1 // Copyright 2017 PDFium Authors. All rights reserved.
2 // Use of this source code is governed by a BSD-style license that can be
3 // found in the LICENSE file.
4
5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
6
7 #include "core/fpdfdoc/cpdf_structtree.h"
8
9 #include "core/fpdfapi/parser/cpdf_array.h"
10 #include "core/fpdfapi/parser/cpdf_document.h"
11 #include "core/fpdfapi/parser/cpdf_number.h"
12 #include "core/fpdfapi/parser/cpdf_reference.h"
13 #include "core/fpdfdoc/cpdf_numbertree.h"
14 #include "core/fpdfdoc/cpdf_structelement.h"
15
16 namespace {
17
IsTagged(const CPDF_Document * pDoc)18 bool IsTagged(const CPDF_Document* pDoc) {
19 const CPDF_Dictionary* pCatalog = pDoc->GetRoot();
20 const CPDF_Dictionary* pMarkInfo = pCatalog->GetDictFor("MarkInfo");
21 return pMarkInfo && pMarkInfo->GetIntegerFor("Marked");
22 }
23
24 } // namespace
25
26 // static
LoadPage(const CPDF_Document * pDoc,const CPDF_Dictionary * pPageDict)27 std::unique_ptr<CPDF_StructTree> CPDF_StructTree::LoadPage(
28 const CPDF_Document* pDoc,
29 const CPDF_Dictionary* pPageDict) {
30 if (!IsTagged(pDoc))
31 return nullptr;
32
33 auto pTree = pdfium::MakeUnique<CPDF_StructTree>(pDoc);
34 pTree->LoadPageTree(pPageDict);
35 return pTree;
36 }
37
CPDF_StructTree(const CPDF_Document * pDoc)38 CPDF_StructTree::CPDF_StructTree(const CPDF_Document* pDoc)
39 : m_pTreeRoot(pDoc->GetRoot()->GetDictFor("StructTreeRoot")),
40 m_pRoleMap(m_pTreeRoot ? m_pTreeRoot->GetDictFor("RoleMap") : nullptr),
41 m_pPage(nullptr) {}
42
~CPDF_StructTree()43 CPDF_StructTree::~CPDF_StructTree() {}
44
LoadPageTree(const CPDF_Dictionary * pPageDict)45 void CPDF_StructTree::LoadPageTree(const CPDF_Dictionary* pPageDict) {
46 m_pPage = pPageDict;
47 if (!m_pTreeRoot)
48 return;
49
50 CPDF_Object* pKids = m_pTreeRoot->GetDirectObjectFor("K");
51 if (!pKids)
52 return;
53
54 uint32_t dwKids = 0;
55 if (pKids->IsDictionary())
56 dwKids = 1;
57 else if (CPDF_Array* pArray = pKids->AsArray())
58 dwKids = pArray->GetCount();
59 else
60 return;
61
62 m_Kids.clear();
63 m_Kids.resize(dwKids);
64 CPDF_Dictionary* pParentTree = m_pTreeRoot->GetDictFor("ParentTree");
65 if (!pParentTree)
66 return;
67
68 CPDF_NumberTree parent_tree(pParentTree);
69 int parents_id = pPageDict->GetIntegerFor("StructParents", -1);
70 if (parents_id < 0)
71 return;
72
73 CPDF_Array* pParentArray = ToArray(parent_tree.LookupValue(parents_id));
74 if (!pParentArray)
75 return;
76
77 std::map<CPDF_Dictionary*, RetainPtr<CPDF_StructElement>> element_map;
78 for (size_t i = 0; i < pParentArray->GetCount(); i++) {
79 if (CPDF_Dictionary* pParent = pParentArray->GetDictAt(i))
80 AddPageNode(pParent, &element_map, 0);
81 }
82 }
83
AddPageNode(CPDF_Dictionary * pDict,std::map<CPDF_Dictionary *,RetainPtr<CPDF_StructElement>> * map,int nLevel)84 RetainPtr<CPDF_StructElement> CPDF_StructTree::AddPageNode(
85 CPDF_Dictionary* pDict,
86 std::map<CPDF_Dictionary*, RetainPtr<CPDF_StructElement>>* map,
87 int nLevel) {
88 static constexpr int kStructTreeMaxRecursion = 32;
89 if (nLevel > kStructTreeMaxRecursion)
90 return nullptr;
91
92 auto it = map->find(pDict);
93 if (it != map->end())
94 return it->second;
95
96 auto pElement = pdfium::MakeRetain<CPDF_StructElement>(this, nullptr, pDict);
97 (*map)[pDict] = pElement;
98 CPDF_Dictionary* pParent = pDict->GetDictFor("P");
99 if (!pParent || pParent->GetStringFor("Type") == "StructTreeRoot") {
100 if (!AddTopLevelNode(pDict, pElement))
101 map->erase(pDict);
102 return pElement;
103 }
104
105 RetainPtr<CPDF_StructElement> pParentElement =
106 AddPageNode(pParent, map, nLevel + 1);
107 bool bSave = false;
108 for (CPDF_StructKid& kid : *pParentElement->GetKids()) {
109 if (kid.m_Type == CPDF_StructKid::Element && kid.m_pDict == pDict) {
110 kid.m_pElement = pElement;
111 bSave = true;
112 }
113 }
114 if (!bSave)
115 map->erase(pDict);
116 return pElement;
117 }
118
AddTopLevelNode(CPDF_Dictionary * pDict,const RetainPtr<CPDF_StructElement> & pElement)119 bool CPDF_StructTree::AddTopLevelNode(
120 CPDF_Dictionary* pDict,
121 const RetainPtr<CPDF_StructElement>& pElement) {
122 CPDF_Object* pObj = m_pTreeRoot->GetDirectObjectFor("K");
123 if (!pObj)
124 return false;
125
126 if (pObj->IsDictionary()) {
127 if (pObj->GetObjNum() != pDict->GetObjNum())
128 return false;
129 m_Kids[0] = pElement;
130 }
131
132 CPDF_Array* pTopKids = pObj->AsArray();
133 if (!pTopKids)
134 return true;
135
136 bool bSave = false;
137 for (size_t i = 0; i < pTopKids->GetCount(); i++) {
138 CPDF_Reference* pKidRef = ToReference(pTopKids->GetObjectAt(i));
139 if (pKidRef && pKidRef->GetRefObjNum() == pDict->GetObjNum()) {
140 m_Kids[i] = pElement;
141 bSave = true;
142 }
143 }
144 return bSave;
145 }
146