Fix CData parsing in CFDE_XMLSyntaxParser.

This CL splits the handling of CData sections out to an individual phase
of the parser. This fixes the issue with the CData parser getting confused
by < characters inside the data section.

BUG=pdfium:90

Review URL: https://codereview.chromium.org/1842633004

3 files changed, 565 insertions, 43 deletions

diff --git a/xfa/fde/xml/fde_xml_imp.cpp b/xfa/fde/xml/fde_xml_imp.cpp
index ef5a7e4ee3..0affe8a621 100644
--- a/xfa/fde/xml/fde_xml_imp.cpp
+++ b/xfa/fde/xml/fde_xml_imp.cpp
@@ -1491,7 +1491,7 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
   FXSYS_assert(m_pStream && m_pBuffer && m_BlockBuffer.IsInitialized());
   int32_t iStreamLength = m_pStream->GetLength();
   int32_t iPos;
-  FX_WCHAR ch;
+
   uint32_t dwStatus = FDE_XMLSYNTAXSTATUS_None;
   while (TRUE) {
     if (m_pStart >= m_pEnd) {
@@ -1516,8 +1516,9 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
       m_pStart = m_pBuffer;
       m_pEnd = m_pBuffer + m_iBufferChars;
     }
+
     while (m_pStart < m_pEnd) {
-      ch = *m_pStart;
+      FX_WCHAR ch = *m_pStart;
       switch (m_dwMode) {
         case FDE_XMLSYNTAXMODE_Text:
           if (ch == L'<') {
@@ -1783,27 +1784,51 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
           m_pStart++;
           break;
         case FDE_XMLSYNTAXMODE_SkipCommentOrDecl:
-          if (ch == '-') {
+          if (FX_wcsnicmp(m_pStart, L"--", 2) == 0) {
+            m_pStart += 2;
             m_dwMode = FDE_XMLSYNTAXMODE_SkipComment;
+          } else if (FX_wcsnicmp(m_pStart, L"[CDATA[", 7) == 0) {
+            m_pStart += 7;
+            m_dwMode = FDE_XMLSYNTAXMODE_SkipCData;
           } else {
             m_dwMode = FDE_XMLSYNTAXMODE_SkipDeclNode;
             m_SkipChar = L'>';
             m_SkipStack.Push(L'>');
           }
           break;
+        case FDE_XMLSYNTAXMODE_SkipCData: {
+          if (FX_wcsnicmp(m_pStart, L"]]>", 3) == 0) {
+            m_pStart += 3;
+            dwStatus = FDE_XMLSYNTAXSTATUS_CData;
+            m_iTextDataLength = m_iDataLength;
+            m_BlockBuffer.Reset();
+            m_pCurrentBlock = m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
+            m_dwMode = FDE_XMLSYNTAXMODE_Text;
+          } else {
+            if (m_iIndexInBlock == m_iAllocStep) {
+              m_pCurrentBlock =
+                  m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
+              if (!m_pCurrentBlock)
+                return FDE_XMLSYNTAXSTATUS_Error;
+            }
+            m_pCurrentBlock[m_iIndexInBlock++] = ch;
+            m_iDataLength++;
+            m_pStart++;
+          }
+          break;
+        }
         case FDE_XMLSYNTAXMODE_SkipDeclNode:
           if (m_SkipChar == L'\'' || m_SkipChar == L'\"') {
             m_pStart++;
-            if (ch != m_SkipChar) {
+            if (ch != m_SkipChar)
               break;
-            }
+
             m_SkipStack.Pop();
             uint32_t* pDWord = m_SkipStack.GetTopElement();
-            if (pDWord == NULL) {
+            if (!pDWord)
               m_dwMode = FDE_XMLSYNTAXMODE_Text;
-            } else {
+            else
               m_SkipChar = (FX_WCHAR)*pDWord;
-            }
           } else {
             switch (ch) {
               case L'<':
@@ -1830,20 +1855,10 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
                 if (ch == m_SkipChar) {
                   m_SkipStack.Pop();
                   uint32_t* pDWord = m_SkipStack.GetTopElement();
-                  if (pDWord == NULL) {
+                  if (!pDWord) {
                     if (m_iDataLength >= 9) {
                       CFX_WideString wsHeader;
                       m_BlockBuffer.GetTextData(wsHeader, 0, 7);
-                      if (wsHeader.Equal(FX_WSTRC(L"[CDATA["))) {
-                        CFX_WideString wsTailer;
-                        m_BlockBuffer.GetTextData(wsTailer, m_iDataLength - 2,
-                                                  2);
-                        if (wsTailer.Equal(FX_WSTRC(L"]]"))) {
-                          m_BlockBuffer.DeleteTextChars(7, TRUE);
-                          m_BlockBuffer.DeleteTextChars(2, FALSE);
-                          dwStatus = FDE_XMLSYNTAXSTATUS_CData;
-                        }
-                      }
                     }
                     m_iTextDataLength = m_iDataLength;
                     m_BlockBuffer.Reset();
@@ -1851,7 +1866,7 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
                         m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
                     m_dwMode = FDE_XMLSYNTAXMODE_Text;
                   } else {
-                    m_SkipChar = (FX_WCHAR)*pDWord;
+                    m_SkipChar = static_cast<FX_WCHAR>(*pDWord);
                   }
                 }
                 break;
@@ -1871,27 +1886,11 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
           }
           break;
         case FDE_XMLSYNTAXMODE_SkipComment:
-          if (ch == L'-') {
-            if (m_iIndexInBlock == m_iAllocStep) {
-              m_pCurrentBlock =
-                  m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
-              if (!m_pCurrentBlock) {
-                return FDE_XMLSYNTAXSTATUS_Error;
-              }
-            }
-            m_pCurrentBlock[m_iIndexInBlock++] = L'-';
-            m_iDataLength++;
-          } else if (ch == L'>') {
-            if (m_iDataLength > 1) {
-              m_BlockBuffer.Reset();
-              m_pCurrentBlock =
-                  m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
-              m_dwMode = FDE_XMLSYNTAXMODE_Text;
-            }
-          } else {
-            m_BlockBuffer.Reset();
-            m_pCurrentBlock = m_BlockBuffer.GetAvailableBlock(m_iIndexInBlock);
+          if (FX_wcsnicmp(m_pStart, L"-->", 3) == 0) {
+            m_pStart += 2;
+            m_dwMode = FDE_XMLSYNTAXMODE_Text;
           }
+
           m_pStart++;
           break;
         case FDE_XMLSYNTAXMODE_TargetData:
@@ -1945,9 +1944,8 @@ uint32_t CFDE_XMLSyntaxParser::DoSyntaxParse() {
         default:
           break;
       }
-      if (dwStatus != FDE_XMLSYNTAXSTATUS_None) {
+      if (dwStatus != FDE_XMLSYNTAXSTATUS_None)
         return dwStatus;
-      }
     }
   }
   return 0;
diff --git a/xfa/fde/xml/fde_xml_imp.h b/xfa/fde/xml/fde_xml_imp.h
index a4361bdfea..0f252c12dd 100644
--- a/xfa/fde/xml/fde_xml_imp.h
+++ b/xfa/fde/xml/fde_xml_imp.h
@@ -309,7 +309,9 @@ class CFDE_BlockBuffer : public CFX_Target {
 #define FDE_XMLSYNTAXMODE_DeclCharData 15
 #define FDE_XMLSYNTAXMODE_SkipComment 16
 #define FDE_XMLSYNTAXMODE_SkipCommentOrDecl 17
-#define FDE_XMLSYNTAXMODE_TargetData 18
+#define FDE_XMLSYNTAXMODE_SkipCData 18
+#define FDE_XMLSYNTAXMODE_TargetData 19
+
 class CFDE_XMLSyntaxParser : public CFX_Target {
  public:
   CFDE_XMLSyntaxParser();
diff --git a/xfa/fde/xml/fde_xml_imp_unittest.cpp b/xfa/fde/xml/fde_xml_imp_unittest.cpp
new file mode 100644
index 0000000000..42119ebe37
--- /dev/null
+++ b/xfa/fde/xml/fde_xml_imp_unittest.cpp
@@ -0,0 +1,522 @@
+// Copyright 2016 PDFium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "xfa/fde/xml/fde_xml_imp.h"
+
+#include "xfa/fgas/crt/fgas_stream.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+TEST(CFDE_XMLSyntaxParser, CData) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![CDATA[\n"
+      L"    if (a[1] < 3)\n"
+      L"      app.alert(\"Tclams\");\n"
+      L"  ]]>\n"
+      L"</script>";
+
+  const FX_WCHAR* cdata =
+      L"\n"
+      L"    if (a[1] < 3)\n"
+      L"      app.alert(\"Tclams\");\n"
+      L"  ";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_CData, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(cdata, data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, CDataWithInnerScript) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![CDATA[\n"
+      L"    if (a[1] < 3)\n"
+      L"      app.alert(\"Tclams\");\n"
+      L"    </script>\n"
+      L"  ]]>\n"
+      L"</script>";
+
+  const FX_WCHAR* cdata =
+      L"\n"
+      L"    if (a[1] < 3)\n"
+      L"      app.alert(\"Tclams\");\n"
+      L"    </script>\n"
+      L"  ";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_CData, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(cdata, data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, ArrowBangArrow) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!>\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, ArrowBangBracketArrow) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![>\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  // Parser walks to end of input.
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, IncompleteCData) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![CDATA>\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  // Parser walks to end of input.
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, UnClosedCData) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![CDATA[\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  // Parser walks to end of input.
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, EmptyCData) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <![CDATA[]]>\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_CData, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, Comment) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!-- A Comment -->\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, IncorrectCommentStart) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!- A Comment -->\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, CommentEmpty) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!---->\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementClose, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, CommentThreeDash) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!--->\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}
+
+TEST(CFDE_XMLSyntaxParser, CommentTwoDash) {
+  const FX_WCHAR* input =
+      L"<script contentType=\"application/x-javascript\">\n"
+      L"  <!-->\n"
+      L"</script>";
+
+  // We * sizeof(FX_WCHAR) because we pass in the uint8_t, not the FX_WCHAR.
+  size_t len = FXSYS_wcslen(input) * sizeof(FX_WCHAR);
+  std::unique_ptr<IFX_Stream> stream(IFX_Stream::CreateStream(
+      reinterpret_cast<uint8_t*>(const_cast<FX_WCHAR*>(input)), len, 0));
+  CFDE_XMLSyntaxParser parser;
+  parser.Init(stream.get(), 256);
+
+  CFX_WideString data;
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementOpen, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_TagName, parser.DoSyntaxParse());
+  parser.GetTagName(data);
+  EXPECT_EQ(L"script", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriName, parser.DoSyntaxParse());
+  parser.GetAttributeName(data);
+  EXPECT_EQ(L"contentType", data);
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_AttriValue, parser.DoSyntaxParse());
+  parser.GetAttributeValue(data);
+  EXPECT_EQ(L"application/x-javascript", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_ElementBreak, parser.DoSyntaxParse());
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_Text, parser.DoSyntaxParse());
+  parser.GetTextData(data);
+  EXPECT_EQ(L"\n  ", data);
+
+  EXPECT_EQ(FDE_XMLSYNTAXSTATUS_EOS, parser.DoSyntaxParse());
+}