Revert "Do not merge content streams to parse them."
This reverts commit 9d3fb11a985a8f2aefbf361c0030c7bce2248b8e.
Reason for revert: Break rendering of http://www.flbb.lu/bio_files/BIO_25_2005.pdf
Original change's description:
> Do not merge content streams to parse them.
>
> Removed the PrepareContent() step in CPDF_ContentParser, as its goal
> was to merge these streams.
>
> In Parse(), now loop through the m_StreamArray and parse each stream
> separately.
>
> Bug: pdfium:1051
> Change-Id: Id07f3efe766828081c8cf8601bf69c4f37e20e13
> Reviewed-on: https://pdfium-review.googlesource.com/33595
> Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
> Reviewed-by: Ryan Harrison <rharrison@chromium.org>
TBR=thestig@chromium.org,hnakashima@chromium.org,rharrison@chromium.org
Change-Id: I8186b2dac8538ca477302bb91e22dcc54124461a
Bug: pdfium:1051
Reviewed-on: https://pdfium-review.googlesource.com/33991
Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
Reviewed-by: Henrique Nakashima <hnakashima@chromium.org>
diff --git a/core/fpdfapi/page/cpdf_contentparser.cpp b/core/fpdfapi/page/cpdf_contentparser.cpp
index 471e707..29b5f12 100644
--- a/core/fpdfapi/page/cpdf_contentparser.cpp
+++ b/core/fpdfapi/page/cpdf_contentparser.cpp
@@ -16,6 +16,7 @@
#include "core/fpdfapi/parser/cpdf_dictionary.h"
#include "core/fpdfapi/parser/cpdf_stream.h"
#include "core/fpdfapi/parser/cpdf_stream_acc.h"
+#include "core/fxcrt/fx_safe_types.h"
#include "core/fxcrt/pauseindicator_iface.h"
#include "third_party/base/ptr_util.h"
@@ -37,11 +38,9 @@
CPDF_Stream* pStream = pContent->AsStream();
if (pStream) {
- RetainPtr<CPDF_StreamAcc> pSingleStream =
- pdfium::MakeRetain<CPDF_StreamAcc>(pStream);
- pSingleStream->LoadAllDataFiltered();
- m_StreamArray.push_back(pSingleStream);
- m_CurrentStage = Stage::kParse;
+ m_pSingleStream = pdfium::MakeRetain<CPDF_StreamAcc>(pStream);
+ m_pSingleStream->LoadAllDataFiltered();
+ m_CurrentStage = Stage::kPrepareContent;
return;
}
@@ -107,10 +106,10 @@
pState->SetFillAlpha(1.0f);
pState->SetSoftMask(nullptr);
}
- RetainPtr<CPDF_StreamAcc> pSingleStream =
- pdfium::MakeRetain<CPDF_StreamAcc>(pForm->GetStream());
- pSingleStream->LoadAllDataFiltered();
- m_StreamArray.push_back(pSingleStream);
+ m_pSingleStream = pdfium::MakeRetain<CPDF_StreamAcc>(pForm->GetStream());
+ m_pSingleStream->LoadAllDataFiltered();
+ m_pData.Reset(m_pSingleStream->GetData());
+ m_Size = m_pSingleStream->GetSize();
}
CPDF_ContentParser::~CPDF_ContentParser() {}
@@ -125,6 +124,9 @@
return true;
}
+ if (m_CurrentStage == Stage::kPrepareContent)
+ m_CurrentStage = PrepareContent();
+
while (m_CurrentStage == Stage::kParse) {
m_CurrentStage = Parse();
if (pPause && pPause->NeedToPauseNow())
@@ -148,12 +150,40 @@
m_StreamArray[m_CurrentOffset]->LoadAllDataFiltered();
m_CurrentOffset++;
- if (m_CurrentOffset >= m_nStreams) {
- m_CurrentOffset = 0;
+ return m_CurrentOffset == m_nStreams ? Stage::kPrepareContent
+ : Stage::kGetContent;
+}
+
+CPDF_ContentParser::Stage CPDF_ContentParser::PrepareContent() {
+ m_CurrentOffset = 0;
+
+ if (m_StreamArray.empty()) {
+ m_pData.Reset(m_pSingleStream->GetData());
+ m_Size = m_pSingleStream->GetSize();
return Stage::kParse;
}
- return Stage::kGetContent;
+ FX_SAFE_UINT32 safeSize = 0;
+ for (const auto& stream : m_StreamArray) {
+ safeSize += stream->GetSize();
+ safeSize += 1;
+ }
+ if (!safeSize.IsValid())
+ return Stage::kComplete;
+
+ m_Size = safeSize.ValueOrDie();
+ m_pData.Reset(
+ std::unique_ptr<uint8_t, FxFreeDeleter>(FX_Alloc(uint8_t, m_Size)));
+
+ uint32_t pos = 0;
+ for (const auto& stream : m_StreamArray) {
+ memcpy(m_pData.Get() + pos, stream->GetData(), stream->GetSize());
+ pos += stream->GetSize();
+ m_pData.Get()[pos++] = ' ';
+ }
+ m_StreamArray.clear();
+
+ return Stage::kParse;
}
CPDF_ContentParser::Stage CPDF_ContentParser::Parse() {
@@ -166,19 +196,12 @@
nullptr, m_parsedSet.get());
m_pParser->GetCurStates()->m_ColorState.SetDefault();
}
+ if (m_CurrentOffset >= m_Size)
+ return Stage::kCheckClip;
- m_CurrentOffset += m_pParser->Parse(
- m_StreamArray[m_CurrentStream]->GetData() + m_CurrentOffset,
- m_StreamArray[m_CurrentStream]->GetSize() - m_CurrentOffset,
- PARSE_STEP_LIMIT);
-
- if (m_CurrentOffset >= m_StreamArray[m_CurrentStream]->GetSize()) {
- m_CurrentOffset = 0;
- ++m_CurrentStream;
- if (m_CurrentStream >= m_nStreams)
- return Stage::kCheckClip;
- }
-
+ m_CurrentOffset +=
+ m_pParser->Parse(m_pData.Get() + m_CurrentOffset,
+ m_Size - m_CurrentOffset, PARSE_STEP_LIMIT);
return Stage::kParse;
}
diff --git a/core/fpdfapi/page/cpdf_contentparser.h b/core/fpdfapi/page/cpdf_contentparser.h
index 92a3e06..f9b491d 100644
--- a/core/fpdfapi/page/cpdf_contentparser.h
+++ b/core/fpdfapi/page/cpdf_contentparser.h
@@ -42,22 +42,26 @@
private:
enum class Stage : uint8_t {
kGetContent = 1,
+ kPrepareContent,
kParse,
kCheckClip,
kComplete,
};
Stage GetContent();
+ Stage PrepareContent();
Stage Parse();
Stage CheckClip();
Stage m_CurrentStage;
UnownedPtr<CPDF_PageObjectHolder> const m_pObjectHolder;
UnownedPtr<CPDF_Type3Char> m_pType3Char; // Only used when parsing forms.
+ RetainPtr<CPDF_StreamAcc> m_pSingleStream;
std::vector<RetainPtr<CPDF_StreamAcc>> m_StreamArray;
+ MaybeOwned<uint8_t, FxFreeDeleter> m_pData;
uint32_t m_nStreams = 0;
+ uint32_t m_Size = 0;
uint32_t m_CurrentOffset = 0;
- uint32_t m_CurrentStream = 0;
// Only used when parsing pages.
std::unique_ptr<std::set<const uint8_t*>> m_parsedSet;
