summaryrefslogtreecommitdiff
path: root/core/fpdfapi/parser
diff options
context:
space:
mode:
authorArtem Strygin <art-snake@yandex-team.ru>2018-06-27 18:15:10 +0000
committerChromium commit bot <commit-bot@chromium.org>2018-06-27 18:15:10 +0000
commit20eca1e383b7dce69cd791d42bda3558a3966301 (patch)
treed3fded7303e8326d3883ca51b6bed2ad2438b6da /core/fpdfapi/parser
parent00ba8bbea0ff57d6f11257736408e530e54ef642 (diff)
downloadpdfium-20eca1e383b7dce69cd791d42bda3558a3966301.tar.xz
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>
Diffstat (limited to 'core/fpdfapi/parser')
-rw-r--r--core/fpdfapi/parser/cpdf_data_avail.cpp12
-rw-r--r--core/fpdfapi/parser/cpdf_data_avail.h1
-rw-r--r--core/fpdfapi/parser/cpdf_document.cpp39
-rw-r--r--core/fpdfapi/parser/cpdf_document.h21
-rw-r--r--core/fpdfapi/parser/cpdf_document_unittest.cpp16
-rw-r--r--core/fpdfapi/parser/cpdf_parser.cpp103
-rw-r--r--core/fpdfapi/parser/cpdf_parser.h29
7 files changed, 137 insertions, 84 deletions
diff --git a/core/fpdfapi/parser/cpdf_data_avail.cpp b/core/fpdfapi/parser/cpdf_data_avail.cpp
index acc01c65ed..3b39e3f54b 100644
--- a/core/fpdfapi/parser/cpdf_data_avail.cpp
+++ b/core/fpdfapi/parser/cpdf_data_avail.cpp
@@ -214,7 +214,7 @@ std::unique_ptr<CPDF_Object> CPDF_DataAvail::GetObject(uint32_t objnum,
std::unique_ptr<CPDF_Object> pRet;
if (pParser) {
const CPDF_ReadValidator::Session read_session(GetValidator().Get());
- pRet = pParser->ParseIndirectObject(nullptr, objnum);
+ pRet = pParser->ParseIndirectObject(objnum);
if (GetValidator()->has_read_problems())
return nullptr;
}
@@ -233,7 +233,7 @@ bool CPDF_DataAvail::CheckInfo() {
}
const CPDF_ReadValidator::Session read_session(GetValidator().Get());
- m_parser.ParseIndirectObject(nullptr, dwInfoObjNum);
+ m_parser.ParseIndirectObject(dwInfoObjNum);
if (GetValidator()->has_read_problems())
return false;
@@ -249,7 +249,7 @@ bool CPDF_DataAvail::CheckRoot() {
}
const CPDF_ReadValidator::Session read_session(GetValidator().Get());
- m_pRoot = ToDictionary(m_parser.ParseIndirectObject(nullptr, dwRootObjNum));
+ m_pRoot = ToDictionary(m_parser.ParseIndirectObject(dwRootObjNum));
if (GetValidator()->has_read_problems())
return false;
@@ -1011,11 +1011,11 @@ CPDF_DataAvail::ParseDocument(const char* password) {
}
auto parser = pdfium::MakeUnique<CPDF_Parser>();
parser->SetPassword(password);
- auto document = pdfium::MakeUnique<CPDF_Document>(std::move(parser));
+ auto document = pdfium::MakeUnique<CPDF_Document>();
CPDF_ReadValidator::Session read_session(GetValidator().Get());
- CPDF_Parser::Error error = document->GetParser()->StartLinearizedParse(
- GetValidator(), document.get());
+ CPDF_Parser::Error error =
+ document->LoadLinearizedDoc(GetValidator(), password);
// Additional check, that all ok.
if (GetValidator()->has_read_problems()) {
diff --git a/core/fpdfapi/parser/cpdf_data_avail.h b/core/fpdfapi/parser/cpdf_data_avail.h
index 8872bb103f..b6396bf82a 100644
--- a/core/fpdfapi/parser/cpdf_data_avail.h
+++ b/core/fpdfapi/parser/cpdf_data_avail.h
@@ -19,6 +19,7 @@
class CPDF_CrossRefAvail;
class CPDF_Dictionary;
+class CPDF_Document;
class CPDF_HintTables;
class CPDF_IndirectObjectHolder;
class CPDF_LinearizedHeader;
diff --git a/core/fpdfapi/parser/cpdf_document.cpp b/core/fpdfapi/parser/cpdf_document.cpp
index 8f727ed14a..91c6b9c6f9 100644
--- a/core/fpdfapi/parser/cpdf_document.cpp
+++ b/core/fpdfapi/parser/cpdf_document.cpp
@@ -22,6 +22,7 @@
#include "core/fpdfapi/parser/cpdf_name.h"
#include "core/fpdfapi/parser/cpdf_number.h"
#include "core/fpdfapi/parser/cpdf_parser.h"
+#include "core/fpdfapi/parser/cpdf_read_validator.h"
#include "core/fpdfapi/parser/cpdf_reference.h"
#include "core/fpdfapi/parser/cpdf_stream.h"
#include "core/fpdfapi/parser/cpdf_string.h"
@@ -185,17 +186,13 @@ std::unique_ptr<CPDF_Dictionary> CalculateFontDesc(
} // namespace
-CPDF_Document::CPDF_Document(std::unique_ptr<CPDF_Parser> pParser)
- : CPDF_IndirectObjectHolder(),
- m_pParser(std::move(pParser)),
+CPDF_Document::CPDF_Document()
+ : ParsedObjectsHolder(),
m_pRootDict(nullptr),
m_iNextPageToTraverse(0),
m_bReachedMaxPageLevel(false),
m_pDocPage(pdfium::MakeUnique<CPDF_DocPageData>(this)),
- m_pDocRender(pdfium::MakeUnique<CPDF_DocRenderData>(this)) {
- if (pParser)
- SetLastObjNum(m_pParser->GetLastObjNum());
-}
+ m_pDocRender(pdfium::MakeUnique<CPDF_DocRenderData>(this)) {}
CPDF_Document::~CPDF_Document() {
CPDF_ModuleMgr::Get()->GetPageModule()->ClearStockFont(this);
@@ -203,7 +200,7 @@ CPDF_Document::~CPDF_Document() {
std::unique_ptr<CPDF_Object> CPDF_Document::ParseIndirectObject(
uint32_t objnum) {
- return m_pParser ? m_pParser->ParseIndirectObject(this, objnum) : nullptr;
+ return m_pParser ? m_pParser->ParseIndirectObject(objnum) : nullptr;
}
void CPDF_Document::LoadDocInternal() {
@@ -218,9 +215,28 @@ void CPDF_Document::LoadDocInternal() {
return;
}
-void CPDF_Document::LoadDoc() {
+bool CPDF_Document::TryInit() {
LoadDocInternal();
LoadPages();
+ return GetRoot() && (GetPageCount() > 0);
+}
+
+CPDF_Parser::Error CPDF_Document::LoadDoc(
+ const RetainPtr<IFX_SeekableReadStream>& pFileAccess,
+ const char* password) {
+ if (!m_pParser)
+ SetParser(pdfium::MakeUnique<CPDF_Parser>(this));
+
+ return m_pParser->StartParse(pFileAccess, password);
+}
+
+CPDF_Parser::Error CPDF_Document::LoadLinearizedDoc(
+ const RetainPtr<CPDF_ReadValidator>& validator,
+ const char* password) {
+ if (!m_pParser)
+ SetParser(pdfium::MakeUnique<CPDF_Parser>(this));
+
+ return m_pParser->StartLinearizedParse(validator, password);
}
void CPDF_Document::LoadPages() {
@@ -309,6 +325,11 @@ void CPDF_Document::ResetTraversal() {
m_pTreeTraversal.clear();
}
+void CPDF_Document::SetParser(std::unique_ptr<CPDF_Parser> pParser) {
+ DCHECK(!m_pParser);
+ m_pParser = std::move(pParser);
+}
+
const CPDF_Dictionary* CPDF_Document::GetPagesDict() const {
const CPDF_Dictionary* pRoot = GetRoot();
return pRoot ? pRoot->GetDictFor("Pages") : nullptr;
diff --git a/core/fpdfapi/parser/cpdf_document.h b/core/fpdfapi/parser/cpdf_document.h
index 8d0aa08009..328f09b6ef 100644
--- a/core/fpdfapi/parser/cpdf_document.h
+++ b/core/fpdfapi/parser/cpdf_document.h
@@ -15,9 +15,10 @@
#include "core/fpdfapi/page/cpdf_image.h"
#include "core/fpdfapi/page/cpdf_page.h"
-#include "core/fpdfapi/parser/cpdf_indirect_object_holder.h"
#include "core/fpdfapi/parser/cpdf_object.h"
+#include "core/fpdfapi/parser/cpdf_parser.h"
#include "core/fpdfdoc/cpdf_linklist.h"
+#include "core/fxcrt/retain_ptr.h"
class CFX_Font;
class CFX_Matrix;
@@ -30,7 +31,9 @@ class CPDF_IccProfile;
class CPDF_LinearizedHeader;
class CPDF_Parser;
class CPDF_Pattern;
+class CPDF_ReadValidator;
class CPDF_StreamAcc;
+class IFX_SeekableReadStream;
class JBig2_DocumentContext;
#define FPDFPERM_MODIFY 0x0008
@@ -38,7 +41,7 @@ class JBig2_DocumentContext;
#define FPDFPERM_FILL_FORM 0x0100
#define FPDFPERM_EXTRACT_ACCESS 0x0200
-class CPDF_Document : public CPDF_IndirectObjectHolder {
+class CPDF_Document : public CPDF_Parser::ParsedObjectsHolder {
public:
// Type from which the XFA extension can subclass itself.
class Extension {
@@ -52,7 +55,7 @@ class CPDF_Document : public CPDF_IndirectObjectHolder {
static const int kPageMaxNum = 0xFFFFF;
- explicit CPDF_Document(std::unique_ptr<CPDF_Parser> pParser);
+ CPDF_Document();
~CPDF_Document() override;
Extension* GetExtension() const { return m_pExtension.get(); }
@@ -96,7 +99,16 @@ class CPDF_Document : public CPDF_IndirectObjectHolder {
RetainPtr<CPDF_StreamAcc> LoadFontFile(const CPDF_Stream* pStream);
RetainPtr<CPDF_IccProfile> LoadIccProfile(const CPDF_Stream* pStream);
- void LoadDoc();
+ // CPDF_Parser::ParsedObjectsHolder overrides:
+ bool TryInit() override;
+
+ CPDF_Parser::Error LoadDoc(
+ const RetainPtr<IFX_SeekableReadStream>& pFileAccess,
+ const char* password);
+ CPDF_Parser::Error LoadLinearizedDoc(
+ const RetainPtr<CPDF_ReadValidator>& validator,
+ const char* password);
+
void LoadPages();
void CreateNewDoc();
@@ -144,6 +156,7 @@ class CPDF_Document : public CPDF_IndirectObjectHolder {
std::set<CPDF_Dictionary*>* pVisited);
bool InsertNewPage(int iPage, CPDF_Dictionary* pPageDict);
void ResetTraversal();
+ void SetParser(std::unique_ptr<CPDF_Parser> pParser);
std::unique_ptr<CPDF_Parser> m_pParser;
UnownedPtr<CPDF_Dictionary> m_pRootDict;
diff --git a/core/fpdfapi/parser/cpdf_document_unittest.cpp b/core/fpdfapi/parser/cpdf_document_unittest.cpp
index 522aa65fc9..c64da6382a 100644
--- a/core/fpdfapi/parser/cpdf_document_unittest.cpp
+++ b/core/fpdfapi/parser/cpdf_document_unittest.cpp
@@ -49,7 +49,7 @@ std::unique_ptr<CPDF_Dictionary> CreateNumberedPage(size_t number) {
class CPDF_TestDocumentForPages : public CPDF_Document {
public:
- CPDF_TestDocumentForPages() : CPDF_Document(nullptr) {
+ CPDF_TestDocumentForPages() : CPDF_Document() {
// Set up test
auto zeroToTwo = pdfium::MakeUnique<CPDF_Array>();
zeroToTwo->AddNew<CPDF_Reference>(
@@ -101,7 +101,7 @@ class CPDF_TestDocumentForPages : public CPDF_Document {
class CPDF_TestDocumentWithPageWithoutPageNum : public CPDF_Document {
public:
- CPDF_TestDocumentWithPageWithoutPageNum() : CPDF_Document(nullptr) {
+ CPDF_TestDocumentWithPageWithoutPageNum() : CPDF_Document() {
// Set up test
auto allPages = pdfium::MakeUnique<CPDF_Array>();
allPages->AddNew<CPDF_Reference>(
@@ -132,7 +132,7 @@ class TestLinearized : public CPDF_LinearizedHeader {
class CPDF_TestDocPagesWithoutKids : public CPDF_Document {
public:
- CPDF_TestDocPagesWithoutKids() : CPDF_Document(nullptr) {
+ CPDF_TestDocPagesWithoutKids() : CPDF_Document() {
CPDF_Dictionary* pagesDict = NewIndirect<CPDF_Dictionary>();
pagesDict->SetNewFor<CPDF_Name>("Type", "Pages");
pagesDict->SetNewFor<CPDF_Number>("Count", 3);
@@ -143,6 +143,11 @@ class CPDF_TestDocPagesWithoutKids : public CPDF_Document {
}
};
+class CPDF_TestDocumentAllowSetParser : public CPDF_Document {
+ public:
+ using CPDF_Document::SetParser;
+};
+
} // namespace
class cpdf_document_test : public testing::Test {
@@ -222,8 +227,9 @@ TEST_F(cpdf_document_test, UseCachedPageObjNumIfHaveNotPagesDict) {
auto linearized = pdfium::MakeUnique<TestLinearized>(dict.get());
auto parser = pdfium::MakeUnique<CPDF_Parser>();
parser->SetLinearizedHeader(std::move(linearized));
- CPDF_Document document(std::move(parser));
- document.LoadDoc();
+ CPDF_TestDocumentAllowSetParser document;
+ document.SetParser(std::move(parser));
+ document.LoadPages();
ASSERT_EQ(page_count, document.GetPageCount());
CPDF_Object* page_stub = document.NewIndirect<CPDF_Dictionary>();
const uint32_t obj_num = page_stub->GetObjNum();
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;
diff --git a/core/fpdfapi/parser/cpdf_parser.h b/core/fpdfapi/parser/cpdf_parser.h
index f859db5d16..1a1f9b0de4 100644
--- a/core/fpdfapi/parser/cpdf_parser.h
+++ b/core/fpdfapi/parser/cpdf_parser.h
@@ -14,6 +14,7 @@
#include <vector>
#include "core/fpdfapi/parser/cpdf_cross_ref_table.h"
+#include "core/fpdfapi/parser/cpdf_indirect_object_holder.h"
#include "core/fpdfapi/parser/cpdf_syntax_parser.h"
#include "core/fxcrt/fx_string.h"
#include "core/fxcrt/fx_system.h"
@@ -23,8 +24,6 @@
class CPDF_Array;
class CPDF_CryptoHandler;
class CPDF_Dictionary;
-class CPDF_Document;
-class CPDF_IndirectObjectHolder;
class CPDF_LinearizedHeader;
class CPDF_Object;
class CPDF_ObjectStream;
@@ -35,6 +34,11 @@ class IFX_SeekableReadStream;
class CPDF_Parser {
public:
+ class ParsedObjectsHolder : public CPDF_IndirectObjectHolder {
+ public:
+ virtual bool TryInit() = 0;
+ };
+
enum Error {
SUCCESS = 0,
FILE_ERROR,
@@ -49,13 +53,14 @@ class CPDF_Parser {
static const size_t kInvalidPos = std::numeric_limits<size_t>::max();
+ explicit CPDF_Parser(ParsedObjectsHolder* holder);
CPDF_Parser();
~CPDF_Parser();
Error StartParse(const RetainPtr<IFX_SeekableReadStream>& pFile,
- CPDF_Document* pDocument);
+ const char* password);
Error StartLinearizedParse(const RetainPtr<CPDF_ReadValidator>& validator,
- CPDF_Document* pDocument);
+ const char* password);
void SetPassword(const char* password) { m_Password = password; }
ByteString GetPassword() const { return m_Password; }
@@ -72,12 +77,11 @@ class CPDF_Parser {
uint32_t GetRootObjNum() const;
uint32_t GetInfoObjNum() const;
const CPDF_Array* GetIDArray() const;
+ CPDF_Dictionary* GetRoot() const;
CPDF_Dictionary* GetEncryptDict() const { return m_pEncryptDict.Get(); }
- std::unique_ptr<CPDF_Object> ParseIndirectObject(
- CPDF_IndirectObjectHolder* pObjList,
- uint32_t objnum);
+ std::unique_ptr<CPDF_Object> ParseIndirectObject(uint32_t objnum);
uint32_t GetLastObjNum() const;
bool IsValidObjectNumber(uint32_t objnum) const;
@@ -96,12 +100,10 @@ class CPDF_Parser {
bool IsXRefStream() const { return m_bXRefStream; }
std::unique_ptr<CPDF_Object> ParseIndirectObjectAt(
- CPDF_IndirectObjectHolder* pObjList,
FX_FILESIZE pos,
uint32_t objnum);
std::unique_ptr<CPDF_Object> ParseIndirectObjectAtByStrict(
- CPDF_IndirectObjectHolder* pObjList,
FX_FILESIZE pos,
uint32_t objnum,
FX_FILESIZE* pResultPos);
@@ -150,7 +152,7 @@ class CPDF_Parser {
ObjectInfo info;
};
- Error StartParseInternal(CPDF_Document* pDocument);
+ Error StartParseInternal();
FX_FILESIZE ParseStartXRef();
bool LoadAllCrossRefV4(FX_FILESIZE pos);
bool LoadAllCrossRefV5(FX_FILESIZE pos);
@@ -161,8 +163,7 @@ class CPDF_Parser {
bool LoadLinearizedAllCrossRefV4(FX_FILESIZE pos);
bool LoadLinearizedAllCrossRefV5(FX_FILESIZE pos);
Error LoadLinearizedMainXRefTable();
- const CPDF_ObjectStream* GetObjectStream(CPDF_IndirectObjectHolder* pObjList,
- uint32_t object_number);
+ const CPDF_ObjectStream* GetObjectStream(uint32_t object_number);
std::unique_ptr<CPDF_LinearizedHeader> ParseLinearizedHeader();
void SetEncryptDictionary(CPDF_Dictionary* pDict);
void ShrinkObjectMap(uint32_t size);
@@ -180,7 +181,6 @@ class CPDF_Parser {
void MergeCrossRefObjectsData(const std::vector<CrossRefObjData>& objects);
std::unique_ptr<CPDF_Object> ParseIndirectObjectAtInternal(
- CPDF_IndirectObjectHolder* pObjList,
FX_FILESIZE pos,
uint32_t objnum,
CPDF_SyntaxParser::ParseType parse_type,
@@ -193,7 +193,8 @@ class CPDF_Parser {
ObjectType GetObjectTypeFromCrossRefStreamType(
int cross_ref_stream_type) const;
- UnownedPtr<CPDF_Document> m_pDocument;
+ std::unique_ptr<ParsedObjectsHolder> m_pOwnedObjectsHolder;
+ UnownedPtr<ParsedObjectsHolder> m_pObjectsHolder;
bool m_bHasParsed;
bool m_bXRefStream;