def test_uses_json_format(self):
KEY = ['a', 1]
VALUE = {'foo': 'bar'}
ENCODED = b'["a", 1]\t{"foo": "bar"}'
self.assertEqual((KEY, VALUE), JSONProtocol().read(ENCODED))
self.assertEqual(ENCODED, JSONProtocol().write(KEY, VALUE))
def test_uses_json_format(self):
KEY = ['a', 1]
VALUE = {'foo': {'bar': 3}, 'baz': None}
ENCODED = '["a", 1]\t{"foo": {"bar": 3}, "baz": null}'
self.assertEqual((KEY, VALUE), JSONProtocol().read(ENCODED))
self.assertEqual(ENCODED, JSONProtocol().write(KEY, VALUE))
def test_numerical_keys_become_strs(self):
# JSON should convert numbers to strings when they are dict keys
self.assertEqual(({
'1': 2
}, {
'3': 4
}),
JSONProtocol().read(JSONProtocol().write({1: 2},
{3: 4})))
def test_bad_keys_and_values(self):
# dictionaries have to have strings as keys
self.assertCantEncode(JSONProtocol(), {(1, 2): 3}, None)
# only unicodes (or bytes in utf-8) are allowed
self.assertCantEncode(JSONProtocol(), '0\xa2', '\xe9')
# sets don't exist in JSON
self.assertCantEncode(JSONProtocol(), set([1]), set())
# Point class has no representation in JSON
self.assertCantEncode(JSONProtocol(), Point(2, 3), Point(1, 4))
def test_encode(self):
linRegFactory = LinearRegressionFactory(11)
linReg = linRegFactory.get_instance()
encoded = linRegFactory.encode(linReg)
protocol = JSONProtocol()
print protocol.write(0, encoded)
def encode_node(node_id, links=None, score=1):
node = {}
if links:
node['links'] = sorted(links.items())
node['score'] = score
x = JSONProtocol()
return x.write(node_id, node) + '\n'
def parse_output(self, protocol=None):
""".. deprecated:: 0.4.2
Parse the output from the given sandboxed job's ``self.stdout``.
This was only useful for testing individual mappers/reducers
without using a runner; normally you'd just use
:py:meth:`runner.stream_output()
<mrjob.runner.MRJobRunner.stream_output()>`
:type protocol: protocol
:param protocol: A protocol instance to use. Defaults to
``JSONProtocol()``.
"""
if self.stdout == sys.stdout:
raise AssertionError('You must call sandbox() first;'
' parse_output() is for testing only.')
log.warning(
'parse_output() is deprecated and will be removed in v0.5.0')
if protocol is None:
protocol = JSONProtocol()
lines = StringIO(self.stdout.getvalue())
return [protocol.read(line) for line in lines]
def test_encode(self):
'''
Test whether algorithm can be json encoded (used as mrjob internal protocol)
'''
layerSizes = [3,2,1]
nnFactory = PredictionNNFactory(layerSizes)
nn = nnFactory.get_instance()
# encode
obj_encoded = nnFactory.encode(nn)
# call json protocol
protocol = JSONProtocol()
protocol.write("test_decode", obj_encoded)
def test_decode(self):
linRegFactory = LinearRegressionFactory(11)
linReg = linRegFactory.get_instance()
obj_encoded = linRegFactory.encode(linReg)
protocol = JSONProtocol()
json_encoded = protocol.write(0, obj_encoded)
obj_encoded = protocol.read(json_encoded)
linRegArr = linRegFactory.decode([obj_encoded[1]])
assert type(linRegArr) == list, "decoded not as a list"
assert type(linRegArr[0]
) == LinearRegression, "decoded not as LinearRegression"
def test_decode(self):
'''
Test whether algorithm can be json encoded (used as mrjob internal protocol)
'''
layerSizes = [3,2,1]
nnFactory = PredictionNNFactory(layerSizes)
nn = nnFactory.get_instance()
# encode
obj_encoded = nnFactory.encode(nn)
# call json protocol
protocol = JSONProtocol()
json_encoded = protocol.write("test_decode", obj_encoded)
obj_encoded = protocol.read(json_encoded)
nnArr = nnFactory.decode([obj_encoded[1]])
assert type(nnArr) == list, "decoded not as a list"
assert type(nnArr[0]) == MultilayerPerceptron, "decoded not as LinearRegression"
#s3_input_path = "s3://joeloren//iceval_out//input//datasets//"
tmp_dir_out = "s3://joeloren/interim_out/"
tmp_dir_in = "s3://joeloren/interim_in/"
tmp_dir_in_relative = "interim_in/"
tmp_dir_out_relative = "interim_out/"
from mrjob.protocol import JSONValueProtocol, JSONProtocol
jvp = JSONValueProtocol()
jp = JSONProtocol()
from boto.s3.connection import S3Connection
import sys
c = S3Connection('AKIAI4OZ3HY56BTOHA3A',
'6isbkZjBM8kt3PIk53EXVIf76VOPxOH8rNleGc6B')
bucket = c.get_bucket("joeloren")
datasets_bucket = c.get_bucket('joel_datasets')
def input_protocol(self):
if self.options.job_to_run != 'stats':
LOG.debug('Reading text input from cdx files')
return RawValueProtocol()
LOG.debug('Reading JSON input from count job')
return JSONProtocol()
["Ut", "pulvinar", "lectus", "quis", "feugiat", "adipiscing"],
["Nunc", "vulputate", "mauris", "congue", "diam", "ultrices", "aliquet"],
["Nulla", "pharetra", "laoreet", "est", "quis", "vestibulum"],
["Quisque", "feugiat", "pharetra", "sagittis"],
["Phasellus", "nulla", "massa", "sodales", "a", "suscipit", "blandit", "facilisis", "eu", "augue"],
["Cras", "mi", "massa", "ullamcorper", "nec", "tristique", "at", "convallis", "quis", "eros"],
["Mauris", "non", "fermentum", "lacus", "vitae", "tristique", "tellus"],
["In", "volutpat", "metus", "augue", "nec", "laoreet", "ante", "hendrerit", "vitae"],
["Vivamus", "id", "lacus", "nec", "orci", "tristique", "vulputate"]
]
logging.basicConfig(level=logging.INFO)
mr_job = MRWordCounter()
## JSONValueProtocol doesn't need a key
#mr_job.stdin = [JSONValueProtocol().write(None, line) for line in TEXT]
## JSONProtocol wants also a key
mr_job.stdin = [JSONProtocol().write(linenum, line) for linenum, line in enumerate(TEXT)]
result = {}
with mr_job.make_runner() as runner:
runner.run()
for line in runner.stream_output():
key, value = mr_job.parse_output_line(line)
#print "Line: ", key, " Count: ", value
result[key] = value
# Print the output in JSON
print json.dumps(result)
def test_round_trip(self):
for k, v in JSON_KEYS_AND_VALUES:
self.assertRoundTripOK(JSONProtocol(), k, v)
NUMBER_RE = re.compile(r"[-?\d']+")
input_file = 'sample_input.txt'
with open(input_file, 'r') as out_file:
data = [x.split() for x in out_file.read().splitlines()]
# print(data)
nodes = {}
for line in data:
nodes[int(line[0])] = [] #Will be written as null
for line in data:
#Check for dangling nodes
if line[1:] == []:
nodes[int(line[0])] = [] #Will be written as null
else:
nodes[int(line[0])].append(int(line[1:][0]))
# print('nodes',nodes)
# unique_nodes = sorted(set(nodes), key = lambda ele: nodes.count(ele))
# print(nodes)
# print(unique_nodes)
unique_node_count = len(nodes.keys())
initial_pagerank = 1 / unique_node_count
j = JSONProtocol()
with open("preprocessed_" + input_file, "wb+") as out_file:
j = JSONProtocol()
for _id, adj in nodes.items():
out_file.write(j.write(_id, (adj, initial_pagerank)))
out_file.write('\n'.encode('utf-8'))
def test_bad_data(self):
self.assertCantDecode(JSONProtocol(), '{@#$@#!^&*$%^')
def INTERNAL_PROTOCOL(self):
return JSONProtocol()
def test_round_trip_with_trailing_tab(self):
for k, v in JSON_KEYS_AND_VALUES:
self.assertRoundTripWithTrailingTabOK(JSONProtocol(), k, v)
def INPUT_PROTOCOL(self):
return JSONProtocol()
def test_tuples_become_lists(self):
# JSON should convert tuples into lists
self.assertEqual(([1, 2], [3, 4]),
JSONProtocol().read(JSONProtocol().write((1, 2),
(3, 4))))
