def persistent_collections_to_dsk(collections,
key=None,
serializers=None,
cache=None,
*args,
**kwargs):
"""
wrapper arount dask.base.collections_to_dsk
*args and **kwargs are passed to collections_to_dsk
"""
dsk = collections_to_dsk(collections, *args, **kwargs)
if key is not None:
dsk, _ = cull(dsk, key)
if serializers is not None:
# load instead of compute
dsk_serialized = get_relevant_keys_from_on_disk_cache(dsk, serializers)
dsk.update(dsk_serialized)
if cache is not None:
# use cache instead of loadind
dsk_cached = get_relevant_keys_from_memory_cache(dsk, cache)
dsk.update(dsk_cached)
# filter again task after function have been replaced by load or values
if key is not None:
dsk, _ = cull(dsk, key)
return dsk
def optimize(dsk, keys, **kwargs):
flatkeys = list(flatten(keys)) if isinstance(keys, list) else [keys]
dsk, dependencies = cull(dsk, flatkeys)
dsk, dependencies = fuse(dsk,
keys,
dependencies=dependencies,
ave_width=_globals.get('fuse_ave_width', 1))
dsk, _ = cull(dsk, keys)
return dsk
def __getitem__(self, geometry):
if isinstance(geometry, BaseGeometry) or getattr(
geometry, "__geo_interface__", None) is not None:
image = GeoImage.__getitem__(self, geometry)
image._ipe_op = self._ipe_op
return image
else:
result = super(IpeImage, self).__getitem__(geometry)
dsk, _ = optimize.cull(result.dask, result.__dask_keys__())
image = super(IpeImage,
self.__class__).__new__(self.__class__, dsk,
result.name, result.chunks,
result.dtype, result.shape)
if all([isinstance(e, slice)
for e in geometry]) and len(geometry) == len(self.shape):
xmin, ymin, xmax, ymax = geometry[2].start, geometry[
1].start, geometry[2].stop, geometry[1].stop
xmin = 0 if xmin is None else xmin
ymin = 0 if ymin is None else ymin
xmax = self.shape[2] if xmax is None else xmax
ymax = self.shape[1] if ymax is None else ymax
g = ops.transform(self.__geo_transform__.fwd,
box(xmin, ymin, xmax, ymax))
image.__geo_interface__ = mapping(g)
image.__geo_transform__ = self.__geo_transform__ + (xmin, ymin)
else:
image.__geo_interface__ = self.__geo_interface__
image.__geo_transform__ = self.__geo_transform__
image._ipe_op = self._ipe_op
return image
def fuse_dag(dag, copy=False):
if copy:
dag = deepcopy(dag)
dag = dict(dag)
dsk, dep = cull(dag, dag.keys())
dsk, dep = fuse(dsk, ave_width=1, rename_keys=False)
return dsk
def test_fuse_getitem():
def load(*args):
pass
dsk = {'x': (load, 'store', 'part', ['a', 'b']),
'y': (getitem, 'x', 'a')}
dsk2 = fuse_getitem(dsk, load, 3)
dsk2 = cull(dsk2, 'y')
assert dsk2 == {'y': (load, 'store', 'part', 'a')}
def test_fuse_getitem():
def load(*args):
pass
dsk = {'x': (load, 'store', 'part', ['a', 'b']), 'y': (getitem, 'x', 'a')}
dsk2 = fuse_getitem(dsk, load, 3)
dsk2 = cull(dsk2, 'y')
assert dsk2 == {'y': (load, 'store', 'part', 'a')}
def _cull(dsk, keys=None):
if not keys:
return dsk
elif "token" not in dsk:
return dsk
if "token" not in keys:
keys.append("token")
dsk1, _ = optimize.cull(dsk, keys)
return dsk1
def test_fuse_selections():
def load(*args):
pass
dsk = {'x': (load, 'store', 'part', ['a', 'b']), 'y': (getitem, 'x', 'a')}
merge = lambda t1, t2: (load, t2[1], t2[2], t1[2])
dsk2 = fuse_selections(dsk, getitem, load, merge)
dsk2 = cull(dsk2, 'y')
assert dsk2 == {'y': (load, 'store', 'part', 'a')}
def test_inline_cull_dependencies():
d = {'a': 1,
'b': 'a',
'c': 'b',
'd': ['a', 'b', 'c'],
'e': (add, (len, 'd'), 'a')}
d2, dependencies = cull(d, ['d', 'e'])
inline(d2, {'b'}, dependencies=dependencies)
def test_fuse_selections():
def load(*args):
pass
dsk = {'x': (load, 'store', 'part', ['a', 'b']),
'y': (getitem, 'x', 'a')}
merge = lambda t1, t2: (load, t2[1], t2[2], t1[2])
dsk2 = fuse_selections(dsk, getitem, load, merge)
dsk2 = cull(dsk2, 'y')
assert dsk2 == {'y': (load, 'store', 'part', 'a')}
def test_cull():
# 'out' depends on 'x' and 'y', but not 'z'
d = {'x': 1, 'y': (inc, 'x'), 'z': (inc, 'x'), 'out': (add, 'y', 10)}
culled = cull(d, 'out')
assert culled == {'x': 1, 'y': (inc, 'x'), 'out': (add, 'y', 10)}
assert cull(d, 'out') == cull(d, ['out'])
assert cull(d, ['out', 'z']) == d
assert cull(d, [['out'], ['z']]) == cull(d, ['out', 'z'])
assert raises(KeyError, lambda: cull(d, 'badkey'))
def test_cull():
# 'out' depends on 'x' and 'y', but not 'z'
d = {'x': 1, 'y': (inc, 'x'), 'z': (inc, 'x'), 'out': (add, 'y', 10)}
culled = cull(d, 'out')
assert culled == {'x': 1, 'y': (inc, 'x'), 'out': (add, 'y', 10)}
assert cull(d, 'out') == cull(d, ['out'])
assert cull(d, ['out', 'z']) == d
assert cull(d, [['out'], ['z']]) == cull(d, ['out', 'z'])
assert raises(KeyError, lambda: cull(d, 'badkey'))
def test_inline_cull_dependencies():
d = {
'a': 1,
'b': 'a',
'c': 'b',
'd': ['a', 'b', 'c'],
'e': (add, (len, 'd'), 'a')
}
d2, dependencies = cull(d, ['d', 'e'])
inline(d2, {'b'}, dependencies=dependencies)
def get(self,
dsk,
keys,
optimize_graph=True,
docker='lensa/dask.mesos',
params={},
threaded=True,
**kwargs): # should be mode instead: threaded / mesos / hybrid
""" Compute dask graph
Parameters
----------
dsk: dict
keys: object, or nested lists of objects
optimize_graph: bool
docker: string, default docker image for computations on mesos
params: dict, mesos options per dask key
threaded: bool, offload task without mesos parameters to threads
Examples
--------
>>> from operator import add # doctest: +SKIP
>>> e = MesosExecutor('127.0.0.1:8787') # doctest: +SKIP
>>> e.get({'x': (add, 1, 2)}, 'x') # doctest: +SKIP
3
See Also
--------
Executor.compute: Compute asynchronous collections
"""
# Optimize Dask
dsk2, dependencies = cull(dsk, keys)
if optimize_graph:
dsk3, dependencies = fuse(dsk2, keys, dependencies)
else:
dsk3 = dsk2
def apply_async(execute_task, args):
key = args[0]
if threaded and key not in params:
logging.info('Task `{}` is calculating in threads'.format(key))
return self.threadpool.submit(execute_task, *args)
options = params.get(key, {})
options['id'] = key
if 'docker' not in options:
options['docker'] = docker
logging.info('Task `{}` is calculating on mesos'.format(key))
return self.submit(execute_task, args, **options)
# Run
queue = Queue(self.zk, str(uuid4()))
result = get_async(apply_async, 1e4, dsk3, keys, queue=queue, **kwargs)
return result
def test_cull():
# 'out' depends on 'x' and 'y', but not 'z'
d = {'x': 1, 'y': (inc, 'x'), 'z': (inc, 'x'), 'out': (add, 'y', 10)}
culled, dependencies = cull(d, 'out')
assert culled == {'x': 1, 'y': (inc, 'x'), 'out': (add, 'y', 10)}
assert dependencies == {'x': [], 'y': ['x'], 'out': ['y']}
assert cull(d, 'out') == cull(d, ['out'])
assert cull(d, ['out', 'z'])[0] == d
assert cull(d, [['out'], ['z']]) == cull(d, ['out', 'z'])
pytest.raises(KeyError, lambda: cull(d, 'badkey'))
def test_cull():
# 'out' depends on 'x' and 'y', but not 'z'
d = {'x': 1, 'y': (inc, 'x'), 'z': (inc, 'x'), 'out': (add, 'y', 10)}
culled, dependencies = cull(d, 'out')
assert culled == {'x': 1, 'y': (inc, 'x'), 'out': (add, 'y', 10)}
assert dependencies == {'x': [], 'y': ['x'], 'out': ['y']}
assert cull(d, 'out') == cull(d, ['out'])
assert cull(d, ['out', 'z'])[0] == d
assert cull(d, [['out'], ['z']]) == cull(d, ['out', 'z'])
pytest.raises(KeyError, lambda: cull(d, 'badkey'))
def __dask_optimize__(cls, dsk, keys):
dsk1, _ = optimize.cull(dsk, keys)
dsk2 = {}
coll = []
for key, val in dsk1.items():
if isinstance(key, tuple) and key[0].startswith('image'):
name, z, x, y = key
dfn, url, token, chunk = val
dsk2[key] = (operator.getitem, "load_urls", (z, x, y))
coll.append([url, token, (z, x, y)])
else:
dsk2[key] = val
dsk2['load_urls'] = (cls.__fetch__, coll)
return dsk2
def get(dsk, keys, optimizations=[], num_workers=None,
docker='lensa/dask.mesos',
zk=os.getenv('ZOOKEEPER_HOST', '127.0.0.1:2181'),
mesos=os.getenv('MESOS_MASTER', '127.0.0.1:5050'),
**kwargs):
"""Mesos get function appropriate for Bags
Parameters
----------
dsk: dict
dask graph
keys: object or list
Desired results from graph
optimizations: list of functions
optimizations to perform on graph before execution
num_workers: int
Number of worker processes (defaults to number of cores)
docker: string
Default docker image name to run the dask in
zk: string
Zookeeper host and port the distributed Queue should connect to
mesos: string
Mesos Master hostname and port the Satyr framework should connect to
"""
pool, kazoo = _globals['pool'], _globals['kazoo']
if pool is None:
pool = Pool(name='dask-pool', master=mesos, processes=num_workers)
pool.start()
cleanup_pool = True
else:
cleanup_pool = False
if kazoo is None:
kazoo = KazooClient(hosts=zk)
kazoo.start()
cleanup_kazoo = True
else:
cleanup_kazoo = False
# Optimize Dask
dsk2, dependencies = cull(dsk, keys)
dsk3, dependencies = fuse(dsk2, keys, dependencies)
dsk4 = pipe(dsk3, *optimizations)
def apply_async(execute_task, args):
key = args[0]
func = args[1][0]
params = func.params if isinstance(func, SatyrPack) else {}
params['id'] = key
if 'docker' not in params:
params['docker'] = docker
return pool.apply_async(execute_task, args, **params)
try:
# Run
queue = Queue(kazoo, str(uuid4()))
result = get_async(apply_async, 1e4, dsk3, keys, queue=queue, **kwargs)
finally:
if cleanup_kazoo:
kazoo.stop()
if cleanup_pool:
pool.stop()
return result
def is_valid(self, layer='_output'):
return 'data_0' in cull(self.graph, layer)[0]
def warp(self, dem=None, proj="EPSG:4326", **kwargs):
"""
Delayed warp across an entire AOI or Image
creates a new dask image by deferring calls to the warp_geometry on chunks
kwargs:
dem (ndarray): optional. A DEM for warping to specific elevation planes
proj (str): optional. An EPSG proj string to project the image data into ("EPSG:32612")
Returns:
image (dask): a warped image as deferred image array (a dask)
"""
try:
img_md = self.ipe.metadata["image"]
x_size = img_md["tileXSize"]
y_size = img_md["tileYSize"]
except (AttributeError, KeyError):
x_size = kwargs.get("chunk_size", 256)
y_size = kwargs.get("chunk_size", 256)
# Create an affine transform to convert between real-world and pixels
if self.proj is None:
from_proj = "EPSG:4326"
else:
from_proj = self.proj
try:
# NOTE: this only works on images that have IPE rpcs metadata
center = wkt.loads(
self.ipe.metadata["image"]["imageBoundsWGS84"]).centroid
g = box(*(center.buffer(self.ipe.metadata["rpcs"]["gsd"] /
2).bounds))
# print "Input GSD (deg):", self.ipe.metadata["rpcs"]["gsd"]
tfm = partial(pyproj.transform, pyproj.Proj(init="EPSG:4326"),
pyproj.Proj(init=proj))
gsd = kwargs.get("gsd", ops.transform(tfm, g).area**0.5)
current_bounds = wkt.loads(
self.ipe.metadata["image"]["imageBoundsWGS84"]).bounds
except (AttributeError, KeyError, TypeError):
tfm = partial(pyproj.transform, pyproj.Proj(init=self.proj),
pyproj.Proj(init=proj))
gsd = kwargs.get("gsd", (ops.transform(tfm, shape(self)).area /
(self.shape[1] * self.shape[2]))**0.5)
current_bounds = self.bounds
tfm = partial(pyproj.transform, pyproj.Proj(init=from_proj),
pyproj.Proj(init=proj))
itfm = partial(pyproj.transform, pyproj.Proj(init=proj),
pyproj.Proj(init=from_proj))
output_bounds = ops.transform(tfm, box(*current_bounds)).bounds
gtf = Affine.from_gdal(output_bounds[0], gsd, 0.0, output_bounds[3],
0.0, -1 * gsd)
ll = ~gtf * (output_bounds[:2])
ur = ~gtf * (output_bounds[2:])
x_chunks = int((ur[0] - ll[0]) / x_size) + 1
y_chunks = int((ll[1] - ur[1]) / y_size) + 1
num_bands = self.shape[0]
try:
dtype = IPE_TO_DTYPE[img_md["dataType"]]
except:
dtype = 'uint8'
daskmeta = {
"dask": {},
"chunks": (num_bands, y_size, x_size),
"dtype": dtype,
"name": "warp-{}".format(self.name),
"shape": (num_bands, y_chunks * y_size, x_chunks * x_size)
}
def px_to_geom(xmin, ymin):
xmax = int(xmin + x_size)
ymax = int(ymin + y_size)
bounds = list((gtf * (xmin, ymax)) + (gtf * (xmax, ymin)))
return box(*bounds)
full_bounds = box(*output_bounds)
dasks = []
if isinstance(dem, GeoImage):
if dem.proj != proj:
dem = dem.warp(proj=proj, dem=dem)
dasks.append(dem.dask)
for y in xrange(y_chunks):
for x in xrange(x_chunks):
xmin = x * x_size
ymin = y * y_size
geometry = px_to_geom(xmin, ymin)
daskmeta["dask"][(daskmeta["name"], 0, y,
x)] = (self._warp, geometry, gsd, dem, proj,
dtype, 5)
daskmeta["dask"], _ = optimize.cull(
sharedict.merge(daskmeta["dask"], *dasks),
list(daskmeta["dask"].keys()))
result = GeoDaskWrapper(daskmeta, self)
result.__geo_interface__ = mapping(full_bounds)
result.__geo_transform__ = AffineTransform(gtf, proj)
return GeoImage.__getitem__(result, box(*output_bounds))
def is_valid(self, layer=512):
return '_graph_input' in cull(self.graph, layer)[0]
def optimize(dsk, keys, **kwargs):
dsk, _ = cull(dsk, keys)
return dsk
