Move CXFA_XMLParser to CFDE_XMLParser

There is nothing XFA specific in the CXFA_XMLParser. This Cl moves it to
the CFDE_XMLParser and co-locates with the other XML code.

Change-Id: I86c12da3f6f5732be54b8019562978f88234e2fc
Reviewed-on: https://pdfium-review.googlesource.com/3432
Reviewed-by: Nicolás Peña <npm@chromium.org>
Commit-Queue: dsinclair <dsinclair@chromium.org>

4 files changed, 215 insertions, 39 deletions

diff --git a/xfa/fde/xml/cfde_xml_parser.cpp b/xfa/fde/xml/cfde_xml_parser.cpp
new file mode 100644
index 0000000000..840c34a5c4
--- /dev/null
+++ b/xfa/fde/xml/cfde_xml_parser.cpp
@@ -0,0 +1,167 @@
+// 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.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#include "xfa/fde/xml/cfde_xml_parser.h"
+
+#include "third_party/base/ptr_util.h"
+
+CFDE_XMLParser::CFDE_XMLParser(CFDE_XMLNode* pParent,
+                               const CFX_RetainPtr<IFGAS_Stream>& pStream)
+    : m_nElementStart(0),
+      m_dwCheckStatus(0),
+      m_dwCurrentCheckStatus(0),
+      m_pStream(pStream),
+      m_pParser(pdfium::MakeUnique<CFDE_XMLSyntaxParser>()),
+      m_pParent(pParent),
+      m_pChild(nullptr),
+      m_syntaxParserResult(FDE_XmlSyntaxResult::None) {
+  ASSERT(m_pParent && m_pStream);
+  m_NodeStack.push(m_pParent);
+  m_pParser->Init(m_pStream, 32 * 1024, 1024 * 1024);
+}
+
+CFDE_XMLParser::~CFDE_XMLParser() {}
+
+int32_t CFDE_XMLParser::DoParser(IFX_Pause* pPause) {
+  if (m_syntaxParserResult == FDE_XmlSyntaxResult::Error)
+    return -1;
+  if (m_syntaxParserResult == FDE_XmlSyntaxResult::EndOfString)
+    return 100;
+
+  int32_t iCount = 0;
+  while (true) {
+    m_syntaxParserResult = m_pParser->DoSyntaxParse();
+    switch (m_syntaxParserResult) {
+      case FDE_XmlSyntaxResult::InstructionOpen:
+        break;
+      case FDE_XmlSyntaxResult::InstructionClose:
+        if (m_pChild) {
+          if (m_pChild->GetType() != FDE_XMLNODE_Instruction) {
+            m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
+            break;
+          }
+        }
+        m_pChild = m_pParent;
+        break;
+      case FDE_XmlSyntaxResult::ElementOpen:
+        if (m_dwCheckStatus != 0x03 && m_NodeStack.size() == 2)
+          m_nElementStart = m_pParser->GetCurrentPos() - 1;
+        break;
+      case FDE_XmlSyntaxResult::ElementBreak:
+        break;
+      case FDE_XmlSyntaxResult::ElementClose:
+        if (m_pChild->GetType() != FDE_XMLNODE_Element) {
+          m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
+          break;
+        }
+        m_pParser->GetTagName(m_ws1);
+        static_cast<CFDE_XMLElement*>(m_pChild)->GetTagName(m_ws2);
+        if (m_ws1.GetLength() > 0 && m_ws1 != m_ws2) {
+          m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
+          break;
+        }
+        if (!m_NodeStack.empty())
+          m_NodeStack.pop();
+        if (m_NodeStack.empty()) {
+          m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
+          break;
+        } else if (m_dwCurrentCheckStatus != 0 && m_NodeStack.size() == 2) {
+          m_nSize[m_dwCurrentCheckStatus - 1] =
+              m_pParser->GetCurrentBinaryPos() -
+              m_nStart[m_dwCurrentCheckStatus - 1];
+          m_dwCurrentCheckStatus = 0;
+        }
+        m_pParent = m_NodeStack.top();
+        m_pChild = m_pParent;
+        iCount++;
+        break;
+      case FDE_XmlSyntaxResult::TargetName:
+        m_pParser->GetTargetName(m_ws1);
+        if (m_ws1 == L"originalXFAVersion" || m_ws1 == L"acrobat") {
+          m_pChild = new CFDE_XMLInstruction(m_ws1);
+          m_pParent->InsertChildNode(m_pChild);
+        } else {
+          m_pChild = nullptr;
+        }
+        m_ws1.clear();
+        break;
+      case FDE_XmlSyntaxResult::TagName:
+        m_pParser->GetTagName(m_ws1);
+        m_pChild = new CFDE_XMLElement(m_ws1);
+        m_pParent->InsertChildNode(m_pChild);
+        m_NodeStack.push(m_pChild);
+        m_pParent = m_pChild;
+
+        if (m_dwCheckStatus != 0x03 && m_NodeStack.size() == 3) {
+          CFX_WideString wsTag;
+          static_cast<CFDE_XMLElement*>(m_pChild)->GetLocalTagName(wsTag);
+          if (wsTag == L"template") {
+            m_dwCheckStatus |= 0x01;
+            m_dwCurrentCheckStatus = 0x01;
+            m_nStart[0] = m_pParser->GetCurrentBinaryPos() -
+                          (m_pParser->GetCurrentPos() - m_nElementStart);
+          } else if (wsTag == L"datasets") {
+            m_dwCheckStatus |= 0x02;
+            m_dwCurrentCheckStatus = 0x02;
+            m_nStart[1] = m_pParser->GetCurrentBinaryPos() -
+                          (m_pParser->GetCurrentPos() - m_nElementStart);
+          }
+        }
+        break;
+      case FDE_XmlSyntaxResult::AttriName:
+        m_pParser->GetAttributeName(m_ws1);
+        break;
+      case FDE_XmlSyntaxResult::AttriValue:
+        if (m_pChild) {
+          m_pParser->GetAttributeName(m_ws2);
+          if (m_pChild->GetType() == FDE_XMLNODE_Element) {
+            static_cast<CFDE_XMLElement*>(m_pChild)->SetString(m_ws1, m_ws2);
+          }
+        }
+        m_ws1.clear();
+        break;
+      case FDE_XmlSyntaxResult::Text:
+        m_pParser->GetTextData(m_ws1);
+        m_pChild = new CFDE_XMLText(m_ws1);
+        m_pParent->InsertChildNode(m_pChild);
+        m_pChild = m_pParent;
+        break;
+      case FDE_XmlSyntaxResult::CData:
+        m_pParser->GetTextData(m_ws1);
+        m_pChild = new CFDE_XMLCharData(m_ws1);
+        m_pParent->InsertChildNode(m_pChild);
+        m_pChild = m_pParent;
+        break;
+      case FDE_XmlSyntaxResult::TargetData:
+        if (m_pChild) {
+          if (m_pChild->GetType() != FDE_XMLNODE_Instruction) {
+            m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
+            break;
+          }
+          if (!m_ws1.IsEmpty()) {
+            static_cast<CFDE_XMLInstruction*>(m_pChild)->AppendData(m_ws1);
+          }
+          m_pParser->GetTargetData(m_ws1);
+          static_cast<CFDE_XMLInstruction*>(m_pChild)->AppendData(m_ws1);
+        }
+        m_ws1.clear();
+        break;
+      default:
+        break;
+    }
+    if (m_syntaxParserResult == FDE_XmlSyntaxResult::Error ||
+        m_syntaxParserResult == FDE_XmlSyntaxResult::EndOfString) {
+      break;
+    }
+    if (pPause && iCount > 500 && pPause->NeedToPauseNow()) {
+      break;
+    }
+  }
+  return (m_syntaxParserResult == FDE_XmlSyntaxResult::Error ||
+          m_NodeStack.size() != 1)
+             ? -1
+             : m_pParser->GetStatus();
+}
diff --git a/xfa/fde/xml/cfde_xml_parser.h b/xfa/fde/xml/cfde_xml_parser.h
new file mode 100644
index 0000000000..cd8ccbe389
--- /dev/null
+++ b/xfa/fde/xml/cfde_xml_parser.h
@@ -0,0 +1,43 @@
+// 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.
+
+// Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
+
+#ifndef XFA_FDE_XML_CFDE_XML_PARSER_H_
+#define XFA_FDE_XML_CFDE_XML_PARSER_H_
+
+#include <memory>
+#include <stack>
+
+#include "xfa/fde/xml/fde_xml_imp.h"
+
+class IFGAS_Stream;
+class IFX_Pause;
+
+class CFDE_XMLParser {
+ public:
+  CFDE_XMLParser(CFDE_XMLNode* pParent,
+                 const CFX_RetainPtr<IFGAS_Stream>& pStream);
+  ~CFDE_XMLParser();
+
+  int32_t DoParser(IFX_Pause* pPause);
+
+  FX_FILESIZE m_nStart[2];
+  size_t m_nSize[2];
+  FX_FILESIZE m_nElementStart;
+  uint16_t m_dwCheckStatus;
+  uint16_t m_dwCurrentCheckStatus;
+
+ private:
+  CFX_RetainPtr<IFGAS_Stream> m_pStream;
+  std::unique_ptr<CFDE_XMLSyntaxParser> m_pParser;
+  CFDE_XMLNode* m_pParent;
+  CFDE_XMLNode* m_pChild;
+  std::stack<CFDE_XMLNode*> m_NodeStack;
+  CFX_WideString m_ws1;
+  CFX_WideString m_ws2;
+  FDE_XmlSyntaxResult m_syntaxParserResult;
+};
+
+#endif  // XFA_FDE_XML_CFDE_XML_PARSER_H_
diff --git a/xfa/fde/xml/fde_xml_imp.cpp b/xfa/fde/xml/fde_xml_imp.cpp
index ade8fc0557..68a4978209 100644
--- a/xfa/fde/xml/fde_xml_imp.cpp
+++ b/xfa/fde/xml/fde_xml_imp.cpp
@@ -13,6 +13,7 @@
 #include "core/fxcrt/fx_safe_types.h"
 #include "third_party/base/ptr_util.h"
 #include "third_party/base/stl_util.h"
+#include "xfa/fde/xml/cfde_xml_parser.h"
 #include "xfa/fgas/crt/fgas_codepage.h"
 
 namespace {
@@ -917,7 +918,7 @@ CFDE_XMLDoc::CFDE_XMLDoc()
 
 CFDE_XMLDoc::~CFDE_XMLDoc() {}
 
-bool CFDE_XMLDoc::LoadXML(std::unique_ptr<IFDE_XMLParser> pXMLParser) {
+bool CFDE_XMLDoc::LoadXML(std::unique_ptr<CFDE_XMLParser> pXMLParser) {
   if (!pXMLParser)
     return false;
 
@@ -1052,34 +1053,6 @@ void CFDE_XMLDoc::SaveXMLNode(const CFX_RetainPtr<IFGAS_Stream>& pXMLStream,
   }
 }
 
-void CFDE_XMLDoc::SaveXML(CFX_RetainPtr<IFGAS_Stream>& pXMLStream,
-                          bool bSaveBOM) {
-  if (!pXMLStream || pXMLStream == m_pStream) {
-    m_pStream->Seek(FX_STREAMSEEK_Begin, 0);
-    pXMLStream = m_pStream;
-  }
-  ASSERT((pXMLStream->GetAccessModes() & FX_STREAMACCESS_Text) != 0);
-  ASSERT((pXMLStream->GetAccessModes() & FX_STREAMACCESS_Write) != 0);
-  uint16_t wCodePage = pXMLStream->GetCodePage();
-  if (wCodePage != FX_CODEPAGE_UTF16LE && wCodePage != FX_CODEPAGE_UTF16BE &&
-      wCodePage != FX_CODEPAGE_UTF8) {
-    wCodePage = FX_CODEPAGE_UTF8;
-    pXMLStream->SetCodePage(wCodePage);
-  }
-  if (bSaveBOM) {
-    pXMLStream->WriteString(L"\xFEFF", 1);
-  }
-  CFDE_XMLNode* pNode = m_pRoot->m_pChild;
-  while (pNode) {
-    SaveXMLNode(pXMLStream, static_cast<CFDE_XMLNode*>(pNode));
-    pNode = pNode->m_pNext;
-  }
-  if (pXMLStream == m_pStream) {
-    int32_t iPos = pXMLStream->GetPosition();
-    pXMLStream->SetLength(iPos);
-  }
-}
-
 CFDE_BlockBuffer::CFDE_BlockBuffer(int32_t iAllocStep)
     : m_iDataLength(0),
       m_iBufferSize(0),
diff --git a/xfa/fde/xml/fde_xml_imp.h b/xfa/fde/xml/fde_xml_imp.h
index 09cdcbacdf..bd88da06cc 100644
--- a/xfa/fde/xml/fde_xml_imp.h
+++ b/xfa/fde/xml/fde_xml_imp.h
@@ -22,8 +22,8 @@ class CFDE_XMLElement;
 class CFDE_XMLText;
 class CFDE_XMLDoc;
 class CFDE_XMLDOMParser;
+class CFDE_XMLParser;
 class CFDE_XMLSyntaxParser;
-class IFDE_XMLParser;
 
 class CFDE_XMLNode {
  public:
@@ -191,27 +191,20 @@ class CFDE_XMLDoc {
   CFDE_XMLDoc();
   ~CFDE_XMLDoc();
 
-  bool LoadXML(std::unique_ptr<IFDE_XMLParser> pXMLParser);
+  bool LoadXML(std::unique_ptr<CFDE_XMLParser> pXMLParser);
   int32_t DoLoad(IFX_Pause* pPause = nullptr);
   void CloseXML();
   CFDE_XMLNode* GetRoot() const { return m_pRoot.get(); }
-  void SaveXML(CFX_RetainPtr<IFGAS_Stream>& pXMLStream, bool bSaveBOM = true);
   void SaveXMLNode(const CFX_RetainPtr<IFGAS_Stream>& pXMLStream,
                    CFDE_XMLNode* pNode);
 
  private:
   int32_t m_iStatus;
   std::unique_ptr<CFDE_XMLNode> m_pRoot;
-  std::unique_ptr<IFDE_XMLParser> m_pXMLParser;
+  std::unique_ptr<CFDE_XMLParser> m_pXMLParser;
   CFX_RetainPtr<IFGAS_Stream> m_pStream;
 };
 
-class IFDE_XMLParser {
- public:
-  virtual ~IFDE_XMLParser() {}
-  virtual int32_t DoParser(IFX_Pause* pPause) = 0;
-};
-
 class CFDE_BlockBuffer {
  public:
   explicit CFDE_BlockBuffer(int32_t iAllocStep = 1024 * 1024);