abelsonlive/input.csv

## input.csv

          
            1
            50 Diamond St. Brooklyn NY

            
              2
              442 George Road, New York, NY

## mrgeo.py
"""MR-based geocoder
"""
from mrjob.job import MRJob
class MRGeocode(MRJob):

    def mapper(self, _, line):
    	id, address = line.split(',')
    	lat, long = geocode(address)
    	line += ",{},{}".format(lat, long)
    	yield id, line

if __name__ == '__main__':
     MRGeocode.run()

## run.sh
# locally
python mrgeo.py input.csv > geocoded.csv
# on EMR
python mrgeo.py input.csv -r emr > geocoded.csv
	"""MR-based geocoder
	"""
	from mrjob.job import MRJob
	class MRGeocode(MRJob):

	def mapper(self, _, line):
	id, address = line.split(',')
	lat, long = geocode(address)
	line += ",{},{}".format(lat, long)
	yield id, line

	if __name__ == '__main__':
	MRGeocode.run()
	# locally
	python mrgeo.py input.csv > geocoded.csv
	# on EMR
	python mrgeo.py input.csv -r emr > geocoded.csv