Revert of Remove CPDF_Object::Release() in favor of direct delete (patchset #11 id:200001 of https://codereview.chromium.org/2384883003/ )

Reason for revert:
Looks like it's blocking the roll.

https://build.chromium.org/p/tryserver.chromium.linux/builders/linux_chromium_compile_dbg_ng/builds/186619

Original issue's description:
> Remove CPDF_Object::Release() in favor of direct delete
>
> Follow-on once we prove Release always deletes in previous CL.
>
> Committed: https://pdfium.googlesource.com/pdfium/+/4de3d095c9d9e961f93750cf1ebd489fd515be12

TBR=thestig@chromium.org,tsepez@chromium.org
# Skipping CQ checks because original CL landed less than 1 days ago.
NOPRESUBMIT=true
NOTREECHECKS=true
NOTRY=true

Review-Url: https://codereview.chromium.org/2478253002

19 files changed, 185 insertions, 130 deletions

diff --git a/core/fpdfapi/parser/cfdf_document.cpp b/core/fpdfapi/parser/cfdf_document.cpp
index dcd4b3c109..d39ec31d3c 100644
--- a/core/fpdfapi/parser/cfdf_document.cpp
+++ b/core/fpdfapi/parser/cfdf_document.cpp
@@ -81,7 +81,7 @@ void CFDF_Document::ParseStream(IFX_SeekableReadStream* pFile, bool bOwnFile) {
       if (CPDF_Dictionary* pMainDict =
               ToDictionary(parser.GetObject(this, 0, 0, true))) {
         m_pRootDict = pMainDict->GetDictFor("Root");
-        delete pMainDict;
+        pMainDict->Release();
       }
       break;
     }
diff --git a/core/fpdfapi/parser/cpdf_array.cpp b/core/fpdfapi/parser/cpdf_array.cpp
index 4000bbc980..e118fd66e6 100644
--- a/core/fpdfapi/parser/cpdf_array.cpp
+++ b/core/fpdfapi/parser/cpdf_array.cpp
@@ -24,7 +24,7 @@ CPDF_Array::~CPDF_Array() {
   m_ObjNum = kInvalidObjNum;
   for (auto& it : m_Objects) {
     if (it && it->GetObjNum() != kInvalidObjNum)
-      delete it;
+      it->Release();
   }
 }
 
@@ -139,9 +139,10 @@ void CPDF_Array::RemoveAt(size_t i, size_t nCount) {
   if (nCount <= 0 || nCount > m_Objects.size() - i)
     return;
 
-  for (size_t j = 0; j < nCount; ++j)
-    delete m_Objects[i + j];
-
+  for (size_t j = 0; j < nCount; ++j) {
+    if (CPDF_Object* p = m_Objects[i + j])
+      p->Release();
+  }
   m_Objects.erase(m_Objects.begin() + i, m_Objects.begin() + i + nCount);
 }
 
@@ -165,7 +166,9 @@ void CPDF_Array::SetAt(size_t i, CPDF_Object* pObj) {
     ASSERT(false);
     return;
   }
-  delete m_Objects[i];
+  if (CPDF_Object* pOld = m_Objects[i])
+    pOld->Release();
+
   m_Objects[i] = pObj;
 }
 
diff --git a/core/fpdfapi/parser/cpdf_array.h b/core/fpdfapi/parser/cpdf_array.h
index 9deb478809..8cfa0333bb 100644
--- a/core/fpdfapi/parser/cpdf_array.h
+++ b/core/fpdfapi/parser/cpdf_array.h
@@ -21,7 +21,6 @@ class CPDF_Array : public CPDF_Object {
   using const_iterator = std::vector<CPDF_Object*>::const_iterator;
 
   CPDF_Array();
-  ~CPDF_Array() override;
 
   // CPDF_Object.
   Type GetType() const override;
@@ -62,12 +61,15 @@ class CPDF_Array : public CPDF_Object {
   const_iterator end() const { return m_Objects.end(); }
 
  protected:
+  ~CPDF_Array() override;
+
   CPDF_Object* CloneNonCyclic(
       bool bDirect,
       std::set<const CPDF_Object*>* pVisited) const override;
 
   std::vector<CPDF_Object*> m_Objects;
 };
+using UniqueArray = std::unique_ptr<CPDF_Array, ReleaseDeleter<CPDF_Object>>;
 
 inline CPDF_Array* ToArray(CPDF_Object* obj) {
   return obj ? obj->AsArray() : nullptr;
@@ -77,12 +79,12 @@ inline const CPDF_Array* ToArray(const CPDF_Object* obj) {
   return obj ? obj->AsArray() : nullptr;
 }
 
-inline std::unique_ptr<CPDF_Array> ToArray(std::unique_ptr<CPDF_Object> obj) {
+inline UniqueArray ToArray(UniqueObject obj) {
   CPDF_Array* pArray = ToArray(obj.get());
   if (!pArray)
     return nullptr;
   obj.release();
-  return std::unique_ptr<CPDF_Array>(pArray);
+  return UniqueArray(pArray);
 }
 
 #endif  // CORE_FPDFAPI_PARSER_CPDF_ARRAY_H_
diff --git a/core/fpdfapi/parser/cpdf_array_unittest.cpp b/core/fpdfapi/parser/cpdf_array_unittest.cpp
index 800afb0f9a..bc9f578021 100644
--- a/core/fpdfapi/parser/cpdf_array_unittest.cpp
+++ b/core/fpdfapi/parser/cpdf_array_unittest.cpp
@@ -10,10 +10,16 @@
 
 #include "testing/gtest/include/gtest/gtest.h"
 
+namespace {
+
+using ScopedArray = std::unique_ptr<CPDF_Array, ReleaseDeleter<CPDF_Array>>;
+
+}  // namespace
+
 TEST(cpdf_array, RemoveAt) {
   {
     int elems[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(elems); ++i)
       arr->AddInteger(elems[i]);
     arr->RemoveAt(3, 3);
@@ -30,7 +36,7 @@ TEST(cpdf_array, RemoveAt) {
   {
     // When the range is out of bound, RemoveAt has no effect.
     int elems[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(elems); ++i)
       arr->AddInteger(elems[i]);
     arr->RemoveAt(8, 5);
@@ -47,7 +53,7 @@ TEST(cpdf_array, RemoveAt) {
 TEST(cpdf_array, InsertAt) {
   {
     int elems[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(elems); ++i)
       arr->InsertAt(i, new CPDF_Number(elems[i]));
     EXPECT_EQ(FX_ArraySize(elems), arr->GetCount());
@@ -66,7 +72,7 @@ TEST(cpdf_array, InsertAt) {
     // an element is inserted at that position while other unfilled
     // positions have nullptr.
     int elems[] = {1, 2};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(elems); ++i)
       arr->InsertAt(i, new CPDF_Number(elems[i]));
     arr->InsertAt(10, new CPDF_Number(10));
@@ -83,10 +89,10 @@ TEST(cpdf_array, Clone) {
   {
     // Basic case.
     int elems[] = {1, 2, 3, 4, 5, 6, 7, 8, 9, 10};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(elems); ++i)
       arr->InsertAt(i, new CPDF_Number(elems[i]));
-    std::unique_ptr<CPDF_Array> arr2(arr->Clone()->AsArray());
+    ScopedArray arr2(arr->Clone()->AsArray());
     EXPECT_EQ(arr->GetCount(), arr2->GetCount());
     for (size_t i = 0; i < FX_ArraySize(elems); ++i) {
       // Clone() always create new objects.
@@ -100,7 +106,7 @@ TEST(cpdf_array, Clone) {
     static const size_t kNumOfRowElems = 5;
     int elems[kNumOfRows][kNumOfRowElems] = {
         {1, 2, 3, 4, 5}, {10, 9, 8, 7, 6}, {11, 12, 13, 14, 15}};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     // Indirect references to indirect objects.
     std::unique_ptr<CPDF_IndirectObjectHolder> obj_holder(
         new CPDF_IndirectObjectHolder());
@@ -118,10 +124,10 @@ TEST(cpdf_array, Clone) {
     ASSERT_EQ(kNumOfRows, arr->GetCount());
     // Not dereferencing reference objects means just creating new references
     // instead of new copies of direct objects.
-    std::unique_ptr<CPDF_Array> arr1(arr->Clone()->AsArray());
+    ScopedArray arr1(arr->Clone()->AsArray());
     EXPECT_EQ(arr->GetCount(), arr1->GetCount());
     // Dereferencing reference objects creates new copies of direct objects.
-    std::unique_ptr<CPDF_Array> arr2(arr->CloneDirectObject()->AsArray());
+    ScopedArray arr2(arr->CloneDirectObject()->AsArray());
     EXPECT_EQ(arr->GetCount(), arr2->GetCount());
     for (size_t i = 0; i < kNumOfRows; ++i) {
       CPDF_Array* arr_elem = arr->GetObjectAt(i)->AsArray();
@@ -165,7 +171,7 @@ TEST(cpdf_array, Clone) {
 TEST(cpdf_array, Iterator) {
   int elems[] = {-23, -11,     3,         455,   2345877,
                  0,   7895330, -12564334, 10000, -100000};
-  std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+  ScopedArray arr(new CPDF_Array);
   for (size_t i = 0; i < FX_ArraySize(elems); ++i)
     arr->InsertAt(i, new CPDF_Number(elems[i]));
   size_t index = 0;
diff --git a/core/fpdfapi/parser/cpdf_data_avail.cpp b/core/fpdfapi/parser/cpdf_data_avail.cpp
index eadbf1e828..a077ebb715 100644
--- a/core/fpdfapi/parser/cpdf_data_avail.cpp
+++ b/core/fpdfapi/parser/cpdf_data_avail.cpp
@@ -83,12 +83,17 @@ CPDF_DataAvail::CPDF_DataAvail(FileAvail* pFileAvail,
 
 CPDF_DataAvail::~CPDF_DataAvail() {
   m_pHintTables.reset();
-  delete m_pLinearized;
-  delete m_pRoot;
-  delete m_pTrailer;
+  if (m_pLinearized)
+    m_pLinearized->Release();
+
+  if (m_pRoot)
+    m_pRoot->Release();
+
+  if (m_pTrailer)
+    m_pTrailer->Release();
 
   for (CPDF_Object* pObject : m_arrayAcroforms)
-    delete pObject;
+    pObject->Release();
 }
 
 void CPDF_DataAvail::SetDocument(CPDF_Document* pDoc) {
@@ -225,7 +230,7 @@ bool CPDF_DataAvail::CheckAcroFormSubObject(DownloadHints* pHints) {
   }
 
   for (CPDF_Object* pObject : m_arrayAcroforms)
-    delete pObject;
+    pObject->Release();
 
   m_arrayAcroforms.clear();
   return true;
@@ -395,7 +400,9 @@ bool CPDF_DataAvail::CheckInfo(DownloadHints* pHints) {
     return false;
   }
 
-  delete pInfo;
+  if (pInfo)
+    pInfo->Release();
+
   m_docStatus =
       (m_bHaveAcroForm ? PDF_DATAAVAIL_ACROFORM : PDF_DATAAVAIL_PAGETREE);
 
@@ -492,7 +499,7 @@ bool CPDF_DataAvail::CheckPage(DownloadHints* pHints) {
     }
 
     if (!pObj->IsDictionary()) {
-      delete pObj;
+      pObj->Release();
       continue;
     }
 
@@ -501,7 +508,7 @@ bool CPDF_DataAvail::CheckPage(DownloadHints* pHints) {
       m_PagesArray.push_back(pObj);
       continue;
     }
-    delete pObj;
+    pObj->Release();
   }
 
   m_PageObjList.RemoveAll();
@@ -517,15 +524,15 @@ bool CPDF_DataAvail::CheckPage(DownloadHints* pHints) {
       continue;
 
     if (!GetPageKids(m_pCurrentParser, pPages)) {
-      delete pPages;
+      pPages->Release();
       while (++i < iPages)
-        delete m_PagesArray[i];
+        m_PagesArray[i]->Release();
 
       m_PagesArray.clear();
       m_docStatus = PDF_DATAAVAIL_ERROR;
       return false;
     }
-    delete pPages;
+    pPages->Release();
   }
 
   m_PagesArray.clear();
@@ -580,12 +587,12 @@ bool CPDF_DataAvail::CheckPages(DownloadHints* pHints) {
   }
 
   if (!GetPageKids(m_pCurrentParser, pPages)) {
-    delete pPages;
+    pPages->Release();
     m_docStatus = PDF_DATAAVAIL_ERROR;
     return false;
   }
 
-  delete pPages;
+  pPages->Release();
   m_docStatus = PDF_DATAAVAIL_PAGE;
   return true;
 }
@@ -756,7 +763,7 @@ bool CPDF_DataAvail::CheckHintTables(DownloadHints* pHints) {
 
   std::unique_ptr<CPDF_HintTables> pHintTables(
       new CPDF_HintTables(this, pDict));
-  std::unique_ptr<CPDF_Object> pHintStream(
+  std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> pHintStream(
       ParseIndirectObjectAt(szHintStart, 0));
   CPDF_Stream* pStream = ToStream(pHintStream.get());
   if (pStream && pHintTables->LoadHintStream(pStream))
@@ -944,11 +951,11 @@ int32_t CPDF_DataAvail::CheckCrossRefStream(DownloadHints* pHints,
       if (pName->GetString() == "XRef") {
         m_Pos += m_parser.m_pSyntax->SavePos();
         xref_offset = pObj->GetDict()->GetIntegerFor("Prev");
-        delete pObj;
+        pObj->Release();
         return 1;
       }
     }
-    delete pObj;
+    pObj->Release();
     return -1;
   }
   pHints->AddSegment(m_Pos, req_size);
@@ -1174,7 +1181,7 @@ bool CPDF_DataAvail::CheckTrailer(DownloadHints* pHints) {
     ScopedFileStream file(FX_CreateMemoryStream(pBuf, (size_t)iSize, false));
     m_syntaxParser.InitParser(file.get(), 0);
 
-    std::unique_ptr<CPDF_Object> pTrailer(
+    std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> pTrailer(
         m_syntaxParser.GetObject(nullptr, 0, 0, true));
     if (!pTrailer) {
       m_Pos += m_syntaxParser.SavePos();
@@ -1259,7 +1266,7 @@ bool CPDF_DataAvail::CheckArrayPageNode(uint32_t dwPageNo,
 
   CPDF_Array* pArray = pPages->AsArray();
   if (!pArray) {
-    delete pPages;
+    pPages->Release();
     m_docStatus = PDF_DATAAVAIL_ERROR;
     return false;
   }
@@ -1274,7 +1281,7 @@ bool CPDF_DataAvail::CheckArrayPageNode(uint32_t dwPageNo,
     pPageNode->m_childNode.Add(pNode);
     pNode->m_dwPageNo = pKid->GetRefObjNum();
   }
-  delete pPages;
+  pPages->Release();
   return true;
 }
 
@@ -1297,12 +1304,12 @@ bool CPDF_DataAvail::CheckUnkownPageNode(uint32_t dwPageNo,
   if (pPage->IsArray()) {
     pPageNode->m_dwPageNo = dwPageNo;
     pPageNode->m_type = PDF_PAGENODE_ARRAY;
-    delete pPage;
+    pPage->Release();
     return true;
   }
 
   if (!pPage->IsDictionary()) {
-    delete pPage;
+    pPage->Release();
     m_docStatus = PDF_DATAAVAIL_ERROR;
     return false;
   }
@@ -1343,11 +1350,11 @@ bool CPDF_DataAvail::CheckUnkownPageNode(uint32_t dwPageNo,
   } else if (type == "Page") {
     pPageNode->m_type = PDF_PAGENODE_PAGE;
   } else {
-    delete pPage;
+    pPage->Release();
     m_docStatus = PDF_DATAAVAIL_ERROR;
     return false;
   }
-  delete pPage;
+  pPage->Release();
   return true;
 }
 
@@ -1435,23 +1442,23 @@ bool CPDF_DataAvail::CheckPageCount(DownloadHints* pHints) {
 
   CPDF_Dictionary* pPagesDict = pPages->GetDict();
   if (!pPagesDict) {
-    delete pPages;
+    pPages->Release();
     m_docStatus = PDF_DATAAVAIL_ERROR;
     return false;
   }
 
   if (!pPagesDict->KeyExist("Kids")) {
-    delete pPages;
+    pPages->Release();
     return true;
   }
 
   int count = pPagesDict->GetIntegerFor("Count");
   if (count > 0) {
-    delete pPages;
+    pPages->Release();
     return true;
   }
 
-  delete pPages;
+  pPages->Release();
   return false;
 }
 
diff --git a/core/fpdfapi/parser/cpdf_dictionary.cpp b/core/fpdfapi/parser/cpdf_dictionary.cpp
index 37efbbc34a..75cb1e859c 100644
--- a/core/fpdfapi/parser/cpdf_dictionary.cpp
+++ b/core/fpdfapi/parser/cpdf_dictionary.cpp
@@ -31,7 +31,7 @@ CPDF_Dictionary::~CPDF_Dictionary() {
   m_ObjNum = kInvalidObjNum;
   for (const auto& it : m_Map) {
     if (it.second && it.second->GetObjNum() != kInvalidObjNum)
-      delete it.second;
+      it.second->Release();
   }
 }
 
@@ -184,7 +184,7 @@ void CPDF_Dictionary::SetFor(const CFX_ByteString& key, CPDF_Object* pObj) {
 
   if (it->second == pObj)
     return;
-  delete it->second;
+  it->second->Release();
 
   if (pObj)
     it->second = pObj;
@@ -208,7 +208,7 @@ void CPDF_Dictionary::RemoveFor(const CFX_ByteString& key) {
   if (it == m_Map.end())
     return;
 
-  delete it->second;
+  it->second->Release();
   m_Map.erase(it);
 }
 
@@ -223,7 +223,7 @@ void CPDF_Dictionary::ReplaceKey(const CFX_ByteString& oldkey,
     return;
 
   if (new_it != m_Map.end()) {
-    delete new_it->second;
+    new_it->second->Release();
     new_it->second = old_it->second;
   } else {
     m_Map.insert(std::make_pair(MaybeIntern(newkey), old_it->second));
diff --git a/core/fpdfapi/parser/cpdf_dictionary.h b/core/fpdfapi/parser/cpdf_dictionary.h
index 4ef2f96ce7..fb8200f78c 100644
--- a/core/fpdfapi/parser/cpdf_dictionary.h
+++ b/core/fpdfapi/parser/cpdf_dictionary.h
@@ -26,7 +26,6 @@ class CPDF_Dictionary : public CPDF_Object {
 
   CPDF_Dictionary();
   explicit CPDF_Dictionary(const CFX_WeakPtr<CFX_ByteStringPool>& pPool);
-  ~CPDF_Dictionary() override;
 
   // CPDF_Object.
   Type GetType() const override;
@@ -89,6 +88,8 @@ class CPDF_Dictionary : public CPDF_Object {
   CFX_WeakPtr<CFX_ByteStringPool> GetByteStringPool() const { return m_pPool; }
 
  protected:
+  ~CPDF_Dictionary() override;
+
   CFX_ByteString MaybeIntern(const CFX_ByteString& str);
   CPDF_Object* CloneNonCyclic(
       bool bDirect,
@@ -98,6 +99,9 @@ class CPDF_Dictionary : public CPDF_Object {
   std::map<CFX_ByteString, CPDF_Object*> m_Map;
 };
 
+using UniqueDictionary =
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Object>>;
+
 inline CPDF_Dictionary* ToDictionary(CPDF_Object* obj) {
   return obj ? obj->AsDictionary() : nullptr;
 }
@@ -106,13 +110,12 @@ inline const CPDF_Dictionary* ToDictionary(const CPDF_Object* obj) {
   return obj ? obj->AsDictionary() : nullptr;
 }
 
-inline std::unique_ptr<CPDF_Dictionary> ToDictionary(
-    std::unique_ptr<CPDF_Object> obj) {
+inline UniqueDictionary ToDictionary(UniqueObject obj) {
   CPDF_Dictionary* pDict = ToDictionary(obj.get());
   if (!pDict)
     return nullptr;
   obj.release();
-  return std::unique_ptr<CPDF_Dictionary>(pDict);
+  return UniqueDictionary(pDict);
 }
 
 #endif  // CORE_FPDFAPI_PARSER_CPDF_DICTIONARY_H_
diff --git a/core/fpdfapi/parser/cpdf_document.cpp b/core/fpdfapi/parser/cpdf_document.cpp
index 64574047e5..ad84a15a97 100644
--- a/core/fpdfapi/parser/cpdf_document.cpp
+++ b/core/fpdfapi/parser/cpdf_document.cpp
@@ -634,7 +634,7 @@ CPDF_Dictionary* CPDF_Document::CreateNewPage(int iPage) {
   pDict->SetNameFor("Type", "Page");
   uint32_t dwObjNum = AddIndirectObject(pDict);
   if (!InsertNewPage(iPage, pDict)) {
-    DeleteIndirectObject(dwObjNum);
+    ReleaseIndirectObject(dwObjNum);
     return nullptr;
   }
   return pDict;
diff --git a/core/fpdfapi/parser/cpdf_document_unittest.cpp b/core/fpdfapi/parser/cpdf_document_unittest.cpp
index c09665b716..9336626f45 100644
--- a/core/fpdfapi/parser/cpdf_document_unittest.cpp
+++ b/core/fpdfapi/parser/cpdf_document_unittest.cpp
@@ -76,7 +76,8 @@ class CPDF_TestDocumentForPages : public CPDF_Document {
   }
 
  private:
-  std::unique_ptr<CPDF_Dictionary> m_pOwnedRootDict;
+  std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>>
+      m_pOwnedRootDict;
 };
 }  // namespace
 
@@ -120,7 +121,7 @@ TEST_F(cpdf_document_test, UseCachedPageObjNumIfHaveNotPagesDict) {
   // 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>());
-  std::unique_ptr<CPDF_Dictionary> dict(new CPDF_Dictionary());
+  ScopedDictionary dict(new CPDF_Dictionary());
   const int page_count = 100;
   dict->SetIntegerFor("N", page_count);
   document.LoadLinearizedDoc(dict.get());
diff --git a/core/fpdfapi/parser/cpdf_indirect_object_holder.cpp b/core/fpdfapi/parser/cpdf_indirect_object_holder.cpp
index 427fac66e4..6e549de5a7 100644
--- a/core/fpdfapi/parser/cpdf_indirect_object_holder.cpp
+++ b/core/fpdfapi/parser/cpdf_indirect_object_holder.cpp
@@ -70,7 +70,7 @@ bool CPDF_IndirectObjectHolder::ReplaceIndirectObjectIfHigherGeneration(
   return true;
 }
 
-void CPDF_IndirectObjectHolder::DeleteIndirectObject(uint32_t objnum) {
+void CPDF_IndirectObjectHolder::ReleaseIndirectObject(uint32_t objnum) {
   CPDF_Object* pObj = GetIndirectObject(objnum);
   if (!pObj || pObj->GetObjNum() == CPDF_Object::kInvalidObjNum)
     return;
diff --git a/core/fpdfapi/parser/cpdf_indirect_object_holder.h b/core/fpdfapi/parser/cpdf_indirect_object_holder.h
index 83a31535c4..da4e942b5d 100644
--- a/core/fpdfapi/parser/cpdf_indirect_object_holder.h
+++ b/core/fpdfapi/parser/cpdf_indirect_object_holder.h
@@ -24,7 +24,7 @@ class CPDF_IndirectObjectHolder {
 
   CPDF_Object* GetIndirectObject(uint32_t objnum) const;
   CPDF_Object* GetOrParseIndirectObject(uint32_t objnum);
-  void DeleteIndirectObject(uint32_t objnum);
+  void ReleaseIndirectObject(uint32_t objnum);
 
   // Take ownership of |pObj|.
   uint32_t AddIndirectObject(CPDF_Object* pObj);
diff --git a/core/fpdfapi/parser/cpdf_object.cpp b/core/fpdfapi/parser/cpdf_object.cpp
index e9c215ce19..cc410d10c8 100644
--- a/core/fpdfapi/parser/cpdf_object.cpp
+++ b/core/fpdfapi/parser/cpdf_object.cpp
@@ -37,6 +37,11 @@ CPDF_Object* CPDF_Object::CloneNonCyclic(
   return Clone();
 }
 
+void CPDF_Object::Release() {
+  CHECK(!m_ObjNum);
+  delete this;
+}
+
 CFX_ByteString CPDF_Object::GetString() const {
   return CFX_ByteString();
 }
diff --git a/core/fpdfapi/parser/cpdf_object.h b/core/fpdfapi/parser/cpdf_object.h
index 8f6491ec72..c888605d72 100644
--- a/core/fpdfapi/parser/cpdf_object.h
+++ b/core/fpdfapi/parser/cpdf_object.h
@@ -38,8 +38,6 @@ class CPDF_Object {
     REFERENCE
   };
 
-  virtual ~CPDF_Object();
-
   virtual Type GetType() const = 0;
   uint32_t GetObjNum() const { return m_ObjNum; }
   uint32_t GetGenNum() const { return m_GenNum; }
@@ -52,6 +50,8 @@ class CPDF_Object {
   virtual CPDF_Object* CloneDirectObject() const;
   virtual CPDF_Object* GetDirect() const;
 
+  void Release();
+
   virtual CFX_ByteString GetString() const;
   virtual CFX_WideString GetUnicodeText() const;
   virtual FX_FLOAT GetNumber() const;
@@ -94,8 +94,10 @@ class CPDF_Object {
   friend class CPDF_Parser;
   friend class CPDF_Reference;
   friend class CPDF_Stream;
+  friend struct std::default_delete<CPDF_Object>;
 
   CPDF_Object() : m_ObjNum(0), m_GenNum(0) {}
+  virtual ~CPDF_Object();
 
   CPDF_Object* CloneObjectNonCyclic(bool bDirect) const;
 
@@ -116,4 +118,6 @@ class CPDF_Object {
   CPDF_Object(const CPDF_Object& src) {}
 };
 
+using UniqueObject = std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>>;
+
 #endif  // CORE_FPDFAPI_PARSER_CPDF_OBJECT_H_
diff --git a/core/fpdfapi/parser/cpdf_object_unittest.cpp b/core/fpdfapi/parser/cpdf_object_unittest.cpp
index 64dc8c63ec..8215226ef3 100644
--- a/core/fpdfapi/parser/cpdf_object_unittest.cpp
+++ b/core/fpdfapi/parser/cpdf_object_unittest.cpp
@@ -22,6 +22,11 @@
 
 namespace {
 
+using ScopedArray = std::unique_ptr<CPDF_Array, ReleaseDeleter<CPDF_Array>>;
+using ScopedDict =
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>>;
+using ScopedStream = std::unique_ptr<CPDF_Stream, ReleaseDeleter<CPDF_Stream>>;
+
 void TestArrayAccessors(const CPDF_Array* arr,
                         size_t index,
                         const char* str_val,
@@ -168,12 +173,14 @@ class PDFObjectsTest : public testing::Test {
   }
 
  protected:
+  using ScopedObj = std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>>;
+
   // m_ObjHolder needs to be declared first and destructed last since it also
   // refers to some objects in m_DirectObjs.
   std::unique_ptr<CPDF_IndirectObjectHolder> m_ObjHolder;
-  std::vector<std::unique_ptr<CPDF_Object>> m_DirectObjs;
+  std::vector<ScopedObj> m_DirectObjs;
   std::vector<int> m_DirectObjTypes;
-  std::vector<std::unique_ptr<CPDF_Object>> m_RefObjs;
+  std::vector<ScopedObj> m_RefObjs;
   CPDF_Dictionary* m_DictObj;
   CPDF_Dictionary* m_StreamDictObj;
   CPDF_Array* m_ArrayObj;
@@ -268,13 +275,13 @@ TEST_F(PDFObjectsTest, GetArray) {
 TEST_F(PDFObjectsTest, Clone) {
   // Check for direct objects.
   for (size_t i = 0; i < m_DirectObjs.size(); ++i) {
-    std::unique_ptr<CPDF_Object> obj(m_DirectObjs[i]->Clone());
+    ScopedObj obj(m_DirectObjs[i]->Clone());
     EXPECT_TRUE(Equal(m_DirectObjs[i].get(), obj.get()));
   }
 
   // Check indirect references.
   for (const auto& it : m_RefObjs) {
-    std::unique_ptr<CPDF_Object> obj(it->Clone());
+    ScopedObj obj(it->Clone());
     EXPECT_TRUE(Equal(it.get(), obj.get()));
   }
 }
@@ -386,7 +393,7 @@ TEST(PDFArrayTest, GetMatrix) {
                       {2.3f, 4.05f, 3, -2, -3, 0.0f},
                       {0.05f, 0.1f, 0.56f, 0.67f, 1.34f, 99.9f}};
   for (size_t i = 0; i < FX_ArraySize(elems); ++i) {
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     CFX_Matrix matrix(elems[i][0], elems[i][1], elems[i][2], elems[i][3],
                       elems[i][4], elems[i][5]);
     for (size_t j = 0; j < 6; ++j)
@@ -407,7 +414,7 @@ TEST(PDFArrayTest, GetRect) {
                       {2.3f, 4.05f, -3, 0.0f},
                       {0.05f, 0.1f, 1.34f, 99.9f}};
   for (size_t i = 0; i < FX_ArraySize(elems); ++i) {
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     CFX_FloatRect rect(elems[i]);
     for (size_t j = 0; j < 4; ++j)
       arr->AddNumber(elems[i][j]);
@@ -423,7 +430,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   {
     // Boolean array.
     const bool vals[] = {true, false, false, true, true};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(vals); ++i)
       arr->InsertAt(i, new CPDF_Boolean(vals[i]));
     for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
@@ -440,7 +447,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   {
     // Integer array.
     const int vals[] = {10, 0, -345, 2089345456, -1000000000, 567, 93658767};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(vals); ++i)
       arr->InsertAt(i, new CPDF_Number(vals[i]));
     for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
@@ -461,7 +468,7 @@ TEST(PDFArrayTest, GetTypeAt) {
                           897.34f, -2.5f, -1.0f, -345.0f, -0.0f};
     const char* const expected_str[] = {
         "0", "0", "10", "10", "0.0345", "897.34", "-2.5", "-1", "-345", "0"};
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
       arr->InsertAt(i, new CPDF_Number(vals[i]));
     }
@@ -480,8 +487,8 @@ TEST(PDFArrayTest, GetTypeAt) {
     // String and name array
     const char* const vals[] = {"this", "adsde$%^", "\r\t",           "\"012",
                                 ".",    "EYREW",    "It is a joke :)"};
-    std::unique_ptr<CPDF_Array> string_array(new CPDF_Array);
-    std::unique_ptr<CPDF_Array> name_array(new CPDF_Array);
+    ScopedArray string_array(new CPDF_Array);
+    ScopedArray name_array(new CPDF_Array);
     for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
       string_array->InsertAt(i, new CPDF_String(vals[i], false));
       name_array->InsertAt(i, new CPDF_Name(vals[i]));
@@ -507,7 +514,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   }
   {
     // Null element array.
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < 3; ++i)
       arr->InsertAt(i, new CPDF_Null);
     for (size_t i = 0; i < 3; ++i) {
@@ -524,7 +531,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   {
     // Array of array.
     CPDF_Array* vals[3];
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < 3; ++i) {
       vals[i] = new CPDF_Array;
       for (size_t j = 0; j < 3; ++j) {
@@ -547,7 +554,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   {
     // Dictionary array.
     CPDF_Dictionary* vals[3];
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < 3; ++i) {
       vals[i] = new CPDF_Dictionary();
       for (size_t j = 0; j < 3; ++j) {
@@ -574,7 +581,7 @@ TEST(PDFArrayTest, GetTypeAt) {
     // Stream array.
     CPDF_Dictionary* vals[3];
     CPDF_Stream* stream_vals[3];
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     for (size_t i = 0; i < 3; ++i) {
       vals[i] = new CPDF_Dictionary();
       for (size_t j = 0; j < 3; ++j) {
@@ -604,7 +611,7 @@ TEST(PDFArrayTest, GetTypeAt) {
   }
   {
     // Mixed array.
-    std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+    ScopedArray arr(new CPDF_Array);
     // Array arr will take ownership of all the objects inserted.
     arr->InsertAt(0, new CPDF_Boolean(true));
     arr->InsertAt(1, new CPDF_Boolean(false));
@@ -669,7 +676,7 @@ TEST(PDFArrayTest, GetTypeAt) {
 TEST(PDFArrayTest, AddNumber) {
   float vals[] = {1.0f,         -1.0f, 0,    0.456734f,
                   12345.54321f, 0.5f,  1000, 0.000045f};
-  std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+  ScopedArray arr(new CPDF_Array);
   for (size_t i = 0; i < FX_ArraySize(vals); ++i)
     arr->AddNumber(vals[i]);
   for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
@@ -680,7 +687,7 @@ TEST(PDFArrayTest, AddNumber) {
 
 TEST(PDFArrayTest, AddInteger) {
   int vals[] = {0, 1, 934435456, 876, 10000, -1, -24354656, -100};
-  std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
+  ScopedArray arr(new CPDF_Array);
   for (size_t i = 0; i < FX_ArraySize(vals); ++i)
     arr->AddInteger(vals[i]);
   for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
@@ -692,8 +699,8 @@ TEST(PDFArrayTest, AddInteger) {
 TEST(PDFArrayTest, AddStringAndName) {
   const char* vals[] = {"",        "a", "ehjhRIOYTTFdfcdnv",  "122323",
                         "$#%^&**", " ", "This is a test.\r\n"};
-  std::unique_ptr<CPDF_Array> string_array(new CPDF_Array);
-  std::unique_ptr<CPDF_Array> name_array(new CPDF_Array);
+  ScopedArray string_array(new CPDF_Array);
+  ScopedArray name_array(new CPDF_Array);
   for (size_t i = 0; i < FX_ArraySize(vals); ++i) {
     string_array->AddString(vals[i]);
     name_array->AddName(vals[i]);
@@ -718,8 +725,8 @@ TEST(PDFArrayTest, AddReferenceAndGetObjectAt) {
   CPDF_Object* indirect_objs[] = {boolean_obj, int_obj,  float_obj,
                                   str_obj,     name_obj, null_obj};
   unsigned int obj_nums[] = {2, 4, 7, 2345, 799887, 1};
-  std::unique_ptr<CPDF_Array> arr(new CPDF_Array);
-  std::unique_ptr<CPDF_Array> arr1(new CPDF_Array);
+  ScopedArray arr(new CPDF_Array);
+  ScopedArray arr1(new CPDF_Array);
   // Create two arrays of references by different AddReference() APIs.
   for (size_t i = 0; i < FX_ArraySize(indirect_objs); ++i) {
     // All the indirect objects inserted will be owned by holder.
@@ -745,7 +752,7 @@ TEST(PDFArrayTest, AddReferenceAndGetObjectAt) {
 
 TEST(PDFArrayTest, CloneDirectObject) {
   CPDF_IndirectObjectHolder objects_holder;
-  std::unique_ptr<CPDF_Array> array(new CPDF_Array);
+  ScopedArray array(new CPDF_Array);
   array->AddReference(&objects_holder, 1234);
   ASSERT_EQ(1U, array->GetCount());
   CPDF_Object* obj = array->GetObjectAt(0);
@@ -756,7 +763,7 @@ TEST(PDFArrayTest, CloneDirectObject) {
   ASSERT_TRUE(cloned_array_object);
   ASSERT_TRUE(cloned_array_object->IsArray());
 
-  std::unique_ptr<CPDF_Array> cloned_array(cloned_array_object->AsArray());
+  ScopedArray cloned_array(cloned_array_object->AsArray());
   ASSERT_EQ(1U, cloned_array->GetCount());
   CPDF_Object* cloned_obj = cloned_array->GetObjectAt(0);
   EXPECT_FALSE(cloned_obj);
@@ -764,7 +771,7 @@ TEST(PDFArrayTest, CloneDirectObject) {
 
 TEST(PDFArrayTest, ConvertIndirect) {
   CPDF_IndirectObjectHolder objects_holder;
-  std::unique_ptr<CPDF_Array> array(new CPDF_Array);
+  ScopedArray array(new CPDF_Array);
   CPDF_Object* pObj = new CPDF_Number(42);
   array->Add(pObj);
   array->ConvertToIndirectObjectAt(0, &objects_holder);
@@ -779,7 +786,7 @@ TEST(PDFArrayTest, ConvertIndirect) {
 
 TEST(PDFDictionaryTest, CloneDirectObject) {
   CPDF_IndirectObjectHolder objects_holder;
-  std::unique_ptr<CPDF_Dictionary> dict(new CPDF_Dictionary());
+  ScopedDict dict(new CPDF_Dictionary());
   dict->SetReferenceFor("foo", &objects_holder, 1234);
   ASSERT_EQ(1U, dict->GetCount());
   CPDF_Object* obj = dict->GetObjectFor("foo");
@@ -790,8 +797,7 @@ TEST(PDFDictionaryTest, CloneDirectObject) {
   ASSERT_TRUE(cloned_dict_object);
   ASSERT_TRUE(cloned_dict_object->IsDictionary());
 
-  std::unique_ptr<CPDF_Dictionary> cloned_dict(
-      cloned_dict_object->AsDictionary());
+  ScopedDict cloned_dict(cloned_dict_object->AsDictionary());
   ASSERT_EQ(1U, cloned_dict->GetCount());
   CPDF_Object* cloned_obj = cloned_dict->GetObjectFor("foo");
   EXPECT_FALSE(cloned_obj);
@@ -801,12 +807,12 @@ TEST(PDFObjectTest, CloneCheckLoop) {
   {
     // Create a dictionary/array pair with a reference loop.
     CPDF_Dictionary* dict_obj = new CPDF_Dictionary();
-    std::unique_ptr<CPDF_Array> arr_obj(new CPDF_Array);
+    ScopedArray arr_obj(new CPDF_Array);
     dict_obj->SetFor("arr", arr_obj.get());
     arr_obj->InsertAt(0, dict_obj);
 
     // Clone this object to see whether stack overflow will be triggered.
-    std::unique_ptr<CPDF_Array> cloned_array(arr_obj->Clone()->AsArray());
+    ScopedArray cloned_array(arr_obj->Clone()->AsArray());
     // Cloned object should be the same as the original.
     ASSERT_TRUE(cloned_array);
     EXPECT_EQ(1u, cloned_array->GetCount());
@@ -819,12 +825,11 @@ TEST(PDFObjectTest, CloneCheckLoop) {
   {
     // Create a dictionary/stream pair with a reference loop.
     CPDF_Dictionary* dict_obj = new CPDF_Dictionary();
-    std::unique_ptr<CPDF_Stream> stream_obj(
-        new CPDF_Stream(nullptr, 0, dict_obj));
+    ScopedStream stream_obj(new CPDF_Stream(nullptr, 0, dict_obj));
     dict_obj->SetFor("stream", stream_obj.get());
 
     // Clone this object to see whether stack overflow will be triggered.
-    std::unique_ptr<CPDF_Stream> cloned_stream(stream_obj->Clone()->AsStream());
+    ScopedStream cloned_stream(stream_obj->Clone()->AsStream());
     // Cloned object should be the same as the original.
     ASSERT_TRUE(cloned_stream);
     CPDF_Object* cloned_dict = cloned_stream->GetDict();
@@ -850,8 +855,7 @@ TEST(PDFObjectTest, CloneCheckLoop) {
     EXPECT_EQ(dict_obj, elem0->AsReference()->GetDirect());
 
     // Clone this object to see whether stack overflow will be triggered.
-    std::unique_ptr<CPDF_Dictionary> cloned_dict(
-        ToDictionary(dict_obj->CloneDirectObject()));
+    ScopedDict cloned_dict(ToDictionary(dict_obj->CloneDirectObject()));
     // Cloned object should be the same as the original.
     ASSERT_TRUE(cloned_dict);
     CPDF_Object* cloned_arr = cloned_dict->GetObjectFor("arr");
@@ -865,7 +869,7 @@ TEST(PDFObjectTest, CloneCheckLoop) {
 
 TEST(PDFDictionaryTest, ConvertIndirect) {
   CPDF_IndirectObjectHolder objects_holder;
-  std::unique_ptr<CPDF_Dictionary> dict(new CPDF_Dictionary);
+  ScopedDict dict(new CPDF_Dictionary);
   CPDF_Object* pObj = new CPDF_Number(42);
   dict->SetFor("clams", pObj);
   dict->ConvertToIndirectObjectFor("clams", &objects_holder);
diff --git a/core/fpdfapi/parser/cpdf_parser.cpp b/core/fpdfapi/parser/cpdf_parser.cpp
index da95cc5f24..cff0f77b32 100644
--- a/core/fpdfapi/parser/cpdf_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_parser.cpp
@@ -61,7 +61,9 @@ CPDF_Parser::CPDF_Parser()
 }
 
 CPDF_Parser::~CPDF_Parser() {
-  delete m_pTrailer;
+  if (m_pTrailer)
+    m_pTrailer->Release();
+
   ReleaseEncryptHandler();
   SetEncryptDictionary(nullptr);
 
@@ -70,10 +72,13 @@ CPDF_Parser::~CPDF_Parser() {
     m_pSyntax->m_pFileAccess = nullptr;
   }
 
-  for (CPDF_Dictionary* trailer : m_Trailers)
-    delete trailer;
+  for (CPDF_Dictionary* trailer : m_Trailers) {
+    if (trailer)
+      trailer->Release();
+  }
 
-  delete m_pLinearized;
+  if (m_pLinearized)
+    m_pLinearized->Release();
 }
 
 uint32_t CPDF_Parser::GetLastObjNum() const {
@@ -352,7 +357,8 @@ bool CPDF_Parser::LoadAllCrossRefV4(FX_FILESIZE xrefpos) {
     CrossRefList.insert(CrossRefList.begin(), xrefpos);
     LoadCrossRefV4(xrefpos, 0, true);
 
-    std::unique_ptr<CPDF_Dictionary> pDict(LoadTrailerV4());
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>> pDict(
+        LoadTrailerV4());
     if (!pDict)
       return false;
 
@@ -406,7 +412,8 @@ bool CPDF_Parser::LoadLinearizedAllCrossRefV4(FX_FILESIZE xrefpos,
     CrossRefList.insert(CrossRefList.begin(), xrefpos);
     LoadCrossRefV4(xrefpos, 0, true);
 
-    std::unique_ptr<CPDF_Dictionary> pDict(LoadTrailerV4());
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>> pDict(
+        LoadTrailerV4());
     if (!pDict)
       return false;
 
@@ -583,14 +590,18 @@ bool CPDF_Parser::LoadAllCrossRefV5(FX_FILESIZE xrefpos) {
 bool CPDF_Parser::RebuildCrossRef() {
   m_ObjectInfo.clear();
   m_SortedOffset.clear();
-  delete m_pTrailer;
-  m_pTrailer = nullptr;
+  if (m_pTrailer) {
+    m_pTrailer->Release();
+    m_pTrailer = nullptr;
+  }
 
   ParserState state = ParserState::kDefault;
+
   int32_t inside_index = 0;
   uint32_t objnum = 0;
   uint32_t gennum = 0;
   int32_t depth = 0;
+
   const uint32_t kBufferSize = 4096;
   std::vector<uint8_t> buffer(kBufferSize);
 
@@ -744,7 +755,8 @@ bool CPDF_Parser::RebuildCrossRef() {
                       CPDF_Object* pRoot = pDict->GetObjectFor("Root");
                       if (pRoot && pRoot->GetDict() &&
                           pRoot->GetDict()->GetObjectFor("Pages")) {
-                        delete m_pTrailer;
+                        if (m_pTrailer)
+                          m_pTrailer->Release();
                         m_pTrailer = ToDictionary(pDict->Clone());
                       }
                     }
@@ -782,7 +794,8 @@ bool CPDF_Parser::RebuildCrossRef() {
                   m_ObjectInfo[objnum].gennum = gennum;
                 }
 
-                delete pObject;
+                if (pObject)
+                  pObject->Release();
               }
               --i;
               state = ParserState::kDefault;
@@ -799,7 +812,7 @@ bool CPDF_Parser::RebuildCrossRef() {
               CPDF_Object* pObj = m_pSyntax->GetObject(m_pDocument, 0, 0, true);
               if (pObj) {
                 if (!pObj->IsDictionary() && !pObj->AsStream()) {
-                  delete pObj;
+                  pObj->Release();
                 } else {
                   CPDF_Stream* pStream = pObj->AsStream();
                   if (CPDF_Dictionary* pTrailer =
@@ -825,11 +838,11 @@ bool CPDF_Parser::RebuildCrossRef() {
                           }
                         }
                       }
-                      delete pObj;
+                      pObj->Release();
                     } else {
                       if (pObj->IsStream()) {
                         m_pTrailer = ToDictionary(pTrailer->Clone());
-                        delete pObj;
+                        pObj->Release();
                       } else {
                         m_pTrailer = pTrailer;
                       }
@@ -846,7 +859,7 @@ bool CPDF_Parser::RebuildCrossRef() {
                       m_pSyntax->RestorePos(dwSavePos);
                     }
                   } else {
-                    delete pObj;
+                    pObj->Release();
                   }
                 }
               }
@@ -1405,7 +1418,7 @@ CPDF_Dictionary* CPDF_Parser::LoadTrailerV4() {
   if (m_pSyntax->GetKeyword() != "trailer")
     return nullptr;
 
-  std::unique_ptr<CPDF_Object> pObj(
+  std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> pObj(
       m_pSyntax->GetObject(m_pDocument, 0, 0, true));
   if (!ToDictionary(pObj.get()))
     return nullptr;
@@ -1457,7 +1470,7 @@ bool CPDF_Parser::IsLinearizedFile(IFX_SeekableReadStream* pFileAccess,
 
     CPDF_Object* pLen = pDict->GetObjectFor("L");
     if (!pLen) {
-      delete m_pLinearized;
+      m_pLinearized->Release();
       m_pLinearized = nullptr;
       return false;
     }
@@ -1473,7 +1486,7 @@ bool CPDF_Parser::IsLinearizedFile(IFX_SeekableReadStream* pFileAccess,
 
     return true;
   }
-  delete m_pLinearized;
+  m_pLinearized->Release();
   m_pLinearized = nullptr;
   return false;
 }
@@ -1582,11 +1595,12 @@ bool CPDF_Parser::LoadLinearizedAllCrossRefV5(FX_FILESIZE xrefpos) {
 CPDF_Parser::Error CPDF_Parser::LoadLinearizedMainXRefTable() {
   uint32_t dwSaveMetadataObjnum = m_pSyntax->m_MetadataObjnum;
   m_pSyntax->m_MetadataObjnum = 0;
+  if (m_pTrailer) {
+    m_pTrailer->Release();
+    m_pTrailer = nullptr;
+  }
 
-  delete m_pTrailer;
-  m_pTrailer = nullptr;
   m_pSyntax->RestorePos(m_LastXRefOffset - m_pSyntax->m_HeaderOffset);
-
   uint8_t ch = 0;
   uint32_t dwCount = 0;
   m_pSyntax->GetNextChar(ch);
diff --git a/core/fpdfapi/parser/cpdf_stream.cpp b/core/fpdfapi/parser/cpdf_stream.cpp
index 11ef1d2dc1..c6e99c84b5 100644
--- a/core/fpdfapi/parser/cpdf_stream.cpp
+++ b/core/fpdfapi/parser/cpdf_stream.cpp
@@ -15,7 +15,7 @@
 CPDF_Stream::CPDF_Stream() {}
 
 CPDF_Stream::CPDF_Stream(uint8_t* pData, uint32_t size, CPDF_Dictionary* pDict)
-    : m_dwSize(size), m_pDict(pDict), m_pDataBuf(pData) {}
+    : m_pDict(pDict), m_dwSize(size), m_pDataBuf(pData) {}
 
 CPDF_Stream::~CPDF_Stream() {
   m_ObjNum = kInvalidObjNum;
diff --git a/core/fpdfapi/parser/cpdf_stream.h b/core/fpdfapi/parser/cpdf_stream.h
index f0ba31924e..73484d8335 100644
--- a/core/fpdfapi/parser/cpdf_stream.h
+++ b/core/fpdfapi/parser/cpdf_stream.h
@@ -20,7 +20,6 @@ class CPDF_Stream : public CPDF_Object {
 
   // Takes ownership of |pData| and |pDict|.
   CPDF_Stream(uint8_t* pData, uint32_t size, CPDF_Dictionary* pDict);
-  ~CPDF_Stream() override;
 
   // CPDF_Object.
   Type GetType() const override;
@@ -48,17 +47,20 @@ class CPDF_Stream : public CPDF_Object {
   bool IsMemoryBased() const { return m_bMemoryBased; }
 
  protected:
+  ~CPDF_Stream() override;
   CPDF_Object* CloneNonCyclic(
       bool bDirect,
       std::set<const CPDF_Object*>* pVisited) const override;
 
+  std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>> m_pDict;
   bool m_bMemoryBased = true;
   uint32_t m_dwSize = 0;
-  std::unique_ptr<CPDF_Dictionary> m_pDict;
   std::unique_ptr<uint8_t, FxFreeDeleter> m_pDataBuf;
   IFX_SeekableReadStream* m_pFile = nullptr;
 };
 
+using UniqueStream = std::unique_ptr<CPDF_Stream, ReleaseDeleter<CPDF_Object>>;
+
 inline CPDF_Stream* ToStream(CPDF_Object* obj) {
   return obj ? obj->AsStream() : nullptr;
 }
@@ -67,12 +69,12 @@ inline const CPDF_Stream* ToStream(const CPDF_Object* obj) {
   return obj ? obj->AsStream() : nullptr;
 }
 
-inline std::unique_ptr<CPDF_Stream> ToStream(std::unique_ptr<CPDF_Object> obj) {
+inline UniqueStream ToStream(UniqueObject obj) {
   CPDF_Stream* pStream = ToStream(obj.get());
   if (!pStream)
     return nullptr;
   obj.release();
-  return std::unique_ptr<CPDF_Stream>(pStream);
+  return UniqueStream(pStream);
 }
 
 #endif  // CORE_FPDFAPI_PARSER_CPDF_STREAM_H_
diff --git a/core/fpdfapi/parser/cpdf_string.h b/core/fpdfapi/parser/cpdf_string.h
index 49834c03f2..efc6d076c7 100644
--- a/core/fpdfapi/parser/cpdf_string.h
+++ b/core/fpdfapi/parser/cpdf_string.h
@@ -16,7 +16,6 @@ class CPDF_String : public CPDF_Object {
   CPDF_String();
   CPDF_String(const CFX_ByteString& str, bool bHex);
   explicit CPDF_String(const CFX_WideString& str);
-  ~CPDF_String() override;
 
   // CPDF_Object.
   Type GetType() const override;
@@ -31,6 +30,8 @@ class CPDF_String : public CPDF_Object {
   bool IsHex() const { return m_bHex; }
 
  protected:
+  ~CPDF_String() override;
+
   CFX_ByteString m_String;
   bool m_bHex;
 };
diff --git a/core/fpdfapi/parser/cpdf_syntax_parser.cpp b/core/fpdfapi/parser/cpdf_syntax_parser.cpp
index da3c8b0c79..178af3bb9b 100644
--- a/core/fpdfapi/parser/cpdf_syntax_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_syntax_parser.cpp
@@ -425,7 +425,8 @@ CPDF_Object* CPDF_SyntaxParser::GetObject(CPDF_IndirectObjectHolder* pObjList,
     int32_t nKeys = 0;
     FX_FILESIZE dwSignValuePos = 0;
 
-    std::unique_ptr<CPDF_Dictionary> pDict(new CPDF_Dictionary(m_pPool));
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>> pDict(
+        new CPDF_Dictionary(m_pPool));
     while (1) {
       CFX_ByteString key = GetNextWord(nullptr);
       if (key.IsEmpty())
@@ -529,7 +530,8 @@ CPDF_Object* CPDF_SyntaxParser::GetObjectForStrict(
   }
 
   if (word == "[") {
-    std::unique_ptr<CPDF_Array> pArray(new CPDF_Array);
+    std::unique_ptr<CPDF_Array, ReleaseDeleter<CPDF_Array>> pArray(
+        new CPDF_Array);
     while (CPDF_Object* pObj = GetObject(pObjList, objnum, gennum, true))
       pArray->Add(pObj);
 
@@ -542,7 +544,8 @@ CPDF_Object* CPDF_SyntaxParser::GetObjectForStrict(
   }
 
   if (word == "<<") {
-    std::unique_ptr<CPDF_Dictionary> pDict(new CPDF_Dictionary(m_pPool));
+    std::unique_ptr<CPDF_Dictionary, ReleaseDeleter<CPDF_Dictionary>> pDict(
+        new CPDF_Dictionary(m_pPool));
     while (1) {
       FX_FILESIZE SavedPos = m_Pos;
       CFX_ByteString key = GetNextWord(nullptr);
@@ -561,7 +564,7 @@ CPDF_Object* CPDF_SyntaxParser::GetObjectForStrict(
         continue;
 
       key = PDF_NameDecode(key);
-      std::unique_ptr<CPDF_Object> obj(
+      std::unique_ptr<CPDF_Object, ReleaseDeleter<CPDF_Object>> obj(
           GetObject(pObjList, objnum, gennum, true));
       if (!obj) {
         uint8_t ch;
@@ -689,7 +692,7 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
 
       // Can't find "endstream" or "endobj".
       if (endStreamOffset < 0 && endObjOffset < 0) {
-        delete pDict;
+        pDict->Release();
         return nullptr;
       }
 
@@ -715,7 +718,7 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
       }
 
       if (len < 0) {
-        delete pDict;
+        pDict->Release();
         return nullptr;
       }
       pDict->SetIntegerFor("Length", len);
@@ -724,7 +727,7 @@ CPDF_Stream* CPDF_SyntaxParser::ReadStream(CPDF_Dictionary* pDict,
   }
 
   if (len < 0) {
-    delete pDict;
+    pDict->Release();
     return nullptr;
   }