// Copyright 2014 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. // Original code copyright 2014 Foxit Software Inc. http://www.foxitsoftware.com #include "public/fpdf_ppo.h" #include #include #include #include #include "core/fpdfapi/parser/cpdf_array.h" #include "core/fpdfapi/parser/cpdf_document.h" #include "core/fpdfapi/parser/cpdf_name.h" #include "core/fpdfapi/parser/cpdf_number.h" #include "core/fpdfapi/parser/cpdf_reference.h" #include "core/fpdfapi/parser/cpdf_stream.h" #include "core/fpdfapi/parser/cpdf_string.h" #include "core/fxcrt/unowned_ptr.h" #include "fpdfsdk/fsdk_define.h" #include "third_party/base/ptr_util.h" namespace { CPDF_Object* PageDictGetInheritableTag(CPDF_Dictionary* pDict, const ByteString& bsSrcTag) { if (!pDict || bsSrcTag.IsEmpty()) return nullptr; if (!pDict->KeyExist("Parent") || !pDict->KeyExist("Type")) return nullptr; CPDF_Object* pType = pDict->GetObjectFor("Type")->GetDirect(); if (!ToName(pType)) return nullptr; if (pType->GetString().Compare("Page")) return nullptr; CPDF_Dictionary* pp = ToDictionary(pDict->GetObjectFor("Parent")->GetDirect()); if (!pp) return nullptr; if (pDict->KeyExist(bsSrcTag)) return pDict->GetObjectFor(bsSrcTag); while (pp) { if (pp->KeyExist(bsSrcTag)) return pp->GetObjectFor(bsSrcTag); if (!pp->KeyExist("Parent")) break; pp = ToDictionary(pp->GetObjectFor("Parent")->GetDirect()); } return nullptr; } bool CopyInheritable(CPDF_Dictionary* pCurPageDict, CPDF_Dictionary* pSrcPageDict, const ByteString& key) { if (pCurPageDict->KeyExist(key)) return true; CPDF_Object* pInheritable = PageDictGetInheritableTag(pSrcPageDict, key); if (!pInheritable) return false; pCurPageDict->SetFor(key, pInheritable->Clone()); return true; } bool ParserPageRangeString(ByteString rangstring, std::vector* pageArray, int nCount) { if (rangstring.IsEmpty()) return true; rangstring.Remove(' '); size_t nLength = rangstring.GetLength(); ByteString cbCompareString("0123456789-,"); for (size_t i = 0; i < nLength; ++i) { if (!cbCompareString.Contains(rangstring[i])) return false; } ByteString cbMidRange; size_t nStringFrom = 0; pdfium::Optional nStringTo = 0; while (nStringTo < nLength) { nStringTo = rangstring.Find(',', nStringFrom); if (!nStringTo.has_value()) nStringTo = nLength; cbMidRange = rangstring.Mid(nStringFrom, nStringTo.value() - nStringFrom); auto nMid = cbMidRange.Find('-'); if (!nMid.has_value()) { uint16_t pageNum = pdfium::base::checked_cast(atoi(cbMidRange.c_str())); if (pageNum <= 0 || pageNum > nCount) return false; pageArray->push_back(pageNum); } else { uint16_t nStartPageNum = pdfium::base::checked_cast( atoi(cbMidRange.Left(nMid.value()).c_str())); if (nStartPageNum == 0) return false; nMid = nMid.value() + 1; size_t nEnd = cbMidRange.GetLength() - nMid.value(); if (nEnd == 0) return false; uint16_t nEndPageNum = pdfium::base::checked_cast( atoi(cbMidRange.Mid(nMid.value(), nEnd).c_str())); if (nStartPageNum < 0 || nStartPageNum > nEndPageNum || nEndPageNum > nCount) { return false; } for (uint16_t i = nStartPageNum; i <= nEndPageNum; ++i) { pageArray->push_back(i); } } nStringFrom = nStringTo.value() + 1; } return true; } } // namespace class CPDF_PageOrganizer { public: CPDF_PageOrganizer(CPDF_Document* pDestPDFDoc, CPDF_Document* pSrcPDFDoc); ~CPDF_PageOrganizer(); bool PDFDocInit(); bool ExportPage(const std::vector& pageNums, int nIndex); private: using ObjectNumberMap = std::map; bool UpdateReference(CPDF_Object* pObj, ObjectNumberMap* pObjNumberMap); uint32_t GetNewObjId(ObjectNumberMap* pObjNumberMap, CPDF_Reference* pRef); UnownedPtr m_pDestPDFDoc; UnownedPtr m_pSrcPDFDoc; }; CPDF_PageOrganizer::CPDF_PageOrganizer(CPDF_Document* pDestPDFDoc, CPDF_Document* pSrcPDFDoc) : m_pDestPDFDoc(pDestPDFDoc), m_pSrcPDFDoc(pSrcPDFDoc) {} CPDF_PageOrganizer::~CPDF_PageOrganizer() {} bool CPDF_PageOrganizer::PDFDocInit() { ASSERT(m_pDestPDFDoc); ASSERT(m_pSrcPDFDoc); CPDF_Dictionary* pNewRoot = m_pDestPDFDoc->GetRoot(); if (!pNewRoot) return false; CPDF_Dictionary* pDocInfoDict = m_pDestPDFDoc->GetInfo(); if (!pDocInfoDict) return false; pDocInfoDict->SetNewFor("Producer", "PDFium", false); ByteString cbRootType = pNewRoot->GetStringFor("Type", ""); if (cbRootType.IsEmpty()) pNewRoot->SetNewFor("Type", "Catalog"); CPDF_Object* pElement = pNewRoot->GetObjectFor("Pages"); CPDF_Dictionary* pNewPages = pElement ? ToDictionary(pElement->GetDirect()) : nullptr; if (!pNewPages) { pNewPages = m_pDestPDFDoc->NewIndirect(); pNewRoot->SetNewFor("Pages", m_pDestPDFDoc.Get(), pNewPages->GetObjNum()); } ByteString cbPageType = pNewPages->GetStringFor("Type", ""); if (cbPageType.IsEmpty()) pNewPages->SetNewFor("Type", "Pages"); if (!pNewPages->GetArrayFor("Kids")) { pNewPages->SetNewFor("Count", 0); pNewPages->SetNewFor( "Kids", m_pDestPDFDoc.Get(), m_pDestPDFDoc->NewIndirect()->GetObjNum()); } return true; } bool CPDF_PageOrganizer::ExportPage(const std::vector& pageNums, int nIndex) { int curpage = nIndex; auto pObjNumberMap = pdfium::MakeUnique(); for (size_t i = 0; i < pageNums.size(); ++i) { CPDF_Dictionary* pCurPageDict = m_pDestPDFDoc->CreateNewPage(curpage); CPDF_Dictionary* pSrcPageDict = m_pSrcPDFDoc->GetPage(pageNums[i] - 1); if (!pSrcPageDict || !pCurPageDict) return false; // Clone the page dictionary for (const auto& it : *pSrcPageDict) { const ByteString& cbSrcKeyStr = it.first; if (cbSrcKeyStr == "Type" || cbSrcKeyStr == "Parent") continue; CPDF_Object* pObj = it.second.get(); pCurPageDict->SetFor(cbSrcKeyStr, pObj->Clone()); } // inheritable item // Even though some entries are required by the PDF spec, there exist // PDFs that omit them. Set some defaults in this case. // 1 MediaBox - required if (!CopyInheritable(pCurPageDict, pSrcPageDict, "MediaBox")) { // Search for "CropBox" in the source page dictionary. // If it does not exist, use the default letter size. CPDF_Object* pInheritable = PageDictGetInheritableTag(pSrcPageDict, "CropBox"); if (pInheritable) { pCurPageDict->SetFor("MediaBox", pInheritable->Clone()); } else { // Make the default size letter size (8.5"x11") CPDF_Array* pArray = pCurPageDict->SetNewFor("MediaBox"); pArray->AddNew(0); pArray->AddNew(0); pArray->AddNew(612); pArray->AddNew(792); } } // 2 Resources - required if (!CopyInheritable(pCurPageDict, pSrcPageDict, "Resources")) { // Use a default empty resources if it does not exist. pCurPageDict->SetNewFor("Resources"); } // 3 CropBox - optional CopyInheritable(pCurPageDict, pSrcPageDict, "CropBox"); // 4 Rotate - optional CopyInheritable(pCurPageDict, pSrcPageDict, "Rotate"); // Update the reference uint32_t dwOldPageObj = pSrcPageDict->GetObjNum(); uint32_t dwNewPageObj = pCurPageDict->GetObjNum(); (*pObjNumberMap)[dwOldPageObj] = dwNewPageObj; UpdateReference(pCurPageDict, pObjNumberMap.get()); ++curpage; } return true; } bool CPDF_PageOrganizer::UpdateReference(CPDF_Object* pObj, ObjectNumberMap* pObjNumberMap) { switch (pObj->GetType()) { case CPDF_Object::REFERENCE: { CPDF_Reference* pReference = pObj->AsReference(); uint32_t newobjnum = GetNewObjId(pObjNumberMap, pReference); if (newobjnum == 0) return false; pReference->SetRef(m_pDestPDFDoc.Get(), newobjnum); break; } case CPDF_Object::DICTIONARY: { CPDF_Dictionary* pDict = pObj->AsDictionary(); auto it = pDict->begin(); while (it != pDict->end()) { const ByteString& key = it->first; CPDF_Object* pNextObj = it->second.get(); ++it; if (key == "Parent" || key == "Prev" || key == "First") continue; if (!pNextObj) return false; if (!UpdateReference(pNextObj, pObjNumberMap)) pDict->RemoveFor(key); } break; } case CPDF_Object::ARRAY: { CPDF_Array* pArray = pObj->AsArray(); for (size_t i = 0; i < pArray->GetCount(); ++i) { CPDF_Object* pNextObj = pArray->GetObjectAt(i); if (!pNextObj) return false; if (!UpdateReference(pNextObj, pObjNumberMap)) return false; } break; } case CPDF_Object::STREAM: { CPDF_Stream* pStream = pObj->AsStream(); CPDF_Dictionary* pDict = pStream->GetDict(); if (!pDict) return false; if (!UpdateReference(pDict, pObjNumberMap)) return false; break; } default: break; } return true; } uint32_t CPDF_PageOrganizer::GetNewObjId(ObjectNumberMap* pObjNumberMap, CPDF_Reference* pRef) { if (!pRef) return 0; uint32_t dwObjnum = pRef->GetRefObjNum(); uint32_t dwNewObjNum = 0; const auto it = pObjNumberMap->find(dwObjnum); if (it != pObjNumberMap->end()) dwNewObjNum = it->second; if (dwNewObjNum) return dwNewObjNum; CPDF_Object* pDirect = pRef->GetDirect(); if (!pDirect) return 0; std::unique_ptr pClone = pDirect->Clone(); if (CPDF_Dictionary* pDictClone = pClone->AsDictionary()) { if (pDictClone->KeyExist("Type")) { ByteString strType = pDictClone->GetStringFor("Type"); if (!FXSYS_stricmp(strType.c_str(), "Pages")) return 4; if (!FXSYS_stricmp(strType.c_str(), "Page")) return 0; } } CPDF_Object* pUnownedClone = m_pDestPDFDoc->AddIndirectObject(std::move(pClone)); dwNewObjNum = pUnownedClone->GetObjNum(); (*pObjNumberMap)[dwObjnum] = dwNewObjNum; if (!UpdateReference(pUnownedClone, pObjNumberMap)) return 0; return dwNewObjNum; } FPDF_EXPORT FPDF_BOOL FPDF_CALLCONV FPDF_ImportPages(FPDF_DOCUMENT dest_doc, FPDF_DOCUMENT src_doc, FPDF_BYTESTRING pagerange, int index) { CPDF_Document* pDestDoc = CPDFDocumentFromFPDFDocument(dest_doc); if (!dest_doc) return false; CPDF_Document* pSrcDoc = CPDFDocumentFromFPDFDocument(src_doc); if (!pSrcDoc) return false; std::vector pageArray; int nCount = pSrcDoc->GetPageCount(); if (pagerange) { if (!ParserPageRangeString(pagerange, &pageArray, nCount)) return false; } else { for (int i = 1; i <= nCount; ++i) { pageArray.push_back(i); } } CPDF_PageOrganizer pageOrg(pDestDoc, pSrcDoc); return pageOrg.PDFDocInit() && pageOrg.ExportPage(pageArray, index); } FPDF_EXPORT FPDF_BOOL FPDF_CALLCONV FPDF_CopyViewerPreferences(FPDF_DOCUMENT dest_doc, FPDF_DOCUMENT src_doc) { CPDF_Document* pDstDoc = CPDFDocumentFromFPDFDocument(dest_doc); if (!pDstDoc) return false; CPDF_Document* pSrcDoc = CPDFDocumentFromFPDFDocument(src_doc); if (!pSrcDoc) return false; CPDF_Dictionary* pSrcDict = pSrcDoc->GetRoot(); pSrcDict = pSrcDict->GetDictFor("ViewerPreferences"); if (!pSrcDict) return false; CPDF_Dictionary* pDstDict = pDstDoc->GetRoot(); if (!pDstDict) return false; pDstDict->SetFor("ViewerPreferences", pSrcDict->CloneDirectObject()); return true; }