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>

8 files changed, 25 insertions, 60 deletions

diff --git a/BUILD.gn b/BUILD.gn
index c4da7d6da6..dfc6d6c833 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -1384,6 +1384,8 @@ if (pdf_enable_xfa) {
       "xfa/fde/ifde_txtedtdorecord.h",
       "xfa/fde/ifde_txtedtengine.h",
       "xfa/fde/ifde_txtedtpage.h",
+      "xfa/fde/xml/cfde_xml_parser.cpp",
+      "xfa/fde/xml/cfde_xml_parser.h",
       "xfa/fde/xml/fde_xml.h",
       "xfa/fde/xml/fde_xml_imp.cpp",
       "xfa/fde/xml/fde_xml_imp.h",
@@ -1756,8 +1758,6 @@ if (pdf_enable_xfa) {
       "xfa/fxfa/parser/cxfa_widetextread.h",
       "xfa/fxfa/parser/cxfa_widgetdata.cpp",
       "xfa/fxfa/parser/cxfa_widgetdata.h",
-      "xfa/fxfa/parser/cxfa_xml_parser.cpp",
-      "xfa/fxfa/parser/cxfa_xml_parser.h",
       "xfa/fxfa/parser/cxfa_xmllocale.cpp",
       "xfa/fxfa/parser/cxfa_xmllocale.h",
       "xfa/fxfa/parser/xfa_basic_data.cpp",
diff --git a/testing/libfuzzer/pdf_xml_fuzzer.cc b/testing/libfuzzer/pdf_xml_fuzzer.cc
index e3cd254344..e255f96f6e 100644
--- a/testing/libfuzzer/pdf_xml_fuzzer.cc
+++ b/testing/libfuzzer/pdf_xml_fuzzer.cc
@@ -10,8 +10,8 @@
 #include "core/fxcrt/fx_safe_types.h"
 #include "core/fxcrt/fx_system.h"
 #include "third_party/base/ptr_util.h"
+#include "xfa/fde/xml/cfde_xml_parser.h"
 #include "xfa/fde/xml/fde_xml_imp.h"
-#include "xfa/fxfa/parser/cxfa_xml_parser.h"
 #include "xfa/fxfa/parser/cxfa_widetextread.h"
 
 namespace {
@@ -57,7 +57,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
     return 0;
 
   auto doc = pdfium::MakeUnique<CFDE_XMLDoc>();
-  if (!doc->LoadXML(pdfium::MakeUnique<CXFA_XMLParser>(doc->GetRoot(), stream)))
+  if (!doc->LoadXML(pdfium::MakeUnique<CFDE_XMLParser>(doc->GetRoot(), stream)))
     return 0;
 
   if (doc->DoLoad(nullptr) < 100)
diff --git a/xfa/fxfa/parser/cxfa_xml_parser.cpp b/xfa/fde/xml/cfde_xml_parser.cpp
index 72564d9fe5..840c34a5c4 100644
--- a/xfa/fxfa/parser/cxfa_xml_parser.cpp
+++ b/xfa/fde/xml/cfde_xml_parser.cpp
@@ -4,11 +4,11 @@
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
-#include "xfa/fxfa/parser/cxfa_xml_parser.h"
+#include "xfa/fde/xml/cfde_xml_parser.h"
 
 #include "third_party/base/ptr_util.h"
 
-CXFA_XMLParser::CXFA_XMLParser(CFDE_XMLNode* pParent,
+CFDE_XMLParser::CFDE_XMLParser(CFDE_XMLNode* pParent,
                                const CFX_RetainPtr<IFGAS_Stream>& pStream)
     : m_nElementStart(0),
       m_dwCheckStatus(0),
@@ -23,9 +23,9 @@ CXFA_XMLParser::CXFA_XMLParser(CFDE_XMLNode* pParent,
   m_pParser->Init(m_pStream, 32 * 1024, 1024 * 1024);
 }
 
-CXFA_XMLParser::~CXFA_XMLParser() {}
+CFDE_XMLParser::~CFDE_XMLParser() {}
 
-int32_t CXFA_XMLParser::DoParser(IFX_Pause* pPause) {
+int32_t CFDE_XMLParser::DoParser(IFX_Pause* pPause) {
   if (m_syntaxParserResult == FDE_XmlSyntaxResult::Error)
     return -1;
   if (m_syntaxParserResult == FDE_XmlSyntaxResult::EndOfString)
diff --git a/xfa/fxfa/parser/cxfa_xml_parser.h b/xfa/fde/xml/cfde_xml_parser.h
index a631df02dc..cd8ccbe389 100644
--- a/xfa/fxfa/parser/cxfa_xml_parser.h
+++ b/xfa/fde/xml/cfde_xml_parser.h
@@ -4,8 +4,8 @@
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
-#ifndef XFA_FXFA_PARSER_CXFA_XML_PARSER_H_
-#define XFA_FXFA_PARSER_CXFA_XML_PARSER_H_
+#ifndef XFA_FDE_XML_CFDE_XML_PARSER_H_
+#define XFA_FDE_XML_CFDE_XML_PARSER_H_
 
 #include <memory>
 #include <stack>
@@ -15,14 +15,13 @@
 class IFGAS_Stream;
 class IFX_Pause;
 
-class CXFA_XMLParser : public IFDE_XMLParser {
+class CFDE_XMLParser {
  public:
-  CXFA_XMLParser(CFDE_XMLNode* pParent,
+  CFDE_XMLParser(CFDE_XMLNode* pParent,
                  const CFX_RetainPtr<IFGAS_Stream>& pStream);
-  ~CXFA_XMLParser() override;
+  ~CFDE_XMLParser();
 
-  // IFDE_XMLParser
-  int32_t DoParser(IFX_Pause* pPause) override;
+  int32_t DoParser(IFX_Pause* pPause);
 
   FX_FILESIZE m_nStart[2];
   size_t m_nSize[2];
@@ -41,4 +40,4 @@ class CXFA_XMLParser : public IFDE_XMLParser {
   FDE_XmlSyntaxResult m_syntaxParserResult;
 };
 
-#endif  // XFA_FXFA_PARSER_CXFA_XML_PARSER_H_
+#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);
diff --git a/xfa/fxfa/parser/cxfa_simple_parser.cpp b/xfa/fxfa/parser/cxfa_simple_parser.cpp
index 574ef3a458..9204660388 100644
--- a/xfa/fxfa/parser/cxfa_simple_parser.cpp
+++ b/xfa/fxfa/parser/cxfa_simple_parser.cpp
@@ -11,12 +11,12 @@
 #include "core/fxcrt/cfx_checksumcontext.h"
 #include "core/fxcrt/fx_ext.h"
 #include "third_party/base/ptr_util.h"
+#include "xfa/fde/xml/cfde_xml_parser.h"
 #include "xfa/fgas/crt/fgas_codepage.h"
 #include "xfa/fxfa/fxfa.h"
 #include "xfa/fxfa/parser/cxfa_document.h"
 #include "xfa/fxfa/parser/cxfa_node.h"
 #include "xfa/fxfa/parser/cxfa_widetextread.h"
-#include "xfa/fxfa/parser/cxfa_xml_parser.h"
 #include "xfa/fxfa/parser/xfa_basic_data.h"
 #include "xfa/fxfa/parser/xfa_utils.h"
 
@@ -298,7 +298,7 @@ int32_t CXFA_SimpleParser::StartParse(
   }
   m_pXMLDoc = pdfium::MakeUnique<CFDE_XMLDoc>();
   auto pNewParser =
-      pdfium::MakeUnique<CXFA_XMLParser>(m_pXMLDoc->GetRoot(), m_pStream);
+      pdfium::MakeUnique<CFDE_XMLParser>(m_pXMLDoc->GetRoot(), m_pStream);
   m_pXMLParser = pNewParser.get();
   if (!m_pXMLDoc->LoadXML(std::move(pNewParser)))
     return XFA_PARSESTATUS_StatusErr;
@@ -334,7 +334,7 @@ int32_t CXFA_SimpleParser::ParseXMLData(const CFX_WideString& wsXML,
   m_pXMLDoc = pdfium::MakeUnique<CFDE_XMLDoc>();
   auto pStream = pdfium::MakeRetain<CXFA_WideTextRead>(wsXML);
   auto pParser =
-      pdfium::MakeUnique<CXFA_XMLParser>(m_pXMLDoc->GetRoot(), pStream);
+      pdfium::MakeUnique<CFDE_XMLParser>(m_pXMLDoc->GetRoot(), pStream);
   pParser->m_dwCheckStatus = 0x03;
   if (!m_pXMLDoc->LoadXML(std::move(pParser)))
     return XFA_PARSESTATUS_StatusErr;
diff --git a/xfa/fxfa/parser/cxfa_simple_parser.h b/xfa/fxfa/parser/cxfa_simple_parser.h
index 5f61ad7d28..350104161c 100644
--- a/xfa/fxfa/parser/cxfa_simple_parser.h
+++ b/xfa/fxfa/parser/cxfa_simple_parser.h
@@ -14,7 +14,7 @@
 
 class CXFA_Document;
 class CXFA_Node;
-class CXFA_XMLParser;
+class CFDE_XMLParser;
 class IFX_SeekableReadStream;
 class IFX_Pause;
 class IFGAS_Stream;
@@ -76,7 +76,7 @@ class CXFA_SimpleParser {
                         CFDE_XMLInstruction* pXMLInstruction,
                         XFA_XDPPACKET ePacketID);
 
-  CXFA_XMLParser* m_pXMLParser;
+  CFDE_XMLParser* m_pXMLParser;
   std::unique_ptr<CFDE_XMLDoc> m_pXMLDoc;
   CFX_RetainPtr<IFGAS_Stream> m_pStream;
   CFX_RetainPtr<IFX_SeekableReadStream> m_pFileRead;