Annotating objectify trees with an XML Schema

[scoder]

lxml.objectify is a great tool for working with XML in a pythonesque data structure. It can easily be combined with (parse-time) XML Schema validation to make sure the data structure that was extracted from the source document is the expected one that results in the expected in-memory tree.

However, validating the document is not always enough, as it does not assure that data types are translated correctly into Python data types by objectify. This is the problem that this bug addresses.

Possible ways to approach this:

1) Use the "generateDS" tool by Dave Kuhlman: http://www.rexx.com/~dkuhlman/generateDS.html

We could try to extract the parser (or reimplement it with lxml) and run an annotation instead of the class generation step.

2) We could also look a bit deeper into the internal handling of XML schema in libxml2, there could be something to start from.

[scoder (scoder)]

libxml2 does not implement the post schema validation infoset, so it's not easy to get to the information that is available in the schema validation. Plus, type information in a schema can be ambiguous, so it's not always possible to infer a single 'correct' type from a schema.