Break xfa_parser_imp apart

xfa/fxfa/parser/cxfa_simple_parser.cpp

-// Copyright 2014 PDFium Authors. All rights reserved.
+// 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/fxfa/parser/xfa_parser_imp.h"
+#include "xfa/fxfa/parser/cxfa_simple_parser.h"
 
-#include <memory>
-
-#include "xfa/fde/xml/fde_xml_imp.h"
 #include "xfa/fgas/crt/fgas_codepage.h"
+#include "xfa/fxfa/include/fxfa.h"
 #include "xfa/fxfa/include/xfa_checksum.h"
-#include "xfa/fxfa/parser/xfa_basic_imp.h"
-#include "xfa/fxfa/parser/xfa_doclayout.h"
+#include "xfa/fxfa/parser/cxfa_xml_parser.h"
 #include "xfa/fxfa/parser/xfa_document.h"
-#include "xfa/fxfa/parser/xfa_localemgr.h"
-#include "xfa/fxfa/parser/xfa_object.h"
-#include "xfa/fxfa/parser/xfa_script.h"
-#include "xfa/fxfa/parser/xfa_utils.h"
 
 CXFA_SimpleParser::CXFA_SimpleParser(CXFA_Document* pFactory,
                                      bool bDocumentParser)
     : m_pXMLParser(nullptr),
       m_pXMLDoc(nullptr),
       m_pStream(nullptr),
       m_pFileRead(nullptr),
       m_pFactory(pFactory),
       m_pRootNode(nullptr),
       m_ePacketID(XFA_XDPPACKET_UNKNOWN),
@@ skipping 1285 matching lines @@
 void CXFA_SimpleParser::CloseParser() {
   if (m_pXMLDoc) {
     m_pXMLDoc->Release();
     m_pXMLDoc = nullptr;
   }
   if (m_pStream) {
     m_pStream->Release();
     m_pStream = nullptr;
   }
 }
-
-CXFA_DocumentParser::CXFA_DocumentParser(CXFA_FFNotify* pNotify)
-    : m_nodeParser(nullptr, TRUE), m_pNotify(pNotify), m_pDocument(nullptr) {}
-
-CXFA_DocumentParser::~CXFA_DocumentParser() {
-  CloseParser();
-}
-
-int32_t CXFA_DocumentParser::StartParse(IFX_FileRead* pStream,
-                                        XFA_XDPPACKET ePacketID) {
-  CloseParser();
-  int32_t nRetStatus = m_nodeParser.StartParse(pStream, ePacketID);
-  if (nRetStatus == XFA_PARSESTATUS_Ready) {
-    m_pDocument.reset(new CXFA_Document(this));
-    m_nodeParser.SetFactory(m_pDocument.get());
-  }
-  return nRetStatus;
-}
-
-int32_t CXFA_DocumentParser::DoParse(IFX_Pause* pPause) {
-  int32_t nRetStatus = m_nodeParser.DoParse(pPause);
-  if (nRetStatus >= XFA_PARSESTATUS_Done) {
-    ASSERT(m_pDocument);
-    m_pDocument->SetRoot(m_nodeParser.GetRootNode());
-  }
-  return nRetStatus;
-}
-
-CFDE_XMLDoc* CXFA_DocumentParser::GetXMLDoc() const {
-  return m_nodeParser.GetXMLDoc();
-}
-
-CXFA_FFNotify* CXFA_DocumentParser::GetNotify() const {
-  return m_pNotify;
-}
-
-CXFA_Document* CXFA_DocumentParser::GetDocument() const {
-  return m_pDocument.get();
-}
-
-void CXFA_DocumentParser::CloseParser() {
-  m_pDocument.reset();
-  m_nodeParser.CloseParser();
-}
-
-CXFA_XMLParser::CXFA_XMLParser(CFDE_XMLNode* pRoot, IFX_Stream* pStream)
-    : m_nElementStart(0),
-      m_dwCheckStatus(0),
-      m_dwCurrentCheckStatus(0),
-      m_pRoot(pRoot),
-      m_pStream(pStream),
-      m_pParser(nullptr),
-      m_pParent(pRoot),
-      m_pChild(nullptr),
-      m_NodeStack(16),
-      m_syntaxParserResult(FDE_XmlSyntaxResult::None) {
-  ASSERT(m_pParent && m_pStream);
-  m_NodeStack.Push(m_pParent);
-  m_pParser = new CFDE_XMLSyntaxParser;
-  m_pParser->Init(m_pStream, 32 * 1024, 1024 * 1024);
-}
-
-CXFA_XMLParser::~CXFA_XMLParser() {
-  if (m_pParser) {
-    m_pParser->Release();
-  }
-  m_NodeStack.RemoveAll();
-  m_ws1.clear();
-  m_ws2.clear();
-}
-
-void CXFA_XMLParser::Release() {
-  delete this;
-}
-
-int32_t CXFA_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.GetSize() == 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;
-        }
-        m_NodeStack.Pop();
-        if (m_NodeStack.GetSize() < 1) {
-          m_syntaxParserResult = FDE_XmlSyntaxResult::Error;
-          break;
-        } else if (m_dwCurrentCheckStatus != 0 && m_NodeStack.GetSize() == 2) {
-          m_nSize[m_dwCurrentCheckStatus - 1] =
-              m_pParser->GetCurrentBinaryPos() -
-              m_nStart[m_dwCurrentCheckStatus - 1];
-          m_dwCurrentCheckStatus = 0;
-        }
-
-        m_pParent = static_cast<CFDE_XMLNode*>(*m_NodeStack.GetTopElement());
-        m_pChild = m_pParent;
-        iCount++;
-        break;
-      case FDE_XmlSyntaxResult::TargetName:
-        m_pParser->GetTargetName(m_ws1);
-        if (m_ws1 == FX_WSTRC(L"originalXFAVersion") ||
-            m_ws1 == FX_WSTRC(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.GetSize() == 3) {
-          CFX_WideString wsTag;
-          static_cast<CFDE_XMLElement*>(m_pChild)->GetLocalTagName(wsTag);
-          if (wsTag == FX_WSTRC(L"template")) {
-            m_dwCheckStatus |= 0x01;
-            m_dwCurrentCheckStatus = 0x01;
-            m_nStart[0] = m_pParser->GetCurrentBinaryPos() -
-                          (m_pParser->GetCurrentPos() - m_nElementStart);
-          } else if (wsTag == FX_WSTRC(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.GetSize() != 1)
-             ? -1
-             : m_pParser->GetStatus();
-}