Immediate next steps

[rufuspollock]

In all cases for library we should focus on a pure parse method and forget the conversion (that can go into the service as needed)

• Shapefile - Shapefile -> Records (GeoJSON) #1 complex as you seem to definitely need multiple files on disk. Could do something with zips and temp directories but this gets very complex very quickly. Have something basically working already with pathes ...
• KML: KML -> Records (GeoJSON) #5 at the moment Fiona does not support this and in any case Fiona needs a path not a stream. Various other options listed in KML -> Records (GeoJSON) #5.

@nigelbabu: my immediate thoughts therefore would be to upgrade https://github.com/okfn/dataproxy to use this library (or alternatively to add dataproxy API compatability to https://github.com/okfn/data-converters-service)

• reinstate xlsx tests #62 reinstate xlsx tests (30m)
• CSV parser should pass through delimiter and quotechar options to messytables #64
• csv parse header_type should be on by default and renamed to guess_types #66
• Convert data if types have been guessed (csv parse) #58 convert data types if guessed (2-4h?)
• Shapefile -> Records (GeoJSON) #1 - shapefile -> records (already working with a simple path - hacking something for a zipfile should not be too hard)
  • real issue is whether we want something pure python in which case we need pyshp
• KML -> Records (GeoJSON) #5 - kml -> records

Other thoughts would be:

• a command line interface (at least excel => csv would be really useful and shapefile => geojson) but that definitely needs an issue and some thinking through.
• researching pdf => text (see recent discussion on school of data list etc) PDF -> ?? #9

[nigelbabu]

Considering that all of the issues mentioned in this bug have got their own issues, I'm closing this meta bug.