Working with Table of Contents in Word Documents using C++

01-Working-with-Table-of-Contents-in-Word-Documents-using-CPP

Working with Table of Contents in Word Documents using C++

Add-Table-Of-Contents.cpp

// Source and output directory paths.
System::String sourceDataDir = u"SourceDirectory\\";
System::String outputDataDir = u"OutputDirectory\\";

// Load the Word file
System::SharedPtr<Document> doc = System::MakeObject<Document>(sourceDataDir + u"Sample 5.docx");

// Create an instance of the DocumentBuilder class
System::SharedPtr<DocumentBuilder> builder = System::MakeObject<DocumentBuilder>(doc);

// Insert a table of contents at the beginning of the document.
builder->InsertTableOfContents(u"\\o \"1-3\" \\h \\z \\u");

// The newly inserted table of contents will be initially empty.
// It needs to be populated by updating the fields in the document.
doc->UpdateFields();

// Output file path
System::String outputPath = outputDataDir + u"AddTOC.docx";

// Save the Word file
doc->Save(outputPath);

Extract-Table-Of-Contents.cpp

// Source direvctory
System::String inputDataDir = u"SourceDirectory\\";

// Load the Word file
System::SharedPtr<Document> doc = System::MakeObject<Document>(inputDataDir + u"SampleTOC.docx");

// Loop through the fields
for (System::SharedPtr<Field> field : System::IterateOver(doc->get_Range()->get_Fields()))
{
	// Get FieldHyperlink fields
	if (field->get_Type() == FieldType::FieldHyperlink)
	{
		System::SharedPtr<FieldHyperlink> hyperlink = System::DynamicCast<FieldHyperlink>(field);

		// Check if field belongs to TOC
		if (hyperlink->get_SubAddress() != nullptr && hyperlink->get_SubAddress().StartsWith(u"_Toc"))
		{
			System::SharedPtr<Paragraph> tocItem = System::DynamicCast<Paragraph>(field->get_FieldStart()->GetAncestor(NodeType::Paragraph));
			std::cout << System::StaticCast<Node>(tocItem)->ToString(SaveFormat::Text).Trim().ToUtf8String() << std::endl;
			std::cout << "------------------" << std::endl;
			if (tocItem != nullptr)
			{
				System::SharedPtr<Bookmark> bm = doc->get_Range()->get_Bookmarks()->idx_get(hyperlink->get_SubAddress());

				// Get the location this TOC Item is pointing to
				System::SharedPtr<Paragraph> pointer = System::DynamicCast<Paragraph>(bm->get_BookmarkStart()->GetAncestor(NodeType::Paragraph));
				std::cout << System::StaticCast<Node>(pointer)->ToString(SaveFormat::Text).ToUtf8String() << std::endl;
			}
		}
	}
}

Remove-Table-Of-Contents.cpp

void RemoveTableOfContents(const System::SharedPtr<Document>& doc, int32_t index)
{
	// Store the FieldStart nodes of TOC fields in the document for quick access.
	std::vector<System::SharedPtr<FieldStart>> fieldStarts;
	// This is a list to store the nodes found inside the specified TOC. They will be removed
	// at the end of this method.
	std::vector<System::SharedPtr<Node>> nodeList;

	for (System::SharedPtr<FieldStart> start : System::IterateOver<System::SharedPtr<FieldStart>>(doc->GetChildNodes(NodeType::FieldStart, true)))
	{
		if (start->get_FieldType() == FieldType::FieldTOC)
		{
			// Add all FieldStarts which are of type FieldTOC.
			fieldStarts.push_back(start);
		}
	}

	// Ensure that the TOC specified by the passed index exists.
	if (index > fieldStarts.size() - 1)
	{
		throw System::ArgumentOutOfRangeException(u"TOC index is out of range");
	}

	bool isRemoving = true;

	// Get the FieldStart of the specified TOC.
	System::SharedPtr<Node> currentNode = fieldStarts[index];

	while (isRemoving)
	{
		// It is safer to store these nodes and delete them all at once later.
		nodeList.push_back(currentNode);
		currentNode = currentNode->NextPreOrder(doc);

		// Once we encounter a FieldEnd node of type FieldTOC then we know we are at the end
		// of the current TOC and we can stop here.
		if (currentNode->get_NodeType() == NodeType::FieldEnd)
		{
			System::SharedPtr<FieldEnd> fieldEnd = System::DynamicCast<FieldEnd>(currentNode);
			if (fieldEnd->get_FieldType() == FieldType::FieldTOC)
			{
				isRemoving = false;
			}
		}
	}

	// Remove all nodes found in the specified TOC.
	for (System::SharedPtr<Node> node : nodeList)
	{
		node->Remove();
	}
}

int main()
{
	// Source and output directory paths.
	System::String sourceDataDir = u"SourceDirectory\\";
	System::String outputDataDir = u"OutputDirectory\\";

	// Open a Word document
	System::SharedPtr<Document> doc = System::MakeObject<Document>(sourceDataDir + u"SampleTOC.docx");

	// Remove the first table of contents from the document.
	RemoveTableOfContents(doc, 0);

	// Output file path
	System::String outputPath = outputDataDir + u"RemoveTOC.docx";

	// Save the Word file
	doc->Save(outputPath);
}