Set default graphics before generating page contents

In this CL, the content generator sets some default graphics states
before processing the page objects. In particular, a default ExtGState
is now set before processing, and the last CTM is now stored right
after parsing finishes: the only command to change matrix is ctm, and
it concatenates, so inverting requires knowing the current value.

Bug: pdfium:779
Change-Id: I35b1c07550ce91839fb0e20fbf717e3e80c9b9d6
Reviewed-on: https://pdfium-review.googlesource.com/7070
Commit-Queue: Nicolás Peña <npm@chromium.org>
Reviewed-by: dsinclair <dsinclair@chromium.org>

5 files changed, 49 insertions, 0 deletions

diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
index eeb6f81721..10b3933742 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
@@ -65,9 +65,24 @@ void CPDF_PageContentGenerator::GenerateContent() {
 
   CPDF_Document* pDoc = m_pDocument.Get();
   std::ostringstream buf;
+
+  // Set the default graphic state values
+  buf << "q\n";
+  if (!m_pObjHolder->GetLastCTM().IsIdentity()) {
+    CFX_Matrix reverse;
+    reverse.SetReverse(m_pObjHolder->GetLastCTM());
+    buf << reverse << " cm\n";
+  }
+  ProcessDefaultGraphics(&buf);
+
+  // Process the page objects
   if (!ProcessPageObjects(&buf))
     return;
 
+  // Return graphics to original state
+  buf << "Q\n";
+
+  // Add buffer to a stream in page's 'Contents'
   CPDF_Dictionary* pPageDict = m_pObjHolder->m_pFormDict.Get();
   CPDF_Object* pContent =
       pPageDict ? pPageDict->GetObjectFor("Contents") : nullptr;
@@ -142,6 +157,7 @@ bool CPDF_PageContentGenerator::ProcessPageObjects(std::ostringstream* buf) {
   for (auto& pPageObj : m_pageObjects) {
     if (m_pObjHolder->IsPage() && !pPageObj->IsDirty())
       continue;
+
     bDirty = true;
     if (CPDF_ImageObject* pImageObject = pPageObj->AsImage())
       ProcessImage(buf, pImageObject);
@@ -304,6 +320,32 @@ void CPDF_PageContentGenerator::ProcessGraphics(std::ostringstream* buf,
   *buf << "/" << PDF_NameEncode(name) << " gs ";
 }
 
+void CPDF_PageContentGenerator::ProcessDefaultGraphics(
+    std::ostringstream* buf) {
+  *buf << "0 0 0 RG 0 0 0 rg 1 w "
+       << static_cast<int>(CFX_GraphStateData::LineCapButt) << " J "
+       << static_cast<int>(CFX_GraphStateData::LineJoinMiter) << " j\n";
+  GraphicsData defaultGraphics;
+  defaultGraphics.fillAlpha = 1.0f;
+  defaultGraphics.strokeAlpha = 1.0f;
+  defaultGraphics.blendType = FXDIB_BLEND_NORMAL;
+  auto it = m_pObjHolder->m_GraphicsMap.find(defaultGraphics);
+  CFX_ByteString name;
+  if (it != m_pObjHolder->m_GraphicsMap.end()) {
+    name = it->second;
+  } else {
+    auto gsDict = pdfium::MakeUnique<CPDF_Dictionary>();
+    gsDict->SetNewFor<CPDF_Number>("ca", defaultGraphics.fillAlpha);
+    gsDict->SetNewFor<CPDF_Number>("CA", defaultGraphics.strokeAlpha);
+    gsDict->SetNewFor<CPDF_Name>("BM", "Normal");
+    CPDF_Object* pDict = m_pDocument->AddIndirectObject(std::move(gsDict));
+    uint32_t dwObjNum = pDict->GetObjNum();
+    name = RealizeResource(dwObjNum, "ExtGState");
+    m_pObjHolder->m_GraphicsMap[defaultGraphics] = name;
+  }
+  *buf << "/" << PDF_NameEncode(name).c_str() << " gs ";
+}
+
 // This method adds text to the buffer, BT begins the text object, ET ends it.
 // Tm sets the text matrix (allows positioning and transforming text).
 // Tf sets the font name (from Font in Resources) and font size.
diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
index 2d90eb4465..5295d8747b 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
@@ -35,6 +35,7 @@ class CPDF_PageContentGenerator {
   void ProcessPath(std::ostringstream* buf, CPDF_PathObject* pPathObj);
   void ProcessImage(std::ostringstream* buf, CPDF_ImageObject* pImageObj);
   void ProcessGraphics(std::ostringstream* buf, CPDF_PageObject* pPageObj);
+  void ProcessDefaultGraphics(std::ostringstream* buf);
   void ProcessText(std::ostringstream* buf, CPDF_TextObject* pTextObj);
   CFX_ByteString RealizeResource(uint32_t dwResourceObjNum,
                                  const CFX_ByteString& bsType);
diff --git a/core/fpdfapi/page/cpdf_contentparser.h b/core/fpdfapi/page/cpdf_contentparser.h
index 58a977301f..b18e07036c 100644
--- a/core/fpdfapi/page/cpdf_contentparser.h
+++ b/core/fpdfapi/page/cpdf_contentparser.h
@@ -29,6 +29,7 @@ class CPDF_ContentParser {
   ~CPDF_ContentParser();
 
   ParseStatus GetStatus() const { return m_Status; }
+  CPDF_StreamContentParser* GetParser() const { return m_pParser.get(); }
   void Start(CPDF_Page* pPage);
   void Start(CPDF_Form* pForm,
              CPDF_AllStates* pGraphicStates,
diff --git a/core/fpdfapi/page/cpdf_pageobjectholder.cpp b/core/fpdfapi/page/cpdf_pageobjectholder.cpp
index 50ef780396..614fa34d2b 100644
--- a/core/fpdfapi/page/cpdf_pageobjectholder.cpp
+++ b/core/fpdfapi/page/cpdf_pageobjectholder.cpp
@@ -8,6 +8,7 @@
 
 #include <algorithm>
 
+#include "core/fpdfapi/page/cpdf_allstates.h"
 #include "core/fpdfapi/page/cpdf_contentparser.h"
 #include "core/fpdfapi/page/cpdf_pageobject.h"
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
@@ -38,6 +39,8 @@ void CPDF_PageObjectHolder::ContinueParse(IFX_Pause* pPause) {
     return;
 
   m_ParseState = CONTENT_PARSED;
+  if (m_pParser->GetParser() && m_pParser->GetParser()->GetCurStates())
+    m_LastCTM = m_pParser->GetParser()->GetCurStates()->m_CTM;
   m_pParser.reset();
 }
 
diff --git a/core/fpdfapi/page/cpdf_pageobjectholder.h b/core/fpdfapi/page/cpdf_pageobjectholder.h
index 4733e06b2c..87ebbc8460 100644
--- a/core/fpdfapi/page/cpdf_pageobjectholder.h
+++ b/core/fpdfapi/page/cpdf_pageobjectholder.h
@@ -55,6 +55,7 @@ class CPDF_PageObjectHolder {
   const CPDF_PageObjectList* GetPageObjectList() const {
     return &m_PageObjectList;
   }
+  const CFX_Matrix& GetLastCTM() const { return m_LastCTM; }
 
   bool BackgroundAlphaNeeded() const { return m_bBackgroundAlphaNeeded; }
   void SetBackgroundAlphaNeeded(bool needed) {
@@ -89,6 +90,7 @@ class CPDF_PageObjectHolder {
   ParseState m_ParseState;
   std::unique_ptr<CPDF_ContentParser> m_pParser;
   CPDF_PageObjectList m_PageObjectList;
+  CFX_Matrix m_LastCTM;
 };
 
 #endif  // CORE_FPDFAPI_PAGE_CPDF_PAGEOBJECTHOLDER_H_