handelaar/an2pw

Akoma Ntoso to PublicWhip converter

an2pw

Work in progress to convert Akoma Ntoso 3.0 sources from data.oireachtas.ie into publicwhip xml

If you find this before it's finished: it is currently in a very alpha state indeed and is spitting limited output to the console. Once all the element types used by the Irish parliament are completed, it'll move to outputting to a file handle (or to stdout) very quickly.

First version is designed to work only for KildareStreet.com; more functions to tweak output for your use case can be added after that.

an2pw's Issues

TODO: <question>

• question elements
• set 'last item was a question' flag so that speeches after questions can be marked as answers

TODO: Set output namespace, revision, timezone with configurable methods

(Ireland as default)

TODO: backlinks to source URIs in output

• calculate how on earth the source site URIs are being calculated
• if it is predictable, add them to every output element
• if it is not predictable, close this issue

PublicWhip XML diffing and revision support

Can we use pre-existing UK parser code to do this? It's in ancient highly-complex python at the moment.

TODO: vote

• Vote elements (for, against, abstain)
• Format as currently output at kildarestreet.com in included HTML

TODO: <speech>

• Attributed speeches
  • First with oirpersonid from source as temp measure
  • Then change to kildarestreet member_id
• Speech items from non-members or witnesses (speaker="unknown")
• Replace not-really-HTML elements within speeches (eg <summary>)
• <table>s within speeches or answers
• tag speeches as answers if last item was a question
• Akosa Ntoso <summary> items rendered as <speech nospeaker="true">

TODO: Concatenate all cttees for a day into one publicwhip output file

• feed arbitrary list of >=1 files into parser
• parser loadfile should check if passed var is an array; if yes, they're committees (document this)
• if 'committees' flag set, demote all of each heading in all AN files by one level
• move cttee prelude matter under artificial subheading 'prelude' as with Dáil
• decide whether to merge AN files in memory before parsing or to walk through them all separately