Remove CFX_XMLDoc and call the parser directly

This CL removes the CFX_XMLDoc and calls the CFX_XMLParser directly from
CXFA_DocumentParser.

Change-Id: I4d715cca90cd15b5e1d79827e0bb2781e873e371
Reviewed-on: https://pdfium-review.googlesource.com/30251
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>
Commit-Queue: dsinclair <dsinclair@chromium.org>

7 files changed, 21 insertions, 77 deletions

diff --git a/BUILD.gn b/BUILD.gn
index 7599b77c6f..87de013c89 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -953,8 +953,6 @@ jumbo_static_library("fxcrt") {
       "core/fxcrt/xml/cfx_xmlattributenode.h",
       "core/fxcrt/xml/cfx_xmlchardata.cpp",
       "core/fxcrt/xml/cfx_xmlchardata.h",
-      "core/fxcrt/xml/cfx_xmldoc.cpp",
-      "core/fxcrt/xml/cfx_xmldoc.h",
       "core/fxcrt/xml/cfx_xmlelement.cpp",
       "core/fxcrt/xml/cfx_xmlelement.h",
       "core/fxcrt/xml/cfx_xmlinstruction.cpp",
diff --git a/core/fxcrt/xml/cfx_xmldoc.cpp b/core/fxcrt/xml/cfx_xmldoc.cpp
deleted file mode 100644
index c14255d432..0000000000
--- a/core/fxcrt/xml/cfx_xmldoc.cpp
+++ /dev/null
@@ -1,31 +0,0 @@
-// Copyright 2017 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 "core/fxcrt/xml/cfx_xmldoc.h"
-
-#include <utility>
-#include <vector>
-
-#include "core/fxcrt/fx_codepage.h"
-#include "core/fxcrt/xml/cfx_xmlchardata.h"
-#include "core/fxcrt/xml/cfx_xmlelement.h"
-#include "core/fxcrt/xml/cfx_xmlinstruction.h"
-#include "core/fxcrt/xml/cfx_xmlnode.h"
-#include "core/fxcrt/xml/cfx_xmltext.h"
-#include "third_party/base/ptr_util.h"
-#include "third_party/base/stl_util.h"
-
-CFX_XMLDoc::CFX_XMLDoc() : m_pRoot(pdfium::MakeUnique<CFX_XMLNode>()) {
-  m_pRoot->AppendChild(new CFX_XMLInstruction(L"xml"));
-}
-
-CFX_XMLDoc::~CFX_XMLDoc() {}
-
-bool CFX_XMLDoc::Load(const RetainPtr<CFX_SeekableStreamProxy>& pStream) {
-  ASSERT(pStream);
-  CFX_XMLParser parser(m_pRoot.get(), pStream);
-  return parser.Parse();
-}
diff --git a/core/fxcrt/xml/cfx_xmldoc.h b/core/fxcrt/xml/cfx_xmldoc.h
deleted file mode 100644
index 915acd325d..0000000000
--- a/core/fxcrt/xml/cfx_xmldoc.h
+++ /dev/null
@@ -1,31 +0,0 @@
-// Copyright 2017 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 CORE_FXCRT_XML_CFX_XMLDOC_H_
-#define CORE_FXCRT_XML_CFX_XMLDOC_H_
-
-#include <memory>
-#include <utility>
-
-#include "core/fxcrt/cfx_seekablestreamproxy.h"
-#include "core/fxcrt/retain_ptr.h"
-#include "core/fxcrt/xml/cfx_xmlnode.h"
-#include "core/fxcrt/xml/cfx_xmlparser.h"
-
-class CFX_XMLDoc {
- public:
-  CFX_XMLDoc();
-  ~CFX_XMLDoc();
-
-  bool Load(const RetainPtr<CFX_SeekableStreamProxy>& pStream);
-
-  std::unique_ptr<CFX_XMLNode> GetTree() { return std::move(m_pRoot); }
-
- private:
-  std::unique_ptr<CFX_XMLNode> m_pRoot;
-};
-
-#endif  // CORE_FXCRT_XML_CFX_XMLDOC_H_
diff --git a/testing/libfuzzer/pdf_xml_fuzzer.cc b/testing/libfuzzer/pdf_xml_fuzzer.cc
index 2ad57cf8bc..a8af779259 100644
--- a/testing/libfuzzer/pdf_xml_fuzzer.cc
+++ b/testing/libfuzzer/pdf_xml_fuzzer.cc
@@ -9,7 +9,6 @@
 #include "core/fxcrt/cfx_seekablestreamproxy.h"
 #include "core/fxcrt/fx_safe_types.h"
 #include "core/fxcrt/fx_system.h"
-#include "core/fxcrt/xml/cfx_xmldoc.h"
 #include "core/fxcrt/xml/cfx_xmlnode.h"
 #include "core/fxcrt/xml/cfx_xmlparser.h"
 #include "third_party/base/ptr_util.h"
@@ -22,11 +21,12 @@ 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);
-  CFX_XMLDoc doc;
-  if (!doc.Load(stream))
+
+  auto root = pdfium::MakeUnique<CFX_XMLNode>();
+  CFX_XMLParser parser(root.get(), stream);
+  if (!parser.Parse())
     return 0;
 
-  auto root = doc.GetTree();
   for (CFX_XMLNode* pXMLNode = root->GetFirstChild(); pXMLNode;
        pXMLNode = pXMLNode->GetNextSibling()) {
     if (pXMLNode->GetType() == FX_XMLNODE_Element)
diff --git a/xfa/fxfa/parser/cxfa_dataexporter.cpp b/xfa/fxfa/parser/cxfa_dataexporter.cpp
index 7773aeb190..1418d030b5 100644
--- a/xfa/fxfa/parser/cxfa_dataexporter.cpp
+++ b/xfa/fxfa/parser/cxfa_dataexporter.cpp
@@ -7,7 +7,6 @@
 #include "xfa/fxfa/parser/cxfa_dataexporter.h"
 
 #include "core/fxcrt/fx_codepage.h"
-#include "core/fxcrt/xml/cfx_xmldoc.h"
 #include "core/fxcrt/xml/cfx_xmlelement.h"
 #include "core/fxcrt/xml/cfx_xmlnode.h"
 #include "third_party/base/stl_util.h"
diff --git a/xfa/fxfa/parser/cxfa_document_parser.cpp b/xfa/fxfa/parser/cxfa_document_parser.cpp
index 707dfc11d5..c2cc1deb88 100644
--- a/xfa/fxfa/parser/cxfa_document_parser.cpp
+++ b/xfa/fxfa/parser/cxfa_document_parser.cpp
@@ -14,7 +14,6 @@
 #include "core/fxcrt/fx_codepage.h"
 #include "core/fxcrt/fx_extension.h"
 #include "core/fxcrt/xml/cfx_xmlchardata.h"
-#include "core/fxcrt/xml/cfx_xmldoc.h"
 #include "core/fxcrt/xml/cfx_xmlelement.h"
 #include "core/fxcrt/xml/cfx_xmlinstruction.h"
 #include "core/fxcrt/xml/cfx_xmlnode.h"
@@ -345,11 +344,10 @@ bool CXFA_DocumentParser::Parse(const RetainPtr<IFX_SeekableStream>& pStream,
     pStreamProxy->SetCodePage(FX_CODEPAGE_UTF8);
   }
 
-  CFX_XMLDoc doc;
-  if (!doc.Load(pStreamProxy))
+  m_pNodeTree = LoadXML(pStreamProxy);
+  if (!m_pNodeTree)
     return false;
 
-  m_pNodeTree = doc.GetTree();
   m_pRootNode = ParseAsXDPPacket(GetDocumentNode(m_pNodeTree.get()), ePacketID);
   return !!m_pRootNode;
 }
@@ -357,13 +355,21 @@ bool CXFA_DocumentParser::Parse(const RetainPtr<IFX_SeekableStream>& pStream,
 CFX_XMLNode* CXFA_DocumentParser::ParseXMLData(const ByteString& wsXML) {
   auto pStream = pdfium::MakeRetain<CFX_SeekableStreamProxy>(
       const_cast<uint8_t*>(wsXML.raw_str()), wsXML.GetLength());
-  CFX_XMLDoc doc;
-  if (doc.Load(pStream))
-    m_pNodeTree = doc.GetTree();
-
+  m_pNodeTree = LoadXML(pStream);
   return m_pNodeTree ? GetDocumentNode(m_pNodeTree.get()) : nullptr;
 }
 
+std::unique_ptr<CFX_XMLNode> CXFA_DocumentParser::LoadXML(
+    const RetainPtr<CFX_SeekableStreamProxy>& pStream) {
+  ASSERT(pStream);
+
+  auto root = pdfium::MakeUnique<CFX_XMLNode>();
+  root->AppendChild(new CFX_XMLInstruction(L"xml"));
+
+  CFX_XMLParser parser(root.get(), pStream);
+  return parser.Parse() ? std::move(root) : nullptr;
+}
+
 void CXFA_DocumentParser::ConstructXFANode(CXFA_Node* pXFANode,
                                            CFX_XMLNode* pXMLNode) {
   XFA_PacketType ePacketID = pXFANode->GetPacketType();
diff --git a/xfa/fxfa/parser/cxfa_document_parser.h b/xfa/fxfa/parser/cxfa_document_parser.h
index e41bf7bf07..8899494c79 100644
--- a/xfa/fxfa/parser/cxfa_document_parser.h
+++ b/xfa/fxfa/parser/cxfa_document_parser.h
@@ -39,6 +39,9 @@ class CXFA_DocumentParser {
   void SetFactory(CXFA_Document* pFactory);
 
  private:
+  std::unique_ptr<CFX_XMLNode> LoadXML(
+      const RetainPtr<CFX_SeekableStreamProxy>& pStream);
+
   CXFA_Node* ParseAsXDPPacket(CFX_XMLNode* pXMLDocumentNode,
                               XFA_PacketType ePacketID);
   CXFA_Node* ParseAsXDPPacket_XDP(CFX_XMLNode* pXMLDocumentNode);