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/data_vector.h"
16 #include "third_party/base/check_op.h"
17
CPDF_StreamAcc(RetainPtr<const CPDF_Stream> pStream)18 CPDF_StreamAcc::CPDF_StreamAcc(RetainPtr<const CPDF_Stream> pStream)
19 : m_pStream(std::move(pStream)) {}
20
21 CPDF_StreamAcc::~CPDF_StreamAcc() = default;
22
LoadAllData(bool bRawAccess,uint32_t estimated_size,bool bImageAcc)23 void CPDF_StreamAcc::LoadAllData(bool bRawAccess,
24 uint32_t estimated_size,
25 bool bImageAcc) {
26 if (bRawAccess) {
27 DCHECK(!estimated_size);
28 DCHECK(!bImageAcc);
29 }
30
31 if (!m_pStream)
32 return;
33
34 bool bProcessRawData = bRawAccess || !m_pStream->HasFilter();
35 if (bProcessRawData)
36 ProcessRawData();
37 else
38 ProcessFilteredData(estimated_size, bImageAcc);
39 }
40
LoadAllDataFiltered()41 void CPDF_StreamAcc::LoadAllDataFiltered() {
42 LoadAllData(false, 0, false);
43 }
44
LoadAllDataFilteredWithEstimatedSize(uint32_t estimated_size)45 void CPDF_StreamAcc::LoadAllDataFilteredWithEstimatedSize(
46 uint32_t estimated_size) {
47 LoadAllData(false, estimated_size, false);
48 }
49
LoadAllDataImageAcc(uint32_t estimated_size)50 void CPDF_StreamAcc::LoadAllDataImageAcc(uint32_t estimated_size) {
51 LoadAllData(false, estimated_size, true);
52 }
53
LoadAllDataRaw()54 void CPDF_StreamAcc::LoadAllDataRaw() {
55 LoadAllData(true, 0, false);
56 }
57
GetStream() const58 RetainPtr<const CPDF_Stream> CPDF_StreamAcc::GetStream() const {
59 return m_pStream;
60 }
61
GetLength1ForTest() const62 int CPDF_StreamAcc::GetLength1ForTest() const {
63 return m_pStream->GetDict()->GetIntegerFor("Length1");
64 }
65
GetImageParam() const66 RetainPtr<const CPDF_Dictionary> CPDF_StreamAcc::GetImageParam() const {
67 return m_pImageParam;
68 }
69
GetSize() const70 uint32_t CPDF_StreamAcc::GetSize() const {
71 return GetSpan().size();
72 }
73
GetSpan() const74 pdfium::span<const uint8_t> CPDF_StreamAcc::GetSpan() const {
75 if (is_owned())
76 return absl::get<DataVector<uint8_t>>(m_Data);
77 if (m_pStream && m_pStream->IsMemoryBased())
78 return m_pStream->GetInMemoryRawData();
79 return {};
80 }
81
KeyForCache() const82 uint64_t CPDF_StreamAcc::KeyForCache() const {
83 return m_pStream ? m_pStream->KeyForCache() : 0;
84 }
85
ComputeDigest() const86 ByteString CPDF_StreamAcc::ComputeDigest() const {
87 uint8_t digest[20];
88 pdfium::span<const uint8_t> span = GetSpan();
89 CRYPT_SHA1Generate(span.data(), span.size(), digest);
90 return ByteString(digest, 20);
91 }
92
DetachData()93 DataVector<uint8_t> CPDF_StreamAcc::DetachData() {
94 if (is_owned())
95 return std::move(absl::get<DataVector<uint8_t>>(m_Data));
96
97 auto span = absl::get<pdfium::span<const uint8_t>>(m_Data);
98 return DataVector<uint8_t>(span.begin(), span.end());
99 }
100
ProcessRawData()101 void CPDF_StreamAcc::ProcessRawData() {
102 if (m_pStream->IsUninitialized())
103 return;
104
105 uint32_t dwSrcSize = m_pStream->GetRawSize();
106 if (dwSrcSize == 0)
107 return;
108
109 if (m_pStream->IsMemoryBased()) {
110 m_Data = m_pStream->GetInMemoryRawData();
111 return;
112 }
113
114 DataVector<uint8_t> data = ReadRawStream();
115 if (data.empty())
116 return;
117
118 m_Data = std::move(data);
119 }
120
ProcessFilteredData(uint32_t estimated_size,bool bImageAcc)121 void CPDF_StreamAcc::ProcessFilteredData(uint32_t estimated_size,
122 bool bImageAcc) {
123 if (m_pStream->IsUninitialized())
124 return;
125
126 uint32_t dwSrcSize = m_pStream->GetRawSize();
127 if (dwSrcSize == 0)
128 return;
129
130 absl::variant<pdfium::span<const uint8_t>, DataVector<uint8_t>> src_data;
131 pdfium::span<const uint8_t> src_span;
132 if (m_pStream->IsMemoryBased()) {
133 src_span = m_pStream->GetInMemoryRawData();
134 src_data = src_span;
135 } else {
136 DataVector<uint8_t> temp_src_data = ReadRawStream();
137 if (temp_src_data.empty())
138 return;
139
140 src_span = pdfium::make_span(temp_src_data);
141 src_data = std::move(temp_src_data);
142 }
143
144 std::unique_ptr<uint8_t, FxFreeDeleter> pDecodedData;
145 uint32_t dwDecodedSize = 0;
146
147 absl::optional<DecoderArray> decoder_array =
148 GetDecoderArray(m_pStream->GetDict());
149 if (!decoder_array.has_value() || decoder_array.value().empty() ||
150 !PDF_DataDecode(src_span, estimated_size, bImageAcc,
151 decoder_array.value(), &pDecodedData, &dwDecodedSize,
152 &m_ImageDecoder, &m_pImageParam)) {
153 m_Data = std::move(src_data);
154 return;
155 }
156
157 if (pDecodedData) {
158 DCHECK_NE(pDecodedData.get(), src_span.data());
159 // TODO(crbug.com/pdfium/1872): Avoid copying.
160 m_Data = DataVector<uint8_t>(pDecodedData.get(),
161 pDecodedData.get() + dwDecodedSize);
162 } else {
163 m_Data = std::move(src_data);
164 }
165 }
166
ReadRawStream() const167 DataVector<uint8_t> CPDF_StreamAcc::ReadRawStream() const {
168 DCHECK(m_pStream);
169 DCHECK(m_pStream->IsFileBased());
170 return m_pStream->ReadAllRawData();
171 }
172