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_structelement.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_name.h"
14 #include "core/fpdfapi/parser/cpdf_number.h"
15 #include "core/fpdfapi/parser/cpdf_object.h"
16 #include "core/fpdfapi/parser/cpdf_reference.h"
17 #include "core/fpdfapi/parser/cpdf_stream.h"
18 #include "core/fpdfdoc/cpdf_structtree.h"
19
20 namespace {
21
GetStructElementType(CPDF_StructTree * pTree,const CPDF_Dictionary * pDict)22 ByteString GetStructElementType(CPDF_StructTree* pTree,
23 const CPDF_Dictionary* pDict) {
24 ByteString type = pDict->GetStringFor("S");
25 if (pTree->GetRoleMap()) {
26 ByteString mapped = pTree->GetRoleMap()->GetStringFor(type);
27 if (!mapped.IsEmpty())
28 type = std::move(mapped);
29 }
30 return type;
31 }
32
33 } // namespace
34
35 CPDF_StructKid::CPDF_StructKid() = default;
36
37 CPDF_StructKid::CPDF_StructKid(const CPDF_StructKid& that) = default;
38
39 CPDF_StructKid::~CPDF_StructKid() = default;
40
CPDF_StructElement(CPDF_StructTree * pTree,CPDF_StructElement * pParent,const CPDF_Dictionary * pDict)41 CPDF_StructElement::CPDF_StructElement(CPDF_StructTree* pTree,
42 CPDF_StructElement* pParent,
43 const CPDF_Dictionary* pDict)
44 : m_pTree(pTree),
45 m_pParent(pParent),
46 m_pDict(pDict),
47 m_Type(GetStructElementType(m_pTree.Get(), m_pDict.Get())) {
48 LoadKids(m_pDict.Get());
49 }
50
51 CPDF_StructElement::~CPDF_StructElement() = default;
52
GetAltText() const53 WideString CPDF_StructElement::GetAltText() const {
54 return GetDict()->GetUnicodeTextFor("Alt");
55 }
56
GetTitle() const57 WideString CPDF_StructElement::GetTitle() const {
58 return GetDict()->GetUnicodeTextFor("T");
59 }
60
CountKids() const61 size_t CPDF_StructElement::CountKids() const {
62 return m_Kids.size();
63 }
64
GetKidIfElement(size_t index) const65 CPDF_StructElement* CPDF_StructElement::GetKidIfElement(size_t index) const {
66 return m_Kids[index].m_Type == CPDF_StructKid::kElement
67 ? m_Kids[index].m_pElement.Get()
68 : nullptr;
69 }
70
LoadKids(const CPDF_Dictionary * pDict)71 void CPDF_StructElement::LoadKids(const CPDF_Dictionary* pDict) {
72 const CPDF_Object* pObj = pDict->GetObjectFor("Pg");
73 uint32_t PageObjNum = 0;
74 if (const CPDF_Reference* pRef = ToReference(pObj))
75 PageObjNum = pRef->GetRefObjNum();
76
77 const CPDF_Object* pKids = pDict->GetDirectObjectFor("K");
78 if (!pKids)
79 return;
80
81 m_Kids.clear();
82 if (const CPDF_Array* pArray = pKids->AsArray()) {
83 m_Kids.resize(pArray->size());
84 for (uint32_t i = 0; i < pArray->size(); i++) {
85 const CPDF_Object* pKid = pArray->GetDirectObjectAt(i);
86 LoadKid(PageObjNum, pKid, &m_Kids[i]);
87 }
88 return;
89 }
90
91 m_Kids.resize(1);
92 LoadKid(PageObjNum, pKids, &m_Kids[0]);
93 }
94
LoadKid(uint32_t PageObjNum,const CPDF_Object * pKidObj,CPDF_StructKid * pKid)95 void CPDF_StructElement::LoadKid(uint32_t PageObjNum,
96 const CPDF_Object* pKidObj,
97 CPDF_StructKid* pKid) {
98 pKid->m_Type = CPDF_StructKid::kInvalid;
99 if (!pKidObj)
100 return;
101
102 if (pKidObj->IsNumber()) {
103 if (m_pTree->GetPage()->GetObjNum() != PageObjNum)
104 return;
105
106 pKid->m_Type = CPDF_StructKid::kPageContent;
107 pKid->m_ContentId = pKidObj->GetInteger();
108 pKid->m_PageObjNum = PageObjNum;
109 return;
110 }
111
112 const CPDF_Dictionary* pKidDict = pKidObj->AsDictionary();
113 if (!pKidDict)
114 return;
115 if (const CPDF_Reference* pRef = ToReference(pKidDict->GetObjectFor("Pg")))
116 PageObjNum = pRef->GetRefObjNum();
117
118 ByteString type = pKidDict->GetStringFor("Type");
119 if ((type == "MCR" || type == "OBJR") &&
120 m_pTree->GetPage()->GetObjNum() != PageObjNum) {
121 return;
122 }
123
124 if (type == "MCR") {
125 pKid->m_Type = CPDF_StructKid::kStreamContent;
126 const CPDF_Reference* pRef = ToReference(pKidDict->GetObjectFor("Stm"));
127 pKid->m_RefObjNum = pRef ? pRef->GetRefObjNum() : 0;
128 pKid->m_PageObjNum = PageObjNum;
129 pKid->m_ContentId = pKidDict->GetIntegerFor("MCID");
130 return;
131 }
132
133 if (type == "OBJR") {
134 pKid->m_Type = CPDF_StructKid::kObject;
135 const CPDF_Reference* pObj = ToReference(pKidDict->GetObjectFor("Obj"));
136 pKid->m_RefObjNum = pObj ? pObj->GetRefObjNum() : 0;
137 pKid->m_PageObjNum = PageObjNum;
138 return;
139 }
140
141 pKid->m_Type = CPDF_StructKid::kElement;
142 pKid->m_pDict.Reset(pKidDict);
143 pKid->m_pElement = nullptr;
144 }
145