def update(self):
with log_errors():
state = {
'all': valmap(len, self.plugin.all),
'nbytes': self.plugin.nbytes
}
for k in ['memory', 'erred', 'released', 'processing']:
state[k] = valmap(len, self.plugin.state[k])
if not state['all'] and not len(self.source.data['all']):
return
d = progress_quads(state)
update(self.source, d)
totals = {
k: sum(state[k].values())
for k in ['all', 'memory', 'erred', 'released']
}
totals['processing'] = totals['all'] - sum(
v for k, v in totals.items() if k != 'all')
self.root.title.text = (
"Progress -- total: %(all)s, "
"in-memory: %(memory)s, processing: %(processing)s, "
"erred: %(erred)s" % totals)
def _createIndiceConverter(
innerIndiceMap: Dict[Indice, int], coeffs: List[Sequence[float]], oi: int,
_ls: List[List[Sequence[float]]], _boole: bool = False
) -> List[Sequence[float]]:
if list(innerIndiceMap.keys())[0] is ():
return coeffs
else:
innerIndiceMapGroup: Dict[int, List[Tuple[Indice, int]]] = compose(
valmap(compose(
lambda v: sorted(v, key=lambda k: k[0]),
map(lambda l: (l[0][1:], l[1])))),
groupby(lambda kv: kv[0][0]))(innerIndiceMap.items())
outArr = list(innerIndiceMapGroup.keys())
inArr = list(valmap(lambda v: v[0][-1], innerIndiceMapGroup).values())
coeff = self.getIndiceTransformCoeffs(outArr, inArr, oi, _ls, _boole)
nextInnerIndiceMapGroup: Dict[int, Dict[Indice, int]] = valmap(dict, innerIndiceMapGroup)
coeffsList = [
_createIndiceConverter(
self.applyIndiceTransform(nextInnerIndiceMap, key, coeff),
[*coeffs, coeff], oi + 1, _ls, _boole
)
for key, nextInnerIndiceMap in nextInnerIndiceMapGroup.items()
]
if allSame(coeffsList):
return coeffsList[0]
else:
raise LinearError
def getIndexMap(
self, insIndex: List[InIndex], outIndex: OutIndex
) -> Dict[Indice, LabelInIndice]:
labeledInsIndex = [
valmap(lambda x, i=i: (i, x), inIndex)
for (i, inIndex) in enumerate(insIndex)
]
return valmap(self.getCorrespondIndex(labeledInsIndex), outIndex)
def createIndexConverter(
self, indexMap: IndexMap, inShape: Shape, outShape: Shape, boole: bool = False
) -> IndexConverter:
if indexMap == {}:
return NullIndexConverter()
if boole and set(indexMap.values()) == {(0,)}:
return UnitIndexConverter(len(self.outAst.shape), 1)
try:
indiceConverters = list(reduce(
lambda ls, i: ls + [self.getIndiceConverter(valmap(lambda v, i=i: v[i], indexMap), ls, boole)],
range(len(inShape)), [[(0, 0, 0, 1)] * len(outShape)]))[1:]
if not indiceConverters:
return ZeroIndexConverter()
genCoeffs = lambda func, func0=identity: [func0([func(c) for c in l]) for l in indiceConverters]
linearCoeffs = genCoeffs(nth(0))
bs = genCoeffs(nth(1), sum)
indexModCoeffs = genCoeffs(nth(2))
indexModValues = genCoeffs(nth(3))
if boole:
return LinearIndiceConverter(linearCoeffs[0], bs[0], inShape[0], indexModCoeffs[0], indexModValues[0])
else:
return LinearIndexConverter(linearCoeffs, bs, inShape, indexModCoeffs, indexModValues)
except LinearError:
if all([s != -1 for s in outShape]):
return FixIndexConverter(indexMap)
else:
raise TransformError("Fail to find a transform.")
def update(self):
with log_errors():
processing = valmap(len, self.scheduler.processing)
workers = list(self.scheduler.workers)
bokeh_addresses = []
for worker in workers:
addr = self.scheduler.get_worker_service_addr(worker, 'bokeh')
bokeh_addresses.append('%s:%d' %
addr if addr is not None else '')
y = list(range(len(workers)))
nprocessing = [processing[w] for w in workers]
processing_color = []
for w in workers:
if w in self.scheduler.idle:
processing_color.append('red')
elif w in self.scheduler.saturated:
processing_color.append('green')
else:
processing_color.append('blue')
nbytes = [self.scheduler.worker_bytes[w] for w in workers]
nbytes_color = []
max_limit = 0
for w, nb in zip(workers, nbytes):
try:
limit = self.scheduler.worker_info[w]['memory_limit']
except KeyError:
limit = 16e9
if limit > max_limit:
max_limit = limit
if nb > limit:
nbytes_color.append('red')
elif nb > limit / 2:
nbytes_color.append('orange')
else:
nbytes_color.append('blue')
now = time()
if any(nprocessing) or self.last + 1 < now:
self.last = now
result = {
'nprocessing': nprocessing,
'nprocessing-half': [np / 2 for np in nprocessing],
'nprocessing-color': processing_color,
'nbytes': nbytes,
'nbytes-half': [nb / 2 for nb in nbytes],
'nbytes-color': nbytes_color,
'bokeh_address': bokeh_addresses,
'worker': workers,
'y': y
}
self.nbytes_figure.title.text = 'Bytes stored: ' + format_bytes(
sum(nbytes))
update(self.source, result)
def applyIndiceTransform(
IndiceMapGroup: Dict[Indice, int], key: int, coeff: Sequence[float]
) -> Dict[Indice, int]:
a, b, c, d = coeff
return valmap(
lambda v: v - (np.floor(a * key) + b + np.floor(c * (key % d))),
IndiceMapGroup,
)
def grouped(self, group, key=lambda x: x.weight):
if group == 'siblings' and not self.parent:
return {self.weight: [self]}
elif group in {'siblings', 'children'}:
agg = self.siblings if group == 'siblings' else self.children
return cc.pipe(((key(x), x) for x in agg),
cc.groupby(lambda x: x[0]),
cc.valmap(lambda x: [y[1] for y in x]))
else:
return {}
def get_interactions():
dates = sorted(set(map(_g('date'), data['interactions'])))
d = t.pipe(data['interactions'],
tc.groupby(lambda i: i.student),
tc.valmap(lambda x: t.pipe(t.groupby(lambda i: i.date,x),
tc.valmap(lambda v: [v[0].time_in, v[0].time_out]))))
mat = [['student'] + dates]
for student, attendance in d.items():
record = [student]
for dt in dates:
if dt in attendance:
record.append(attendance[dt])
elif dt in data['students'][student].absences:
record.append(('',''))
else:
record.append((None,None))
mat.append(record)
return {'interactions': mat}
def indice2Arr(indice: Indice) -> Union[np.ndarray, float]:
whichIn = insDispatcher(indice)[0]
doFuncs: Dict[str, bool] = valmap(lambda v: bool(v(indice)[0]), funcsIndex)
whichFunc = list(valfilter(identity, doFuncs).keys())
if not whichFunc:
func = lambda x: x
else:
if whichFunc[0] not in f2:
raise ValueError("No function called {}.".format(whichFunc[0]))
func = f2[whichFunc[0]]
return func(
self.getInputElement(inArrs[whichIn], tuple(indexConverters2[whichIn](indice)))
)
def getIndexConverters(
self, indexMap: Dict[Indice, LabelInIndice]
) -> List[IndexConverter]:
return [
self.getIndexConverter(
i,
compose(
valmap(lambda v: v[1]),
valfilter(lambda v, i=i: v[0] == i),
)(indexMap),
)
for i in range(len(self.inAsts))
]
def embedding_groups(
node_list: List[T],
persona_embedding_list: List[np.ndarray]) -> Dict[T, List[np.ndarray]]:
"""
Utility function, which given aligned list of nodes and embedding lists from the model.predict function,
obtain a dictionary from base graph nodes to a list of embeddings. The order of the embeddings for the
base nodes is not ordered, and the order may differ on different calls.
:param node_list: list of base nodes, which is duplicated
:param persona_embedding_list: corresponding embeddings
:return: dictionary mapping base nodes to all their embeddings
"""
return pipe(
zip(node_list, persona_embedding_list),
groupby(0),
valmap(lambda x: list(map(getter(1), x))),
)
def kfold(df, n_splits, random_state=0):
kf = KFold(n_splits, random_state=random_state, shuffle=True)
pos_df = df[df['label'] == 1]
neg_df = df[df['label'] == 0]
dataset = TellusDataset(df, has_y=True)
splieted = pipe(
zip(kf.split(pos_df), kf.split(neg_df)),
map(lambda x: {
"train_pos": pos_df.index[x[0][0]],
"val_pos": pos_df.index[x[0][1]],
"train_neg": neg_df.index[x[1][0]],
"val_neg": neg_df.index[x[1][1]],
}),
map(valmap(lambda x: Subset(dataset, x))),
list
)
return splieted
selected_bin = 0
dist = data_input
seen_dists_dict = {}
cycle_len_dict = {}
while not cc.valfilter(lambda x: x >= 2, seen_dists_dict):
dist_tuple = tuple(dist)
if dist_tuple not in seen_dists_dict:
seen_dists_dict[dist_tuple] = 1
else:
seen_dists_dict[dist_tuple] += 1
if dist_tuple not in cycle_len_dict:
cycle_len_dict[dist_tuple] = 0
cycle_len_dict = cc.valmap(lambda x: x + 1, cycle_len_dict)
selected_bin = argmax(dist)
selected_bin_val = dist[selected_bin]
dist[selected_bin] = 0
while selected_bin_val > 0:
selected_bin = (selected_bin + 1) % len(data_input)
dist[selected_bin] += 1
selected_bin_val -= 1
cycle_len_dict = cc.valmap(lambda x: x - 1, cycle_len_dict)
cycles = cc.valfilter(lambda x: x >= 2, seen_dists_dict)
key = tuple(cycles.keys())[0]
print(key)
},
'Embarked': {
'funcs': [
delayed(lambda df: df['Embarked']),
delayed(lambda s: s.fillna(s.mode()[0])),
delayed(label_encode),
delayed(one_hot),
]
},
}
train_df = delayed(pd.read_csv)('/store/kaggle/titanic/train.csv')
preprocessed_train_df = pipe(
preprocess_params,
valmap(lambda x: compose(*reversed(x['funcs']))(train_df)),
delayed(pd.DataFrame),
)
train_dataset = delayed(TitanicDataset)(
x_df=preprocessed_train_df,
y_df=delayed(lambda x: x['Survived'])(train_df),
)
val_dataset = delayed(TitanicDataset)(
x_df=preprocessed_train_df,
y_df=None,
)
train_result = delayed(train)(
model_path='/store/kaggle/titanic/model.pt',
# In[29]:
clf.fit(docs, labels)
clf.predict(docs)
# In[30]:
def get_step_by_name(pipe, name):
return [trans for name_, trans in pipe.steps if name_.startswith(name)][0]
# In[31]:
cnt_vects_pipe = get_step_by_name(tfidf_pipe, "cnt_vects")
cnt_vects = [
get_step_by_name(pipe, "cnt_vect_")
for _name, pipe in cnt_vects_pipe.transformer_list
]
vocabulary_map = pipe(
enumerate(concat(cnt_vect.vocabulary_ for cnt_vect in cnt_vects)),
groupby(get(1)),
valmap(lambda vals: list(pluck(0, vals))),
)
vocabulary_map
# In[ ]:
def getInsDispatcher(self, indexMap: Dict[Indice, LabelInIndice]) -> IndexConverter:
return self.createIndexConverter(
valmap(lambda x: (x[0],), indexMap), (len(self.inAsts),), self.outAst.shape, True
)
block_formatter = apply_formatters_to_dict(BLOCK_FORMATTERS)
SYNCING_FORMATTERS = {
'startingBlock': to_integer_if_hex,
'currentBlock': to_integer_if_hex,
'highestBlock': to_integer_if_hex,
'knownStates': to_integer_if_hex,
'pulledStates': to_integer_if_hex,
}
syncing_formatter = apply_formatters_to_dict(SYNCING_FORMATTERS)
TRANSACTION_POOL_CONTENT_FORMATTERS = {
'pending': compose(
keymap(to_ascii_if_bytes),
valmap(transaction_formatter),
),
'queued': compose(
keymap(to_ascii_if_bytes),
valmap(transaction_formatter),
),
}
transaction_pool_content_formatter = apply_formatters_to_dict(
TRANSACTION_POOL_CONTENT_FORMATTERS)
TRANSACTION_POOL_INSPECT_FORMATTERS = {
'pending': keymap(to_ascii_if_bytes),
'queued': keymap(to_ascii_if_bytes),
}
remaining.extend(current.children)
tree_val_re = re.compile('([a-z]{1,}).*\(([0-9]{1,})\)')
input_file = cc.pipe(open(r'day07.in'), list)
data_input = cc.pipe(
sys.stdin.readlines(), cc.map(lambda x: x.replace('\n', '')),
cc.map(lambda x: x.split('->')),
cc.map(lambda x: (x[0], [] if len(x) == 1 else cc.pipe(
x[1], lambda x: x.split(','), cc.map(str.strip), list))), list)
tree_val_dict = cc.pipe(
data_input, cc.map(cc.first),
cc.map(lambda x: [tree_val_re.match(x).group(y) for y in (1, 2)]), dict,
cc.valmap(int))
tree_mapping_dict = cc.pipe(
data_input, cc.map(lambda x: (tree_val_re.match(x[0]).group(1), x[1])),
dict)
root = cc.pipe(
tree_mapping_dict.keys(),
cc.filter(lambda x: x not in cc.concat(tree_mapping_dict.values())),
cc.first)
tree = Tree(root, tree_mapping_dict, tree_val_dict)
unbalanced = tree.find_unbalanced()
unbalanced_self_weight = unbalanced.weight - sum(x.weight
for x in unbalanced.children)
def getFuncsMap(self, funcsIndex: Dict[str, IndexMap]) -> Dict[str, IndexConverter]:
return valmap(lambda v: self.createIndexConverter(v, (2,), self.outAst.shape, True), funcsIndex)
SYNCING_FORMATTERS = {
'startingBlock': to_integer_if_hex,
'currentBlock': to_integer_if_hex,
'highestBlock': to_integer_if_hex,
'knownStates': to_integer_if_hex,
'pulledStates': to_integer_if_hex,
}
syncing_formatter = apply_formatters_to_dict(SYNCING_FORMATTERS)
TRANSACTION_POOL_CONTENT_FORMATTERS = {
'pending': compose(
keymap(to_ascii_if_bytes),
valmap(transaction_formatter),
),
'queued': compose(
keymap(to_ascii_if_bytes),
valmap(transaction_formatter),
),
}
transaction_pool_content_formatter = apply_formatters_to_dict(
TRANSACTION_POOL_CONTENT_FORMATTERS
)
TRANSACTION_POOL_INSPECT_FORMATTERS = {
'pending': keymap(to_ascii_if_bytes),
def process(workbook: Any, content: str) -> Any:
"""Process Storage-Groups worksheet
:param workbook:
:param content:
:return:
"""
worksheet_name = 'Storage-Groups'
worksheet = workbook.get_sheet_by_name(worksheet_name)
headers = list(concat([
get_parser_header(PORT_TMPL),
get_parser_header(STORAGEGROUP_TMPL)[3:],
]))
RowTuple = namedtuple('RowTuple', headers) # pylint: disable=invalid-name
build_header(worksheet, headers)
cmd_storagegroup_out = run_parser_over(content, STORAGEGROUP_TMPL)
cmd_port_out = run_parser_over(content, PORT_TMPL)
common_columns = (0, 1)
server_names_grouped = compose(
valmap(
compose(list, set, map(last))),
groupby(
itemgetter(*common_columns))
)(cmd_port_out)
cmd_port_relevant = map(
juxt(
compose(first, first),
compose(second, first),
second)
)(server_names_grouped.items())
common_columns_getter = itemgetter(*common_columns)
cmd_merged_out = join(
common_columns_getter, cmd_port_relevant,
common_columns_getter, cmd_storagegroup_out)
cmd_merged_out = sorted(cmd_merged_out)
rows = list(map(
compose(
list,
concat,
juxt(
first,
compose(
drop(3),
second)))
)(cmd_merged_out))
portcmd = {(array, grp) for array, grp, *other in rows}
strgp = {(array, grp) for array, grp, *other in cmd_storagegroup_out}
no_server_groups = strgp - portcmd
storage_list = list(filter(
lambda storage_gr: any(
fnmatch(str((storage_gr[0], storage_gr[1])), str(ctrlServer))
for ctrlServer in no_server_groups),
cmd_storagegroup_out))
storage_list = check_empty_arrays(
list(unique(storage_list + rows, key=itemgetter(0, 1))))
final_col, final_row = 0, 0
for row_n, row_tuple in enumerate(map(RowTuple._make, storage_list), 2):
for col_n, col_value in \
enumerate(row_tuple._asdict().values(), ord('A')):
cell = worksheet['{}{}'.format(chr(col_n), row_n)]
if isinstance(col_value, str):
cell.value = str.strip(col_value)
else:
cell.alignment = Alignment(wrapText=True)
cell.value = '\n'.join(col_value)
style_value_cell(cell)
set_cell_to_number(cell)
final_col = col_n
final_row = row_n
sheet_process_output(
worksheet,
'StorageGroupsTable',
'Storage-Groups',
final_col,
final_row)
return [[lun_map[0], lun_map[1], lun_map[4]] for lun_map in storage_list]
print(acc1)
# List comprehensions version (more Pythonic):
acc2 = [(name, balance) for (id, name, balance, gender) in accounts
if balance > 150]
print(acc2)
# II. SPLIT-APPLY-COMBINE WITH `GROUPBY` AND `REDUCEBY`:
# 1. Split the dataset into groups by some property
# 2. Reduce each of the groups with some synopsis function
# In Memory Split-Apply-Combine
# SELECT gender, SUM(balance) FROM accounts GROUP BY gender;
print(groupby(get(3), accounts))
# {'M': [(2, 'Bob', 200, 'M'), (3, 'Charlie', 150, 'M'), (4, 'Dennis', 50, 'M')], 'F': [(1, 'Alice', 100, 'F'), (5, 'Edith', 300, 'F')]}
print(pipe(accounts, groupby(get(3)), valmap(compose(sum, pluck(2)))))
# {'M': 400, 'F': 400} (pluck item )
# Streaming Split-Apply-Combine
# the groupby operation is not streaming and so this approach is limited
# to datasets that can fit comfortably into memory.
# to achieves streaming split-apply-combine use `reduceby()`
# The `reduceby` operation takes a key function,
# like `get(3)` or `lambda x: x[3]`, and a binary operator like
# `add` or `lesser = lambda acc, x: acc if acc < x else x`.
# It successively applies the key function to each item in succession,
# accumulating running totals for each key by combining each new value
# with the previous using the binary operator.
