def _run(self):
red_out, out_url, fd_list = self.connect_output()
red_in = iter(ReduceReader(self))
params = self.params
if self.ext_reduce:
path = self.path('EXT_REDUCE')
external.prepare(self.reduce, self.ext_params, path)
self.reduce = FunctionType(external.ext_reduce.func_code,
globals=external.__dict__)
self.insert_globals([self.reduce])
Message("Starting reduce")
self.init(red_in, params)
self.reduce(red_in, red_out, params)
Message("Reduce done")
self.close_output(fd_list)
external.close_ext()
if self.save:
OutputURL(util.ddfs_save(self.blobs, self.jobname, self.master))
Message("Results pushed to DDFS")
else:
index, index_url = self.reduce_index
safe_update(index, {'%d %s' % (self.id, out_url): True})
OutputURL(index_url)
def _run(self):
entries = self.track_status(self, "%s entries reduced")
red_out, out_url, fd_list = self.connect_output()
params = self.params
if self.ext_reduce:
external.prepare(self.reduce, self.ext_params, self.path('ext.reduce'))
self.reduce = FunctionType(external.ext_reduce.func_code,
globals=external.__dict__)
self.insert_globals([self.reduce])
total_size = sum(size for fd, size, url in self.connected_inputs)
Status("Input is %s" % (util.format_size(total_size)))
self.init(entries, params)
if util.argcount(self.reduce) < 3:
for k, v in self.reduce(entries, *(params, )):
red_out.add(k, v)
else:
self.reduce(entries, red_out, params)
self.close_output(fd_list)
external.close_ext()
if self.save:
OutputURL(util.ddfs_save(self.blobs, self.jobname, self.master))
Status("Results pushed to DDFS")
else:
index, index_url = self.reduce_index
f = file(index, 'w')
print >> f, '%d %s' % (self.id, out_url)
sync(f)
f.close()
OutputURL(index_url)
def _run(self):
entries = self.track_status(self, "%s entries reduced")
red_out, out_url, fd_list = self.connect_output()
params = self.params
if self.ext_reduce:
external.prepare(self.reduce, self.ext_params, self.path('ext.reduce'))
self.reduce = FunctionType(external.ext_reduce.func_code,
globals=external.__dict__)
self.insert_globals([self.reduce])
total_size = sum(size for fd, size, url in self.connected_inputs)
Message("Input is %s" % (util.format_size(total_size)))
self.init(entries, params)
self.reduce(entries, red_out, params)
self.close_output(fd_list)
external.close_ext()
if self.save:
OutputURL(util.ddfs_save(self.blobs, self.jobname, self.master))
Message("Results pushed to DDFS")
else:
index, index_url = self.reduce_index
safe_update(index, ['%d %s' % (self.id, out_url)])
OutputURL(index_url)
def _run(self):
if len(self.inputs) != 1:
TaskFailed("Map can only handle one input. Got: %s" % ' '.join(self.inputs))
if self.ext_map:
external.prepare(self.map, self.ext_params, self.path('EXT_MAP'))
self.map = FunctionType(external.ext_map.func_code,
globals=external.__dict__)
self.insert_globals([self.map])
partitions = [MapOutput(self, i) for i in xrange(self.num_partitions)]
reader, sze, url = self.connect_input(self.inputs[0])
params = self.params
self.init(reader, params)
entries = (self.map(entry, params) for entry in reader)
for kvs in self.track_status(entries, "%s entries mapped"):
for k, v in kvs:
p = self.partition(k, self.num_partitions, params)
partitions[p].add(k, v)
external.close_ext()
urls = {}
for i, partition in enumerate(partitions):
partition.close()
urls['%d %s' % (i, partition.url)] = True
index, index_url = self.map_index
safe_update(index, urls)
if self.save and not self.reduce:
if self.ispartitioned:
TaskFailed("Storing partitioned outputs in DDFS is not yet supported")
else:
OutputURL(util.ddfs_save(self.blobs, self.jobname, self.master))
Message("Results pushed to DDFS")
else:
OutputURL(index_url)
def _run(self):
if len(self.inputs) != 1:
TaskFailed("Map takes 1 input, got: %s" % ' '.join(self.inputs))
if self.save and not self.reduce and self.ispartitioned:
TaskFailed("Storing partitioned outputs in DDFS is not yet supported")
if self.ext_map:
external.prepare(self.map, self.ext_params, self.path('ext.map'))
self.map = FunctionType(external.ext_map.func_code,
globals=external.__dict__)
self.insert_globals([self.map])
entries = self.track_status(self, "%s entries mapped")
params = self.params
outputs = [MapOutput(self, i)
for i in xrange(max(1, int(self.jobdict['partitions'])))]
self.init(entries, params)
for entry in entries:
for k, v in self.map(entry, params):
outputs[self.partition(k, len(outputs), params)].add(k, v)
external.close_ext()
index, index_url = self.map_index
f = file(index, 'w')
for i, output in enumerate(outputs):
print >> f, '%d %s' % (i, output.url)
output.close()
sync(f)
f.close()
if self.save and not self.reduce:
OutputURL(util.ddfs_save(self.blobs, self.jobname, self.master))
Status("Results pushed to DDFS")
else:
OutputURL(index_url)