Cleanup the tagged code

This CL removes the IPDF_Struct* classes in favour of their only implementation.
The tagged code was split out into files matching the classes they contain. The
friendship between CPDF_StructTree and CPDF_StructElement was broken in favour
of accessors.

Bug: pdfium:672
Change-Id: Iade83b608fb7168b3b0f41338d10d5fd8ab91a6e
Reviewed-on: https://pdfium-review.googlesource.com/3820
Reviewed-by: Tom Sepez <tsepez@chromium.org>
Reviewed-by: Nicolás Peña <npm@chromium.org>
Commit-Queue: dsinclair <dsinclair@chromium.org>

7 files changed, 330 insertions, 370 deletions

diff --git a/BUILD.gn b/BUILD.gn
index 46e020fbb2..67be8cde74 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -325,6 +325,10 @@ static_library("fpdfdoc") {
     "core/fpdfdoc/cpdf_occontext.h",
     "core/fpdfdoc/cpdf_pagelabel.cpp",
     "core/fpdfdoc/cpdf_pagelabel.h",
+    "core/fpdfdoc/cpdf_structelement.cpp",
+    "core/fpdfdoc/cpdf_structelement.h",
+    "core/fpdfdoc/cpdf_structtree.cpp",
+    "core/fpdfdoc/cpdf_structtree.h",
     "core/fpdfdoc/cpdf_variabletext.cpp",
     "core/fpdfdoc/cpdf_variabletext.h",
     "core/fpdfdoc/cpdf_viewerpreferences.cpp",
@@ -353,11 +357,8 @@ static_library("fpdfdoc") {
     "core/fpdfdoc/csection.h",
     "core/fpdfdoc/ctypeset.cpp",
     "core/fpdfdoc/ctypeset.h",
-    "core/fpdfdoc/doc_tagged.cpp",
-    "core/fpdfdoc/fpdf_tagged.h",
     "core/fpdfdoc/ipdf_formnotify.h",
     "core/fpdfdoc/ipvt_fontmap.h",
-    "core/fpdfdoc/tagged_int.h",
   ]
   configs += [ ":pdfium_core_config" ]
   deps = [
diff --git a/core/fpdfdoc/doc_tagged.cpp b/core/fpdfdoc/cpdf_structelement.cpp
index 418fab411a..418f75b3a9 100644
--- a/core/fpdfdoc/doc_tagged.cpp
+++ b/core/fpdfdoc/cpdf_structelement.cpp
@@ -1,33 +1,49 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
+// 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 <map>
-#include <memory>
-#include <utility>
+#include "core/fpdfdoc/cpdf_structelement.h"
 
 #include "core/fpdfapi/parser/cpdf_array.h"
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
-#include "core/fpdfapi/parser/cpdf_document.h"
 #include "core/fpdfapi/parser/cpdf_name.h"
 #include "core/fpdfapi/parser/cpdf_number.h"
+#include "core/fpdfapi/parser/cpdf_object.h"
 #include "core/fpdfapi/parser/cpdf_reference.h"
 #include "core/fpdfapi/parser/cpdf_stream.h"
-#include "core/fpdfdoc/cpdf_numbertree.h"
-#include "core/fpdfdoc/fpdf_tagged.h"
-#include "core/fpdfdoc/tagged_int.h"
-#include "third_party/base/ptr_util.h"
+#include "core/fpdfdoc/cpdf_structtree.h"
+#include "third_party/base/stl_util.h"
 
 namespace {
 
 const int nMaxRecursion = 32;
 
-bool IsTagged(const CPDF_Document* pDoc) {
-  CPDF_Dictionary* pCatalog = pDoc->GetRoot();
-  CPDF_Dictionary* pMarkInfo = pCatalog->GetDictFor("MarkInfo");
-  return pMarkInfo && pMarkInfo->GetIntegerFor("Marked");
+CPDF_Dictionary* FindAttrDict(CPDF_Object* pAttrs,
+                              const CFX_ByteStringC& owner,
+                              float nLevel = 0.0F) {
+  if (nLevel > nMaxRecursion)
+    return nullptr;
+  if (!pAttrs)
+    return nullptr;
+
+  CPDF_Dictionary* pDict = nullptr;
+  if (pAttrs->IsDictionary()) {
+    pDict = pAttrs->AsDictionary();
+  } else if (CPDF_Stream* pStream = pAttrs->AsStream()) {
+    pDict = pStream->GetDict();
+  } else if (CPDF_Array* pArray = pAttrs->AsArray()) {
+    for (uint32_t i = 0; i < pArray->GetCount(); i++) {
+      CPDF_Object* pElement = pArray->GetDirectObjectAt(i);
+      pDict = FindAttrDict(pElement, owner, nLevel + 1);
+      if (pDict)
+        return pDict;
+    }
+  }
+  if (pDict && pDict->GetStringFor("O") == owner)
+    return pDict;
+  return nullptr;
 }
 
 }  // namespace
@@ -41,134 +57,7 @@ CPDF_StructKid::CPDF_StructKid()
 
 CPDF_StructKid::CPDF_StructKid(const CPDF_StructKid& that) = default;
 
-CPDF_StructKid::~CPDF_StructKid() {}
-
-// static
-std::unique_ptr<IPDF_StructTree> IPDF_StructTree::LoadPage(
-    const CPDF_Document* pDoc,
-    const CPDF_Dictionary* pPageDict) {
-  if (!IsTagged(pDoc))
-    return nullptr;
-
-  auto pTree = pdfium::MakeUnique<CPDF_StructTree>(pDoc);
-  pTree->LoadPageTree(pPageDict);
-  return std::move(pTree);
-}
-
-CPDF_StructTree::CPDF_StructTree(const CPDF_Document* pDoc)
-    : m_pTreeRoot(pDoc->GetRoot()->GetDictFor("StructTreeRoot")),
-      m_pRoleMap(m_pTreeRoot ? m_pTreeRoot->GetDictFor("RoleMap") : nullptr),
-      m_pPage(nullptr) {}
-
-CPDF_StructTree::~CPDF_StructTree() {}
-
-int CPDF_StructTree::CountTopElements() const {
-  return pdfium::CollectionSize<int>(m_Kids);
-}
-
-IPDF_StructElement* CPDF_StructTree::GetTopElement(int i) const {
-  return m_Kids[i].Get();
-}
-
-void CPDF_StructTree::LoadPageTree(const CPDF_Dictionary* pPageDict) {
-  m_pPage = pPageDict;
-  if (!m_pTreeRoot)
-    return;
-
-  CPDF_Object* pKids = m_pTreeRoot->GetDirectObjectFor("K");
-  if (!pKids)
-    return;
-
-  uint32_t dwKids = 0;
-  if (pKids->IsDictionary())
-    dwKids = 1;
-  else if (CPDF_Array* pArray = pKids->AsArray())
-    dwKids = pArray->GetCount();
-  else
-    return;
-
-  m_Kids.clear();
-  m_Kids.resize(dwKids);
-  CPDF_Dictionary* pParentTree = m_pTreeRoot->GetDictFor("ParentTree");
-  if (!pParentTree)
-    return;
-
-  CPDF_NumberTree parent_tree(pParentTree);
-  int parents_id = pPageDict->GetIntegerFor("StructParents", -1);
-  if (parents_id < 0)
-    return;
-
-  CPDF_Array* pParentArray = ToArray(parent_tree.LookupValue(parents_id));
-  if (!pParentArray)
-    return;
-
-  std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>> element_map;
-  for (size_t i = 0; i < pParentArray->GetCount(); i++) {
-    if (CPDF_Dictionary* pParent = pParentArray->GetDictAt(i))
-      AddPageNode(pParent, &element_map);
-  }
-}
-
-CFX_RetainPtr<CPDF_StructElement> CPDF_StructTree::AddPageNode(
-    CPDF_Dictionary* pDict,
-    std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>>* map,
-    int nLevel) {
-  if (nLevel > nMaxRecursion)
-    return nullptr;
-
-  auto it = map->find(pDict);
-  if (it != map->end())
-    return it->second;
-
-  auto pElement = pdfium::MakeRetain<CPDF_StructElement>(this, nullptr, pDict);
-  (*map)[pDict] = pElement;
-  CPDF_Dictionary* pParent = pDict->GetDictFor("P");
-  if (!pParent || pParent->GetStringFor("Type") == "StructTreeRoot") {
-    if (!AddTopLevelNode(pDict, pElement))
-      map->erase(pDict);
-    return pElement;
-  }
-
-  CFX_RetainPtr<CPDF_StructElement> pParentElement =
-      AddPageNode(pParent, map, nLevel + 1);
-  bool bSave = false;
-  for (CPDF_StructKid& kid : *pParentElement->GetKids()) {
-    if (kid.m_Type == CPDF_StructKid::Element && kid.m_pDict == pDict) {
-      kid.m_pElement = pElement;
-      bSave = true;
-    }
-  }
-  if (!bSave)
-    map->erase(pDict);
-  return pElement;
-}
-
-bool CPDF_StructTree::AddTopLevelNode(
-    CPDF_Dictionary* pDict,
-    const CFX_RetainPtr<CPDF_StructElement>& pElement) {
-  CPDF_Object* pObj = m_pTreeRoot->GetDirectObjectFor("K");
-  if (!pObj)
-    return false;
-
-  if (pObj->IsDictionary()) {
-    if (pObj->GetObjNum() != pDict->GetObjNum())
-      return false;
-    m_Kids[0] = pElement;
-  }
-  if (CPDF_Array* pTopKids = pObj->AsArray()) {
-    bool bSave = false;
-    for (size_t i = 0; i < pTopKids->GetCount(); i++) {
-      CPDF_Reference* pKidRef = ToReference(pTopKids->GetObjectAt(i));
-      if (pKidRef && pKidRef->GetRefObjNum() == pDict->GetObjNum()) {
-        m_Kids[i] = pElement;
-        bSave = true;
-      }
-    }
-    if (!bSave)
-      return false;
-  }
-  return true;
-}
+CPDF_StructKid::~CPDF_StructKid() = default;
 
 CPDF_StructElement::CPDF_StructElement(CPDF_StructTree* pTree,
                                        CPDF_StructElement* pParent,
@@ -177,43 +66,26 @@ CPDF_StructElement::CPDF_StructElement(CPDF_StructTree* pTree,
       m_pParent(pParent),
       m_pDict(pDict),
       m_Type(pDict->GetStringFor("S")) {
-  if (pTree->m_pRoleMap) {
-    CFX_ByteString mapped = pTree->m_pRoleMap->GetStringFor(m_Type);
+  if (pTree->GetRoleMap()) {
+    CFX_ByteString mapped = pTree->GetRoleMap()->GetStringFor(m_Type);
     if (!mapped.IsEmpty())
       m_Type = mapped;
   }
   LoadKids(pDict);
 }
 
-IPDF_StructTree* CPDF_StructElement::GetTree() const {
-  return m_pTree;
-}
-
-const CFX_ByteString& CPDF_StructElement::GetType() const {
-  return m_Type;
-}
-
-IPDF_StructElement* CPDF_StructElement::GetParent() const {
-  return m_pParent;
-}
-
-CPDF_Dictionary* CPDF_StructElement::GetDict() const {
-  return m_pDict;
-}
+CPDF_StructElement::~CPDF_StructElement() = default;
 
 int CPDF_StructElement::CountKids() const {
   return pdfium::CollectionSize<int>(m_Kids);
 }
 
-IPDF_StructElement* CPDF_StructElement::GetKidIfElement(int index) const {
-  if (m_Kids[index].m_Type != CPDF_StructKid::Element)
-    return nullptr;
-
-  return m_Kids[index].m_pElement.Get();
+CPDF_StructElement* CPDF_StructElement::GetKidIfElement(int index) const {
+  return m_Kids[index].m_Type == CPDF_StructKid::Element
+             ? m_Kids[index].m_pElement.Get()
+             : nullptr;
 }
 
-CPDF_StructElement::~CPDF_StructElement() {}
-
 void CPDF_StructElement::LoadKids(CPDF_Dictionary* pDict) {
   CPDF_Object* pObj = pDict->GetObjectFor("Pg");
   uint32_t PageObjNum = 0;
@@ -231,11 +103,13 @@ void CPDF_StructElement::LoadKids(CPDF_Dictionary* pDict) {
       CPDF_Object* pKid = pArray->GetDirectObjectAt(i);
       LoadKid(PageObjNum, pKid, &m_Kids[i]);
     }
-  } else {
-    m_Kids.resize(1);
-    LoadKid(PageObjNum, pKids, &m_Kids[0]);
+    return;
   }
+
+  m_Kids.resize(1);
+  LoadKid(PageObjNum, pKids, &m_Kids[0]);
 }
+
 void CPDF_StructElement::LoadKid(uint32_t PageObjNum,
                                  CPDF_Object* pKidObj,
                                  CPDF_StructKid* pKid) {
@@ -244,9 +118,9 @@ void CPDF_StructElement::LoadKid(uint32_t PageObjNum,
     return;
 
   if (pKidObj->IsNumber()) {
-    if (m_pTree->m_pPage && m_pTree->m_pPage->GetObjNum() != PageObjNum) {
+    if (m_pTree->GetPage() && m_pTree->GetPage()->GetObjNum() != PageObjNum)
       return;
-    }
+
     pKid->m_Type = CPDF_StructKid::PageContent;
     pKid->m_ContentId = pKidObj->GetInteger();
     pKid->m_PageObjNum = PageObjNum;
@@ -256,96 +130,70 @@ void CPDF_StructElement::LoadKid(uint32_t PageObjNum,
   CPDF_Dictionary* pKidDict = pKidObj->AsDictionary();
   if (!pKidDict)
     return;
-
   if (CPDF_Reference* pRef = ToReference(pKidDict->GetObjectFor("Pg")))
     PageObjNum = pRef->GetRefObjNum();
 
   CFX_ByteString type = pKidDict->GetStringFor("Type");
+  if ((type == "MCR" || type == "OBJR") && m_pTree->GetPage() &&
+      m_pTree->GetPage()->GetObjNum() != PageObjNum) {
+    return;
+  }
+
   if (type == "MCR") {
-    if (m_pTree->m_pPage && m_pTree->m_pPage->GetObjNum() != PageObjNum) {
-      return;
-    }
     pKid->m_Type = CPDF_StructKid::StreamContent;
     CPDF_Reference* pRef = ToReference(pKidDict->GetObjectFor("Stm"));
     pKid->m_RefObjNum = pRef ? pRef->GetRefObjNum() : 0;
     pKid->m_PageObjNum = PageObjNum;
     pKid->m_ContentId = pKidDict->GetIntegerFor("MCID");
-  } else if (type == "OBJR") {
-    if (m_pTree->m_pPage && m_pTree->m_pPage->GetObjNum() != PageObjNum) {
-      return;
-    }
+    return;
+  }
+
+  if (type == "OBJR") {
     pKid->m_Type = CPDF_StructKid::Object;
     CPDF_Reference* pObj = ToReference(pKidDict->GetObjectFor("Obj"));
     pKid->m_RefObjNum = pObj ? pObj->GetRefObjNum() : 0;
     pKid->m_PageObjNum = PageObjNum;
-  } else {
-    pKid->m_Type = CPDF_StructKid::Element;
-    pKid->m_pDict = pKidDict;
-    if (!m_pTree->m_pPage) {
-      pKid->m_pElement =
-          pdfium::MakeRetain<CPDF_StructElement>(m_pTree, this, pKidDict);
-    } else {
-      pKid->m_pElement = nullptr;
-    }
+    return;
   }
-}
-static CPDF_Dictionary* FindAttrDict(CPDF_Object* pAttrs,
-                                     const CFX_ByteStringC& owner,
-                                     float nLevel = 0.0F) {
-  if (nLevel > nMaxRecursion)
-    return nullptr;
-  if (!pAttrs)
-    return nullptr;
 
-  CPDF_Dictionary* pDict = nullptr;
-  if (pAttrs->IsDictionary()) {
-    pDict = pAttrs->AsDictionary();
-  } else if (CPDF_Stream* pStream = pAttrs->AsStream()) {
-    pDict = pStream->GetDict();
-  } else if (CPDF_Array* pArray = pAttrs->AsArray()) {
-    for (uint32_t i = 0; i < pArray->GetCount(); i++) {
-      CPDF_Object* pElement = pArray->GetDirectObjectAt(i);
-      pDict = FindAttrDict(pElement, owner, nLevel + 1);
-      if (pDict)
-        return pDict;
-    }
+  pKid->m_Type = CPDF_StructKid::Element;
+  pKid->m_pDict = pKidDict;
+  if (m_pTree->GetPage()) {
+    pKid->m_pElement = nullptr;
+    return;
   }
-  if (pDict && pDict->GetStringFor("O") == owner)
-    return pDict;
-  return nullptr;
+
+  pKid->m_pElement =
+      pdfium::MakeRetain<CPDF_StructElement>(m_pTree, this, pKidDict);
 }
+
 CPDF_Object* CPDF_StructElement::GetAttr(const CFX_ByteStringC& owner,
                                          const CFX_ByteStringC& name,
                                          bool bInheritable,
                                          float fLevel) {
-  if (fLevel > nMaxRecursion) {
+  if (fLevel > nMaxRecursion)
     return nullptr;
-  }
+
   if (bInheritable) {
-    CPDF_Object* pAttr = GetAttr(owner, name, false);
-    if (pAttr) {
+    if (CPDF_Object* pAttr = GetAttr(owner, name, false))
       return pAttr;
-    }
-    if (!m_pParent) {
+    if (!m_pParent)
       return nullptr;
-    }
     return m_pParent->GetAttr(owner, name, true, fLevel + 1);
   }
-  CPDF_Object* pA = m_pDict->GetDirectObjectFor("A");
-  if (pA) {
-    CPDF_Dictionary* pAttrDict = FindAttrDict(pA, owner);
-    if (pAttrDict) {
-      CPDF_Object* pAttr = pAttrDict->GetDirectObjectFor(CFX_ByteString(name));
-      if (pAttr) {
-        return pAttr;
-      }
+
+  if (CPDF_Object* pA = m_pDict->GetDirectObjectFor("A")) {
+    if (CPDF_Dictionary* dict = FindAttrDict(pA, owner)) {
+      if (CPDF_Object* attr = dict->GetDirectObjectFor(CFX_ByteString(name)))
+        return attr;
     }
   }
+
   CPDF_Object* pC = m_pDict->GetDirectObjectFor("C");
   if (!pC)
     return nullptr;
 
-  CPDF_Dictionary* pClassMap = m_pTree->m_pTreeRoot->GetDictFor("ClassMap");
+  CPDF_Dictionary* pClassMap = m_pTree->GetTreeRoot()->GetDictFor("ClassMap");
   if (!pClassMap)
     return nullptr;
 
@@ -358,12 +206,14 @@ CPDF_Object* CPDF_StructElement::GetAttr(const CFX_ByteStringC& owner,
     }
     return nullptr;
   }
+
   CFX_ByteString class_name = pC->GetString();
   CPDF_Dictionary* pClassDict = pClassMap->GetDictFor(class_name);
   if (pClassDict && pClassDict->GetStringFor("O") == owner)
     return pClassDict->GetDirectObjectFor(CFX_ByteString(name));
   return nullptr;
 }
+
 CPDF_Object* CPDF_StructElement::GetAttr(const CFX_ByteStringC& owner,
                                          const CFX_ByteStringC& name,
                                          bool bInheritable,
@@ -372,11 +222,11 @@ CPDF_Object* CPDF_StructElement::GetAttr(const CFX_ByteStringC& owner,
   CPDF_Array* pArray = ToArray(pAttr);
   if (!pArray || subindex == -1)
     return pAttr;
-
   if (subindex >= static_cast<int>(pArray->GetCount()))
     return pAttr;
   return pArray->GetDirectObjectAt(subindex);
 }
+
 CFX_ByteString CPDF_StructElement::GetName(const CFX_ByteStringC& owner,
                                            const CFX_ByteStringC& name,
                                            const CFX_ByteStringC& default_value,
@@ -396,10 +246,11 @@ FX_ARGB CPDF_StructElement::GetColor(const CFX_ByteStringC& owner,
   CPDF_Array* pArray = ToArray(GetAttr(owner, name, bInheritable, subindex));
   if (!pArray)
     return default_value;
-  return 0xff000000 | ((int)(pArray->GetNumberAt(0) * 255) << 16) |
-         ((int)(pArray->GetNumberAt(1) * 255) << 8) |
-         (int)(pArray->GetNumberAt(2) * 255);
+  return 0xff000000 | (static_cast<int>(pArray->GetNumberAt(0) * 255) << 16) |
+         (static_cast<int>(pArray->GetNumberAt(1) * 255) << 8) |
+         static_cast<int>(pArray->GetNumberAt(2) * 255);
 }
+
 float CPDF_StructElement::GetNumber(const CFX_ByteStringC& owner,
                                     const CFX_ByteStringC& name,
                                     float default_value,
@@ -408,6 +259,7 @@ float CPDF_StructElement::GetNumber(const CFX_ByteStringC& owner,
   CPDF_Object* pAttr = GetAttr(owner, name, bInheritable, subindex);
   return ToNumber(pAttr) ? pAttr->GetNumber() : default_value;
 }
+
 int CPDF_StructElement::GetInteger(const CFX_ByteStringC& owner,
                                    const CFX_ByteStringC& name,
                                    int default_value,
diff --git a/core/fpdfdoc/tagged_int.h b/core/fpdfdoc/cpdf_structelement.h
index cafcbd42aa..b227397431 100644
--- a/core/fpdfdoc/tagged_int.h
+++ b/core/fpdfdoc/cpdf_structelement.h
@@ -1,23 +1,25 @@
-// Copyright 2014 PDFium Authors. All rights reserved.
+// 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_FPDFDOC_TAGGED_INT_H_
-#define CORE_FPDFDOC_TAGGED_INT_H_
+#ifndef CORE_FPDFDOC_CPDF_STRUCTELEMENT_H_
+#define CORE_FPDFDOC_CPDF_STRUCTELEMENT_H_
 
-#include <map>
-#include <memory>
 #include <vector>
 
-#include "core/fpdfdoc/fpdf_tagged.h"
 #include "core/fxcrt/cfx_retain_ptr.h"
-#include "third_party/base/stl_util.h"
+#include "core/fxcrt/fx_string.h"
+#include "core/fxge/fx_dib.h"
 
+class CPDF_Dictionary;
+class CPDF_Object;
 class CPDF_StructElement;
+class CPDF_StructTree;
 
-struct CPDF_StructKid {
+class CPDF_StructKid {
+ public:
   CPDF_StructKid();
   CPDF_StructKid(const CPDF_StructKid& that);
   ~CPDF_StructKid();
@@ -31,69 +33,42 @@ struct CPDF_StructKid {
   uint32_t m_ContentId;   // For PageContent, StreamContent.
 };
 
-class CPDF_StructTree final : public IPDF_StructTree {
- public:
-  explicit CPDF_StructTree(const CPDF_Document* pDoc);
-  ~CPDF_StructTree() override;
-
-  // IPDF_StructTree:
-  int CountTopElements() const override;
-  IPDF_StructElement* GetTopElement(int i) const override;
-
-  void LoadPageTree(const CPDF_Dictionary* pPageDict);
-  CFX_RetainPtr<CPDF_StructElement> AddPageNode(
-      CPDF_Dictionary* pElement,
-      std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>>* map,
-      int nLevel = 0);
-  bool AddTopLevelNode(CPDF_Dictionary* pDict,
-                       const CFX_RetainPtr<CPDF_StructElement>& pElement);
-
- protected:
-  const CPDF_Dictionary* const m_pTreeRoot;
-  const CPDF_Dictionary* const m_pRoleMap;
-  const CPDF_Dictionary* m_pPage;
-  std::vector<CFX_RetainPtr<CPDF_StructElement>> m_Kids;
-
-  friend class CPDF_StructElement;
-};
-
-class CPDF_StructElement final : public CFX_Retainable,
-                                 public IPDF_StructElement {
+class CPDF_StructElement : public CFX_Retainable {
  public:
   template <typename T, typename... Args>
   friend CFX_RetainPtr<T> pdfium::MakeRetain(Args&&... args);
 
-  // IPDF_StructElement
-  IPDF_StructTree* GetTree() const override;
-  const CFX_ByteString& GetType() const override;
-  IPDF_StructElement* GetParent() const override;
-  CPDF_Dictionary* GetDict() const override;
-  int CountKids() const override;
-  IPDF_StructElement* GetKidIfElement(int index) const override;
+  CPDF_StructTree* GetTree() const { return m_pTree; }
+  const CFX_ByteString& GetType() const { return m_Type; }
+  CPDF_StructElement* GetParent() const { return m_pParent; }
+  CPDF_Dictionary* GetDict() const { return m_pDict; }
+
+  int CountKids() const;
+  CPDF_StructElement* GetKidIfElement(int index) const;
   CPDF_Object* GetAttr(const CFX_ByteStringC& owner,
                        const CFX_ByteStringC& name,
                        bool bInheritable = false,
-                       float fLevel = 0.0F) override;
+                       float fLevel = 0.0F);
   CFX_ByteString GetName(const CFX_ByteStringC& owner,
                          const CFX_ByteStringC& name,
                          const CFX_ByteStringC& default_value,
                          bool bInheritable = false,
-                         int subindex = -1) override;
+                         int subindex = -1);
   FX_ARGB GetColor(const CFX_ByteStringC& owner,
                    const CFX_ByteStringC& name,
                    FX_ARGB default_value,
                    bool bInheritable = false,
-                   int subindex = -1) override;
+                   int subindex = -1);
   float GetNumber(const CFX_ByteStringC& owner,
                   const CFX_ByteStringC& name,
                   float default_value,
                   bool bInheritable = false,
-                  int subindex = -1) override;
+                  int subindex = -1);
   int GetInteger(const CFX_ByteStringC& owner,
                  const CFX_ByteStringC& name,
                  int default_value,
                  bool bInheritable = false,
-                 int subindex = -1) override;
+                 int subindex = -1);
 
   std::vector<CPDF_StructKid>* GetKids() { return &m_Kids; }
   void LoadKids(CPDF_Dictionary* pDict);
@@ -116,4 +91,4 @@ class CPDF_StructElement final : public CFX_Retainable,
   std::vector<CPDF_StructKid> m_Kids;
 };
 
-#endif  // CORE_FPDFDOC_TAGGED_INT_H_
+#endif  // CORE_FPDFDOC_CPDF_STRUCTELEMENT_H_
diff --git a/core/fpdfdoc/cpdf_structtree.cpp b/core/fpdfdoc/cpdf_structtree.cpp
new file mode 100644
index 0000000000..51ad2c775c
--- /dev/null
+++ b/core/fpdfdoc/cpdf_structtree.cpp
@@ -0,0 +1,155 @@
+// 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/fpdfdoc/cpdf_structtree.h"
+
+#include "core/fpdfapi/parser/cpdf_array.h"
+#include "core/fpdfapi/parser/cpdf_document.h"
+#include "core/fpdfapi/parser/cpdf_number.h"
+#include "core/fpdfapi/parser/cpdf_reference.h"
+#include "core/fpdfdoc/cpdf_numbertree.h"
+#include "core/fpdfdoc/cpdf_structelement.h"
+#include "third_party/base/stl_util.h"
+
+namespace {
+
+const int nMaxRecursion = 32;
+
+bool IsTagged(const CPDF_Document* pDoc) {
+  CPDF_Dictionary* pCatalog = pDoc->GetRoot();
+  CPDF_Dictionary* pMarkInfo = pCatalog->GetDictFor("MarkInfo");
+  return pMarkInfo && pMarkInfo->GetIntegerFor("Marked");
+}
+
+}  // namespace
+
+// static
+std::unique_ptr<CPDF_StructTree> CPDF_StructTree::LoadPage(
+    const CPDF_Document* pDoc,
+    const CPDF_Dictionary* pPageDict) {
+  if (!IsTagged(pDoc))
+    return nullptr;
+
+  auto pTree = pdfium::MakeUnique<CPDF_StructTree>(pDoc);
+  pTree->LoadPageTree(pPageDict);
+  return pTree;
+}
+
+CPDF_StructTree::CPDF_StructTree(const CPDF_Document* pDoc)
+    : m_pTreeRoot(pDoc->GetRoot()->GetDictFor("StructTreeRoot")),
+      m_pRoleMap(m_pTreeRoot ? m_pTreeRoot->GetDictFor("RoleMap") : nullptr),
+      m_pPage(nullptr) {}
+
+CPDF_StructTree::~CPDF_StructTree() {}
+
+int CPDF_StructTree::CountTopElements() const {
+  return pdfium::CollectionSize<int>(m_Kids);
+}
+
+CPDF_StructElement* CPDF_StructTree::GetTopElement(int i) const {
+  return m_Kids[i].Get();
+}
+
+void CPDF_StructTree::LoadPageTree(const CPDF_Dictionary* pPageDict) {
+  m_pPage = pPageDict;
+  if (!m_pTreeRoot)
+    return;
+
+  CPDF_Object* pKids = m_pTreeRoot->GetDirectObjectFor("K");
+  if (!pKids)
+    return;
+
+  uint32_t dwKids = 0;
+  if (pKids->IsDictionary())
+    dwKids = 1;
+  else if (CPDF_Array* pArray = pKids->AsArray())
+    dwKids = pArray->GetCount();
+  else
+    return;
+
+  m_Kids.clear();
+  m_Kids.resize(dwKids);
+  CPDF_Dictionary* pParentTree = m_pTreeRoot->GetDictFor("ParentTree");
+  if (!pParentTree)
+    return;
+
+  CPDF_NumberTree parent_tree(pParentTree);
+  int parents_id = pPageDict->GetIntegerFor("StructParents", -1);
+  if (parents_id < 0)
+    return;
+
+  CPDF_Array* pParentArray = ToArray(parent_tree.LookupValue(parents_id));
+  if (!pParentArray)
+    return;
+
+  std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>> element_map;
+  for (size_t i = 0; i < pParentArray->GetCount(); i++) {
+    if (CPDF_Dictionary* pParent = pParentArray->GetDictAt(i))
+      AddPageNode(pParent, &element_map);
+  }
+}
+
+CFX_RetainPtr<CPDF_StructElement> CPDF_StructTree::AddPageNode(
+    CPDF_Dictionary* pDict,
+    std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>>* map,
+    int nLevel) {
+  if (nLevel > nMaxRecursion)
+    return nullptr;
+
+  auto it = map->find(pDict);
+  if (it != map->end())
+    return it->second;
+
+  auto pElement = pdfium::MakeRetain<CPDF_StructElement>(this, nullptr, pDict);
+  (*map)[pDict] = pElement;
+  CPDF_Dictionary* pParent = pDict->GetDictFor("P");
+  if (!pParent || pParent->GetStringFor("Type") == "StructTreeRoot") {
+    if (!AddTopLevelNode(pDict, pElement))
+      map->erase(pDict);
+    return pElement;
+  }
+
+  CFX_RetainPtr<CPDF_StructElement> pParentElement =
+      AddPageNode(pParent, map, nLevel + 1);
+  bool bSave = false;
+  for (CPDF_StructKid& kid : *pParentElement->GetKids()) {
+    if (kid.m_Type == CPDF_StructKid::Element && kid.m_pDict == pDict) {
+      kid.m_pElement = pElement;
+      bSave = true;
+    }
+  }
+  if (!bSave)
+    map->erase(pDict);
+  return pElement;
+}
+
+bool CPDF_StructTree::AddTopLevelNode(
+    CPDF_Dictionary* pDict,
+    const CFX_RetainPtr<CPDF_StructElement>& pElement) {
+  CPDF_Object* pObj = m_pTreeRoot->GetDirectObjectFor("K");
+  if (!pObj)
+    return false;
+
+  if (pObj->IsDictionary()) {
+    if (pObj->GetObjNum() != pDict->GetObjNum())
+      return false;
+    m_Kids[0] = pElement;
+  }
+
+  CPDF_Array* pTopKids = pObj->AsArray();
+  if (!pTopKids)
+    return true;
+
+  bool bSave = false;
+  for (size_t i = 0; i < pTopKids->GetCount(); i++) {
+    CPDF_Reference* pKidRef = ToReference(pTopKids->GetObjectAt(i));
+    if (pKidRef && pKidRef->GetRefObjNum() == pDict->GetObjNum()) {
+      m_Kids[i] = pElement;
+      bSave = true;
+    }
+  }
+  return bSave;
+}
diff --git a/core/fpdfdoc/cpdf_structtree.h b/core/fpdfdoc/cpdf_structtree.h
new file mode 100644
index 0000000000..20bf41e7ee
--- /dev/null
+++ b/core/fpdfdoc/cpdf_structtree.h
@@ -0,0 +1,51 @@
+// 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_FPDFDOC_CPDF_STRUCTTREE_H_
+#define CORE_FPDFDOC_CPDF_STRUCTTREE_H_
+
+#include <map>
+#include <memory>
+#include <vector>
+
+#include "core/fxcrt/cfx_retain_ptr.h"
+
+class CPDF_Dictionary;
+class CPDF_Document;
+class CPDF_StructElement;
+
+class CPDF_StructTree {
+ public:
+  static std::unique_ptr<CPDF_StructTree> LoadPage(
+      const CPDF_Document* pDoc,
+      const CPDF_Dictionary* pPageDict);
+
+  explicit CPDF_StructTree(const CPDF_Document* pDoc);
+  ~CPDF_StructTree();
+
+  int CountTopElements() const;
+  CPDF_StructElement* GetTopElement(int i) const;
+
+  void LoadPageTree(const CPDF_Dictionary* pPageDict);
+  CFX_RetainPtr<CPDF_StructElement> AddPageNode(
+      CPDF_Dictionary* pElement,
+      std::map<CPDF_Dictionary*, CFX_RetainPtr<CPDF_StructElement>>* map,
+      int nLevel = 0);
+  bool AddTopLevelNode(CPDF_Dictionary* pDict,
+                       const CFX_RetainPtr<CPDF_StructElement>& pElement);
+
+  const CPDF_Dictionary* GetRoleMap() const { return m_pRoleMap; }
+  const CPDF_Dictionary* GetPage() const { return m_pPage; }
+  const CPDF_Dictionary* GetTreeRoot() const { return m_pTreeRoot; }
+
+ private:
+  const CPDF_Dictionary* const m_pTreeRoot;
+  const CPDF_Dictionary* const m_pRoleMap;
+  const CPDF_Dictionary* m_pPage;
+  std::vector<CFX_RetainPtr<CPDF_StructElement>> m_Kids;
+};
+
+#endif  // CORE_FPDFDOC_CPDF_STRUCTTREE_H_
diff --git a/core/fpdfdoc/fpdf_tagged.h b/core/fpdfdoc/fpdf_tagged.h
deleted file mode 100644
index 5e7b1827c5..0000000000
--- a/core/fpdfdoc/fpdf_tagged.h
+++ /dev/null
@@ -1,75 +0,0 @@
-// Copyright 2014 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_FPDFDOC_FPDF_TAGGED_H_
-#define CORE_FPDFDOC_FPDF_TAGGED_H_
-
-#include <memory>
-
-#include "core/fxge/fx_dib.h"
-
-class CPDF_Dictionary;
-class CPDF_Document;
-class IPDF_StructElement;
-
-class IPDF_StructTree {
- public:
-  static std::unique_ptr<IPDF_StructTree> LoadPage(
-      const CPDF_Document* pDoc,
-      const CPDF_Dictionary* pPageDict);
-
-
-  virtual int CountTopElements() const = 0;
-  virtual IPDF_StructElement* GetTopElement(int i) const = 0;
-
- protected:
-  friend std::default_delete<IPDF_StructTree>;
-  virtual ~IPDF_StructTree() {}
-};
-
-class IPDF_StructElement {
- public:
-  virtual IPDF_StructTree* GetTree() const = 0;
-  virtual const CFX_ByteString& GetType() const = 0;
-  virtual IPDF_StructElement* GetParent() const = 0;
-  virtual CPDF_Dictionary* GetDict() const = 0;
-  virtual int CountKids() const = 0;
-  virtual IPDF_StructElement* GetKidIfElement(int index) const = 0;
-
-  virtual CPDF_Object* GetAttr(const CFX_ByteStringC& owner,
-                               const CFX_ByteStringC& name,
-                               bool bInheritable = false,
-                               float fLevel = 0.0F) = 0;
-
-  virtual CFX_ByteString GetName(const CFX_ByteStringC& owner,
-                                 const CFX_ByteStringC& name,
-                                 const CFX_ByteStringC& default_value,
-                                 bool bInheritable = false,
-                                 int subindex = -1) = 0;
-
-  virtual FX_ARGB GetColor(const CFX_ByteStringC& owner,
-                           const CFX_ByteStringC& name,
-                           FX_ARGB default_value,
-                           bool bInheritable = false,
-                           int subindex = -1) = 0;
-
-  virtual float GetNumber(const CFX_ByteStringC& owner,
-                          const CFX_ByteStringC& name,
-                          float default_value,
-                          bool bInheritable = false,
-                          int subindex = -1) = 0;
-
-  virtual int GetInteger(const CFX_ByteStringC& owner,
-                         const CFX_ByteStringC& name,
-                         int default_value,
-                         bool bInheritable = false,
-                         int subindex = -1) = 0;
-
- protected:
-  virtual ~IPDF_StructElement() {}
-};
-
-#endif  // CORE_FPDFDOC_FPDF_TAGGED_H_
diff --git a/fpdfsdk/fpdf_structtree.cpp b/fpdfsdk/fpdf_structtree.cpp
index 8a93d2299d..96d40b41c2 100644
--- a/fpdfsdk/fpdf_structtree.cpp
+++ b/fpdfsdk/fpdf_structtree.cpp
@@ -8,17 +8,18 @@
 
 #include "core/fpdfapi/page/cpdf_page.h"
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
-#include "core/fpdfdoc/fpdf_tagged.h"
+#include "core/fpdfdoc/cpdf_structelement.h"
+#include "core/fpdfdoc/cpdf_structtree.h"
 #include "fpdfsdk/fsdk_define.h"
 
 namespace {
 
-IPDF_StructTree* ToStructTree(FPDF_STRUCTTREE struct_tree) {
-  return reinterpret_cast<IPDF_StructTree*>(struct_tree);
+CPDF_StructTree* ToStructTree(FPDF_STRUCTTREE struct_tree) {
+  return reinterpret_cast<CPDF_StructTree*>(struct_tree);
 }
 
-IPDF_StructElement* ToStructTreeElement(FPDF_STRUCTELEMENT struct_element) {
-  return reinterpret_cast<IPDF_StructElement*>(struct_element);
+CPDF_StructElement* ToStructTreeElement(FPDF_STRUCTELEMENT struct_element) {
+  return reinterpret_cast<CPDF_StructElement*>(struct_element);
 }
 
 unsigned long WideStringToBuffer(const CFX_WideString& str,
@@ -40,23 +41,23 @@ DLLEXPORT FPDF_STRUCTTREE STDCALL FPDF_StructTree_GetForPage(FPDF_PAGE page) {
   CPDF_Page* pPage = CPDFPageFromFPDFPage(page);
   if (!pPage)
     return nullptr;
-  return IPDF_StructTree::LoadPage(pPage->m_pDocument, pPage->m_pFormDict)
+  return CPDF_StructTree::LoadPage(pPage->m_pDocument, pPage->m_pFormDict)
       .release();
 }
 
 DLLEXPORT void STDCALL FPDF_StructTree_Close(FPDF_STRUCTTREE struct_tree) {
-  std::unique_ptr<IPDF_StructTree>(ToStructTree(struct_tree));
+  std::unique_ptr<CPDF_StructTree>(ToStructTree(struct_tree));
 }
 
 DLLEXPORT int STDCALL
 FPDF_StructTree_CountChildren(FPDF_STRUCTTREE struct_tree) {
-  IPDF_StructTree* tree = ToStructTree(struct_tree);
+  CPDF_StructTree* tree = ToStructTree(struct_tree);
   return tree ? tree->CountTopElements() : -1;
 }
 
 DLLEXPORT FPDF_STRUCTELEMENT STDCALL
 FPDF_StructTree_GetChildAtIndex(FPDF_STRUCTTREE struct_tree, int index) {
-  IPDF_StructTree* tree = ToStructTree(struct_tree);
+  CPDF_StructTree* tree = ToStructTree(struct_tree);
   if (!tree || index < 0 || index >= tree->CountTopElements())
     return nullptr;
   return tree->GetTopElement(index);
@@ -66,7 +67,7 @@ DLLEXPORT unsigned long STDCALL
 FPDF_StructElement_GetAltText(FPDF_STRUCTELEMENT struct_element,
                               void* buffer,
                               unsigned long buflen) {
-  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  CPDF_StructElement* elem = ToStructTreeElement(struct_element);
   return (elem && elem->GetDict())
              ? WideStringToBuffer(elem->GetDict()->GetUnicodeTextFor("Alt"),
                                   buffer, buflen)
@@ -77,21 +78,21 @@ DLLEXPORT unsigned long STDCALL
 FPDF_StructElement_GetType(FPDF_STRUCTELEMENT struct_element,
                            void* buffer,
                            unsigned long buflen) {
-  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  CPDF_StructElement* elem = ToStructTreeElement(struct_element);
   return elem ? WideStringToBuffer(elem->GetType().UTF8Decode(), buffer, buflen)
               : 0;
 }
 
 DLLEXPORT int STDCALL
 FPDF_StructElement_CountChildren(FPDF_STRUCTELEMENT struct_element) {
-  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  CPDF_StructElement* elem = ToStructTreeElement(struct_element);
   return elem ? elem->CountKids() : -1;
 }
 
 DLLEXPORT FPDF_STRUCTELEMENT STDCALL
 FPDF_StructElement_GetChildAtIndex(FPDF_STRUCTELEMENT struct_element,
                                    int index) {
-  IPDF_StructElement* elem = ToStructTreeElement(struct_element);
+  CPDF_StructElement* elem = ToStructTreeElement(struct_element);
   if (!elem || index < 0 || index >= elem->CountKids())
     return nullptr;