def _reduce_with_count(pairwise, iterator, accumulator=None):
"""Return both the result of the reduction and the number of elements.
Parameters
----------
pairwise : function (a -> b -> a)
The function with which to reduce the `iterator` sequence.
iterator : iterable
The sequence being reduced.
accumulator : type "a", optional
An initial value with which to perform the reduction.
Returns
-------
result : type "a"
The result of the reduce operation.
count : int
The number of elements that were accumulated.
Examples
--------
>>> x = [5, 6, 7]
>>> _reduce_with_count(np.add, x)
(18, 3)
"""
def new_pairwise(a, b):
(elem1, c1), (elem2, c2) = a, b
return pairwise(elem1, elem2), c2
new_iter = zip(iterator, it.count(1))
new_acc = (0, accumulator)
return tz.reduce(new_pairwise, new_iter, new_acc)
def _reduce_with_count(pairwise, iterator, accumulator=None):
"""Return both the result of the reduction and the number of elements.
Parameters
----------
pairwise : function (a -> b -> a)
The function with which to reduce the `iterator` sequence.
iterator : iterable
The sequence being reduced.
accumulator : type "a", optional
An initial value with which to perform the reduction.
Returns
-------
result : type "a"
The result of the reduce operation.
count : int
The number of elements that were accumulated.
Examples
--------
>>> x = [5, 6, 7]
>>> _reduce_with_count(np.add, x)
(18, 3)
"""
def new_pairwise(a, b):
(elem1, c1), (elem2, c2) = a, b
return pairwise(elem1, elem2), c2
new_iter = it.izip(iterator, it.count(1))
new_acc = (0, accumulator)
return tlz.reduce(new_pairwise, new_iter, new_acc)
def put_in(keys, coll, val):
"""Inverse of get_in, but does type promotion in the case of lists"""
if keys:
holder = reduce(operator.getitem, keys[:-1], coll)
if isinstance(holder, tuple):
holder = list(holder)
coll = put_in(keys[:-1], coll, holder)
holder[keys[-1]] = val
else:
coll = val
return coll
def getFuncsIndex(self):
for elt in ast.walk(self.node):
if isinstance(elt, ast.Name):
indice = self.getIndice(elt)
if hasattr(elt, "func"):
self.funcs = merge_with(lambda x: reduce(add, x, []),
self.funcs, {elt.func: [indice]})
for key in self.funcs.keys():
self.funcsIndex[key] = itemmap(lambda kv, key=key:
(kv[0],
(int(kv[0] in self.funcs[key]), )),
self.index)
def basic_validate(net, criterion, val_batches):
print('running validation ... ', end='')
net.eval()
start = time()
with torch.no_grad():
validate_fn = val_step(compute_loss(net, criterion))
n_data, tot_loss = reduce(lambda a, b: (a[0] + b[0], a[1] + b[1]),
starmap(validate_fn, val_batches), (0, 0))
val_loss = tot_loss / n_data
print('validation finished in {}'.format(
timedelta(seconds=int(time() - start))))
print('validation loss: {:.4f} ... '.format(val_loss))
return {'loss': val_loss}
def write(values, filepath):
"""Writes a sequence of values to a filepath
Args:
filepath (str): Full path of raster file to write
values (sequence): Values to write
Returns:
tuple: (filepath, bytes written)
"""
with open(filepath, 'w+') as handle:
return (filepath, reduce(add, map(handle.write, values)))
def verify_merkle_proof(root: Hash32, item: Hashable, item_index: int,
proof: MerkleProof) -> bool:
"""Verify a Merkle proof against a root hash."""
leaf = keccak(item)
branch_indices = get_branch_indices(item_index, len(proof))
node_orderers = [
identity if branch_index % 2 == 0 else reversed
for branch_index in branch_indices
]
proof_root = reduce(
lambda n1, n2_and_order: _calc_parent_hash(*n2_and_order[1]
([n1, n2_and_order[0]])),
zip(proof, node_orderers),
leaf,
)
return proof_root == root
def basic_validate(net, criterion, val_batches):
print('running validation ... ', end='')
net.eval()
start = time()
with torch.no_grad():
validate_fn = val_step(compute_loss(net, criterion))
n_data, tot_loss = reduce(
lambda a, b: (a[0]+b[0], a[1]+b[1]),
starmap(validate_fn, val_batches),
(0, 0)
)
val_loss = tot_loss / n_data
print(
'validation finished in {} '.format(
timedelta(seconds=int(time()-start)))
)
print('validation loss: {:.4f} ... '.format(val_loss))
return {'loss': val_loss}
def __getitem__(self, key: str) -> iter:
if key not in self.map:
raise ValueError(
dedent("""\
Key '{}' is invalid!
Valid keys: {}
""".format(
key,
reduce(lambda k1, k2: '{}, {}'.format(k1, k2),
map(lambda k: "'{}'".format(k), self.map)))))
ref = self.map[key]
if 'api' not in ref:
ref['api'] = 'dbpy' # default api
api = ref['api']
# load reader
if key not in self.cache:
print("Loading '{}' reader...".format(key))
if api not in ('dbpy', 'stpy'):
raise ValueError("Invalid api type '{}'!".format(api))
if 'id' not in ref:
ref['id'] = key # default id
id = ref['id']
if api == 'dbpy':
self.cache[key] = fromiter(
read_syncdatalist_float(id, self.hi_tag,
tuple(map(int,
self.low_tags))),
'float')
if api == 'stpy':
self.cache[key] = StorageWrapper(*map(int, self.runs),
beamline=self.beamline,
id=id)
if 'deco' not in ref:
ref['deco'] = identity # default deco
print('Loaded!')
data = self.cache[key]
deco = ref['deco'] if hasattr(ref['deco'], '__call__') else eval(
ref['deco'])
if api == 'dbpy':
return map(deco, data)
if api == 'stpy':
return map(compose(deco, data.__getitem__), self.low_tags)
def symmetric(datemap):
"""Returns a sequence of dates that are common to all map values if
all datemap values are represented, else Exception.
Args:
datemap: {key: [datestrings,]}
Returns:
Sequence of date strings or Exception
Example:
>>> common({"reds": [ds3, ds1, ds2],
"blues": [ds2, ds3, ds1]})
[2, 3, 1]
>>>
>>> common({"reds": [ds3, ds1],
"blues": [ds2, ds3, ds1]})
Exception: reds:[3, 1] does not match blues:[2, 3, 1]
"""
def check(a, b):
"""Reducer for efficiently comparing two unordered sequences.
Executes in linear(On) time.
Args:
a: {k:[datestring1, datestring2...]}
b: {k:[datestring2, datestring1...]}
Returns:
b if a == b, else Exception with details
"""
if f.seqeq(second(a), second(b)):
return b
else:
msg = ('assymetric dates detected - {} != {}'
.format(first(a), first(b)))
msga = '{}{}'.format(first(a), second(a))
msgb = '{}{}'.format(first(b), second(b))
raise Exception('\n\n'.join([msg, msga, msgb]))
return second(reduce(check, datemap.items()))
def chips(x, y, acquired, ubids, url, resource='/chips'):
"""Returns chips from a Chipmunk url given x, y, date range and ubid sequence
Args:
x (int): projection coordinate x
y (int): projection coordinate y
acquired (str): ISO8601 daterange '2012-01-01/2014-01-03'
ubids (sequence): sequence of ubids
url (str): protocol://host:port/path
resource (str): /chips/resource/path (default: /chips)
Returns:
tuple: chips
Example:
>>> chipmunk.chips(url='http://host:port/path',
x=123456,
y=789456,
acquired='2012-01-01/2014-01-03',
ubids=['LE07_SRB1', 'LT05_SRB1'])
(LE07_SRB1_DATE1, LT05_SRB1_DATE2, LE07_SRB1_DATE2, ...)
"""
url = '{}{}'.format(url, resource)
params = [{'x': x, 'y': y, 'acquired': acquired, 'ubid': u } for u in ubids]
def request(url, params):
r = requests.get(url=url, params=params)
body = r.json()
if not r.ok:
logger.error("{} at {} for {}".format(body, url, params))
return None
else:
return body
responses = [request(url=url, params=p) for p in params]
return tuple(reduce(add, filter(lambda x: type(x) in [list, tuple], responses), []))
def multi_basic_validate(net, criterion, val_batches):
print('running validation ... ', end='')
net.eval()
start = time()
with torch.no_grad():
validate_fn = multi_val_step(compute_loss(net, criterion))
n, tot_loss, tot_p, tot_r, tot_f = reduce(
lambda a, b:
(a[0] + b[0], a[1] + b[1], a[2] + b[2], a[3] + b[3], a[4] + b[4]),
starmap(validate_fn, val_batches), (0, 0, 0, 0, 0))
val_loss = tot_loss / n
val_p = tot_p / n
val_r = tot_r / n
val_f = tot_f / n
print(
'validation finished in {} '.format(
timedelta(seconds=int(time() - start))))
print('validation precision: {:.4f} ... '.format(val_loss))
print('validation p: {:.4f} ... '.format(val_p))
print('validation r: {:.4f} ... '.format(val_r))
print('validation f: {:.4f} ... '.format(val_f))
return {'loss': val_loss}
def flip_keys(dods):
"""Accepts a dictionary of dictionaries and flips the outer and inner keys.
All inner dictionaries must have a consistent set of keys or key Exception
is raised.
Args:
dods: dict of dicts
Returns:
dict of dicts with inner and outer keys flipped
Example:
>>> dods = {"reds": {(0, 0): [110, 110, 234, 664],
(0, 1): [23, 887, 110, 111]},
"greens": {(0, 0): [120, 112, 224, 624],
(0, 1): [33, 387, 310, 511]},
"blues": {(0, 0): [128, 412, 244, 654],
(0, 1): [73, 987, 119, 191]},
>>> flip_keys(dods)
{(0, 0): {"reds": [110, 110, 234, 664],
"greens": [120, 112, 224, 624],
"blues": [128, 412, 244, 654], ... },
(0, 1), {"reds": [23, 887, 110, 111],
"greens": [33, 387, 310, 511],
"blues": [73, 987, 119, 191], ... }}
"""
def flip(innerkeys, outerkeys, inputs):
for ik in innerkeys:
yield({ik: {ok: inputs[ok][ik] for ok in outerkeys}})
outerkeys = set(dods.keys())
innerkeys = set(reduce(lambda accum, v: accum + v,
[list(dods[ok].keys()) for ok in outerkeys]))
return merge(flip(innerkeys, outerkeys, dods))
def flow(*args):
return lambda k, v, p: reduce(
lambda tup, f: f(tup[0], tup[1], tup[2]) + (tup[2], ), args,
(k, v, p))[0:2]
def logging_once(self, step):
"""Log information at one step"""
do = self.valid
# Print console info
self.logger.info(
'[step=%04d]'
'[dslt=%6.3f]'
'[btpt=%6.3f],'
'[cs1t=%6.3f]'
'[cs2t=%6.3f]'
'[ebpt=%6.3f],'
'[ss1t=%6.3f]'
'[ss2t=%6.3f]'
'[expt=%6.3f]', #
step,
do.mt.dslt.avg,
do.mt.btpt.avg,
do.mt.cs1t.avg,
do.mt.cs2t.avg,
do.mt.ebpt.avg,
do.mt.ss1t.avg,
do.mt.ss2t.avg,
do.mt.expt.avg)
self.logger.info(
'[step=%04d]'
'[step-1-attn-loss=%8.1fb]'
'[step-2-attn-loss=%8.1fb]'
'[step-3-attn-loss=%8.1fb]'
'[step-1-area=%5.3f/%5.3f]'
'[step-2-area=%5.3f/%5.3f]'
'[step-3-area=%5.3f/%5.3f]', #
step,
do.mt.s1al.avg,
do.mt.s2al.avg,
do.mt.s3al.avg,
do.mt.c1aa.avg,
do.mt.s1aa.avg,
do.mt.c2aa.avg,
do.mt.s2aa.avg,
do.mt.c3aa.avg,
do.mt.s3aa.avg)
avg_embd_bytes = np.mean([msg.size for msg in do.embedded_messages])
avg_extt_bytes = np.mean([msg.size for msg in do.extracted_messages])
total_pixels = tz.reduce(operator.mul, do.cover_images.shape[2:])
self.logger.info(
'[embd-payload=%6db/%6dP(%5.2f)]'
'[extt-payload=%6db/%8dP(%5.2f)],'
'[msg-diff=%10.3fb/%10db(%05.2f%%)]'
'[msg-EDis=%10.3fb/%10db(%05.2f%%)]', #
avg_embd_bytes * 8,
total_pixels,
avg_embd_bytes * 8 / total_pixels,
avg_extt_bytes * 8,
total_pixels,
avg_extt_bytes * 8 / total_pixels,
do.mt.msg_diff.avg,
avg_embd_bytes * 8,
do.mt.msg_diff.avg / (avg_embd_bytes * 8) * 100,
do.mt.msg_edis.avg,
avg_embd_bytes * 8,
do.mt.msg_edis.avg / (avg_embd_bytes * 8) * 100)
def mask(v):
return reduce(logical_or, (__mask(*reg, v=v) for reg in regs))
