Implement CPDF_CrossRefTable

Change-Id: I5ac61ab323adb5eec2de8660064fff95ee877b5e
Reviewed-on: https://pdfium-review.googlesource.com/35432
Reviewed-by: dsinclair <dsinclair@chromium.org>
Commit-Queue: Art Snake <art-snake@yandex-team.ru>

5 files changed, 228 insertions, 68 deletions

diff --git a/BUILD.gn b/BUILD.gn
index 938bfa9f9b..6c26519f6e 100644
--- a/BUILD.gn
+++ b/BUILD.gn
@@ -605,6 +605,8 @@ jumbo_static_library("fpdfapi") {
     "core/fpdfapi/parser/cpdf_boolean.h",
     "core/fpdfapi/parser/cpdf_cross_ref_avail.cpp",
     "core/fpdfapi/parser/cpdf_cross_ref_avail.h",
+    "core/fpdfapi/parser/cpdf_cross_ref_table.cpp",
+    "core/fpdfapi/parser/cpdf_cross_ref_table.h",
     "core/fpdfapi/parser/cpdf_crypto_handler.cpp",
     "core/fpdfapi/parser/cpdf_crypto_handler.h",
     "core/fpdfapi/parser/cpdf_data_avail.cpp",
diff --git a/core/fpdfapi/parser/cpdf_cross_ref_table.cpp b/core/fpdfapi/parser/cpdf_cross_ref_table.cpp
new file mode 100644
index 0000000000..770c483dca
--- /dev/null
+++ b/core/fpdfapi/parser/cpdf_cross_ref_table.cpp
@@ -0,0 +1,129 @@
+// Copyright 2018 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.
+
+#include "core/fpdfapi/parser/cpdf_cross_ref_table.h"
+
+#include <utility>
+
+#include "core/fpdfapi/parser/cpdf_dictionary.h"
+
+CPDF_CrossRefTable::CPDF_CrossRefTable() = default;
+
+CPDF_CrossRefTable::CPDF_CrossRefTable(std::unique_ptr<CPDF_Dictionary> trailer)
+    : trailer_(std::move(trailer)) {}
+
+CPDF_CrossRefTable::~CPDF_CrossRefTable() = default;
+
+void CPDF_CrossRefTable::AddCompressed(uint32_t obj_num,
+                                       uint32_t archive_obj_num) {
+  auto& info = objects_info_[obj_num];
+  if (info.gennum > 0)
+    return;
+
+  if (info.type == ObjectType::kObjStream)
+    return;
+
+  info.type = ObjectType::kCompressed;
+  info.archive_obj_num = archive_obj_num;
+  info.gennum = 0;
+
+  objects_info_[archive_obj_num].type = ObjectType::kObjStream;
+}
+
+void CPDF_CrossRefTable::AddNormal(uint32_t obj_num,
+                                   uint16_t gen_num,
+                                   FX_FILESIZE pos) {
+  auto& info = objects_info_[obj_num];
+  if (info.gennum > gen_num)
+    return;
+
+  if (info.type == ObjectType::kCompressed && gen_num == 0)
+    return;
+
+  if (info.type != ObjectType::kObjStream)
+    info.type = ObjectType::kNormal;
+
+  info.gennum = gen_num;
+  info.pos = pos;
+}
+
+void CPDF_CrossRefTable::SetFree(uint32_t obj_num) {
+  auto& info = objects_info_[obj_num];
+  info.type = ObjectType::kFree;
+  info.gennum = 0xFFFF;
+  info.pos = 0;
+}
+
+void CPDF_CrossRefTable::SetTrailer(std::unique_ptr<CPDF_Dictionary> trailer) {
+  trailer_ = std::move(trailer);
+}
+
+const CPDF_CrossRefTable::ObjectInfo* CPDF_CrossRefTable::GetObjectInfo(
+    uint32_t obj_num) const {
+  const auto it = objects_info_.find(obj_num);
+  return it != objects_info_.end() ? &it->second : nullptr;
+}
+
+void CPDF_CrossRefTable::Update(
+    std::unique_ptr<CPDF_CrossRefTable> new_cross_ref) {
+  UpdateInfo(std::move(new_cross_ref->objects_info_));
+  UpdateTrailer(std::move(new_cross_ref->trailer_));
+}
+
+void CPDF_CrossRefTable::ShrinkObjectMap(uint32_t objnum) {
+  if (objnum == 0) {
+    objects_info_.clear();
+    return;
+  }
+
+  objects_info_.erase(objects_info_.lower_bound(objnum), objects_info_.end());
+
+  if (!pdfium::ContainsKey(objects_info_, objnum - 1))
+    objects_info_[objnum - 1].pos = 0;
+}
+
+void CPDF_CrossRefTable::UpdateInfo(
+    std::map<uint32_t, ObjectInfo>&& new_objects_info) {
+  auto cur_it = objects_info_.begin();
+  auto new_it = new_objects_info.begin();
+  while (cur_it != objects_info_.end() && new_it != new_objects_info.end()) {
+    if (cur_it->first == new_it->first) {
+      if (cur_it->second.type == ObjectType::kObjStream &&
+          new_it->second.type == ObjectType::kNormal) {
+        new_it->second.type = ObjectType::kObjStream;
+      }
+      ++cur_it;
+      ++new_it;
+    } else if (cur_it->first < new_it->first) {
+      new_objects_info.insert(new_it, *cur_it);
+      ++cur_it;
+    } else {
+      new_it = new_objects_info.lower_bound(cur_it->first);
+    }
+  }
+  for (; cur_it != objects_info_.end(); ++cur_it) {
+    new_objects_info.insert(new_objects_info.end(), *cur_it);
+  }
+  objects_info_ = std::move(new_objects_info);
+}
+
+void CPDF_CrossRefTable::UpdateTrailer(
+    std::unique_ptr<CPDF_Dictionary> new_trailer) {
+  if (!new_trailer)
+    return;
+
+  if (!trailer_) {
+    trailer_ = std::move(new_trailer);
+    return;
+  }
+
+  new_trailer->SetFor("XRefStm", trailer_->RemoveFor("XRefStm"));
+  new_trailer->SetFor("Prev", trailer_->RemoveFor("Prev"));
+
+  for (auto it = new_trailer->begin(); it != new_trailer->end();) {
+    const ByteString key = it->first;
+    ++it;
+    trailer_->SetFor(key, new_trailer->RemoveFor(key));
+  }
+}
diff --git a/core/fpdfapi/parser/cpdf_cross_ref_table.h b/core/fpdfapi/parser/cpdf_cross_ref_table.h
new file mode 100644
index 0000000000..ade1b336b2
--- /dev/null
+++ b/core/fpdfapi/parser/cpdf_cross_ref_table.h
@@ -0,0 +1,68 @@
+// Copyright 2018 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.
+
+#ifndef CORE_FPDFAPI_PARSER_CPDF_CROSS_REF_TABLE_H_
+#define CORE_FPDFAPI_PARSER_CPDF_CROSS_REF_TABLE_H_
+
+#include <map>
+#include <memory>
+
+#include "core/fxcrt/fx_system.h"
+
+class CPDF_Dictionary;
+
+class CPDF_CrossRefTable {
+ public:
+  enum class ObjectType : uint8_t {
+    kFree = 0x00,
+    kNormal = 0x01,
+    kNotCompressed = kNormal,
+    kCompressed = 0x02,
+    kObjStream = 0xFF,
+    kNull = kObjStream,
+  };
+
+  struct ObjectInfo {
+    ObjectInfo() : pos(0), type(ObjectType::kFree), gennum(0) {}
+    // if type is ObjectType::kCompressed the archive_obj_num should be used.
+    // if type is ObjectType::kNotCompressed the pos should be used.
+    // In other cases its are unused.
+    union {
+      FX_FILESIZE pos;
+      uint32_t archive_obj_num;
+    };
+    ObjectType type;
+    uint16_t gennum;
+  };
+
+  CPDF_CrossRefTable();
+  explicit CPDF_CrossRefTable(std::unique_ptr<CPDF_Dictionary> trailer);
+  ~CPDF_CrossRefTable();
+
+  void AddCompressed(uint32_t obj_num, uint32_t archive_obj_num);
+  void AddNormal(uint32_t obj_num, uint16_t gen_num, FX_FILESIZE pos);
+  void SetFree(uint32_t obj_num);
+
+  const CPDF_Dictionary* trailer() const { return trailer_.get(); }
+  void SetTrailer(std::unique_ptr<CPDF_Dictionary> trailer);
+
+  const ObjectInfo* GetObjectInfo(uint32_t obj_num) const;
+
+  const std::map<uint32_t, ObjectInfo>& objects_info() const {
+    return objects_info_;
+  }
+
+  void Update(std::unique_ptr<CPDF_CrossRefTable> new_cross_ref);
+
+  void ShrinkObjectMap(uint32_t objnum);
+
+ private:
+  void UpdateInfo(std::map<uint32_t, ObjectInfo>&& new_objects_info);
+  void UpdateTrailer(std::unique_ptr<CPDF_Dictionary> new_trailer);
+
+  std::unique_ptr<CPDF_Dictionary> trailer_;
+  std::map<uint32_t, ObjectInfo> objects_info_;
+};
+
+#endif  // CORE_FPDFAPI_PARSER_CPDF_CROSS_REF_TABLE_H_
diff --git a/core/fpdfapi/parser/cpdf_parser.cpp b/core/fpdfapi/parser/cpdf_parser.cpp
index 3a7afd0b49..c7a3fe16c8 100644
--- a/core/fpdfapi/parser/cpdf_parser.cpp
+++ b/core/fpdfapi/parser/cpdf_parser.cpp
@@ -670,6 +670,7 @@ bool CPDF_Parser::LoadAllCrossRefV5(FX_FILESIZE xrefpos) {
 }
 
 bool CPDF_Parser::RebuildCrossRef() {
+  CPDF_CrossRefTable cross_ref_table;
   m_ObjectInfo.clear();
   m_TrailerData->Clear();
 
@@ -826,23 +827,14 @@ bool CPDF_Parser::RebuildCrossRef() {
                       CPDF_Object* pRoot = pDict->GetObjectFor("Root");
                       if (pRoot && pRoot->GetDict() &&
                           pRoot->GetDict()->GetObjectFor("Pages")) {
-                        m_TrailerData->SetMainTrailer(
-                            ToDictionary(pDict->Clone()));
+                        cross_ref_table.Update(
+                            pdfium::MakeUnique<CPDF_CrossRefTable>(
+                                ToDictionary(pDict->Clone())));
                       }
                     }
                   }
                 }
-                ObjectInfo& info = m_ObjectInfo[objnum];
-                if (pObject || !info.pos) {
-                  info.pos = obj_pos;
-                  info.type = ObjectType::kNotCompressed;
-                  // The newer version of object should be located after old
-                  // version.
-                  // Do not worry about gennum in this case, because we read
-                  // file
-                  // in front order.
-                  info.gennum = gennum;
-                }
+                cross_ref_table.AddNormal(objnum, gennum, obj_pos);
               }
               state = ParserState::kDefault;
               break;
@@ -859,42 +851,21 @@ bool CPDF_Parser::RebuildCrossRef() {
               if (!pObj)
                 m_pSyntax->SetPos(current_char_pos);
 
-              {
-                if (pObj && (pObj->IsDictionary() || pObj->IsStream())) {
-                  if (CPDF_Dictionary* pTrailer =
-                          pObj->IsStream() ? pObj->AsStream()->GetDict()
-                                           : pObj->AsDictionary()) {
-                    if (GetTrailer()) {
-                      CPDF_Object* pRoot = pTrailer->GetObjectFor("Root");
-                      CPDF_Reference* pRef = ToReference(pRoot);
-                      if (!pRoot ||
-                          (pRef && IsValidObjectNumber(pRef->GetRefObjNum()) &&
-                           m_ObjectInfo[pRef->GetRefObjNum()].pos != 0)) {
-                        // This is newer version of trailer. Merge it with old.
-                        for (auto it = pTrailer->begin();
-                             it != pTrailer->end();) {
-                          DCHECK(it->second->IsInline());
-                          const ByteString key = it->first;
-                          ++it;
-                          GetTrailer()->SetFor(key, pTrailer->RemoveFor(key));
-                        }
-                      }
-                    } else {
-                      m_TrailerData->SetMainTrailer(
-                          ToDictionary(pObj->IsStream() ? pTrailer->Clone()
-                                                        : std::move(pObj)));
-
-                      FX_FILESIZE dwSavePos = m_pSyntax->GetPos();
-                      ByteString strWord = m_pSyntax->GetKeyword();
-                      if (!strWord.Compare("startxref")) {
-                        bool bNumber;
-                        ByteString bsOffset = m_pSyntax->GetNextWord(&bNumber);
-                        if (bNumber)
-                          m_LastXRefOffset = FXSYS_atoi(bsOffset.c_str());
-                      }
-                      m_pSyntax->SetPos(dwSavePos);
-                    }
-                  }
+              if (pObj) {
+                cross_ref_table.Update(pdfium::MakeUnique<CPDF_CrossRefTable>(
+                    ToDictionary(pObj->IsStream()
+                                     ? pObj->AsStream()->GetDict()->Clone()
+                                     : std::move(pObj))));
+
+                FX_FILESIZE dwSavePos = m_pSyntax->GetPos();
+                ByteString strWord = m_pSyntax->GetKeyword();
+                if (!strWord.Compare("startxref")) {
+                  bool bNumber;
+                  ByteString bsOffset = m_pSyntax->GetNextWord(&bNumber);
+                  if (bNumber)
+                    last_xref = FXSYS_atoi(bsOffset.c_str());
+                } else {
+                  m_pSyntax->SetPos(dwSavePos);
                 }
               }
             }
@@ -972,8 +943,14 @@ bool CPDF_Parser::RebuildCrossRef() {
   else if (last_trailer == -1 || last_xref < last_obj)
     last_trailer = m_pSyntax->m_FileLen;
 
+  if (cross_ref_table.trailer()) {
+    m_TrailerData->SetMainTrailer(
+        ToDictionary(cross_ref_table.trailer()->Clone()));
+    m_ObjectInfo = cross_ref_table.objects_info();
+  }
   // Resore default buffer size.
   m_pSyntax->SetReadBufferSize(CPDF_ModuleMgr::kFileBufSize);
+
   return GetTrailer() && !m_ObjectInfo.empty();
 }
 
diff --git a/core/fpdfapi/parser/cpdf_parser.h b/core/fpdfapi/parser/cpdf_parser.h
index 3151da61c9..f859db5d16 100644
--- a/core/fpdfapi/parser/cpdf_parser.h
+++ b/core/fpdfapi/parser/cpdf_parser.h
@@ -13,6 +13,7 @@
 #include <set>
 #include <vector>
 
+#include "core/fpdfapi/parser/cpdf_cross_ref_table.h"
 #include "core/fpdfapi/parser/cpdf_syntax_parser.h"
 #include "core/fxcrt/fx_string.h"
 #include "core/fxcrt/fx_system.h"
@@ -113,25 +114,8 @@ class CPDF_Parser {
   void SetLinearizedHeader(std::unique_ptr<CPDF_LinearizedHeader> pLinearized);
 
  protected:
-  enum class ObjectType : uint8_t {
-    kFree = 0x00,
-    kNotCompressed = 0x01,
-    kCompressed = 0x02,
-    kNull = 0xFF,
-  };
-
-  struct ObjectInfo {
-    ObjectInfo() : pos(0), type(ObjectType::kFree), gennum(0) {}
-    // if type is ObjectType::kCompressed the archive_obj_num should be used.
-    // if type is ObjectType::kNotCompressed the pos should be used.
-    // In other cases its are unused.
-    union {
-      FX_FILESIZE pos;
-      FX_FILESIZE archive_obj_num;
-    };
-    ObjectType type;
-    uint16_t gennum;
-  };
+  using ObjectType = CPDF_CrossRefTable::ObjectType;
+  using ObjectInfo = CPDF_CrossRefTable::ObjectInfo;
 
   std::unique_ptr<CPDF_SyntaxParser> m_pSyntax;
   std::map<uint32_t, ObjectInfo> m_ObjectInfo;