Fix a bug that occurs when an object has the same object number with the root object

core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp

Index: core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
diff --git a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
index 2f834ea400b8f3be2e1f3a313dabf30676746180..52dda954d2a68d73b4dafddab670d9b664b5d5f7 100644
--- a/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
+++ b/core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp
@@ -1002,7 +1002,7 @@ FX_BOOL CPDF_Parser::LoadCrossRefV5(FX_FILESIZE pos, FX_FILESIZE& prev, FX_BOOL
     if (!pStream) {
         return FALSE;
     }
-    if (m_pDocument) {
+    if (m_pDocument && m_pDocument->GetRoot()->GetObjNum() != pStream->m_ObjNum) {

[Bo Xu, 2014/12/15 17:49:25]

Could GetRoot() return null?

[jun_fang, 2014/12/15 18:46:48]

m_pRootDict is returned from GetRoot().It's assigned to a root dictionary during
the process of loading documents. If m_pDocument is not NULL, that means the
document is loaded successfully. Otherwise, it can't get here. 

There are two ways to load a document, FPDF_LoadCustomDocument and
FPDFAvail_GetDocument. In both of them, 'm_pDocument->GetRoot() == NULL' is
checked in the functions of CPDF_Parser::StartParse and
CPDF_Parser::StartAsynParse.   

[Tom Sepez, 2014/12/15 19:33:43]

How about returning false if this is the case? Is there a reason to continue?

         m_pDocument->InsertIndirectObject(pStream->m_ObjNum, pStream);
     }
     if (pStream->GetType() != PDFOBJ_STREAM) {