Method makeStructTreeRoot

src/pdf/PDFTag.cpp:222–302

}

PDFIndirectReference PDFTagTree::makeStructTreeRoot(PDFDocumentImpl* document) {
  if (!root || can_discard(root.get())) {
    return PDFIndirectReference();
  }

  PDFIndirectReference ref = document->reserveRef();

  auto pageCount = static_cast<uint32_t>(document->pageCount());

  // Build the StructTreeRoot.
  PDFDictionary structTreeRoot("StructTreeRoot");
  structTreeRoot.insertRef("K", PrepareTagTreeToEmit(ref, root.get(), document));
  structTreeRoot.insertInt("ParentTreeNextKey", static_cast<int>(pageCount));

  // Build the parent tree, which consists of two things:
  // (1) For each page, a mapping from the marked content IDs on
  // each page to their corresponding tags
  // (2) For each annotation, an indirect reference to that
  // annotation's struct tree element.
  PDFDictionary parentTree("ParentTree");
  auto parentTreeNums = MakePDFArray();

  // First, one entry per page.
  DEBUG_ASSERT(static_cast<uint32_t>(marksPerPage.size()) <= pageCount);
  for (size_t j = 0; j < marksPerPage.size(); ++j) {
    auto pageMarks = marksPerPage[j];
    PDFArray markToTagArray;
    for (PDFTagNode* mark : pageMarks) {
      DEBUG_ASSERT(mark->ref);
      markToTagArray.appendRef(mark->ref);
    }
    parentTreeNums->appendInt(static_cast<int>(j));
    parentTreeNums->appendRef(document->emit(markToTagArray));
  }

  // Then, one entry per annotation.
  for (size_t j = 0; j < parentTreeAnnotationNodeIds.size(); ++j) {
    int nodeId = parentTreeAnnotationNodeIds[j];
    int structParentKey = FirstAnnotationStructParentKey + static_cast<int>(j);

    auto iter = nodeMap.find(nodeId);
    if (iter == nodeMap.end()) {
      continue;
    }
    auto tag = iter->second;
    parentTreeNums->appendInt(structParentKey);
    parentTreeNums->appendRef(tag->ref);
  }

  parentTree.insertObject("Nums", std::move(parentTreeNums));
  structTreeRoot.insertRef("ParentTree", document->emit(parentTree));

  // Build the IDTree, a mapping from every unique ID string to
  // a reference to its corresponding structure element node.
  if (!IDTreeEntries.empty()) {
    std::sort(IDTreeEntries.begin(), IDTreeEntries.end(),
              [](const IDTreeEntry& a, const IDTreeEntry& b) { return a.nodeId < b.nodeId; });