json-to-schema.py

from data import ROOT

class SchemaBuilder(object):
    def __init__(self, prefix=''):
        self.last_key = ''
        self.prefix = prefix
        self.results = {}
    
    def parse(self, obj, my_key='Root'):
        if obj is None: return 'Any'
        if isinstance(obj, list):
            if not obj: return 'List[Any]'
            return 'List[%s]' % self.parse(obj[0], my_key)
        if isinstance(obj, dict):
            my_name = self.prefix + my_key[0].upper() + my_key[1:]
            parts = ['val %s: %s' % (k, self.parse(obj[k], k)) for k in obj]
            self.results[my_name] = 'case class %s(%s)' % (my_name, ',\n'.join(parts)
            )
            return my_name
        if isinstance(obj, (unicode, str)): return 'String'
        if isinstance(obj, float): return 'Double'
        if isinstance(obj, int): return 'Int'
        return 'N/A'

builder = SchemaBuilder('Reddit')
builder.parse(ROOT)
for v in builder.results.values():
    print v, '\n'

Feeling too lazy to make a schema for your random JSON document? But also too lazy to work with an untyped interface? Then this is for you! It automatically infers a passable set of case classes corresponding to a JSON document!