= TODO =

== Code improvement ==

* Refactor plain text parser to a class. There is too much duplication
  of that.
* Use shared_ptr for librevenge::RVNGInputStream.

== Existing formats improvement ==

== All formats ==

* Check handling of images.
* Check if the format supports headings, text language, embedded fonts or
  any other feature that librevenge can handle now. If yes, implement
  support for it.

=== FB2 ===

* Improve parsing and handling of footnotes. Drop footnote identifier
  explicitly given as <title> or <subtitle>.
* Handle sub- and superscript.

=== LRF ===

* Handle metadata.
* Parse Canvas.
* Try to reconstruct tables from canvas blocks.
* Try to reconstruct footnotes.
* Handle embedded fonts.
* Handle numbered lists.
* Handle headers and footers.
* Do not create separate page spans if page properties have not changed.(?)

=== PLK ===

* Handle font sizes.

=== PML ===

* Handle sub- and superscript.
* Handle large font.

=== TD ===

* Handle large font.
* Handle images.

== New formats ==

=== Documented / reverse-engineered ===

* TEI Lite (?)

=== Undocumented ===

* ISilo 1-3
* TomeRaider 2 and 3

=== Unknown ===

* HandStory eBook
* Newton eBook
* TeBR
