Move the CFX_XMLParser out of CXFA_SimpleParser

This CL removes the CFX_XMLParser from CXFA_SimpleParser and constructs
it in the CFX_XMLDoc.

Change-Id: Id69b396f091892170a4c4654ac1d0fd65d1efbb6
Reviewed-on: https://pdfium-review.googlesource.com/29970
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>
Commit-Queue: dsinclair <dsinclair@chromium.org>

5 files changed, 19 insertions, 41 deletions

diff --git a/core/fxcrt/xml/cfx_xmldoc.cpp b/core/fxcrt/xml/cfx_xmldoc.cpp
index 236ab05c37..64ed5a98d0 100644
--- a/core/fxcrt/xml/cfx_xmldoc.cpp
+++ b/core/fxcrt/xml/cfx_xmldoc.cpp
@@ -18,24 +18,17 @@
 #include "third_party/base/ptr_util.h"
 #include "third_party/base/stl_util.h"
 
-CFX_XMLDoc::CFX_XMLDoc()
-    : m_iStatus(0), m_pRoot(pdfium::MakeUnique<CFX_XMLNode>()) {
+CFX_XMLDoc::CFX_XMLDoc(const RetainPtr<CFX_SeekableStreamProxy>& pStream)
+    : m_iStatus(0),
+      m_pRoot(pdfium::MakeUnique<CFX_XMLNode>()),
+      m_pXMLParser(pdfium::MakeUnique<CFX_XMLParser>(m_pRoot.get(), pStream)) {
+  ASSERT(pStream);
+
   m_pRoot->AppendChild(new CFX_XMLInstruction(L"xml"));
 }
 
 CFX_XMLDoc::~CFX_XMLDoc() {}
 
-bool CFX_XMLDoc::LoadXML(std::unique_ptr<CFX_XMLParser> pXMLParser) {
-  if (!pXMLParser)
-    return false;
-
-  m_iStatus = 0;
-  m_pStream.Reset();
-  m_pRoot->DeleteChildren();
-  m_pXMLParser = std::move(pXMLParser);
-  return true;
-}
-
 int32_t CFX_XMLDoc::DoLoad() {
   if (m_iStatus < 100)
     m_iStatus = m_pXMLParser->DoParser();
@@ -46,4 +39,3 @@ int32_t CFX_XMLDoc::DoLoad() {
 void CFX_XMLDoc::CloseXML() {
   m_pXMLParser.reset();
 }
-
diff --git a/core/fxcrt/xml/cfx_xmldoc.h b/core/fxcrt/xml/cfx_xmldoc.h
index 8ab6b8fea4..07be2d4e84 100644
--- a/core/fxcrt/xml/cfx_xmldoc.h
+++ b/core/fxcrt/xml/cfx_xmldoc.h
@@ -16,20 +16,19 @@
 
 class CFX_XMLDoc {
  public:
-  CFX_XMLDoc();
+  explicit CFX_XMLDoc(const RetainPtr<CFX_SeekableStreamProxy>& pStream);
   ~CFX_XMLDoc();
 
-  bool LoadXML(std::unique_ptr<CFX_XMLParser> pXMLParser);
   int32_t DoLoad();
   void CloseXML();
 
   CFX_XMLNode* GetRoot() const { return m_pRoot.get(); }
+  CFX_XMLParser* GetParser() const { return m_pXMLParser.get(); }
 
  private:
   int32_t m_iStatus;
   std::unique_ptr<CFX_XMLNode> m_pRoot;
   std::unique_ptr<CFX_XMLParser> m_pXMLParser;
-  RetainPtr<CFX_SeekableStreamProxy> m_pStream;
 };
 
 #endif  // CORE_FXCRT_XML_CFX_XMLDOC_H_
diff --git a/testing/libfuzzer/pdf_xml_fuzzer.cc b/testing/libfuzzer/pdf_xml_fuzzer.cc
index 3a2889421f..072f86c702 100644
--- a/testing/libfuzzer/pdf_xml_fuzzer.cc
+++ b/testing/libfuzzer/pdf_xml_fuzzer.cc
@@ -22,10 +22,7 @@ extern "C" int LLVMFuzzerTestOneInput(const uint8_t* data, size_t size) {
   RetainPtr<CFX_SeekableStreamProxy> stream =
       pdfium::MakeRetain<CFX_SeekableStreamProxy>(const_cast<uint8_t*>(data),
                                                   size);
-  auto doc = pdfium::MakeUnique<CFX_XMLDoc>();
-  if (!doc->LoadXML(pdfium::MakeUnique<CFX_XMLParser>(doc->GetRoot(), stream)))
-    return 0;
-
+  auto doc = pdfium::MakeUnique<CFX_XMLDoc>(stream);
   if (doc->DoLoad() < 100)
     return 0;
 
diff --git a/xfa/fxfa/parser/cxfa_simple_parser.cpp b/xfa/fxfa/parser/cxfa_simple_parser.cpp
index f30ab6b631..de9f74f291 100644
--- a/xfa/fxfa/parser/cxfa_simple_parser.cpp
+++ b/xfa/fxfa/parser/cxfa_simple_parser.cpp
@@ -350,12 +350,8 @@ int32_t CXFA_SimpleParser::StartParse(
       wCodePage != FX_CODEPAGE_UTF8) {
     m_pStream->SetCodePage(FX_CODEPAGE_UTF8);
   }
-  m_pXMLDoc = pdfium::MakeUnique<CFX_XMLDoc>();
-  auto pNewParser =
-      pdfium::MakeUnique<CFX_XMLParser>(m_pXMLDoc->GetRoot(), m_pStream);
-  m_pXMLParser = pNewParser.get();
-  if (!m_pXMLDoc->LoadXML(std::move(pNewParser)))
-    return XFA_PARSESTATUS_StatusErr;
+
+  m_pXMLDoc = pdfium::MakeUnique<CFX_XMLDoc>(m_pStream);
 
   m_bParseStarted = true;
   m_ePacketID = ePacketID;
@@ -373,7 +369,6 @@ int32_t CXFA_SimpleParser::DoParse() {
     return iRet / 2;
 
   m_pRootNode = ParseAsXDPPacket(GetDocumentNode(m_pXMLDoc.get()), m_ePacketID);
-  m_pXMLParser.Release();
   m_pXMLDoc->CloseXML();
   m_pStream.Reset();
 
@@ -385,15 +380,11 @@ int32_t CXFA_SimpleParser::DoParse() {
 
 CFX_XMLNode* CXFA_SimpleParser::ParseXMLData(const ByteString& wsXML) {
   CloseParser();
-  m_pXMLDoc = pdfium::MakeUnique<CFX_XMLDoc>();
 
   auto pStream = pdfium::MakeRetain<CFX_SeekableStreamProxy>(
       const_cast<uint8_t*>(wsXML.raw_str()), wsXML.GetLength());
-  auto pParser =
-      pdfium::MakeUnique<CFX_XMLParser>(m_pXMLDoc->GetRoot(), pStream);
-  pParser->m_dwCheckStatus = 0x03;
-  if (!m_pXMLDoc->LoadXML(std::move(pParser)))
-    return nullptr;
+  m_pXMLDoc = pdfium::MakeUnique<CFX_XMLDoc>(pStream);
+  m_pXMLDoc->GetParser()->m_dwCheckStatus = 0x03;
 
   int32_t iRet = m_pXMLDoc->DoLoad();
   if (iRet < 0 || iRet >= 100)
@@ -662,16 +653,17 @@ CXFA_Node* CXFA_SimpleParser::ParseAsXDPPacket_Form(
   CFX_XMLElement* pXMLDocumentElement =
       static_cast<CFX_XMLElement*>(pXMLDocumentNode);
   WideString wsChecksum = pXMLDocumentElement->GetString(L"checksum");
-  if (wsChecksum.GetLength() != 28 || m_pXMLParser->m_dwCheckStatus != 0x03) {
+  if (wsChecksum.GetLength() != 28 ||
+      m_pXMLDoc->GetParser()->m_dwCheckStatus != 0x03) {
     return nullptr;
   }
 
   auto pChecksum = pdfium::MakeUnique<CFX_ChecksumContext>();
   pChecksum->StartChecksum();
-  pChecksum->UpdateChecksum(m_pFileRead, m_pXMLParser->m_nStart[0],
-                            m_pXMLParser->m_nSize[0]);
-  pChecksum->UpdateChecksum(m_pFileRead, m_pXMLParser->m_nStart[1],
-                            m_pXMLParser->m_nSize[1]);
+  pChecksum->UpdateChecksum(m_pFileRead, m_pXMLDoc->GetParser()->m_nStart[0],
+                            m_pXMLDoc->GetParser()->m_nSize[0]);
+  pChecksum->UpdateChecksum(m_pFileRead, m_pXMLDoc->GetParser()->m_nStart[1],
+                            m_pXMLDoc->GetParser()->m_nSize[1]);
   pChecksum->FinishChecksum();
   ByteString bsCheck = pChecksum->GetChecksum();
   if (bsCheck != wsChecksum.UTF8Encode())
diff --git a/xfa/fxfa/parser/cxfa_simple_parser.h b/xfa/fxfa/parser/cxfa_simple_parser.h
index c1015469c9..1300a8c4f2 100644
--- a/xfa/fxfa/parser/cxfa_simple_parser.h
+++ b/xfa/fxfa/parser/cxfa_simple_parser.h
@@ -16,7 +16,6 @@ class CXFA_Node;
 class CFX_XMLDoc;
 class CFX_XMLInstruction;
 class CFX_XMLNode;
-class CFX_XMLParser;
 class IFX_SeekableStream;
 class CFX_SeekableStreamProxy;
 
@@ -73,7 +72,6 @@ class CXFA_SimpleParser {
                         XFA_PacketType ePacketID);
 
   std::unique_ptr<CFX_XMLDoc> m_pXMLDoc;
-  UnownedPtr<CFX_XMLParser> m_pXMLParser;  // Owned by |m_pXMLDoc|
   RetainPtr<CFX_SeekableStreamProxy> m_pStream;
   RetainPtr<IFX_SeekableStream> m_pFileRead;
   UnownedPtr<CXFA_Document> m_pFactory;