1 // Copyright 2016 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/fpdfapi/parser/cpdf_stream_acc.h"
8
9 #include <utility>
10
11 #include "core/fdrm/fx_crypt.h"
12 #include "core/fpdfapi/parser/cpdf_dictionary.h"
13 #include "core/fpdfapi/parser/cpdf_stream.h"
14 #include "core/fpdfapi/parser/fpdf_parser_decode.h"
15 #include "core/fxcrt/check_op.h"
16 #include "core/fxcrt/compiler_specific.h"
17 #include "core/fxcrt/data_vector.h"
18
CPDF_StreamAcc(RetainPtr<const CPDF_Stream> pStream)19 CPDF_StreamAcc::CPDF_StreamAcc(RetainPtr<const CPDF_Stream> pStream)
20 : m_pStream(std::move(pStream)) {}
21
22 CPDF_StreamAcc::~CPDF_StreamAcc() = default;
23
LoadAllData(bool bRawAccess,uint32_t estimated_size,bool bImageAcc)24 void CPDF_StreamAcc::LoadAllData(bool bRawAccess,
25 uint32_t estimated_size,
26 bool bImageAcc) {
27 if (bRawAccess) {
28 DCHECK(!estimated_size);
29 DCHECK(!bImageAcc);
30 }
31
32 if (!m_pStream)
33 return;
34
35 bool bProcessRawData = bRawAccess || !m_pStream->HasFilter();
36 if (bProcessRawData)
37 ProcessRawData();
38 else
39 ProcessFilteredData(estimated_size, bImageAcc);
40 }
41
LoadAllDataFiltered()42 void CPDF_StreamAcc::LoadAllDataFiltered() {
43 LoadAllData(false, 0, false);
44 }
45
LoadAllDataFilteredWithEstimatedSize(uint32_t estimated_size)46 void CPDF_StreamAcc::LoadAllDataFilteredWithEstimatedSize(
47 uint32_t estimated_size) {
48 LoadAllData(false, estimated_size, false);
49 }
50
LoadAllDataImageAcc(uint32_t estimated_size)51 void CPDF_StreamAcc::LoadAllDataImageAcc(uint32_t estimated_size) {
52 LoadAllData(false, estimated_size, true);
53 }
54
LoadAllDataRaw()55 void CPDF_StreamAcc::LoadAllDataRaw() {
56 LoadAllData(true, 0, false);
57 }
58
GetStream() const59 RetainPtr<const CPDF_Stream> CPDF_StreamAcc::GetStream() const {
60 return m_pStream;
61 }
62
GetLength1ForTest() const63 int CPDF_StreamAcc::GetLength1ForTest() const {
64 return m_pStream->GetDict()->GetIntegerFor("Length1");
65 }
66
GetImageParam() const67 RetainPtr<const CPDF_Dictionary> CPDF_StreamAcc::GetImageParam() const {
68 return m_pImageParam;
69 }
70
GetSize() const71 uint32_t CPDF_StreamAcc::GetSize() const {
72 return GetSpan().size();
73 }
74
GetSpan() const75 pdfium::span<const uint8_t> CPDF_StreamAcc::GetSpan() const {
76 if (is_owned())
77 return absl::get<DataVector<uint8_t>>(m_Data);
78 if (m_pStream && m_pStream->IsMemoryBased())
79 return m_pStream->GetInMemoryRawData();
80 return {};
81 }
82
KeyForCache() const83 uint64_t CPDF_StreamAcc::KeyForCache() const {
84 return m_pStream ? m_pStream->KeyForCache() : 0;
85 }
86
ComputeDigest() const87 DataVector<uint8_t> CPDF_StreamAcc::ComputeDigest() const {
88 return CRYPT_SHA1Generate(GetSpan());
89 }
90
DetachData()91 DataVector<uint8_t> CPDF_StreamAcc::DetachData() {
92 if (is_owned())
93 return std::move(absl::get<DataVector<uint8_t>>(m_Data));
94
95 auto span = absl::get<pdfium::raw_span<const uint8_t>>(m_Data);
96 return DataVector<uint8_t>(span.begin(), span.end());
97 }
98
ProcessRawData()99 void CPDF_StreamAcc::ProcessRawData() {
100 uint32_t dwSrcSize = m_pStream->GetRawSize();
101 if (dwSrcSize == 0)
102 return;
103
104 if (m_pStream->IsMemoryBased()) {
105 m_Data = m_pStream->GetInMemoryRawData();
106 return;
107 }
108
109 DataVector<uint8_t> data = ReadRawStream();
110 if (data.empty())
111 return;
112
113 m_Data = std::move(data);
114 }
115
ProcessFilteredData(uint32_t estimated_size,bool bImageAcc)116 void CPDF_StreamAcc::ProcessFilteredData(uint32_t estimated_size,
117 bool bImageAcc) {
118 uint32_t dwSrcSize = m_pStream->GetRawSize();
119 if (dwSrcSize == 0)
120 return;
121
122 absl::variant<pdfium::raw_span<const uint8_t>, DataVector<uint8_t>> src_data;
123 pdfium::span<const uint8_t> src_span;
124 if (m_pStream->IsMemoryBased()) {
125 src_span = m_pStream->GetInMemoryRawData();
126 src_data = src_span;
127 } else {
128 DataVector<uint8_t> temp_src_data = ReadRawStream();
129 if (temp_src_data.empty())
130 return;
131
132 src_span = pdfium::make_span(temp_src_data);
133 src_data = std::move(temp_src_data);
134 }
135
136 std::optional<DecoderArray> decoder_array =
137 GetDecoderArray(m_pStream->GetDict());
138 if (!decoder_array.has_value() || decoder_array.value().empty()) {
139 m_Data = std::move(src_data);
140 return;
141 }
142
143 std::optional<PDFDataDecodeResult> result = PDF_DataDecode(
144 src_span, estimated_size, bImageAcc, decoder_array.value());
145 if (!result.has_value()) {
146 m_Data = std::move(src_data);
147 return;
148 }
149
150 m_ImageDecoder = std::move(result.value().image_encoding);
151 m_pImageParam = std::move(result.value().image_params);
152
153 if (result.value().data.empty()) {
154 m_Data = std::move(src_data);
155 return;
156 }
157
158 m_Data = std::move(result.value().data);
159 }
160
ReadRawStream() const161 DataVector<uint8_t> CPDF_StreamAcc::ReadRawStream() const {
162 DCHECK(m_pStream);
163 DCHECK(m_pStream->IsFileBased());
164 return m_pStream->ReadAllRawData();
165 }
166