Unify some code

core/fpdfapi/parser/cpdf_parser.cpp

Index: core/fpdfapi/parser/cpdf_parser.cpp
diff --git a/core/fpdfapi/parser/cpdf_parser.cpp b/core/fpdfapi/parser/cpdf_parser.cpp
index cff0f77b3247107d4a33d724480115955f432e67..eef5bc38444ceea36af6b9bd3bd6de012f6759d9 100644
--- a/core/fpdfapi/parser/cpdf_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_parser.cpp
@@ -12,6 +12,7 @@
 #include "core/fpdfapi/parser/cpdf_crypto_handler.h"
 #include "core/fpdfapi/parser/cpdf_dictionary.h"
 #include "core/fpdfapi/parser/cpdf_document.h"
+#include "core/fpdfapi/parser/cpdf_linearized.h"
 #include "core/fpdfapi/parser/cpdf_number.h"
 #include "core/fpdfapi/parser/cpdf_reference.h"
 #include "core/fpdfapi/parser/cpdf_security_handler.h"
@@ -54,8 +55,6 @@ CPDF_Parser::CPDF_Parser()
       m_pTrailer(nullptr),
       m_pEncryptDict(nullptr),
       m_bVersionUpdated(false),
-      m_pLinearized(nullptr),
-      m_dwFirstPageNo(0),
       m_dwXrefStartObjNum(0) {
   m_pSyntax.reset(new CPDF_SyntaxParser);
 }
@@ -76,9 +75,6 @@ CPDF_Parser::~CPDF_Parser() {
     if (trailer)
       trailer->Release();
   }
-
-  if (m_pLinearized)
-    m_pLinearized->Release();
 }
 
 uint32_t CPDF_Parser::GetLastObjNum() const {
@@ -1414,6 +1410,10 @@ CPDF_Object* CPDF_Parser::ParseIndirectObjectAtByStrict(
   return pObj;
 }
 
+uint32_t CPDF_Parser::GetFirstPageNo() const {
+  return m_pLinearized ? m_pLinearized->GetFirstPageNo() : 0;
+}
+
 CPDF_Dictionary* CPDF_Parser::LoadTrailerV4() {
   if (m_pSyntax->GetKeyword() != "trailer")
     return nullptr;
@@ -1460,35 +1460,12 @@ bool CPDF_Parser::IsLinearizedFile(IFX_SeekableReadStream* pFileAccess,
     return false;
   }
 
-  m_pLinearized = m_pSyntax->GetObject(nullptr, objnum, gennum, true);
+  m_pLinearized = CPDF_Linearized::CreateForObject(
+      UniqueObject(m_pSyntax->GetObject(nullptr, objnum, gennum, true)));
   if (!m_pLinearized)
     return false;
-
-  CPDF_Dictionary* pDict = m_pLinearized->GetDict();
-  if (pDict && pDict->GetObjectFor("Linearized")) {
-    m_pSyntax->GetNextWord(nullptr);
-
-    CPDF_Object* pLen = pDict->GetObjectFor("L");
-    if (!pLen) {
-      m_pLinearized->Release();
-      m_pLinearized = nullptr;
-      return false;
-    }
-
-    if (pLen->GetInteger() != (int)pFileAccess->GetSize())
-      return false;
-
-    if (CPDF_Number* pNo = ToNumber(pDict->GetObjectFor("P")))
-      m_dwFirstPageNo = pNo->GetInteger();
-
-    if (CPDF_Number* pTable = ToNumber(pDict->GetObjectFor("T")))
-      m_LastXRefOffset = pTable->GetInteger();
-
-    return true;
-  }
-  m_pLinearized->Release();
-  m_pLinearized = nullptr;
-  return false;
+  m_LastXRefOffset = m_pLinearized->GetLastXRefOffset();
+  return true;
 }
 
 CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
@@ -1511,6 +1488,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
   m_bHasParsed = true;
   m_pDocument = pDocument;
 
+  m_pSyntax->GetNextWord(nullptr);

[Lei Zhang, 2016/11/04 01:10:15]

This is the same call that used to be in CPDF_Parser::IsLinearizedFile(), right?

I don't know if this is the right place to call GetNextWord(), or if it should
remain in IsLinearizedFile().

[snake, 2016/11/04 22:39:14]

Moved back for now. Will think how will be better.

[Lei Zhang, 2016/11/04 23:02:15]

There's still a subtle difference. In the old code, the "Linearized" dictionary
just has to exist. Now it has to parse correctly before we call GetNextWord().
I'm not sure if this matters.

   FX_FILESIZE dwFirstXRefOffset = m_pSyntax->SavePos();
 
   bool bXRefRebuilt = false;
@@ -1537,7 +1515,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
   if (eRet != SUCCESS)
     return eRet;
 
-  m_pDocument->LoadLinearizedDoc(m_pLinearized->GetDict());
+  m_pDocument->LoadLinearizedDoc(m_pLinearized.get());
   if (!m_pDocument->GetRoot() || m_pDocument->GetPageCount() == 0) {
     if (bXRefRebuilt)
       return FORMAT_ERROR;
@@ -1550,7 +1528,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
     if (eRet != SUCCESS)
       return eRet;
 
-    m_pDocument->LoadLinearizedDoc(m_pLinearized->GetDict());
+    m_pDocument->LoadLinearizedDoc(m_pLinearized.get());
     if (!m_pDocument->GetRoot())
       return FORMAT_ERROR;
   }