def get_graph(*, _limit=None, _print=False): graph = bonobo.Graph() trunk = graph.add_chain(bonobo.JsonReader('datasets/theaters.json'), *((bonobo.Limit(_limit),) if _limit else ())) if _print: graph.add_chain(bonobo.PrettyPrinter(), _input=trunk.output) graph.add_chain(bonobo.JsonWriter('theaters.json', fs='fs.output'), _input=trunk.output) graph.add_chain(bonobo.LdjsonWriter('theaters.ldjson', fs='fs.output'), _input=trunk.output) return graph
def get_graph(*, _limit=None, _print=False): graph = bonobo.Graph() trunk = graph.add_chain(bonobo.JsonReader("theaters.json", fs="fs.static"), *((bonobo.Limit(_limit), ) if _limit else ())) if _print: graph.add_chain(bonobo.PrettyPrinter(), _input=trunk.output) graph.add_chain(bonobo.JsonWriter("theaters.output.json", fs="fs.output"), _input=trunk.output) graph.add_chain(bonobo.LdjsonWriter("theaters.output.ldjson", fs="fs.output"), _input=trunk.output) return graph
import bonobo as bb url = 'https://data.toulouse-metropole.fr/explore/dataset/theatres-et-salles-de-spectacles/download?format=json&timezone=Europe/Berlin&use_labels_for_header=true' graph = bb.Graph(bb.JsonReader(path=url), print) if __name__ == '__main__': bb.run(graph)
import bonobo from bonobo import Bag from bonobo.commands.run import get_default_services def get_fields(**row): return Bag(**row['fields']) graph = bonobo.Graph( bonobo.JsonReader('datasets/theaters.json'), get_fields, bonobo.PrettyPrinter(), ) if __name__ == '__main__': bonobo.run(graph, services=get_default_services(__file__))