Fix a regression for parsing '/' - commit a568ff2d was too strict.

core/src/fpdfapi/fpdf_parser/fpdf_parser_parser.cpp

 // Copyright 2014 PDFium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
 
 #include "parser_int.h"
 
 #include <set>
 #include <utility>
@@ skipping 2107 matching lines @@
         break;
       }
       if (key[0] != '/')
         continue;
 
       ++nKeys;
       key = PDF_NameDecode(key);
       if (key.IsEmpty())
         continue;
 
-      CFX_ByteStringC keyNoSlash(key.c_str() + 1, key.GetLength() - 1);
-      if (keyNoSlash.IsEmpty())
-        continue;
-
       if (key == FX_BSTRC("/Contents"))
         dwSignValuePos = m_Pos;
 
       CPDF_Object* pObj = GetObject(pObjList, objnum, gennum);
       if (!pObj)
         continue;
 
+      CFX_ByteStringC keyNoSlash(key.c_str() + 1, key.GetLength() - 1);
       // TODO(thestig): Remove this conditional once CPDF_Dictionary has a
       // better underlying map implementation.
       if (nKeys < 32) {
         pDict->SetAt(keyNoSlash, pObj);
       } else {
         pDict->AddValue(keyNoSlash, pObj);
       }
     }
 
     if (IsSignatureDict(pDict.get())) {
@@ skipping 2801 matching lines @@
   if (!m_pLinearizedDict)
     return -1;
   CPDF_Array* pRange = m_pLinearizedDict->GetArray(FX_BSTRC("H"));
   if (!pRange)
     return -1;
   CPDF_Object* pStreamLen = pRange->GetElementValue(1);
   if (!pStreamLen)
     return -1;
   return pStreamLen->GetInteger();
 }