def groupby(data,
keys=None,
size=None,
min_size=None,
max_size=None,
contiguous=False):
"""
return list of (keys, values) pairs where
group by the set of keys
values IS LIST OF ALL data that has those keys
contiguous - MAINTAIN THE ORDER OF THE DATA, STARTING THE NEW GROUP WHEN THE SELECTOR CHANGES
"""
if size != None or min_size != None or max_size != None:
if size != None:
max_size = size
return groupby_min_max_size(data, min_size=min_size, max_size=max_size)
if isinstance(data, Cube):
return data.groupby(keys)
keys = listwrap(keys)
def get_keys(d):
output = Dict()
for k in keys:
output[k] = d[k]
return output
if contiguous:
try:
if not data:
return wrap([])
agg = DictList()
acc = DictList()
curr_key = value2key(keys, data[0])
for d in data:
key = value2key(keys, d)
if key != curr_key:
agg.append((get_keys(acc[0]), acc))
curr_key = key
acc = [d]
else:
acc.append(d)
agg.append((get_keys(acc[0]), acc))
return wrap(agg)
except Exception, e:
Log.error("Problem grouping contiguous values", e)
def groupby(data, keys=None, size=None, min_size=None, max_size=None, contiguous=False):
"""
return list of (keys, values) pairs where
group by the set of keys
values IS LIST OF ALL data that has those keys
contiguous - MAINTAIN THE ORDER OF THE DATA, STARTING THE NEW GROUP WHEN THE SELECTOR CHANGES
"""
if size != None or min_size != None or max_size != None:
if size != None:
max_size = size
return groupby_min_max_size(data, min_size=min_size, max_size=max_size)
if isinstance(data, Cube):
return data.groupby(keys)
if not isinstance(keys, (tuple, list)):
keys = (keys,)
def get_keys(d):
output = Dict()
for k in keys:
output[k] = d[k]
return output
if contiguous:
try:
if not data:
return wrap([])
agg = DictList()
acc = DictList()
curr_key = value2key(keys, data[0])
for d in data:
key = value2key(keys, d)
if key != curr_key:
agg.append((get_keys(acc[0]), acc))
curr_key = key
acc = [d]
else:
acc.append(d)
agg.append((get_keys(acc[0]), acc))
return wrap(agg)
except Exception, e:
Log.error("Problem grouping contiguous values", e)
key = value2key(keys, d)
if key != curr_key:
agg.append((get_keys(acc[0]), acc))
curr_key = key
acc = [d]
else:
acc.append(d)
agg.append((get_keys(acc[0]), acc))
return wrap(agg)
except Exception, e:
Log.error("Problem grouping contiguous values", e)
try:
agg = {}
for d in data:
key = value2key(keys, d)
pair = agg.get(key)
if pair is None:
pair = (get_keys(d), DictList())
agg[key] = pair
pair[1].append(d)
return agg.values()
except Exception, e:
Log.error("Problem grouping", e)
def groupby_size(data, size):
if hasattr(data, "next"):
iterator = data
elif hasattr(data, "__iter__"):
key = value2key(keys, d)
if key != curr_key:
agg.append((get_keys(acc[0]), acc))
curr_key = key
acc = [d]
else:
acc.append(d)
agg.append((get_keys(acc[0]), acc))
return wrap(agg)
except Exception, e:
Log.error("Problem grouping contiguous values", e)
try:
agg = {}
for d in data:
key = value2key(keys, d)
pair = agg.get(key)
if pair is None:
pair = (get_keys(d), DictList())
agg[key] = pair
pair[1].append(d)
return agg.values()
except Exception, e:
Log.error("Problem grouping", e)
def groupby_size(data, size):
if hasattr(data, "next"):
iterator = data
elif hasattr(data, "__iter__"):
