Unify CPDF_Document loading methods.

Change-Id: Ibf7aee942027adace7ec0831aefe0fe8c28e41cc
Reviewed-on: https://pdfium-review.googlesource.com/35610
Commit-Queue: Art Snake <art-snake@yandex-team.ru>
Reviewed-by: dsinclair <dsinclair@chromium.org>

5 files changed, 33 insertions, 30 deletions

diff --git a/core/fpdfapi/parser/cpdf_document.cpp b/core/fpdfapi/parser/cpdf_document.cpp
index 393a6f6785..8f727ed14a 100644
--- a/core/fpdfapi/parser/cpdf_document.cpp
+++ b/core/fpdfapi/parser/cpdf_document.cpp
@@ -191,9 +191,6 @@ CPDF_Document::CPDF_Document(std::unique_ptr<CPDF_Parser> pParser)
       m_pRootDict(nullptr),
       m_iNextPageToTraverse(0),
       m_bReachedMaxPageLevel(false),
-      m_bLinearized(false),
-      m_iFirstPageNo(0),
-      m_dwFirstPageObjNum(0),
       m_pDocPage(pdfium::MakeUnique<CPDF_DocPageData>(this)),
       m_pDocRender(pdfium::MakeUnique<CPDF_DocRenderData>(this)) {
   if (pParser)
@@ -226,17 +223,18 @@ void CPDF_Document::LoadDoc() {
   LoadPages();
 }
 
-void CPDF_Document::LoadLinearizedDoc(
-    const CPDF_LinearizedHeader* pLinearizationParams) {
-  m_bLinearized = true;
-  LoadDocInternal();
-  m_PageList.resize(pLinearizationParams->GetPageCount());
-  m_iFirstPageNo = pLinearizationParams->GetFirstPageNo();
-  m_dwFirstPageObjNum = pLinearizationParams->GetFirstPageObjNum();
-}
-
 void CPDF_Document::LoadPages() {
-  m_PageList.resize(RetrievePageCount());
+  const CPDF_LinearizedHeader* linearized_header =
+      m_pParser->GetLinearizedHeader();
+  if (!linearized_header) {
+    m_PageList.resize(RetrievePageCount());
+    return;
+  }
+
+  m_PageList.resize(linearized_header->GetPageCount());
+  DCHECK(linearized_header->GetFirstPageNo() < m_PageList.size());
+  m_PageList[linearized_header->GetFirstPageNo()] =
+      linearized_header->GetFirstPageObjNum();
 }
 
 CPDF_Dictionary* CPDF_Document::TraversePDFPages(int iPage,
@@ -329,15 +327,12 @@ CPDF_Dictionary* CPDF_Document::GetPageDictionary(int iPage) {
   if (!pdfium::IndexInBounds(m_PageList, iPage))
     return nullptr;
 
-  if (m_bLinearized && iPage == m_iFirstPageNo) {
-    if (CPDF_Dictionary* pDict =
-            ToDictionary(GetOrParseIndirectObject(m_dwFirstPageObjNum))) {
-      return pDict;
-    }
+  const uint32_t objnum = m_PageList[iPage];
+  if (objnum) {
+    CPDF_Dictionary* result = ToDictionary(GetOrParseIndirectObject(objnum));
+    if (result)
+      return result;
   }
-  uint32_t objnum = m_PageList[iPage];
-  if (objnum)
-    return ToDictionary(GetOrParseIndirectObject(objnum));
 
   CPDF_Dictionary* pPages = GetPagesDict();
   if (!pPages)
diff --git a/core/fpdfapi/parser/cpdf_document.h b/core/fpdfapi/parser/cpdf_document.h
index db659b38c8..8d0aa08009 100644
--- a/core/fpdfapi/parser/cpdf_document.h
+++ b/core/fpdfapi/parser/cpdf_document.h
@@ -97,7 +97,6 @@ class CPDF_Document : public CPDF_IndirectObjectHolder {
   RetainPtr<CPDF_IccProfile> LoadIccProfile(const CPDF_Stream* pStream);
 
   void LoadDoc();
-  void LoadLinearizedDoc(const CPDF_LinearizedHeader* pLinearizationParams);
   void LoadPages();
 
   void CreateNewDoc();
@@ -159,9 +158,6 @@ class CPDF_Document : public CPDF_IndirectObjectHolder {
   // Index of the next page that will be traversed from the page tree.
   int m_iNextPageToTraverse;
   bool m_bReachedMaxPageLevel;
-  bool m_bLinearized;
-  int m_iFirstPageNo;
-  uint32_t m_dwFirstPageObjNum;
   uint32_t m_ParsedPageCount = 0;
   std::unique_ptr<CPDF_DocPageData> m_pDocPage;
   std::unique_ptr<CPDF_DocRenderData> m_pDocRender;
diff --git a/core/fpdfapi/parser/cpdf_document_unittest.cpp b/core/fpdfapi/parser/cpdf_document_unittest.cpp
index fa52d3bc7f..522aa65fc9 100644
--- a/core/fpdfapi/parser/cpdf_document_unittest.cpp
+++ b/core/fpdfapi/parser/cpdf_document_unittest.cpp
@@ -215,13 +215,15 @@ TEST_F(cpdf_document_test, UseCachedPageObjNumIfHaveNotPagesDict) {
   // ObjNum can be added in CPDF_DataAvail::IsPageAvail, and PagesDict
   // can be not exists in this case.
   // (case, when hint table is used to page check in CPDF_DataAvail).
-  CPDF_Document document(pdfium::MakeUnique<CPDF_Parser>());
   auto dict = pdfium::MakeUnique<CPDF_Dictionary>();
   dict->SetNewFor<CPDF_Boolean>("Linearized", true);
   const int page_count = 100;
   dict->SetNewFor<CPDF_Number>("N", page_count);
-  TestLinearized linearized(dict.get());
-  document.LoadLinearizedDoc(&linearized);
+  auto linearized = pdfium::MakeUnique<TestLinearized>(dict.get());
+  auto parser = pdfium::MakeUnique<CPDF_Parser>();
+  parser->SetLinearizedHeader(std::move(linearized));
+  CPDF_Document document(std::move(parser));
+  document.LoadDoc();
   ASSERT_EQ(page_count, document.GetPageCount());
   CPDF_Object* page_stub = document.NewIndirect<CPDF_Dictionary>();
   const uint32_t obj_num = page_stub->GetObjNum();
diff --git a/core/fpdfapi/parser/cpdf_parser.cpp b/core/fpdfapi/parser/cpdf_parser.cpp
index 6cdd6ab789..599b04199d 100644
--- a/core/fpdfapi/parser/cpdf_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_parser.cpp
@@ -1287,6 +1287,11 @@ uint32_t CPDF_Parser::GetFirstPageNo() const {
   return m_pLinearized ? m_pLinearized->GetFirstPageNo() : 0;
 }
 
+void CPDF_Parser::SetLinearizedHeader(
+    std::unique_ptr<CPDF_LinearizedHeader> pLinearized) {
+  m_pLinearized = std::move(pLinearized);
+}
+
 std::unique_ptr<CPDF_Dictionary> CPDF_Parser::LoadTrailerV4() {
   if (m_pSyntax->GetKeyword() != "trailer")
     return nullptr;
@@ -1354,7 +1359,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
   if (eRet != SUCCESS)
     return eRet;
 
-  m_pDocument->LoadLinearizedDoc(m_pLinearized.get());
+  m_pDocument->LoadDoc();
   if (!m_pDocument->GetRoot() || m_pDocument->GetPageCount() == 0) {
     if (bXRefRebuilt)
       return FORMAT_ERROR;
@@ -1367,7 +1372,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
     if (eRet != SUCCESS)
       return eRet;
 
-    m_pDocument->LoadLinearizedDoc(m_pLinearized.get());
+    m_pDocument->LoadDoc();
     if (!m_pDocument->GetRoot())
       return FORMAT_ERROR;
   }
diff --git a/core/fpdfapi/parser/cpdf_parser.h b/core/fpdfapi/parser/cpdf_parser.h
index 705fad9bdb..b74f1d52ab 100644
--- a/core/fpdfapi/parser/cpdf_parser.h
+++ b/core/fpdfapi/parser/cpdf_parser.h
@@ -106,6 +106,11 @@ class CPDF_Parser {
       FX_FILESIZE* pResultPos);
 
   uint32_t GetFirstPageNo() const;
+  const CPDF_LinearizedHeader* GetLinearizedHeader() const {
+    return m_pLinearized.get();
+  }
+
+  void SetLinearizedHeader(std::unique_ptr<CPDF_LinearizedHeader> pLinearized);
 
  protected:
   enum class ObjectType : uint8_t {