Home | History | Annotate | Download | only in parser 
      1 // Copyright 2016 PDFium Authors. All rights reserved.
      2 // Use of this source code is governed by a BSD-style license that can be
      3 // found in the LICENSE file.
      4 
      5 // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com
      6 
      7 #include "core/fpdfapi/parser/cpdf_linearized_header.h"
      8 
      9 #include <algorithm>
     10 #include <limits>
     11 #include <utility>
     12 
     13 #include "core/fpdfapi/parser/cpdf_array.h"
     14 #include "core/fpdfapi/parser/cpdf_dictionary.h"
     15 #include "core/fpdfapi/parser/cpdf_number.h"
     16 #include "core/fpdfapi/parser/cpdf_syntax_parser.h"
     17 #include "third_party/base/ptr_util.h"
     18 
     19 namespace {
     20 
     21 constexpr FX_FILESIZE kLinearizedHeaderOffset = 9;
     22 constexpr size_t kMaxInt = static_cast<size_t>(std::numeric_limits<int>::max());
     23 
     24 template <class T>
     25 bool IsValidNumericDictionaryValue(const CPDF_Dictionary* pDict,
     26                                    const char* key,
     27                                    T min_value,
     28                                    bool must_exist = true) {
     29   if (!pDict->KeyExist(key))
     30     return !must_exist;
     31   const CPDF_Number* pNum = ToNumber(pDict->GetObjectFor(key));
     32   if (!pNum || !pNum->IsInteger())
     33     return false;
     34   const int raw_value = pNum->GetInteger();
     35   if (!pdfium::base::IsValueInRangeForNumericType<T>(raw_value))
     36     return false;
     37   return static_cast<T>(raw_value) >= min_value;
     38 }
     39 
     40 bool IsLinearizedHeaderValid(const CPDF_LinearizedHeader* header,
     41                              FX_FILESIZE file_size) {
     42   ASSERT(header);
     43   return header->GetFileSize() == file_size &&
     44          static_cast<int>(header->GetFirstPageNo()) >= 0 &&
     45          header->GetFirstPageNo() < kMaxInt &&
     46          header->GetMainXRefTableFirstEntryOffset() < file_size &&
     47          header->GetPageCount() > 0 &&
     48          header->GetFirstPageEndOffset() < file_size &&
     49          header->GetLastXRefOffset() < file_size &&
     50          header->GetHintStart() < file_size;
     51 }
     52 
     53 }  // namespace
     54 
     55 // static
     56 std::unique_ptr<CPDF_LinearizedHeader> CPDF_LinearizedHeader::Parse(
     57     CPDF_SyntaxParser* parser) {
     58   parser->SetPos(kLinearizedHeaderOffset);
     59 
     60   const auto pDict = ToDictionary(
     61       parser->GetIndirectObject(nullptr, CPDF_SyntaxParser::ParseType::kLoose));
     62 
     63   if (!pDict || !pDict->KeyExist("Linearized") ||
     64       !IsValidNumericDictionaryValue<FX_FILESIZE>(pDict.get(), "L", 1) ||
     65       !IsValidNumericDictionaryValue<uint32_t>(pDict.get(), "P", 0, false) ||
     66       !IsValidNumericDictionaryValue<FX_FILESIZE>(pDict.get(), "T", 1) ||
     67       !IsValidNumericDictionaryValue<uint32_t>(pDict.get(), "N", 0) ||
     68       !IsValidNumericDictionaryValue<FX_FILESIZE>(pDict.get(), "E", 1) ||
     69       !IsValidNumericDictionaryValue<uint32_t>(pDict.get(), "O", 1)) {
     70     return nullptr;
     71   }
     72   // Move parser to the start of the xref table for the documents first page.
     73   // (skpping endobj keyword)
     74   if (parser->GetNextWord(nullptr) != "endobj")
     75     return nullptr;
     76 
     77   auto result = pdfium::WrapUnique(
     78       new CPDF_LinearizedHeader(pDict.get(), parser->GetPos()));
     79 
     80   if (!IsLinearizedHeaderValid(result.get(),
     81                                parser->GetFileAccess()->GetSize())) {
     82     return nullptr;
     83   }
     84   return result;
     85 }
     86 
     87 CPDF_LinearizedHeader::CPDF_LinearizedHeader(const CPDF_Dictionary* pDict,
     88                                              FX_FILESIZE szLastXRefOffset)
     89     : m_szFileSize(pDict->GetIntegerFor("L")),
     90       m_dwFirstPageNo(pDict->GetIntegerFor("P")),
     91       m_szMainXRefTableFirstEntryOffset(pDict->GetIntegerFor("T")),
     92       m_PageCount(pDict->GetIntegerFor("N")),
     93       m_szFirstPageEndOffset(pDict->GetIntegerFor("E")),
     94       m_FirstPageObjNum(pDict->GetIntegerFor("O")),
     95       m_szLastXRefOffset(szLastXRefOffset) {
     96   const CPDF_Array* pHintStreamRange = pDict->GetArrayFor("H");
     97   const size_t nHintStreamSize =
     98       pHintStreamRange ? pHintStreamRange->GetCount() : 0;
     99   if (nHintStreamSize == 2 || nHintStreamSize == 4) {
    100     m_szHintStart = std::max(pHintStreamRange->GetIntegerAt(0), 0);
    101     const FX_SAFE_UINT32 safe_hint_length = pHintStreamRange->GetIntegerAt(1);
    102     if (safe_hint_length.IsValid())
    103       m_HintLength = safe_hint_length.ValueOrDie();
    104   }
    105 }
    106 
    107 CPDF_LinearizedHeader::~CPDF_LinearizedHeader() {}
    108 
    109 bool CPDF_LinearizedHeader::HasHintTable() const {
    110   return GetPageCount() > 1 && GetHintStart() > 0 && GetHintLength() > 0;
    111 }
    112