Much of the potential source text for hyperbooks comes from already existing texts, either on paper or in electronic format. Either way, adding this text to a hypertext system would require extensive transcribing unless a method of importing it is available.
Most systems can import pure ASCII text; some can also read word processor formatting codes or special "markup codes." These are predefined text strings used to indicate a particular structural or typographic element. For instance, suppose we had written a long essay in WordPerfect, and wished to move it to a hypertext platform, dividing it into multiple nodes at the same time. By including a code (eg. //title//) before each line we wanted to use as a node title, the system could chunk the document for us. Further codes (or "tags") might be used to mark anchors for automatic link generation. First, however, our hypertext system would have to understand two structures: the WordPerfect file format, and our markup language.
Attempts have been made to develop standards for markup languages and file interchange formats. Examples include SGML, DIF, DCA, and ODA. For a comparison of these, see Beaujardière.
Markup languages may encode information pertaining to:
The counterpart to importing is, of course, exporting--the ability to move text from the hyperbook to an external file format. Here, too, we wish to retain as much formatting and structural information as possible. Hypertext systems are usually weaker at exporting than importing.
This entire discussion assumes that there is a clear and defined "inside" and "outside" relative to the hyperbook. Indexing systems, such as SmarText, create a hypertext structure from existing documents, leaving them in their native formats. The grandiose Xanadu aims to turn all text into a world hypertext library. It is fundamental to this system that it does so without requiring multiple copies of each document.