How we extracted the text content and metadata from various kinds of documents.