Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- import csv
- from dumbo import main
- def load_contratos_mapper(contratos_files):
- contratos = {}
- try:
- # Read table - medal|prize|country|year
- with open(contratos_files) as f:
- reader = csv.reader(f, delimiter=';', quotechar='"', doublequote=False)
- reader.next()
- for line in reader:
- contratos[(line[0], line[1], line[2], lin[3], line[4])] = line[5]
- except:
- pass
- return contratos
- class Join_contratos_reduce:
- def __init__(self):
- self.contratos = load_contratos('./Contratos_por_municipio.csv')
- def __call__(self, key, values):
- try:
- values.groupby([1])[5].sum()
- yield values
- except:
- pass
- def runner(job):
- inout_opts = [("inputformat", "text"), ("outputformat", "text")]
- o1 = job.additer(load_contratos_mapper, Join_contratos_reduce, opts=inout_opts)
- if __name__ == "__main__":
- main(runner)
Add Comment
Please, Sign In to add comment