");
+  }
+    
+  @Test
+  public void testParseBadLink2() throws FHIRFormatError, IOException {
+    Assertions.assertThrows(FHIRException.class, () -> new XhtmlParser().parse(BaseTestingUtilities.loadTestResource("xhtml", "bad-link.html"), "div"));
+  }
+
+
+
 }

org.hl7.fhir.validation/src/test/java/org/hl7/fhir/comparison/tests/ComparisonTests.java

@@ -33,6 +33,7 @@ import org.hl7.fhir.r5.comparison.ProfileComparer.ProfileComparison;
 import org.hl7.fhir.r5.comparison.ValueSetComparer;
 import org.hl7.fhir.r5.comparison.ValueSetComparer.ValueSetComparison;
 import org.hl7.fhir.r5.conformance.ProfileUtilities;
+import org.hl7.fhir.r5.context.BaseWorkerContext;
 import org.hl7.fhir.r5.context.IWorkerContext;
 import org.hl7.fhir.r5.formats.IParser.OutputStyle;
 import org.hl7.fhir.r5.formats.JsonParser;
@@ -109,7 +110,11 @@ public class ComparisonTests {
       context = TestingUtilities.context();
       FilesystemPackageCacheManager pcm = new FilesystemPackageCacheManager(true, ToolsVersion.TOOLS_VERSION);
       NpmPackage npm = pcm.loadPackage("hl7.fhir.us.core#3.1.0");
+      BaseWorkerContext bc = (BaseWorkerContext) context;
+      boolean dupl = bc.isAllowLoadingDuplicates();
+      bc.setAllowLoadingDuplicates(true);
       context.loadFromPackage(npm, new R4ToR5Loader(new String[] { "CapabilityStatement", "StructureDefinition", "ValueSet", "CodeSystem", "SearchParameter", "OperationDefinition", "Questionnaire","ConceptMap","StructureMap", "NamingSystem"}, new NullLoaderKnowledgeProvider()));
+      bc.setAllowLoadingDuplicates(dupl);
     }
 
     if (!new File(Utilities.path("[tmp]", "comparison")).exists()) {