Replacing invalid xml character by unicode escape character (microsof…

…t#437) * Replace invalid xml character by unicode escape character
navin22 · Feb 7, 2017 · 4f19865 · 4f19865
1 parent 2f16af1
commit 4f19865
Show file tree

Hide file tree

Showing 17 changed files with 50 additions and 88 deletions.
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/TrxResource.Designer.cs b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/TrxResource.Designer.cs
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/TrxResource.resx b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/TrxResource.resx
@@ -211,8 +211,4 @@ Error Details: {1}:{2}</value>
   <data name="TrxLoggerResultsFileOverwriteWarning" xml:space="preserve">
     <value>WARNING: Overwriting results file: {0}</value>
   </data>
-  <data name="PlaceHolderForInvalidChar" xml:space="preserve">
-    <value>?</value>
-    <comment>Place holder for invalid xml character in trx file</comment>
-  </data>
 </root>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.cs.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.cs.xlf
@@ -298,11 +298,6 @@ Podrobnosti o chybě: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.de.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.de.xlf
@@ -298,11 +298,6 @@ Fehlerdetails: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.es.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.es.xlf
@@ -298,11 +298,6 @@ Detalles del error: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.fr.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.fr.xlf
@@ -298,11 +298,6 @@ Détails de l’erreur : {1} : {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.it.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.it.xlf
@@ -298,11 +298,6 @@ Dettagli errore: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ja.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ja.xlf
@@ -298,11 +298,6 @@ Error Details: {1}:{2}</source>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ko.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ko.xlf
@@ -298,11 +298,6 @@ Error Details: {1}:{2}</source>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.pl.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.pl.xlf
@@ -298,11 +298,6 @@ Szczegóły błędu: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.pt-BR.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.pt-BR.xlf
@@ -298,11 +298,6 @@ Detalhes do erro: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ru.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.ru.xlf
@@ -298,11 +298,6 @@ Error Details: {1}:{2}</source>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.tr.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.tr.xlf
@@ -298,11 +298,6 @@ Hata ayrıntıları: {1}: {2}</target>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.zh-Hans.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.zh-Hans.xlf
@@ -298,11 +298,6 @@ Error Details: {1}:{2}</source>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.zh-Hant.xlf b/src/Microsoft.TestPlatform.Extensions.TrxLogger/Resources/xlf/TrxResource.zh-Hant.xlf
@@ -298,11 +298,6 @@ Error Details: {1}:{2}</source>
         <target state="new">WARNING: Overwriting results file: {0}</target>
         <note></note>
       </trans-unit>
-      <trans-unit id="PlaceHolderForInvalidChar">
-        <source>?</source>
-        <target state="new">?</target>
-        <note>Place holder for invalid xml character in trx file</note>
-      </trans-unit>
     </body>
   </file>
 </xliff>
diff --git a/src/Microsoft.TestPlatform.Extensions.TrxLogger/XML/XmlPersistence.cs b/src/Microsoft.TestPlatform.Extensions.TrxLogger/XML/XmlPersistence.cs
@@ -700,13 +700,23 @@ private static string RemoveInvalidXmlChar(string str)
                 // From xml spec (http://www.w3.org/TR/xml/#charsets) valid chars: 
                 // #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD] | [#x10000-#x10FFFF]  
 
-                string invalidChar = @"[^\x09\x0A\x0D\x20-\uD7FF\uE000-\uFFFD\u10000-\u10FFFF]";
-                return Regex.Replace(str, invalidChar, TrxLoggerResources.PlaceHolderForInvalidChar);
+                // we are handling only #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
+                // because C# support unicode character in range \u0000 to \uFFFF
+                MatchEvaluator evaluator = new MatchEvaluator(ReplaceInvalidCharacterWithUniCodeEscapeSequence);
+                string invalidChar = @"[^\x09\x0A\x0D\x20-\uD7FF\uE000-\uFFFD]";
+                return Regex.Replace(str, invalidChar, evaluator);
             }
 
             return str;
         }
 
+        private static string ReplaceInvalidCharacterWithUniCodeEscapeSequence(Match match)
+        {
+            char x = match.Value[0];
+            return String.Format(@"\u{0:x4}", (ushort)x);
+
+        }
+
         private XmlNode EnsureLocationExists(XmlElement xml, string location, string nameSpaceUri)
         {
             XmlNode node = this.LocationToXmlNode(xml, location);

diff --git a/test/Microsoft.TestPlatform.Extensions.TrxLogger.UnitTests/XmlPersistenceTests.cs b/test/Microsoft.TestPlatform.Extensions.TrxLogger.UnitTests/XmlPersistenceTests.cs
@@ -5,36 +5,66 @@ namespace Microsoft.TestPlatform.Extensions.TrxLogger.UnitTests
 {
     using Microsoft.TestPlatform.Extensions.TrxLogger.XML;
     using Microsoft.VisualStudio.TestTools.UnitTesting;
-    using System.IO;
 
     [TestClass]
     public class XmlPersistenceTests
     {
         [TestMethod]
-        public void SaveObjectShouldRemoveInvalidCharacter()
+        public void SaveObjectShouldReplaceInvalidCharacter()
         {
             XmlPersistence xmlPersistence = new XmlPersistence();
             var node = xmlPersistence.CreateRootElement("TestRun");
 
-            string strWithInvalidCharForXml = "This string has these \0 \v invalid characters";
+            // we are handling only #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
+            char[] invalidXmlCharacterArray = new char[7];
+            invalidXmlCharacterArray[0] = (char)0x5;
+            invalidXmlCharacterArray[1] = (char)0xb;
+            invalidXmlCharacterArray[2] = (char)0xf;
+            invalidXmlCharacterArray[3] = (char)0xd800;
+            invalidXmlCharacterArray[4] = (char)0xdc00;
+            invalidXmlCharacterArray[5] = (char)0xfffe;
+            invalidXmlCharacterArray[6] = (char)0x0;
 
+            string strWithInvalidCharForXml = new string(invalidXmlCharacterArray);
             xmlPersistence.SaveObject(strWithInvalidCharForXml, node, null, "dummy");
 
-            string expectedResult = "This string has these ? ? invalid characters";
+            string expectedResult = "\\u0005\\u000b\\u000f\\ud800\\udc00\\ufffe\\u0000";
             Assert.AreEqual(string.Compare(expectedResult, node.InnerXml), 0);
         }
 
         [TestMethod]
-        public void SaveObjectShouldDoesNotRemoveValidCharacter()
+        public void SaveObjectShouldNotReplaceValidCharacter()
         {
             XmlPersistence xmlPersistence = new XmlPersistence();
             var node = xmlPersistence.CreateRootElement("TestRun");
 
-            string strWithInvalidCharForXml = "This string has these \\0 \v invalid characters";
+            // we are handling only #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
+            char[] validXmlCharacterArray = new char[8];
+            validXmlCharacterArray[0] = (char)0x9;
+            validXmlCharacterArray[1] = (char)0xa;
+            validXmlCharacterArray[2] = (char)0xd;
+            validXmlCharacterArray[3] = (char)0x20;
+            validXmlCharacterArray[4] = (char)0xc123;
+            validXmlCharacterArray[5] = (char)0xe000;
+            validXmlCharacterArray[6] = (char)0xea12;
+            validXmlCharacterArray[7] = (char)0xfffd;
 
-            xmlPersistence.SaveObject(strWithInvalidCharForXml, node, null, "dummy");
+            string strWithValidCharForXml = new string(validXmlCharacterArray);
+
+            xmlPersistence.SaveObject(strWithValidCharForXml, node, null, "dummy");
+
+            string expectedResult = "\t\n\r 섣�";
+            Assert.AreEqual(string.Compare(expectedResult, node.InnerXml), 0);
+        }
 
-            string expectedResult = "This string has these \\0 ? invalid characters";
+        [TestMethod]
+        public void SaveObjectShouldReplaceOnlyInvalidCharacter()
+        {
+            XmlPersistence xmlPersistence = new XmlPersistence();
+            var node = xmlPersistence.CreateRootElement("TestRun");
+            string strWithInvalidCharForXml = "This string has these \0 \v invalid characters";
+            xmlPersistence.SaveObject(strWithInvalidCharForXml, node, null, "dummy");
+            string expectedResult = "This string has these \\u0000 \\u000b invalid characters";
             Assert.AreEqual(string.Compare(expectedResult, node.InnerXml), 0);
         }
     }