Purpose of the library

This is a proposed boost::xml API. It is modular to allow implementations only to provide subsets of the full set of specs, and let users only choose the features they need. (E.g. dom vs reader)

The entire library is parametrized for the (unicode) character type. Proper unicode handling is outside the scope of this library.

This implementation uses libxml2 (http://xmlsoft.org/).

Primary author

Stefan Seefeld (stefan at seefeld dot name)

Structure of the library

boost::dom

This provides (or aims to provide) a DOM-like API to create and manipulate an in-memory representation of XML documents. Documents can be created either by parsing xml files, or by building a DOM tree 'manually'. Instantiated documents can then be traversed, queried (XPATH), and written out to a file.

boost::reader

This provides a stream-based interface to XML, where a document is accessed one token at a time.

Structure of the source tree

boost/xml/dom boost/xml/dom.hpp : the DOM API

boost/xml/reader boost/xml/reader.hpp : the Reader API

doc/ : Reference Manual (generated using Synopsis)

tests/ : Sample applets and tests

TODO

general

• Think about ways to let users 'plug in' their own unicode type, together with conversion traits.

dom

• Set up test suite.
• Clarify and cleanup access to 'private' libxml2 references.
• Fix node iterators to skip internal nodes (such as xinclude markers).
• Let xpath::find return variant or some other tool to access multiple types. (Right now only node_set is supported.)
• Complete tree manipulation API.
• Add validation modules (dtd, relaxng, schema (?) )

reader

• Use boost::variant and static_visitor for typed callbacks ?
• Etc.