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/page/cpdf_contentparser.h"
8
9 #include <utility>
10
11 #include "constants/page_object.h"
12 #include "core/fpdfapi/font/cpdf_type3char.h"
13 #include "core/fpdfapi/page/cpdf_allstates.h"
14 #include "core/fpdfapi/page/cpdf_page.h"
15 #include "core/fpdfapi/page/cpdf_pageobject.h"
16 #include "core/fpdfapi/page/cpdf_path.h"
17 #include "core/fpdfapi/parser/cpdf_array.h"
18 #include "core/fpdfapi/parser/cpdf_dictionary.h"
19 #include "core/fpdfapi/parser/cpdf_stream.h"
20 #include "core/fpdfapi/parser/cpdf_stream_acc.h"
21 #include "core/fxcrt/fixed_try_alloc_zeroed_data_vector.h"
22 #include "core/fxcrt/fx_safe_types.h"
23 #include "core/fxcrt/pauseindicator_iface.h"
24 #include "core/fxcrt/span_util.h"
25 #include "core/fxcrt/stl_util.h"
26 #include "core/fxge/cfx_fillrenderoptions.h"
27 #include "third_party/base/check.h"
28 #include "third_party/base/check_op.h"
29
CPDF_ContentParser(CPDF_Page * pPage)30 CPDF_ContentParser::CPDF_ContentParser(CPDF_Page* pPage)
31 : m_CurrentStage(Stage::kGetContent), m_pPageObjectHolder(pPage) {
32 DCHECK(pPage);
33 if (!pPage->GetDocument()) {
34 m_CurrentStage = Stage::kComplete;
35 return;
36 }
37
38 RetainPtr<CPDF_Object> pContent =
39 pPage->GetMutableDict()->GetMutableDirectObjectFor(
40 pdfium::page_object::kContents);
41 if (!pContent) {
42 HandlePageContentFailure();
43 return;
44 }
45
46 const CPDF_Stream* pStream = pContent->AsStream();
47 if (pStream) {
48 HandlePageContentStream(pStream);
49 return;
50 }
51
52 const CPDF_Array* pArray = pContent->AsArray();
53 if (pArray && HandlePageContentArray(pArray))
54 return;
55
56 HandlePageContentFailure();
57 }
58
CPDF_ContentParser(RetainPtr<const CPDF_Stream> pStream,CPDF_PageObjectHolder * pPageObjectHolder,const CPDF_AllStates * pGraphicStates,const CFX_Matrix * pParentMatrix,CPDF_Type3Char * pType3Char,CPDF_Form::RecursionState * recursion_state)59 CPDF_ContentParser::CPDF_ContentParser(
60 RetainPtr<const CPDF_Stream> pStream,
61 CPDF_PageObjectHolder* pPageObjectHolder,
62 const CPDF_AllStates* pGraphicStates,
63 const CFX_Matrix* pParentMatrix,
64 CPDF_Type3Char* pType3Char,
65 CPDF_Form::RecursionState* recursion_state)
66 : m_CurrentStage(Stage::kParse),
67 m_pPageObjectHolder(pPageObjectHolder),
68 m_pType3Char(pType3Char) {
69 DCHECK(m_pPageObjectHolder);
70 CFX_Matrix form_matrix =
71 m_pPageObjectHolder->GetDict()->GetMatrixFor("Matrix");
72 if (pGraphicStates)
73 form_matrix.Concat(pGraphicStates->m_CTM);
74
75 RetainPtr<const CPDF_Array> pBBox =
76 m_pPageObjectHolder->GetDict()->GetArrayFor("BBox");
77 CFX_FloatRect form_bbox;
78 CPDF_Path ClipPath;
79 if (pBBox) {
80 form_bbox = pBBox->GetRect();
81 ClipPath.Emplace();
82 ClipPath.AppendFloatRect(form_bbox);
83 ClipPath.Transform(form_matrix);
84 if (pParentMatrix)
85 ClipPath.Transform(*pParentMatrix);
86
87 form_bbox = form_matrix.TransformRect(form_bbox);
88 if (pParentMatrix)
89 form_bbox = pParentMatrix->TransformRect(form_bbox);
90 }
91
92 RetainPtr<CPDF_Dictionary> pResources =
93 m_pPageObjectHolder->GetMutableDict()->GetMutableDictFor("Resources");
94 m_pParser = std::make_unique<CPDF_StreamContentParser>(
95 m_pPageObjectHolder->GetDocument(),
96 m_pPageObjectHolder->GetMutablePageResources(),
97 m_pPageObjectHolder->GetMutableResources(), pParentMatrix,
98 m_pPageObjectHolder, std::move(pResources), form_bbox, pGraphicStates,
99 recursion_state);
100 m_pParser->GetCurStates()->m_CTM = form_matrix;
101 m_pParser->GetCurStates()->m_ParentMatrix = form_matrix;
102 if (ClipPath.HasRef()) {
103 m_pParser->GetCurStates()->m_ClipPath.AppendPathWithAutoMerge(
104 ClipPath, CFX_FillRenderOptions::FillType::kWinding);
105 }
106 if (m_pPageObjectHolder->GetTransparency().IsGroup()) {
107 CPDF_GeneralState* pState = &m_pParser->GetCurStates()->m_GeneralState;
108 pState->SetBlendType(BlendMode::kNormal);
109 pState->SetStrokeAlpha(1.0f);
110 pState->SetFillAlpha(1.0f);
111 pState->SetSoftMask(nullptr);
112 }
113 m_pSingleStream = pdfium::MakeRetain<CPDF_StreamAcc>(std::move(pStream));
114 m_pSingleStream->LoadAllDataFiltered();
115 m_Data = m_pSingleStream->GetSpan();
116 }
117
118 CPDF_ContentParser::~CPDF_ContentParser() = default;
119
120 // Returning |true| means that there is more content to be processed and
121 // Continue() should be called again. Returning |false| means that we've
122 // completed the parse and Continue() is complete.
Continue(PauseIndicatorIface * pPause)123 bool CPDF_ContentParser::Continue(PauseIndicatorIface* pPause) {
124 while (m_CurrentStage == Stage::kGetContent) {
125 m_CurrentStage = GetContent();
126 if (pPause && pPause->NeedToPauseNow())
127 return true;
128 }
129
130 if (m_CurrentStage == Stage::kPrepareContent)
131 m_CurrentStage = PrepareContent();
132
133 while (m_CurrentStage == Stage::kParse) {
134 m_CurrentStage = Parse();
135 if (pPause && pPause->NeedToPauseNow())
136 return true;
137 }
138
139 if (m_CurrentStage == Stage::kCheckClip)
140 m_CurrentStage = CheckClip();
141
142 DCHECK_EQ(m_CurrentStage, Stage::kComplete);
143 return false;
144 }
145
GetContent()146 CPDF_ContentParser::Stage CPDF_ContentParser::GetContent() {
147 DCHECK_EQ(m_CurrentStage, Stage::kGetContent);
148 DCHECK(m_pPageObjectHolder->IsPage());
149 RetainPtr<const CPDF_Array> pContent =
150 m_pPageObjectHolder->GetDict()->GetArrayFor(
151 pdfium::page_object::kContents);
152 RetainPtr<const CPDF_Stream> pStreamObj = ToStream(
153 pContent ? pContent->GetDirectObjectAt(m_CurrentOffset) : nullptr);
154 m_StreamArray[m_CurrentOffset] =
155 pdfium::MakeRetain<CPDF_StreamAcc>(std::move(pStreamObj));
156 m_StreamArray[m_CurrentOffset]->LoadAllDataFiltered();
157 m_CurrentOffset++;
158
159 return m_CurrentOffset == m_nStreams ? Stage::kPrepareContent
160 : Stage::kGetContent;
161 }
162
PrepareContent()163 CPDF_ContentParser::Stage CPDF_ContentParser::PrepareContent() {
164 m_CurrentOffset = 0;
165
166 if (m_StreamArray.empty()) {
167 m_Data = m_pSingleStream->GetSpan();
168 return Stage::kParse;
169 }
170
171 FX_SAFE_UINT32 safe_size = 0;
172 for (const auto& stream : m_StreamArray) {
173 m_StreamSegmentOffsets.push_back(safe_size.ValueOrDie());
174 safe_size += stream->GetSize();
175 safe_size += 1;
176 if (!safe_size.IsValid())
177 return Stage::kComplete;
178 }
179
180 const size_t buffer_size = safe_size.ValueOrDie();
181 FixedTryAllocZeroedDataVector<uint8_t> buffer(buffer_size);
182 if (buffer.empty()) {
183 m_Data.emplace<pdfium::span<const uint8_t>>();
184 return Stage::kComplete;
185 }
186
187 size_t pos = 0;
188 auto data_span = buffer.writable_span();
189 for (const auto& stream : m_StreamArray) {
190 fxcrt::spancpy(data_span.subspan(pos), stream->GetSpan());
191 pos += stream->GetSize();
192 data_span[pos++] = ' ';
193 }
194 m_StreamArray.clear();
195 m_Data = std::move(buffer);
196 return Stage::kParse;
197 }
198
Parse()199 CPDF_ContentParser::Stage CPDF_ContentParser::Parse() {
200 if (!m_pParser) {
201 m_RecursionState.parsed_set.clear();
202 m_RecursionState.form_count = 0;
203 m_pParser = std::make_unique<CPDF_StreamContentParser>(
204 m_pPageObjectHolder->GetDocument(),
205 m_pPageObjectHolder->GetMutablePageResources(), nullptr, nullptr,
206 m_pPageObjectHolder, m_pPageObjectHolder->GetMutableResources(),
207 m_pPageObjectHolder->GetBBox(), nullptr, &m_RecursionState);
208 m_pParser->GetCurStates()->m_ColorState.SetDefault();
209 }
210 if (m_CurrentOffset >= GetData().size())
211 return Stage::kCheckClip;
212
213 if (m_StreamSegmentOffsets.empty())
214 m_StreamSegmentOffsets.push_back(0);
215
216 static constexpr uint32_t kParseStepLimit = 100;
217 m_CurrentOffset += m_pParser->Parse(GetData(), m_CurrentOffset,
218 kParseStepLimit, m_StreamSegmentOffsets);
219 return Stage::kParse;
220 }
221
CheckClip()222 CPDF_ContentParser::Stage CPDF_ContentParser::CheckClip() {
223 if (m_pType3Char) {
224 m_pType3Char->InitializeFromStreamData(m_pParser->IsColored(),
225 m_pParser->GetType3Data());
226 }
227
228 for (auto& pObj : *m_pPageObjectHolder) {
229 if (!pObj->m_ClipPath.HasRef())
230 continue;
231 if (pObj->m_ClipPath.GetPathCount() != 1)
232 continue;
233 if (pObj->m_ClipPath.GetTextCount() > 0)
234 continue;
235
236 CPDF_Path ClipPath = pObj->m_ClipPath.GetPath(0);
237 if (!ClipPath.IsRect() || pObj->IsShading())
238 continue;
239
240 CFX_PointF point0 = ClipPath.GetPoint(0);
241 CFX_PointF point2 = ClipPath.GetPoint(2);
242 CFX_FloatRect old_rect(point0.x, point0.y, point2.x, point2.y);
243 if (old_rect.Contains(pObj->GetRect()))
244 pObj->m_ClipPath.SetNull();
245 }
246 return Stage::kComplete;
247 }
248
HandlePageContentStream(const CPDF_Stream * pStream)249 void CPDF_ContentParser::HandlePageContentStream(const CPDF_Stream* pStream) {
250 m_pSingleStream =
251 pdfium::MakeRetain<CPDF_StreamAcc>(pdfium::WrapRetain(pStream));
252 m_pSingleStream->LoadAllDataFiltered();
253 m_CurrentStage = Stage::kPrepareContent;
254 }
255
HandlePageContentArray(const CPDF_Array * pArray)256 bool CPDF_ContentParser::HandlePageContentArray(const CPDF_Array* pArray) {
257 m_nStreams = fxcrt::CollectionSize<uint32_t>(*pArray);
258 if (m_nStreams == 0)
259 return false;
260
261 m_StreamArray.resize(m_nStreams);
262 return true;
263 }
264
HandlePageContentFailure()265 void CPDF_ContentParser::HandlePageContentFailure() {
266 m_CurrentStage = Stage::kComplete;
267 }
268
GetData() const269 pdfium::span<const uint8_t> CPDF_ContentParser::GetData() const {
270 if (is_owned())
271 return absl::get<FixedTryAllocZeroedDataVector<uint8_t>>(m_Data).span();
272 return absl::get<pdfium::span<const uint8_t>>(m_Data);
273 }
274