1 // Copyright 2017 The PDFium Authors
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 <utility>
10
11 #include "core/fpdfapi/parser/cpdf_array.h"
12 #include "core/fpdfapi/parser/cpdf_dictionary.h"
13 #include "core/fpdfapi/parser/cpdf_document.h"
14 #include "core/fpdfapi/parser/cpdf_number.h"
15 #include "core/fpdfapi/parser/cpdf_reference.h"
16 #include "core/fpdfdoc/cpdf_numbertree.h"
17 #include "core/fpdfdoc/cpdf_structelement.h"
18
19 namespace {
20
IsTagged(const CPDF_Document * pDoc)21 bool IsTagged(const CPDF_Document* pDoc) {
22 RetainPtr<const CPDF_Dictionary> pMarkInfo =
23 pDoc->GetRoot()->GetDictFor("MarkInfo");
24 return pMarkInfo && pMarkInfo->GetIntegerFor("Marked");
25 }
26
27 } // namespace
28
29 // static
LoadPage(const CPDF_Document * pDoc,RetainPtr<const CPDF_Dictionary> pPageDict)30 std::unique_ptr<CPDF_StructTree> CPDF_StructTree::LoadPage(
31 const CPDF_Document* pDoc,
32 RetainPtr<const CPDF_Dictionary> pPageDict) {
33 if (!IsTagged(pDoc))
34 return nullptr;
35
36 auto pTree = std::make_unique<CPDF_StructTree>(pDoc);
37 pTree->LoadPageTree(std::move(pPageDict));
38 return pTree;
39 }
40
CPDF_StructTree(const CPDF_Document * pDoc)41 CPDF_StructTree::CPDF_StructTree(const CPDF_Document* pDoc)
42 : m_pTreeRoot(pDoc->GetRoot()->GetDictFor("StructTreeRoot")),
43 m_pRoleMap(m_pTreeRoot ? m_pTreeRoot->GetDictFor("RoleMap") : nullptr) {}
44
45 CPDF_StructTree::~CPDF_StructTree() = default;
46
GetRoleMapNameFor(const ByteString & type) const47 ByteString CPDF_StructTree::GetRoleMapNameFor(const ByteString& type) const {
48 if (m_pRoleMap) {
49 ByteString mapped = m_pRoleMap->GetNameFor(type);
50 if (!mapped.IsEmpty())
51 return mapped;
52 }
53 return type;
54 }
55
LoadPageTree(RetainPtr<const CPDF_Dictionary> pPageDict)56 void CPDF_StructTree::LoadPageTree(RetainPtr<const CPDF_Dictionary> pPageDict) {
57 m_pPage = std::move(pPageDict);
58 if (!m_pTreeRoot)
59 return;
60
61 RetainPtr<const CPDF_Object> pKids = m_pTreeRoot->GetDirectObjectFor("K");
62 if (!pKids)
63 return;
64
65 size_t kids_count;
66 if (pKids->IsDictionary())
67 kids_count = 1;
68 else if (const CPDF_Array* pArray = pKids->AsArray())
69 kids_count = pArray->size();
70 else
71 return;
72
73 m_Kids.clear();
74 m_Kids.resize(kids_count);
75
76 RetainPtr<const CPDF_Dictionary> pParentTree =
77 m_pTreeRoot->GetDictFor("ParentTree");
78 if (!pParentTree)
79 return;
80
81 CPDF_NumberTree parent_tree(std::move(pParentTree));
82 int parents_id = m_pPage->GetIntegerFor("StructParents", -1);
83 if (parents_id < 0)
84 return;
85
86 RetainPtr<const CPDF_Array> pParentArray =
87 ToArray(parent_tree.LookupValue(parents_id));
88 if (!pParentArray)
89 return;
90
91 StructElementMap element_map;
92 for (size_t i = 0; i < pParentArray->size(); i++) {
93 RetainPtr<const CPDF_Dictionary> pParent = pParentArray->GetDictAt(i);
94 if (pParent)
95 AddPageNode(std::move(pParent), &element_map, 0);
96 }
97 }
98
AddPageNode(RetainPtr<const CPDF_Dictionary> pDict,StructElementMap * map,int nLevel)99 RetainPtr<CPDF_StructElement> CPDF_StructTree::AddPageNode(
100 RetainPtr<const CPDF_Dictionary> pDict,
101 StructElementMap* map,
102 int nLevel) {
103 static constexpr int kStructTreeMaxRecursion = 32;
104 if (nLevel > kStructTreeMaxRecursion)
105 return nullptr;
106
107 auto it = map->find(pDict);
108 if (it != map->end())
109 return it->second;
110
111 RetainPtr<const CPDF_Dictionary> key(pDict);
112 auto pElement = pdfium::MakeRetain<CPDF_StructElement>(this, pDict);
113 (*map)[key] = pElement;
114 RetainPtr<const CPDF_Dictionary> pParent = pDict->GetDictFor("P");
115 if (!pParent || pParent->GetNameFor("Type") == "StructTreeRoot") {
116 if (!AddTopLevelNode(pDict, pElement))
117 map->erase(key);
118 return pElement;
119 }
120
121 RetainPtr<CPDF_StructElement> pParentElement =
122 AddPageNode(std::move(pParent), map, nLevel + 1);
123 if (!pParentElement)
124 return pElement;
125
126 if (!pParentElement->UpdateKidIfElement(pDict, pElement.Get())) {
127 map->erase(key);
128 return pElement;
129 }
130
131 pElement->SetParent(pParentElement.Get());
132 return pElement;
133 }
134
AddTopLevelNode(const CPDF_Dictionary * pDict,const RetainPtr<CPDF_StructElement> & pElement)135 bool CPDF_StructTree::AddTopLevelNode(
136 const CPDF_Dictionary* pDict,
137 const RetainPtr<CPDF_StructElement>& pElement) {
138 RetainPtr<const CPDF_Object> pObj = m_pTreeRoot->GetDirectObjectFor("K");
139 if (!pObj)
140 return false;
141
142 if (pObj->IsDictionary()) {
143 if (pObj->GetObjNum() != pDict->GetObjNum())
144 return false;
145 m_Kids[0] = pElement;
146 }
147
148 const CPDF_Array* pTopKids = pObj->AsArray();
149 if (!pTopKids)
150 return true;
151
152 bool bSave = false;
153 for (size_t i = 0; i < pTopKids->size(); i++) {
154 RetainPtr<const CPDF_Reference> pKidRef =
155 ToReference(pTopKids->GetObjectAt(i));
156 if (pKidRef && pKidRef->GetRefObjNum() == pDict->GetObjNum()) {
157 m_Kids[i] = pElement;
158 bSave = true;
159 }
160 }
161 return bSave;
162 }
163