1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537
|
# Copyright 2022- ECMWF.
#
# This software is licensed under the terms of the Apache Licence Version 2.0
# which can be obtained at http://www.apache.org/licenses/LICENSE-2.0.
# In applying this licence, ECMWF does not waive the privileges and immunities
# granted to it by virtue of its status as an intergovernmental organisation
# nor does it submit to any jurisdiction.
# flake8: noqa: F405
# ruff: noqa: F403
from .common import *
class RaggedArray:
@staticmethod
def empty(ndim=1):
if ndim > 0:
data = []
for _ in range(ndim - 1):
data = [data]
else:
data = None
return RaggedArray(data, ndim)
def __init__(self, data, ndim=None):
self.data = data
if ndim is None:
self.ndim = 0
while True:
try:
data = data[0]
except (IndexError, TypeError):
break
else:
self.ndim += 1
else:
self.ndim = ndim
def __repr__(self):
return f"RaggedArray({self.data})"
def __getitem__(self, index):
try:
value = self.data[index[0]]
except (IndexError, TypeError):
try:
value = self.data[index]
except TypeError:
if index in ((), None):
value = self.data
else:
raise IndexError(f"{index}")
else:
for i in range(1, len(index)):
value = value[index[i]]
return value
def __setitem__(self, index, value):
data = self.data
try:
for i in range(0, len(index) - 1):
data = data[index[i]]
except (IndexError, TypeError):
try:
self.data[index] = value
except TypeError:
if index in ((), None):
self.data = value
else:
raise IndexError(f"{index}")
else:
data[index[-1]] = value
def __bool__(self):
if isinstance(self.data, list):
def recurse(data):
if not isinstance(data, list):
return True
elif len(data) == 0:
return False
elif len(data) > 1:
return True
else:
return recurse(data[0])
return recurse(self.data)
else:
return bool(self.data)
def insert(self, index, value):
data = self.data
try:
for i in range(0, len(index) - 1):
data = data[index[i]]
data.insert(index[-1], value)
except (IndexError, TypeError, AttributeError):
try:
data.insert(index, value)
except (IndexError, TypeError):
if index[i] == len(data):
data.append([])
data = data[index[i]]
data.insert(index[-1], value)
else:
raise IndexError(f"{index}")
except AttributeError:
if index in ((), None):
self.data = value
else:
raise IndexError(f"{index}")
class SingletonDict: # TODO: UserDict?
"""A dict which returns the same value, no matter the key.
This is a helper class to optimize storage of key counts in leaf nodes
where the counts remain the same for every replication. By using this class
instead of the build-in dict, we save space without having to special-case
access to the leaf nodes' counts in the rest of the code.
"""
def __init__(self):
self._has_value = False
def __getitem__(self, key):
if self._has_value:
return self._value
else:
raise KeyError(key)
def __setitem__(self, key, value):
# TODO: allow to set/update value only once
self._value = value
self._has_value = True
def get_datetime(
view: Union["View", abc.Mapping],
rank: Optional[Union[int, slice]] = None,
prefix: str = "",
year: Optional[int] = None,
month: Optional[int] = None,
) -> NDArray:
"""
Returns an array of type `datetime64` derived from datetime-related keys/values.
"""
date = get_date(view, rank, prefix, year, month)
time = get_time(view, rank, prefix)
if date is np.ma.masked or time is np.ma.masked:
datetime = np.ma.masked
else:
datetime = date + time
return datetime
def get_date(
view: Union["View", abc.Mapping],
rank: Optional[Union[int, slice]] = None,
prefix: str = "",
year: Optional[int] = None,
month: Optional[int] = None,
) -> NDArray:
"""
Returns a `datetime64` array derived from date-related keys/values.
"""
names = ("year", "month", "day")
values = (year, month, None)
dtypes = ("M8[Y]", "m8[M]", "m8[D]")
offsets = (1970, 1, 1)
parts = []
for name, value, dtype, offset in zip(names, values, dtypes, offsets):
if prefix:
name = name.capitalize()
key = Key(prefix + name, rank if isinstance(rank, int) else None)
value = view[key.string] if value is None else value
part = ensure_masked_value(value, dtype)
parts.append(part - offset)
for i, part in enumerate(parts):
if part is np.ma.masked:
date = np.ma.masked
break
if isinstance(rank, slice):
slice_ = decrement(rank)
parts[i] = part[slice_]
else:
date = sum(parts)
return date
def get_time(
view: Union["View", abc.Mapping],
rank: Optional[Union[int, slice]] = None,
prefix: str = "",
) -> NDArray:
"""
Returns a `timedelta64` array derived from time-related keys/values.
"""
names = ("hour", "minute", "second")
dtypes = ("m8[h]", "m8[m]", "m8[s]")
has_seconds = True
parts = []
for name, dtype in zip(names, dtypes):
if prefix:
name = name.capitalize()
key = Key(prefix + name, rank if isinstance(rank, int) else None)
try:
value = view[key.string]
except (NotFoundError, KeyError) as error:
if name in ("second", "Second"):
has_seconds = False
break
else:
raise error
if name == "second":
if value is np.ma.masked:
part = value
else:
array = ensure_masked_array(value)
if array.dtype == np.dtype("f8"):
dtype = "m8[ms]"
array.fill_value = missing_of(int)
array.data[array.mask] = array.fill_value
array = (array * 1000).astype(dtype)
array.fill_value = np.array("NaT", dtype)
elif array.dtype == np.dtype("i8"):
array = array.view(dtype)
else:
array = cast(MaskedArray, array.astype(dtype))
if array.dtype == np.dtype("m8[s]"):
max_value = np.array(60 - 1, array.dtype)
elif array.dtype == np.dtype("m8[ms]"):
max_value = np.array(60 * 1000 - 1, array.dtype)
else:
assert False
array[:] = np.ma.where(array > max_value, max_value, array)
if isinstance(value, abc.Iterable):
part = array
else:
part = array[0]
else:
part = ensure_masked_value(value, dtype)
parts.append(part)
if not prefix:
key = Key(
"secondsWithinAMinuteMicrosecond", rank if isinstance(rank, int) else None
)
try:
value = view[key.string]
except (NotFoundError, KeyError):
pass
else:
array = ensure_masked_array(value)
useconds = np.ma.empty(array.shape, dtype="m8[us]")
useconds.fill_value = np.array("NaT", dtype="m8[us]")
useconds[:] = array * 1000000
max_useconds = np.array(60 * 1000000 - 1, dtype="m8[us]") # [1]
useconds[:] = np.ma.where(useconds > max_useconds, max_useconds, useconds)
if isinstance(value, abc.Iterable):
part = useconds
else:
part = useconds[0]
if has_seconds:
parts.pop() # [2]
parts.append(part)
for i, part in enumerate(parts):
if part is np.ma.masked:
time = np.ma.masked
break
if isinstance(rank, slice):
slice_ = decrement(rank)
parts[i] = part[slice_]
else:
time = sum(parts)
return time
# [1] Note that 'seconds...Microsecond' is a floating-point number where the
# whole part is seconds and the decimal part is microseconds.
#
# [2] Because of [1], if the message contains both keys, the 'second' and the
# 'seconds...Microsecond', use the latter. (Yes, such messages do exist!)
def set_datetime(
view: Union["View", abc.MutableMapping],
value: Union[DateLike, np.ndarray],
rank: Optional[int] = None,
prefix: str = "",
) -> None:
set_date(view, value, rank, prefix)
set_time(view, value, rank, prefix)
def set_date(
view: Union["View", abc.MutableMapping],
value: Union[DateLike, np.ndarray],
rank: Optional[int] = None,
prefix: str = "",
) -> None:
names = ["year", "month", "day"]
keys = []
for i, name in enumerate(names):
if prefix:
name = prefix + name.capitalize()
keys.append(Key(name, rank).string)
year, month, day = keys
if isinstance(value, np.ndarray):
view[year] = value.astype("M8[Y]").astype(int) + 1970
view[month] = value.astype("M8[M]").astype(int) % 12 + 1
view[day] = (value.astype("M8[D]") - value.astype("M8[M]")).astype(int) + 1
else:
date = ensure_date(value)
view[year] = date.year
view[month] = date.month
view[day] = date.day
def set_time(
view: Union["View", abc.MutableMapping],
value: Union[TimeLike, np.ndarray],
rank: Optional[int] = None,
prefix: str = "",
) -> None:
keys = ["hour", "minute", "second"]
for i, name in enumerate(keys):
if prefix:
name = prefix + name.capitalize()
keys[i] = Key(name, rank).string
hour, minute, second = keys
if isinstance(value, np.ndarray):
delta = (value - value.astype("M8[D]")).astype("timedelta64[s]").view(int)
view[hour] = delta // 3600
view[minute] = (delta % 3600) // 60
try:
view[second] = delta % 60
except (NotFoundError, KeyError):
pass
else:
time = ensure_time(value)
view[hour] = time.hour
view[minute] = time.minute
try:
view[second] = time.second
except (NotFoundError, KeyError):
pass
def ensure_date(value: DateLike) -> dt.date:
"""Takes a date-like value and converts it to `dt.date`."""
date: dt.date
if isinstance(value, dt.date):
date = value
elif isinstance(value, dt.datetime):
date = value.date()
elif isinstance(value, np.datetime64):
date = value.astype("M8[D]").item()
elif isinstance(value, str):
try:
date = dt.datetime.fromisoformat(value).date()
except ValueError:
try:
date = dt.datetime.strptime(value, "%Y%m%d").date()
except ValueError:
raise ValueError(f"Cannot convert string '{value}' to `dt.date`")
else:
raise TypeError("Cannot convert value of type `%s` to `dt.time`" % type(value))
return date
def ensure_time(value: TimeLike) -> dt.time:
"""Takes a time-like value and converts it to `dt.time`."""
if isinstance(value, dt.time):
time = value
elif isinstance(value, dt.datetime):
time = value.time()
elif isinstance(value, dt.timedelta):
time = (dt.datetime.min + value).time()
elif isinstance(value, np.datetime64):
time = value.astype("M8[s]").item().time()
elif isinstance(value, np.timedelta64):
total_seconds = value.astype("m8[s]").astype(int)
hours = total_seconds // 3600
minutes = (total_seconds % 3600) // 60
seconds = total_seconds % 60
time = dt.time(hours, minutes, seconds)
elif isinstance(value, str):
for format in ["%H:%M:%S", "%H:%M", "%H%M%S", "%H%M"]:
try:
time = dt.datetime.strptime(value, format).time()
except ValueError:
continue
else:
break
else:
raise ValueError(f"Cannot convert string '{value}' to `dt.time`")
else:
raise TypeError("Cannot convert value of type `%s` to `dt.time`" % type(value))
return time
def ensure_array(
value: Any, dtype: Optional[DTypeLike] = None
) -> Union[NDArray, MaskedArray]:
"""Takes a value and converts it to `np.ndarray`."""
if isinstance(value, np.ndarray):
array = value
if dtype:
dtype = np.dtype(dtype)
if dtype.type in (np.datetime64, np.timedelta64):
if array.dtype.type == np.int64:
array = array.view(dtype)
else:
array = array.astype(dtype)
else:
array = array.astype(dtype, copy=False)
elif isinstance(value, abc.Iterable):
if isinstance(value, str):
array = np.array([value], dtype=dtype)
else:
array = np.array(value, dtype=dtype)
else:
array = np.array([value], dtype=dtype)
return array
def ensure_masked_value(value: Any, dtype: Optional[DTypeLike]) -> MaskedArray:
if value is np.ma.masked:
return value
else:
array = ensure_masked_array(value, dtype)
if isinstance(value, abc.Iterable):
return array
else:
assert array.size == 1
return array[0]
def ensure_masked_array(value: Any, dtype: Optional[DTypeLike] = None) -> MaskedArray:
"""Takes a value and converts it to `np.ma.MaskedArray`."""
array = ensure_array(value, dtype)
if isinstance(value, np.ma.core.MaskedConstant):
masked_array = cast(np.ma.MaskedArray, array.ravel()) # [1]
elif isinstance(array, np.ma.MaskedArray):
masked_array = array
else:
try:
missing = missing_of(array.dtype)
except ValueError:
masked_array = np.ma.array(array, mask=False, copy=False) # [2]
else:
masked_array = np.ma.masked_equal(array, missing, copy=False)
if dtype and np.dtype(dtype).type in (np.datetime64, np.timedelta64):
masked_array.fill_value = np.array("NaT", dtype)
return masked_array
# [1] We are converting MaskedConstant (i.e., np.ma.masked) into 1d array
# so that we can do a slice assignment. This reduces number of code paths
# later on.
#
# [2] Assume no masked values for non-native types which cannot accomodate
# missing values, such as int16, float32, etc..
def decrement(value: Union[int, range, slice]) -> Union[int, range, slice]:
"""Decrements a value by one."""
if isinstance(value, int):
value -= 1
elif isinstance(value, range):
value = range(value.start - 1, value.stop - 1, value.step)
elif isinstance(value, slice):
start = None if value.start is None else value.start - 1
stop = None if value.stop is None else value.stop - 1
value = slice(start, stop, value.step)
else:
raise TypeError("Cannot decrement value of type `%s`" % type(value))
return value
def flatten(items):
"""Flattens arbitrarily nested sequences.
Examples:
>>> list(flatten([1, (2, (3, 4), [[5]])]))
[1, 2, 3, 4, 5]
>>> list(flatten(1))
[1]
"""
try:
for outer in items:
try:
for inner in flatten(outer):
yield inner
except TypeError:
yield outer
except TypeError:
yield items
MISSING_OF = {
str: "",
int: CODES_MISSING_LONG,
float: CODES_MISSING_DOUBLE,
np.int32: CODES_MISSING_LONG,
np.int64: CODES_MISSING_LONG,
np.float64: CODES_MISSING_DOUBLE,
np.dtype("i4"): CODES_MISSING_LONG,
np.dtype("i8"): CODES_MISSING_LONG,
np.dtype("f8"): CODES_MISSING_DOUBLE,
}
def missing_of(obj: Any) -> Union[int, float, str]:
"""Returns corresponding missing value for the given object or type."""
try:
missing = MISSING_OF[obj]
except (KeyError, TypeError):
if isinstance(obj, np.ndarray):
try:
missing = missing_of(obj.dtype)
except KeyError:
missing = None
elif isinstance(obj, np.dtype):
if obj.type == np.str_:
missing = ""
elif obj.type in (np.datetime64, np.timedelta64):
missing = np.array(CODES_MISSING_LONG, dtype=obj)
else:
missing = None
elif hasattr(obj, "__iter__") and len(obj) > 0:
missing = missing_of(type(obj[0]))
elif not isinstance(obj, type):
missing = missing_of(type(obj))
else:
missing = None
if missing is None:
raise ValueError("Object %s has no corresponding missing value" % obj)
return missing
|