Rework of loading of CPDF_Document.

Improve CPDF_Document interface.
Fix relationship between CPDF_Document and CPDF_Parser.
This CL changes CPDF_Document to internally create the CPDF_Parser
and removes the need for the CPDF_Parser to know about the CPDF_Document.


Change-Id: Iec7aef19575c90f30b9a6c919dfd4f4417e4caf2
Reviewed-on: https://pdfium-review.googlesource.com/35630
Commit-Queue: Art Snake <art-snake@yandex-team.ru>
Reviewed-by: dsinclair <dsinclair@chromium.org>

1 files changed, 57 insertions, 46 deletions

diff --git a/core/fpdfapi/parser/cpdf_parser.cpp b/core/fpdfapi/parser/cpdf_parser.cpp
index c7a3fe16c8..3986f3684f 100644
--- a/core/fpdfapi/parser/cpdf_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_parser.cpp
@@ -47,6 +47,13 @@ uint32_t GetVarInt(const uint8_t* p, int32_t n) {
   return result;
 }
 
+class ObjectsHolderStub : public CPDF_Parser::ParsedObjectsHolder {
+ public:
+  ObjectsHolderStub() = default;
+  ~ObjectsHolderStub() override = default;
+  bool TryInit() override { return true; }
+};
+
 }  // namespace
 
 class CPDF_Parser::TrailerData {
@@ -117,12 +124,20 @@ class CPDF_Parser::TrailerData {
   uint32_t last_root_obj_num_ = CPDF_Object::kInvalidObjNum;
 };
 
-CPDF_Parser::CPDF_Parser()
+CPDF_Parser::CPDF_Parser(ParsedObjectsHolder* holder)
     : m_pSyntax(pdfium::MakeUnique<CPDF_SyntaxParser>()),
+      m_pObjectsHolder(holder),
       m_bHasParsed(false),
       m_bXRefStream(false),
       m_FileVersion(0),
-      m_TrailerData(pdfium::MakeUnique<TrailerData>()) {}
+      m_TrailerData(pdfium::MakeUnique<TrailerData>()) {
+  if (!holder) {
+    m_pOwnedObjectsHolder = pdfium::MakeUnique<ObjectsHolderStub>();
+    m_pObjectsHolder = m_pOwnedObjectsHolder.get();
+  }
+}
+
+CPDF_Parser::CPDF_Parser() : CPDF_Parser(nullptr) {}
 
 CPDF_Parser::~CPDF_Parser() {
   ReleaseEncryptHandler();
@@ -225,20 +240,19 @@ bool CPDF_Parser::ParseFileVersion() {
 
 CPDF_Parser::Error CPDF_Parser::StartParse(
     const RetainPtr<IFX_SeekableReadStream>& pFileAccess,
-    CPDF_Document* pDocument) {
+    const char* password) {
   if (!InitSyntaxParser(
           pdfium::MakeRetain<CPDF_ReadValidator>(pFileAccess, nullptr)))
     return FORMAT_ERROR;
-  return StartParseInternal(pDocument);
+  SetPassword(password);
+  return StartParseInternal();
 }
 
-CPDF_Parser::Error CPDF_Parser::StartParseInternal(CPDF_Document* pDocument) {
+CPDF_Parser::Error CPDF_Parser::StartParseInternal() {
   ASSERT(!m_bHasParsed);
   m_bHasParsed = true;
   m_bXRefStream = false;
 
-  m_pDocument = pDocument;
-
   bool bXRefRebuilt = false;
 
   m_LastXRefOffset = ParseStartXRef();
@@ -262,8 +276,7 @@ CPDF_Parser::Error CPDF_Parser::StartParseInternal(CPDF_Document* pDocument) {
   if (eRet != SUCCESS)
     return eRet;
 
-  m_pDocument->LoadDoc();
-  if (!m_pDocument->GetRoot() || m_pDocument->GetPageCount() == 0) {
+  if (!GetRoot() || !m_pObjectsHolder->TryInit()) {
     if (bXRefRebuilt)
       return FORMAT_ERROR;
 
@@ -275,8 +288,8 @@ CPDF_Parser::Error CPDF_Parser::StartParseInternal(CPDF_Document* pDocument) {
     if (eRet != SUCCESS)
       return eRet;
 
-    m_pDocument->LoadDoc();
-    if (!m_pDocument->GetRoot())
+    m_pObjectsHolder->TryInit();
+    if (!GetRoot())
       return FORMAT_ERROR;
   }
   if (GetRootObjNum() == CPDF_Object::kInvalidObjNum) {
@@ -290,7 +303,7 @@ CPDF_Parser::Error CPDF_Parser::StartParseInternal(CPDF_Document* pDocument) {
   }
   if (m_pSecurityHandler && !m_pSecurityHandler->IsMetadataEncrypted()) {
     CPDF_Reference* pMetadata =
-        ToReference(m_pDocument->GetRoot()->GetObjectFor("Metadata"));
+        ToReference(GetRoot()->GetObjectFor("Metadata"));
     if (pMetadata)
       m_MetadataObjnum = pMetadata->GetRefObjNum();
   }
@@ -330,7 +343,8 @@ CPDF_Parser::Error CPDF_Parser::SetEncryptHandler() {
     if (CPDF_Dictionary* pEncryptDict = pEncryptObj->AsDictionary()) {
       SetEncryptDictionary(pEncryptDict);
     } else if (CPDF_Reference* pRef = pEncryptObj->AsReference()) {
-      pEncryptObj = m_pDocument->GetOrParseIndirectObject(pRef->GetRefObjNum());
+      pEncryptObj =
+          m_pObjectsHolder->GetOrParseIndirectObject(pRef->GetRefObjNum());
       if (pEncryptObj)
         SetEncryptDictionary(pEncryptObj->GetDict());
     }
@@ -955,8 +969,7 @@ bool CPDF_Parser::RebuildCrossRef() {
 }
 
 bool CPDF_Parser::LoadCrossRefV5(FX_FILESIZE* pos, bool bMainXRef) {
-  std::unique_ptr<CPDF_Object> pObject(
-      ParseIndirectObjectAt(m_pDocument.Get(), *pos, 0));
+  std::unique_ptr<CPDF_Object> pObject(ParseIndirectObjectAt(*pos, 0));
   if (!pObject)
     return false;
 
@@ -965,14 +978,12 @@ bool CPDF_Parser::LoadCrossRefV5(FX_FILESIZE* pos, bool bMainXRef) {
     return false;
 
   CPDF_Object* pUnownedObject = pObject.get();
-  if (m_pDocument) {
-    const CPDF_Dictionary* pRootDict = m_pDocument->GetRoot();
-    if (pRootDict && pRootDict->GetObjNum() == objnum)
-      return false;
-    if (!m_pDocument->ReplaceIndirectObjectIfHigherGeneration(
-            objnum, std::move(pObject))) {
-      return false;
-    }
+  const CPDF_Dictionary* pRootDict = GetRoot();
+  if (pRootDict && pRootDict->GetObjNum() == objnum)
+    return false;
+  if (!m_pObjectsHolder->ReplaceIndirectObjectIfHigherGeneration(
+          objnum, std::move(pObject))) {
+    return false;
   }
 
   CPDF_Stream* pStream = pUnownedObject->AsStream();
@@ -1104,6 +1115,12 @@ const CPDF_Array* CPDF_Parser::GetIDArray() const {
   return GetTrailer() ? GetTrailer()->GetArrayFor("ID") : nullptr;
 }
 
+CPDF_Dictionary* CPDF_Parser::GetRoot() const {
+  CPDF_Object* obj =
+      m_pObjectsHolder->GetOrParseIndirectObject(GetRootObjNum());
+  return obj ? obj->GetDict() : nullptr;
+}
+
 CPDF_Dictionary* CPDF_Parser::GetTrailer() const {
   return m_TrailerData->GetMainTrailer();
 }
@@ -1121,7 +1138,6 @@ uint32_t CPDF_Parser::GetRootObjNum() const {
 }
 
 std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObject(
-    CPDF_IndirectObjectHolder* pObjList,
     uint32_t objnum) {
   if (!IsValidObjectNumber(objnum))
     return nullptr;
@@ -1135,22 +1151,20 @@ std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObject(
     FX_FILESIZE pos = m_ObjectInfo[objnum].pos;
     if (pos <= 0)
       return nullptr;
-    return ParseIndirectObjectAt(pObjList, pos, objnum);
+    return ParseIndirectObjectAt(pos, objnum);
   }
   if (GetObjectType(objnum) != ObjectType::kCompressed)
     return nullptr;
 
   const CPDF_ObjectStream* pObjStream =
-      GetObjectStream(pObjList, m_ObjectInfo[objnum].pos);
+      GetObjectStream(m_ObjectInfo[objnum].pos);
   if (!pObjStream)
     return nullptr;
 
-  return pObjStream->ParseObject(pObjList, objnum);
+  return pObjStream->ParseObject(m_pObjectsHolder.Get(), objnum);
 }
 
-const CPDF_ObjectStream* CPDF_Parser::GetObjectStream(
-    CPDF_IndirectObjectHolder* pObjList,
-    uint32_t object_number) {
+const CPDF_ObjectStream* CPDF_Parser::GetObjectStream(uint32_t object_number) {
   // Prevent circular parsing the same object.
   if (pdfium::ContainsKey(m_ParsingObjNums, object_number))
     return nullptr;
@@ -1167,7 +1181,7 @@ const CPDF_ObjectStream* CPDF_Parser::GetObjectStream(
     return nullptr;
 
   std::unique_ptr<CPDF_Object> object =
-      ParseIndirectObjectAt(pObjList, object_pos, object_number);
+      ParseIndirectObjectAt(object_pos, object_number);
   if (!object)
     return nullptr;
 
@@ -1180,22 +1194,21 @@ const CPDF_ObjectStream* CPDF_Parser::GetObjectStream(
 }
 
 std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObjectAt(
-    CPDF_IndirectObjectHolder* pObjList,
     FX_FILESIZE pos,
     uint32_t objnum) {
   return ParseIndirectObjectAtInternal(
-      pObjList, pos, objnum, CPDF_SyntaxParser::ParseType::kLoose, nullptr);
+      pos, objnum, CPDF_SyntaxParser::ParseType::kLoose, nullptr);
 }
 
 std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObjectAtInternal(
-    CPDF_IndirectObjectHolder* pObjList,
     FX_FILESIZE pos,
     uint32_t objnum,
     CPDF_SyntaxParser::ParseType parse_type,
     FX_FILESIZE* pResultPos) {
   const FX_FILESIZE saved_pos = m_pSyntax->GetPos();
   m_pSyntax->SetPos(pos);
-  auto result = m_pSyntax->GetIndirectObject(pObjList, parse_type);
+  auto result =
+      m_pSyntax->GetIndirectObject(m_pObjectsHolder.Get(), parse_type);
 
   if (pResultPos)
     *pResultPos = m_pSyntax->GetPos();
@@ -1215,12 +1228,11 @@ std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObjectAtInternal(
 }
 
 std::unique_ptr<CPDF_Object> CPDF_Parser::ParseIndirectObjectAtByStrict(
-    CPDF_IndirectObjectHolder* pObjList,
     FX_FILESIZE pos,
     uint32_t objnum,
     FX_FILESIZE* pResultPos) {
   return ParseIndirectObjectAtInternal(
-      pObjList, pos, objnum, CPDF_SyntaxParser::ParseType::kStrict, pResultPos);
+      pos, objnum, CPDF_SyntaxParser::ParseType::kStrict, pResultPos);
 }
 
 uint32_t CPDF_Parser::GetFirstPageNo() const {
@@ -1236,7 +1248,7 @@ std::unique_ptr<CPDF_Dictionary> CPDF_Parser::LoadTrailerV4() {
   if (m_pSyntax->GetKeyword() != "trailer")
     return nullptr;
 
-  return ToDictionary(m_pSyntax->GetObjectBody(m_pDocument.Get()));
+  return ToDictionary(m_pSyntax->GetObjectBody(m_pObjectsHolder.Get()));
 }
 
 uint32_t CPDF_Parser::GetPermissions() const {
@@ -1258,8 +1270,9 @@ std::unique_ptr<CPDF_LinearizedHeader> CPDF_Parser::ParseLinearizedHeader() {
 
 CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
     const RetainPtr<CPDF_ReadValidator>& validator,
-    CPDF_Document* pDocument) {
+    const char* password) {
   ASSERT(!m_bHasParsed);
+  SetPassword(password);
   m_bXRefStream = false;
   m_LastXRefOffset = 0;
 
@@ -1268,10 +1281,9 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
 
   m_pLinearized = ParseLinearizedHeader();
   if (!m_pLinearized)
-    return StartParseInternal(std::move(pDocument));
+    return StartParseInternal();
 
   m_bHasParsed = true;
-  m_pDocument = pDocument;
 
   m_LastXRefOffset = m_pLinearized->GetLastXRefOffset();
   FX_FILESIZE dwFirstXRefOffset = m_LastXRefOffset;
@@ -1299,8 +1311,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
   if (eRet != SUCCESS)
     return eRet;
 
-  m_pDocument->LoadDoc();
-  if (!m_pDocument->GetRoot() || m_pDocument->GetPageCount() == 0) {
+  if (!GetRoot() || !m_pObjectsHolder->TryInit()) {
     if (bXRefRebuilt)
       return FORMAT_ERROR;
 
@@ -1312,8 +1323,8 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
     if (eRet != SUCCESS)
       return eRet;
 
-    m_pDocument->LoadDoc();
-    if (!m_pDocument->GetRoot())
+    m_pObjectsHolder->TryInit();
+    if (!GetRoot())
       return FORMAT_ERROR;
   }
 
@@ -1329,7 +1340,7 @@ CPDF_Parser::Error CPDF_Parser::StartLinearizedParse(
 
   if (m_pSecurityHandler && m_pSecurityHandler->IsMetadataEncrypted()) {
     if (CPDF_Reference* pMetadata =
-            ToReference(m_pDocument->GetRoot()->GetObjectFor("Metadata")))
+            ToReference(GetRoot()->GetObjectFor("Metadata")))
       m_MetadataObjnum = pMetadata->GetRefObjNum();
   }
   return SUCCESS;