Modified CPDF_PageContentGenerator to work with both Form and Page

CPDF_PageContentGenerator currently only works for CPDF_Page objects.
However, CPDF_Form is also a CPDF_PageObjectHolder like CPDF_Page, and
content streams can be generated for form objects too.

This CL modifies the content generator to work with both forms and
pages. The content generator will later be used on annotation objects'
CPDF_Form to simplify the code for annotation AP stream generation.

Change-Id: I1c50bdf0329d1f5788db23286ac72750355a10ed
Reviewed-on: https://pdfium-review.googlesource.com/6837
Commit-Queue: Nicolás Peña <npm@chromium.org>
Reviewed-by: dsinclair <dsinclair@chromium.org>
Reviewed-by: Nicolás Peña <npm@chromium.org>

7 files changed, 105 insertions, 41 deletions

diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
index 8b39a7a136..508581f551 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.cpp
@@ -49,9 +49,10 @@ bool GetColor(const CPDF_Color* pColor, float* rgb) {
 
 }  // namespace
 
-CPDF_PageContentGenerator::CPDF_PageContentGenerator(CPDF_Page* pPage)
-    : m_pPage(pPage), m_pDocument(m_pPage->m_pDocument.Get()) {
-  for (const auto& pObj : *pPage->GetPageObjectList()) {
+CPDF_PageContentGenerator::CPDF_PageContentGenerator(
+    CPDF_PageObjectHolder* pObjHolder)
+    : m_pObjHolder(pObjHolder), m_pDocument(pObjHolder->m_pDocument.Get()) {
+  for (const auto& pObj : *pObjHolder->GetPageObjectList()) {
     if (pObj)
       m_pageObjects.emplace_back(pObj.get());
   }
@@ -60,12 +61,14 @@ CPDF_PageContentGenerator::CPDF_PageContentGenerator(CPDF_Page* pPage)
 CPDF_PageContentGenerator::~CPDF_PageContentGenerator() {}
 
 void CPDF_PageContentGenerator::GenerateContent() {
+  ASSERT(m_pObjHolder->IsPage());
+
   CPDF_Document* pDoc = m_pDocument.Get();
   std::ostringstream buf;
   if (!ProcessPageObjects(&buf))
     return;
 
-  CPDF_Dictionary* pPageDict = m_pPage->m_pFormDict.Get();
+  CPDF_Dictionary* pPageDict = m_pObjHolder->m_pFormDict.Get();
   CPDF_Object* pContent =
       pPageDict ? pPageDict->GetObjectFor("Contents") : nullptr;
   CPDF_Stream* pStream = pDoc->NewIndirect<CPDF_Stream>();
@@ -111,14 +114,15 @@ CFX_ByteString CPDF_PageContentGenerator::RealizeResource(
     uint32_t dwResourceObjNum,
     const CFX_ByteString& bsType) {
   ASSERT(dwResourceObjNum);
-  if (!m_pPage->m_pResources) {
-    m_pPage->m_pResources = m_pDocument->NewIndirect<CPDF_Dictionary>();
-    m_pPage->m_pFormDict->SetNewFor<CPDF_Reference>(
-        "Resources", m_pDocument.Get(), m_pPage->m_pResources->GetObjNum());
+  if (!m_pObjHolder->m_pResources) {
+    m_pObjHolder->m_pResources = m_pDocument->NewIndirect<CPDF_Dictionary>();
+    m_pObjHolder->m_pFormDict->SetNewFor<CPDF_Reference>(
+        "Resources", m_pDocument.Get(),
+        m_pObjHolder->m_pResources->GetObjNum());
   }
-  CPDF_Dictionary* pResList = m_pPage->m_pResources->GetDictFor(bsType);
+  CPDF_Dictionary* pResList = m_pObjHolder->m_pResources->GetDictFor(bsType);
   if (!pResList)
-    pResList = m_pPage->m_pResources->SetNewFor<CPDF_Dictionary>(bsType);
+    pResList = m_pObjHolder->m_pResources->SetNewFor<CPDF_Dictionary>(bsType);
 
   CFX_ByteString name;
   int idnum = 1;
@@ -137,7 +141,7 @@ CFX_ByteString CPDF_PageContentGenerator::RealizeResource(
 bool CPDF_PageContentGenerator::ProcessPageObjects(std::ostringstream* buf) {
   bool bDirty = false;
   for (auto& pPageObj : m_pageObjects) {
-    if (!pPageObj->IsDirty())
+    if (m_pObjHolder->IsPage() && !pPageObj->IsDirty())
       continue;
     bDirty = true;
     if (CPDF_ImageObject* pImageObject = pPageObj->AsImage())
@@ -277,8 +281,8 @@ void CPDF_PageContentGenerator::ProcessGraphics(std::ostringstream* buf,
   }
 
   CFX_ByteString name;
-  auto it = m_pPage->m_GraphicsMap.find(graphD);
-  if (it != m_pPage->m_GraphicsMap.end()) {
+  auto it = m_pObjHolder->m_GraphicsMap.find(graphD);
+  if (it != m_pObjHolder->m_GraphicsMap.end()) {
     name = it->second;
   } else {
     auto gsDict = pdfium::MakeUnique<CPDF_Dictionary>();
@@ -296,7 +300,7 @@ void CPDF_PageContentGenerator::ProcessGraphics(std::ostringstream* buf,
     CPDF_Object* pDict = m_pDocument->AddIndirectObject(std::move(gsDict));
     uint32_t dwObjNum = pDict->GetObjNum();
     name = RealizeResource(dwObjNum, "ExtGState");
-    m_pPage->m_GraphicsMap[graphD] = name;
+    m_pObjHolder->m_GraphicsMap[graphD] = name;
   }
   *buf << "/" << PDF_NameEncode(name).c_str() << " gs ";
 }
@@ -321,9 +325,9 @@ void CPDF_PageContentGenerator::ProcessText(std::ostringstream* buf,
   else
     return;
   fontD.baseFont = pFont->GetBaseFont();
-  auto it = m_pPage->m_FontsMap.find(fontD);
+  auto it = m_pObjHolder->m_FontsMap.find(fontD);
   CFX_ByteString dictName;
-  if (it != m_pPage->m_FontsMap.end()) {
+  if (it != m_pObjHolder->m_FontsMap.end()) {
     dictName = it->second;
   } else {
     uint32_t dwObjNum = pFont->GetFontDict()->GetObjNum();
@@ -337,7 +341,7 @@ void CPDF_PageContentGenerator::ProcessText(std::ostringstream* buf,
       dwObjNum = pDict->GetObjNum();
     }
     dictName = RealizeResource(dwObjNum, "Font");
-    m_pPage->m_FontsMap[fontD] = dictName;
+    m_pObjHolder->m_FontsMap[fontD] = dictName;
   }
   *buf << "/" << PDF_NameEncode(dictName).c_str() << " "
        << pTextObj->GetFontSize() << " Tf ";
diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
index 19a2faa6bf..2d90eb4465 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator.h
@@ -16,22 +16,22 @@
 
 class CPDF_Document;
 class CPDF_ImageObject;
-class CPDF_Page;
 class CPDF_PageObject;
+class CPDF_PageObjectHolder;
 class CPDF_PathObject;
 class CPDF_TextObject;
 
 class CPDF_PageContentGenerator {
  public:
-  explicit CPDF_PageContentGenerator(CPDF_Page* pPage);
+  explicit CPDF_PageContentGenerator(CPDF_PageObjectHolder* pObjHolder);
   ~CPDF_PageContentGenerator();
 
   void GenerateContent();
+  bool ProcessPageObjects(std::ostringstream* buf);
 
  private:
   friend class CPDF_PageContentGeneratorTest;
 
-  bool ProcessPageObjects(std::ostringstream* buf);
   void ProcessPath(std::ostringstream* buf, CPDF_PathObject* pPathObj);
   void ProcessImage(std::ostringstream* buf, CPDF_ImageObject* pImageObj);
   void ProcessGraphics(std::ostringstream* buf, CPDF_PageObject* pPageObj);
@@ -39,7 +39,7 @@ class CPDF_PageContentGenerator {
   CFX_ByteString RealizeResource(uint32_t dwResourceObjNum,
                                  const CFX_ByteString& bsType);
 
-  CFX_UnownedPtr<CPDF_Page> const m_pPage;
+  CFX_UnownedPtr<CPDF_PageObjectHolder> const m_pObjHolder;
   CFX_UnownedPtr<CPDF_Document> const m_pDocument;
   std::vector<CFX_UnownedPtr<CPDF_PageObject>> m_pageObjects;
 };
diff --git a/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp b/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
index 91389eef5f..ef186666f1 100644
--- a/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
+++ b/core/fpdfapi/edit/cpdf_pagecontentgenerator_unittest.cpp
@@ -4,8 +4,12 @@
 
 #include "core/fpdfapi/edit/cpdf_pagecontentgenerator.h"
 
+#include <memory>
+#include <utility>
+
 #include "core/fpdfapi/cpdf_modulemgr.h"
 #include "core/fpdfapi/font/cpdf_font.h"
+#include "core/fpdfapi/page/cpdf_form.h"
 #include "core/fpdfapi/page/cpdf_page.h"
 #include "core/fpdfapi/page/cpdf_pathobject.h"
 #include "core/fpdfapi/page/cpdf_textobject.h"
@@ -34,7 +38,7 @@ class CPDF_PageContentGeneratorTest : public testing::Test {
   CPDF_Dictionary* TestGetResource(CPDF_PageContentGenerator* pGen,
                                    const CFX_ByteString& type,
                                    const CFX_ByteString& name) {
-    return pGen->m_pPage->m_pResources->GetDictFor(type)->GetDictFor(name);
+    return pGen->m_pObjHolder->m_pResources->GetDictFor(type)->GetDictFor(name);
   }
 
   void TestProcessText(CPDF_PageContentGenerator* pGen,
@@ -273,3 +277,47 @@ TEST_F(CPDF_PageContentGeneratorTest, ProcessText) {
   EXPECT_EQ("FontDescriptor", fontDesc->GetStringFor("Type"));
   EXPECT_EQ("Helvetica", fontDesc->GetStringFor("FontName"));
 }
+
+TEST_F(CPDF_PageContentGeneratorTest, ProcessEmptyForm) {
+  auto pDoc = pdfium::MakeUnique<CPDF_Document>(nullptr);
+  pDoc->CreateNewDoc();
+  auto pDict = pdfium::MakeUnique<CPDF_Dictionary>();
+  auto pStream = pdfium::MakeUnique<CPDF_Stream>(nullptr, 0, std::move(pDict));
+
+  // Create an empty form.
+  auto pTestForm =
+      pdfium::MakeUnique<CPDF_Form>(pDoc.get(), nullptr, pStream.get());
+  pTestForm->ParseContent(nullptr, nullptr, nullptr);
+  ASSERT_TRUE(pTestForm->IsParsed());
+
+  // The generated stream for the empty form should be an empty string.
+  CPDF_PageContentGenerator generator(pTestForm.get());
+  std::ostringstream buf;
+  generator.ProcessPageObjects(&buf);
+  EXPECT_EQ("", CFX_ByteString(buf));
+}
+
+TEST_F(CPDF_PageContentGeneratorTest, ProcessFormWithPath) {
+  auto pDoc = pdfium::MakeUnique<CPDF_Document>(nullptr);
+  pDoc->CreateNewDoc();
+  auto pDict = pdfium::MakeUnique<CPDF_Dictionary>();
+  const char content[] =
+      "q 3.102 4.67 m 5.45 0.29 l 4.24 3.15 4.65 2.98 3.456 0.24 c 3.102 4.67 "
+      "l h f Q\n";
+  size_t buf_len = FX_ArraySize(content);
+  std::unique_ptr<uint8_t, FxFreeDeleter> buf(FX_Alloc(uint8_t, buf_len));
+  memcpy(buf.get(), content, buf_len);
+  auto pStream = pdfium::MakeUnique<CPDF_Stream>(std::move(buf), buf_len,
+                                                 std::move(pDict));
+
+  // Create a form with a non-empty stream.
+  auto pTestForm =
+      pdfium::MakeUnique<CPDF_Form>(pDoc.get(), nullptr, pStream.get());
+  pTestForm->ParseContent(nullptr, nullptr, nullptr);
+  ASSERT_TRUE(pTestForm->IsParsed());
+
+  CPDF_PageContentGenerator generator(pTestForm.get());
+  std::ostringstream process_buf;
+  generator.ProcessPageObjects(&process_buf);
+  EXPECT_EQ(content, CFX_ByteString(process_buf));
+}
diff --git a/core/fpdfapi/page/cpdf_page.cpp b/core/fpdfapi/page/cpdf_page.cpp
index 075aa9023d..dfbd342f13 100644
--- a/core/fpdfapi/page/cpdf_page.cpp
+++ b/core/fpdfapi/page/cpdf_page.cpp
@@ -74,6 +74,10 @@ CPDF_Page::CPDF_Page(CPDF_Document* pDocument,
 
 CPDF_Page::~CPDF_Page() {}
 
+bool CPDF_Page::IsPage() const {
+  return true;
+}
+
 void CPDF_Page::StartParse() {
   if (m_ParseState == CONTENT_PARSED || m_ParseState == CONTENT_PARSING)
     return;
diff --git a/core/fpdfapi/page/cpdf_page.h b/core/fpdfapi/page/cpdf_page.h
index 0a080a58ae..82df60ce9f 100644
--- a/core/fpdfapi/page/cpdf_page.h
+++ b/core/fpdfapi/page/cpdf_page.h
@@ -7,7 +7,6 @@
 #ifndef CORE_FPDFAPI_PAGE_CPDF_PAGE_H_
 #define CORE_FPDFAPI_PAGE_CPDF_PAGE_H_
 
-#include <map>
 #include <memory>
 
 #include "core/fpdfapi/page/cpdf_pageobjectholder.h"
@@ -21,21 +20,6 @@ class CPDF_Object;
 class CPDF_PageRenderCache;
 class CPDF_PageRenderContext;
 
-// These structs are used to keep track of resources that have already been
-// generated in the page.
-struct GraphicsData {
-  float fillAlpha;
-  float strokeAlpha;
-  int blendType;
-  bool operator<(const GraphicsData& other) const;
-};
-
-struct FontData {
-  CFX_ByteString baseFont;
-  CFX_ByteString type;
-  bool operator<(const FontData& other) const;
-};
-
 class CPDF_Page : public CPDF_PageObjectHolder {
  public:
   class View {};  // Caller implements as desired, empty here due to layering.
@@ -45,6 +29,9 @@ class CPDF_Page : public CPDF_PageObjectHolder {
             bool bPageCache);
   ~CPDF_Page() override;
 
+  // CPDF_PageObjectHolder
+  bool IsPage() const override;
+
   void ParseContent();
 
   CFX_Matrix GetDisplayMatrix(int xPos,
@@ -68,9 +55,6 @@ class CPDF_Page : public CPDF_PageObjectHolder {
   View* GetView() const { return m_pView; }
   void SetView(View* pView) { m_pView = pView; }
 
-  std::map<GraphicsData, CFX_ByteString> m_GraphicsMap;
-  std::map<FontData, CFX_ByteString> m_FontsMap;
-
  private:
   void StartParse();
 
diff --git a/core/fpdfapi/page/cpdf_pageobjectholder.cpp b/core/fpdfapi/page/cpdf_pageobjectholder.cpp
index 53dfd75d5d..50ef780396 100644
--- a/core/fpdfapi/page/cpdf_pageobjectholder.cpp
+++ b/core/fpdfapi/page/cpdf_pageobjectholder.cpp
@@ -25,6 +25,10 @@ CPDF_PageObjectHolder::CPDF_PageObjectHolder(CPDF_Document* pDoc,
 
 CPDF_PageObjectHolder::~CPDF_PageObjectHolder() {}
 
+bool CPDF_PageObjectHolder::IsPage() const {
+  return false;
+}
+
 void CPDF_PageObjectHolder::ContinueParse(IFX_Pause* pPause) {
   if (!m_pParser)
     return;
diff --git a/core/fpdfapi/page/cpdf_pageobjectholder.h b/core/fpdfapi/page/cpdf_pageobjectholder.h
index c95ae50520..4733e06b2c 100644
--- a/core/fpdfapi/page/cpdf_pageobjectholder.h
+++ b/core/fpdfapi/page/cpdf_pageobjectholder.h
@@ -7,6 +7,7 @@
 #ifndef CORE_FPDFAPI_PAGE_CPDF_PAGEOBJECTHOLDER_H_
 #define CORE_FPDFAPI_PAGE_CPDF_PAGEOBJECTHOLDER_H_
 
+#include <map>
 #include <memory>
 #include <vector>
 
@@ -25,11 +26,28 @@ class CPDF_ContentParser;
 #define PDFTRANS_ISOLATED 0x0200
 #define PDFTRANS_KNOCKOUT 0x0400
 
+// These structs are used to keep track of resources that have already been
+// generated in the page object holder.
+struct GraphicsData {
+  float fillAlpha;
+  float strokeAlpha;
+  int blendType;
+  bool operator<(const GraphicsData& other) const;
+};
+
+struct FontData {
+  CFX_ByteString baseFont;
+  CFX_ByteString type;
+  bool operator<(const FontData& other) const;
+};
+
 class CPDF_PageObjectHolder {
  public:
   CPDF_PageObjectHolder(CPDF_Document* pDoc, CPDF_Dictionary* pFormDict);
   virtual ~CPDF_PageObjectHolder();
 
+  virtual bool IsPage() const;
+
   void ContinueParse(IFX_Pause* pPause);
   bool IsParsed() const { return m_ParseState == CONTENT_PARSED; }
 
@@ -56,6 +74,8 @@ class CPDF_PageObjectHolder {
   CFX_UnownedPtr<CPDF_Document> m_pDocument;
   CFX_UnownedPtr<CPDF_Dictionary> m_pPageResources;
   CFX_UnownedPtr<CPDF_Dictionary> m_pResources;
+  std::map<GraphicsData, CFX_ByteString> m_GraphicsMap;
+  std::map<FontData, CFX_ByteString> m_FontsMap;
   CFX_FloatRect m_BBox;
   int m_Transparency;