Fill m_ContentStream field in CPDF_PageObject.

From the comment of CPDF_PageObject::GetContentStream():

"""
Get what content stream the object was parsed from in its page.

This number is the index of the content stream in the "Contents" array,
or 0 if there is a single content stream. If the object is newly
created, -1 is returned.

If the object is spread among more than one content stream, this is
the index of the last one.
"""

Bug: pdfium:1051
Change-Id: I9f7804af4f263dda0422e9542e025e3320ff7c31
Reviewed-on: https://pdfium-review.googlesource.com/34250
Commit-Queue: Henrique Nakashima <hnakashima@chromium.org>
Reviewed-by: dsinclair <dsinclair@chromium.org>

16 files changed, 93 insertions, 28 deletions

diff --git a/core/fpdfapi/page/cpdf_contentparser.cpp b/core/fpdfapi/page/cpdf_contentparser.cpp
index 77cfade1da..2bb376e841 100644
--- a/core/fpdfapi/page/cpdf_contentparser.cpp
+++ b/core/fpdfapi/page/cpdf_contentparser.cpp
@@ -164,11 +164,13 @@ CPDF_ContentParser::Stage CPDF_ContentParser::PrepareContent() {
 
   FX_SAFE_UINT32 safeSize = 0;
   for (const auto& stream : m_StreamArray) {
+    m_StreamSegmentOffsets.push_back(safeSize.ValueOrDie());
+
     safeSize += stream->GetSize();
     safeSize += 1;
+    if (!safeSize.IsValid())
+      return Stage::kComplete;
   }
-  if (!safeSize.IsValid())
-    return Stage::kComplete;
 
   m_Size = safeSize.ValueOrDie();
   m_pData.Reset(
@@ -198,9 +200,12 @@ CPDF_ContentParser::Stage CPDF_ContentParser::Parse() {
   if (m_CurrentOffset >= m_Size)
     return Stage::kCheckClip;
 
-  m_CurrentOffset +=
-      m_pParser->Parse(m_pData.Get() + m_CurrentOffset,
-                       m_Size - m_CurrentOffset, PARSE_STEP_LIMIT);
+  if (m_StreamSegmentOffsets.empty())
+    m_StreamSegmentOffsets.push_back(0);
+
+  m_CurrentOffset += m_pParser->Parse(m_pData.Get() + m_CurrentOffset,
+                                      m_Size - m_CurrentOffset,
+                                      PARSE_STEP_LIMIT, m_StreamSegmentOffsets);
   return Stage::kParse;
 }
 
diff --git a/core/fpdfapi/page/cpdf_contentparser.h b/core/fpdfapi/page/cpdf_contentparser.h
index f9b491defa..b5db5d5e13 100644
--- a/core/fpdfapi/page/cpdf_contentparser.h
+++ b/core/fpdfapi/page/cpdf_contentparser.h
@@ -58,6 +58,7 @@ class CPDF_ContentParser {
   UnownedPtr<CPDF_Type3Char> m_pType3Char;  // Only used when parsing forms.
   RetainPtr<CPDF_StreamAcc> m_pSingleStream;
   std::vector<RetainPtr<CPDF_StreamAcc>> m_StreamArray;
+  std::vector<uint32_t> m_StreamSegmentOffsets;
   MaybeOwned<uint8_t, FxFreeDeleter> m_pData;
   uint32_t m_nStreams = 0;
   uint32_t m_Size = 0;
diff --git a/core/fpdfapi/page/cpdf_formobject.cpp b/core/fpdfapi/page/cpdf_formobject.cpp
index eca92ca0ff..22ac0d3823 100644
--- a/core/fpdfapi/page/cpdf_formobject.cpp
+++ b/core/fpdfapi/page/cpdf_formobject.cpp
@@ -10,9 +10,12 @@
 
 #include "core/fpdfapi/page/cpdf_form.h"
 
-CPDF_FormObject::CPDF_FormObject(std::unique_ptr<CPDF_Form> pForm,
+CPDF_FormObject::CPDF_FormObject(int32_t content_stream,
+                                 std::unique_ptr<CPDF_Form> pForm,
                                  const CFX_Matrix& matrix)
-    : m_pForm(std::move(pForm)), m_FormMatrix(matrix) {}
+    : CPDF_PageObject(content_stream),
+      m_pForm(std::move(pForm)),
+      m_FormMatrix(matrix) {}
 
 CPDF_FormObject::~CPDF_FormObject() {}
 
diff --git a/core/fpdfapi/page/cpdf_formobject.h b/core/fpdfapi/page/cpdf_formobject.h
index c723cc0eaf..b229dce444 100644
--- a/core/fpdfapi/page/cpdf_formobject.h
+++ b/core/fpdfapi/page/cpdf_formobject.h
@@ -16,7 +16,9 @@ class CPDF_Form;
 
 class CPDF_FormObject : public CPDF_PageObject {
  public:
-  CPDF_FormObject(std::unique_ptr<CPDF_Form> pForm, const CFX_Matrix& matrix);
+  CPDF_FormObject(int32_t content_stream,
+                  std::unique_ptr<CPDF_Form> pForm,
+                  const CFX_Matrix& matrix);
   ~CPDF_FormObject() override;
 
   // CPDF_PageObject:
diff --git a/core/fpdfapi/page/cpdf_imageobject.cpp b/core/fpdfapi/page/cpdf_imageobject.cpp
index 3b5a740155..516a6e8cad 100644
--- a/core/fpdfapi/page/cpdf_imageobject.cpp
+++ b/core/fpdfapi/page/cpdf_imageobject.cpp
@@ -12,7 +12,10 @@
 #include "core/fpdfapi/page/cpdf_image.h"
 #include "core/fpdfapi/parser/cpdf_document.h"
 
-CPDF_ImageObject::CPDF_ImageObject() {}
+CPDF_ImageObject::CPDF_ImageObject(int32_t content_stream)
+    : CPDF_PageObject(content_stream) {}
+
+CPDF_ImageObject::CPDF_ImageObject() : CPDF_ImageObject(kNoContentStream) {}
 
 CPDF_ImageObject::~CPDF_ImageObject() {
   MaybePurgeCache();
diff --git a/core/fpdfapi/page/cpdf_imageobject.h b/core/fpdfapi/page/cpdf_imageobject.h
index 16a506e537..d54ef8d818 100644
--- a/core/fpdfapi/page/cpdf_imageobject.h
+++ b/core/fpdfapi/page/cpdf_imageobject.h
@@ -16,6 +16,7 @@ class CPDF_Image;
 
 class CPDF_ImageObject : public CPDF_PageObject {
  public:
+  explicit CPDF_ImageObject(int32_t content_stream);
   CPDF_ImageObject();
   ~CPDF_ImageObject() override;
 
diff --git a/core/fpdfapi/page/cpdf_pageobject.cpp b/core/fpdfapi/page/cpdf_pageobject.cpp
index 8bb5bf5978..604309f6e6 100644
--- a/core/fpdfapi/page/cpdf_pageobject.cpp
+++ b/core/fpdfapi/page/cpdf_pageobject.cpp
@@ -6,7 +6,12 @@
 
 #include "core/fpdfapi/page/cpdf_pageobject.h"
 
-CPDF_PageObject::CPDF_PageObject() : m_bDirty(false) {}
+constexpr int32_t CPDF_PageObject::kNoContentStream;
+
+CPDF_PageObject::CPDF_PageObject(int32_t content_stream)
+    : m_bDirty(false), m_ContentStream(content_stream) {}
+
+CPDF_PageObject::CPDF_PageObject() : CPDF_PageObject(kNoContentStream) {}
 
 CPDF_PageObject::~CPDF_PageObject() {}
 
diff --git a/core/fpdfapi/page/cpdf_pageobject.h b/core/fpdfapi/page/cpdf_pageobject.h
index d23cd971f8..39e7629541 100644
--- a/core/fpdfapi/page/cpdf_pageobject.h
+++ b/core/fpdfapi/page/cpdf_pageobject.h
@@ -28,6 +28,9 @@ class CPDF_PageObject : public CPDF_GraphicStates {
     FORM,
   };
 
+  static constexpr int32_t kNoContentStream = -1;
+
+  explicit CPDF_PageObject(int32_t content_stream);
   CPDF_PageObject();
   ~CPDF_PageObject() override;
 
@@ -59,6 +62,14 @@ class CPDF_PageObject : public CPDF_GraphicStates {
   }
   FX_RECT GetBBox(const CFX_Matrix* pMatrix) const;
 
+  // Get what content stream the object was parsed from in its page. This number
+  // is the index of the content stream in the "Contents" array, or 0 if there
+  // is a single content stream. If the object is newly created,
+  // kNoContentStream is returned.
+  // If the object is spread among more than one content stream, this is the
+  // index of the last stream.
+  int32_t GetContentStream() const { return m_ContentStream; }
+
   float m_Left;
   float m_Right;
   float m_Top;
@@ -73,6 +84,7 @@ class CPDF_PageObject : public CPDF_GraphicStates {
   void operator=(const CPDF_PageObject& src) = delete;
 
   bool m_bDirty;
+  int32_t m_ContentStream;
 };
 
 #endif  // CORE_FPDFAPI_PAGE_CPDF_PAGEOBJECT_H_
diff --git a/core/fpdfapi/page/cpdf_pathobject.cpp b/core/fpdfapi/page/cpdf_pathobject.cpp
index d8c2cb8741..0882dc17f0 100644
--- a/core/fpdfapi/page/cpdf_pathobject.cpp
+++ b/core/fpdfapi/page/cpdf_pathobject.cpp
@@ -6,7 +6,10 @@
 
 #include "core/fpdfapi/page/cpdf_pathobject.h"
 
-CPDF_PathObject::CPDF_PathObject() : m_FillType(0), m_bStroke(false) {}
+CPDF_PathObject::CPDF_PathObject(int32_t content_stream)
+    : CPDF_PageObject(content_stream), m_FillType(0), m_bStroke(false) {}
+
+CPDF_PathObject::CPDF_PathObject() : CPDF_PathObject(kNoContentStream) {}
 
 CPDF_PathObject::~CPDF_PathObject() {}
 
diff --git a/core/fpdfapi/page/cpdf_pathobject.h b/core/fpdfapi/page/cpdf_pathobject.h
index 58499b5a53..5155c40c3d 100644
--- a/core/fpdfapi/page/cpdf_pathobject.h
+++ b/core/fpdfapi/page/cpdf_pathobject.h
@@ -14,6 +14,7 @@
 
 class CPDF_PathObject : public CPDF_PageObject {
  public:
+  explicit CPDF_PathObject(int32_t content_stream);
   CPDF_PathObject();
   ~CPDF_PathObject() override;
 
diff --git a/core/fpdfapi/page/cpdf_shadingobject.cpp b/core/fpdfapi/page/cpdf_shadingobject.cpp
index 1b16ac4862..725e2e4591 100644
--- a/core/fpdfapi/page/cpdf_shadingobject.cpp
+++ b/core/fpdfapi/page/cpdf_shadingobject.cpp
@@ -9,9 +9,10 @@
 #include "core/fpdfapi/page/cpdf_shadingpattern.h"
 #include "core/fpdfapi/parser/cpdf_document.h"
 
-CPDF_ShadingObject::CPDF_ShadingObject(CPDF_ShadingPattern* pattern,
+CPDF_ShadingObject::CPDF_ShadingObject(int32_t content_stream,
+                                       CPDF_ShadingPattern* pattern,
                                        const CFX_Matrix& matrix)
-    : m_pShading(pattern), m_Matrix(matrix) {}
+    : CPDF_PageObject(content_stream), m_pShading(pattern), m_Matrix(matrix) {}
 
 CPDF_ShadingObject::~CPDF_ShadingObject() {}
 
diff --git a/core/fpdfapi/page/cpdf_shadingobject.h b/core/fpdfapi/page/cpdf_shadingobject.h
index 80e062c729..69b606749d 100644
--- a/core/fpdfapi/page/cpdf_shadingobject.h
+++ b/core/fpdfapi/page/cpdf_shadingobject.h
@@ -15,7 +15,9 @@ class CPDF_ShadingPattern;
 
 class CPDF_ShadingObject : public CPDF_PageObject {
  public:
-  CPDF_ShadingObject(CPDF_ShadingPattern* pattern, const CFX_Matrix& matrix);
+  CPDF_ShadingObject(int32_t content_stream,
+                     CPDF_ShadingPattern* pattern,
+                     const CFX_Matrix& matrix);
   ~CPDF_ShadingObject() override;
 
   // CPDF_PageObject:
diff --git a/core/fpdfapi/page/cpdf_streamcontentparser.cpp b/core/fpdfapi/page/cpdf_streamcontentparser.cpp
index 7562fb37a6..0cc81f1e9b 100644
--- a/core/fpdfapi/page/cpdf_streamcontentparser.cpp
+++ b/core/fpdfapi/page/cpdf_streamcontentparser.cpp
@@ -785,7 +785,8 @@ void CPDF_StreamContentParser::AddForm(CPDF_Stream* pStream) {
   CFX_Matrix matrix = m_pCurStates->m_CTM;
   matrix.Concat(m_mtContentToUser);
 
-  auto pFormObj = pdfium::MakeUnique<CPDF_FormObject>(std::move(form), matrix);
+  auto pFormObj = pdfium::MakeUnique<CPDF_FormObject>(GetCurrentStreamIndex(),
+                                                      std::move(form), matrix);
   if (!m_pObjectHolder->BackgroundAlphaNeeded() &&
       pFormObj->form()->BackgroundAlphaNeeded()) {
     m_pObjectHolder->SetBackgroundAlphaNeeded(true);
@@ -800,14 +801,16 @@ CPDF_ImageObject* CPDF_StreamContentParser::AddImage(
   if (!pStream)
     return nullptr;
 
-  auto pImageObj = pdfium::MakeUnique<CPDF_ImageObject>();
+  auto pImageObj =
+      pdfium::MakeUnique<CPDF_ImageObject>(GetCurrentStreamIndex());
   pImageObj->SetImage(
       pdfium::MakeRetain<CPDF_Image>(m_pDocument.Get(), std::move(pStream)));
   return AddImageObject(std::move(pImageObj));
 }
 
 CPDF_ImageObject* CPDF_StreamContentParser::AddImage(uint32_t streamObjNum) {
-  auto pImageObj = pdfium::MakeUnique<CPDF_ImageObject>();
+  auto pImageObj =
+      pdfium::MakeUnique<CPDF_ImageObject>(GetCurrentStreamIndex());
   pImageObj->SetImage(m_pDocument->LoadImageFromPageData(streamObjNum));
   return AddImageObject(std::move(pImageObj));
 }
@@ -817,7 +820,8 @@ CPDF_ImageObject* CPDF_StreamContentParser::AddImage(
   if (!pImage)
     return nullptr;
 
-  auto pImageObj = pdfium::MakeUnique<CPDF_ImageObject>();
+  auto pImageObj =
+      pdfium::MakeUnique<CPDF_ImageObject>(GetCurrentStreamIndex());
   pImageObj->SetImage(
       m_pDocument->GetPageData()->GetImage(pImage->GetStream()->GetObjNum()));
 
@@ -1084,7 +1088,8 @@ void CPDF_StreamContentParser::Handle_ShadeFill() {
 
   CFX_Matrix matrix = m_pCurStates->m_CTM;
   matrix.Concat(m_mtContentToUser);
-  auto pObj = pdfium::MakeUnique<CPDF_ShadingObject>(pShading, matrix);
+  auto pObj = pdfium::MakeUnique<CPDF_ShadingObject>(GetCurrentStreamIndex(),
+                                                     pShading, matrix);
   SetGraphicStates(pObj.get(), false, false, false);
   CFX_FloatRect bbox =
       pObj->m_ClipPath.HasRef() ? pObj->m_ClipPath.GetClipBox() : m_BBox;
@@ -1219,7 +1224,7 @@ void CPDF_StreamContentParser::AddTextObject(ByteString* pStrs,
       pFont->IsType3Font() ? TextRenderingMode::MODE_FILL
                            : m_pCurStates->m_TextState.GetTextMode();
   {
-    auto pText = pdfium::MakeUnique<CPDF_TextObject>();
+    auto pText = pdfium::MakeUnique<CPDF_TextObject>(GetCurrentStreamIndex());
     m_pLastTextObject = pText.get();
     SetGraphicStates(m_pLastTextObject.Get(), true, true, true);
     if (TextRenderingModeIsStrokeMode(text_mode)) {
@@ -1258,6 +1263,12 @@ void CPDF_StreamContentParser::AddTextObject(ByteString* pStrs,
   }
 }
 
+int32_t CPDF_StreamContentParser::GetCurrentStreamIndex() {
+  auto it = std::upper_bound(m_StreamStartOffsets.begin(),
+                             m_StreamStartOffsets.end(), m_pSyntax->GetPos());
+  return (it - m_StreamStartOffsets.begin()) - 1;
+}
+
 void CPDF_StreamContentParser::Handle_ShowText() {
   ByteString str = GetString(0);
   if (str.IsEmpty()) {
@@ -1456,7 +1467,8 @@ void CPDF_StreamContentParser::AddPathObject(int FillType, bool bStroke) {
   CFX_Matrix matrix = m_pCurStates->m_CTM;
   matrix.Concat(m_mtContentToUser);
   if (bStroke || FillType) {
-    auto pPathObj = pdfium::MakeUnique<CPDF_PathObject>();
+    auto pPathObj =
+        pdfium::MakeUnique<CPDF_PathObject>(GetCurrentStreamIndex());
     pPathObj->m_bStroke = bStroke;
     pPathObj->m_FillType = FillType;
     pPathObj->m_Path = Path;
@@ -1474,22 +1486,27 @@ void CPDF_StreamContentParser::AddPathObject(int FillType, bool bStroke) {
   }
 }
 
-uint32_t CPDF_StreamContentParser::Parse(const uint8_t* pData,
-                                         uint32_t dwSize,
-                                         uint32_t max_cost) {
+uint32_t CPDF_StreamContentParser::Parse(
+    const uint8_t* pData,
+    uint32_t dwSize,
+    uint32_t max_cost,
+    const std::vector<uint32_t>& stream_start_offsets) {
   if (m_ParsedSet->size() > kMaxFormLevel ||
       pdfium::ContainsKey(*m_ParsedSet, pData))
     return dwSize;
 
+  m_StreamStartOffsets = stream_start_offsets;
+
   pdfium::ScopedSetInsertion<const uint8_t*> scopedInsert(m_ParsedSet.Get(),
                                                           pData);
 
-  uint32_t InitObjCount = m_pObjectHolder->GetPageObjectList()->size();
+  uint32_t init_obj_count = m_pObjectHolder->GetPageObjectList()->size();
   CPDF_StreamParser syntax(pdfium::make_span(pData, dwSize),
                            m_pDocument->GetByteStringPool());
   CPDF_StreamParserAutoClearer auto_clearer(&m_pSyntax, &syntax);
   while (1) {
-    uint32_t cost = m_pObjectHolder->GetPageObjectList()->size() - InitObjCount;
+    uint32_t cost =
+        m_pObjectHolder->GetPageObjectList()->size() - init_obj_count;
     if (max_cost && cost >= max_cost) {
       break;
     }
diff --git a/core/fpdfapi/page/cpdf_streamcontentparser.h b/core/fpdfapi/page/cpdf_streamcontentparser.h
index 438be024cc..adcb2a5b47 100644
--- a/core/fpdfapi/page/cpdf_streamcontentparser.h
+++ b/core/fpdfapi/page/cpdf_streamcontentparser.h
@@ -43,7 +43,10 @@ class CPDF_StreamContentParser {
                            std::set<const uint8_t*>* parsedSet);
   ~CPDF_StreamContentParser();
 
-  uint32_t Parse(const uint8_t* pData, uint32_t dwSize, uint32_t max_cost);
+  uint32_t Parse(const uint8_t* pData,
+                 uint32_t dwSize,
+                 uint32_t max_cost,
+                 const std::vector<uint32_t>& stream_start_offsets);
   CPDF_PageObjectHolder* GetPageObjectHolder() const {
     return m_pObjectHolder.Get();
   }
@@ -127,6 +130,7 @@ class CPDF_StreamContentParser {
 
   std::vector<float> GetColors() const;
   std::vector<float> GetNamedColors() const;
+  int32_t GetCurrentStreamIndex();
 
   void Handle_CloseFillStrokePath();
   void Handle_FillStrokePath();
@@ -230,6 +234,7 @@ class CPDF_StreamContentParser {
   std::vector<std::unique_ptr<CPDF_AllStates>> m_StateStack;
   float m_Type3Data[6];
   ContentParam m_ParamBuf[kParamBufSize];
+  std::vector<uint32_t> m_StreamStartOffsets;
 };
 
 #endif  // CORE_FPDFAPI_PAGE_CPDF_STREAMCONTENTPARSER_H_
diff --git a/core/fpdfapi/page/cpdf_textobject.cpp b/core/fpdfapi/page/cpdf_textobject.cpp
index 36a4722773..e678d5fc10 100644
--- a/core/fpdfapi/page/cpdf_textobject.cpp
+++ b/core/fpdfapi/page/cpdf_textobject.cpp
@@ -18,7 +18,10 @@ CPDF_TextObjectItem::CPDF_TextObjectItem() : m_CharCode(0) {}
 
 CPDF_TextObjectItem::~CPDF_TextObjectItem() = default;
 
-CPDF_TextObject::CPDF_TextObject() {}
+CPDF_TextObject::CPDF_TextObject(int32_t content_stream)
+    : CPDF_PageObject(content_stream) {}
+
+CPDF_TextObject::CPDF_TextObject() : CPDF_TextObject(kNoContentStream) {}
 
 CPDF_TextObject::~CPDF_TextObject() {
   // Move m_CharCodes to a local variable so it will be captured in crash dumps,
diff --git a/core/fpdfapi/page/cpdf_textobject.h b/core/fpdfapi/page/cpdf_textobject.h
index a6fc62369c..d3b6dcc3de 100644
--- a/core/fpdfapi/page/cpdf_textobject.h
+++ b/core/fpdfapi/page/cpdf_textobject.h
@@ -25,6 +25,7 @@ class CPDF_TextObjectItem {
 
 class CPDF_TextObject : public CPDF_PageObject {
  public:
+  explicit CPDF_TextObject(int32_t content_stream);
   CPDF_TextObject();
   ~CPDF_TextObject() override;