from __future__ import annotations
from packaging.version import Version
import numpy as np
from toolz import memoize
from datashader.glyphs.glyph import Glyph
from datashader.utils import isreal, ngjit
from numba import cuda
try:
import cudf
from ..transfer_functions._cuda_utils import cuda_args
except Exception:
cudf = None
cuda_args = None
try:
import spatialpandas
except Exception:
spatialpandas = None
def values(s):
if isinstance(s, cudf.Series):
if Version(cudf.__version__) >= Version("22.02"):
return s.to_cupy(na_value=np.nan)
else:
return s.to_gpu_array(fillna=np.nan)
else:
return s.values
class _GeometryLike(Glyph):
def __init__(self, geometry):
self.geometry = geometry
@property
def ndims(self):
return 1
@property
def inputs(self):
return (self.geometry,)
@property
def geom_dtypes(self):
if spatialpandas:
from spatialpandas.geometry import GeometryDtype
return (GeometryDtype,)
else:
return () # Empty tuple
def validate(self, in_dshape):
if not isinstance(in_dshape[str(self.geometry)], self.geom_dtypes):
raise ValueError(
'{col} must be an array with one of the following types: {typs}'.format(
col=self.geometry,
typs=', '.join(typ.__name__ for typ in self.geom_dtypes)
))
@property
def x_label(self):
return 'x'
@property
def y_label(self):
return 'y'
def required_columns(self):
return [self.geometry]
def compute_x_bounds(self, df):
bounds = df[self.geometry].array.total_bounds_x
return self.maybe_expand_bounds(bounds)
def compute_y_bounds(self, df):
bounds = df[self.geometry].array.total_bounds_y
return self.maybe_expand_bounds(bounds)
@memoize
def compute_bounds_dask(self, ddf):
total_bounds = ddf[self.geometry].total_bounds
x_extents = (total_bounds[0], total_bounds[2])
y_extents = (total_bounds[1], total_bounds[3])
return (self.maybe_expand_bounds(x_extents),
self.maybe_expand_bounds(y_extents))
class _PointLike(Glyph):
"""Shared methods between Point and Line"""
def __init__(self, x, y):
self.x = x
self.y = y
@property
def ndims(self):
return 1
@property
def inputs(self):
return (self.x, self.y)
def validate(self, in_dshape):
if not isreal(in_dshape.measure[str(self.x)]):
raise ValueError('x must be real')
elif not isreal(in_dshape.measure[str(self.y)]):
raise ValueError('y must be real')
@property
def x_label(self):
return self.x
@property
def y_label(self):
return self.y
def required_columns(self):
return [self.x, self.y]
def compute_x_bounds(self, df):
bounds = self._compute_bounds(df[self.x])
return self.maybe_expand_bounds(bounds)
def compute_y_bounds(self, df):
bounds = self._compute_bounds(df[self.y])
return self.maybe_expand_bounds(bounds)
@memoize
def compute_bounds_dask(self, ddf):
r = ddf.map_partitions(lambda df: np.array([[
np.nanmin(df[self.x].values).item(),
np.nanmax(df[self.x].values).item(),
np.nanmin(df[self.y].values).item(),
np.nanmax(df[self.y].values).item()]]
)).compute()
x_extents = np.nanmin(r[:, 0]), np.nanmax(r[:, 1])
y_extents = np.nanmin(r[:, 2]), np.nanmax(r[:, 3])
return (self.maybe_expand_bounds(x_extents),
self.maybe_expand_bounds(y_extents))
[docs]class Point(_PointLike):
"""A point, with center at ``x`` and ``y``.
Points map each record to a single bin.
Points falling exactly on the upper bounds are treated as a special case,
mapping into the previous bin rather than being cropped off.
Parameters
----------
x, y : str
Column names for the x and y coordinates of each point.
"""
@memoize
def _build_extend(self, x_mapper, y_mapper, info, append, _antialias_stage_2, _antialias_stage_2_funcs):
x_name = self.x
y_name = self.y
@ngjit
@self.expand_aggs_and_cols(append)
def _perform_extend_points(i, sx, tx, sy, ty, xmin, xmax,
ymin, ymax, xs, ys, xxmax, yymax,
*aggs_and_cols):
x = xs[i]
y = ys[i]
# points outside bounds are dropped; remainder
# are mapped onto pixels
if (xmin <= x <= xmax) and (ymin <= y <= ymax):
xx = int(x_mapper(x) * sx + tx)
yy = int(y_mapper(y) * sy + ty)
xi, yi = (xxmax-1 if xx >= xxmax else xx,
yymax-1 if yy >= yymax else yy)
append(i, xi, yi, *aggs_and_cols)
@ngjit
@self.expand_aggs_and_cols(append)
def extend_cpu(sx, tx, sy, ty, xmin, xmax, ymin, ymax, xs, ys,
xxmax, yymax, *aggs_and_cols):
for i in range(xs.shape[0]):
_perform_extend_points(i, sx, tx, sy, ty, xmin, xmax, ymin, ymax,
xs, ys, xxmax, yymax, *aggs_and_cols)
@cuda.jit
@self.expand_aggs_and_cols(append)
def extend_cuda(sx, tx, sy, ty, xmin, xmax, ymin, ymax, xs, ys,
xxmax, yymax, *aggs_and_cols):
i = cuda.grid(1)
if i < xs.shape[0]:
_perform_extend_points(i, sx, tx, sy, ty, xmin, xmax, ymin, ymax,
xs, ys, xxmax, yymax, *aggs_and_cols)
def extend(aggs, df, vt, bounds):
yymax, xxmax = aggs[0].shape[:2]
aggs_and_cols = aggs + info(df, aggs[0].shape[:2])
sx, tx, sy, ty = vt
xmin, xmax, ymin, ymax = bounds
if cudf and isinstance(df, cudf.DataFrame):
xs = values(df[x_name])
ys = values(df[y_name])
do_extend = extend_cuda[cuda_args(xs.shape[0])]
else:
xs = df[x_name].values
ys = df[y_name].values
do_extend = extend_cpu
do_extend(
sx, tx, sy, ty, xmin, xmax, ymin, ymax, xs, ys, xxmax, yymax, *aggs_and_cols
)
return extend
class MultiPointGeometry(_GeometryLike):
# spatialpandas must be available if a MultiPointGeometry object is created.
@property
def geom_dtypes(self):
from spatialpandas.geometry import PointDtype, MultiPointDtype
return PointDtype, MultiPointDtype
@memoize
def _build_extend(self, x_mapper, y_mapper, info, append, _antialias_stage_2, _antialias_stage_2_funcs):
geometry_name = self.geometry
@ngjit
@self.expand_aggs_and_cols(append)
def _perform_extend_points(
i, j, sx, tx, sy, ty, xmin, xmax, ymin, ymax, values, *aggs_and_cols
):
x = values[j]
y = values[j + 1]
# points outside bounds are dropped; remainder
# are mapped onto pixels
if (xmin <= x <= xmax) and (ymin <= y <= ymax):
xx = int(x_mapper(x) * sx + tx)
yy = int(y_mapper(y) * sy + ty)
xi, yi = (xx - 1 if x == xmax else xx,
yy - 1 if y == ymax else yy)
append(i, xi, yi, *aggs_and_cols)
@ngjit
@self.expand_aggs_and_cols(append)
def extend_point_cpu(
sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, missing, eligible_inds, *aggs_and_cols
):
for i in eligible_inds:
if missing[i] is True:
continue
_perform_extend_points(
i, 2 * i, sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, *aggs_and_cols
)
@ngjit
@self.expand_aggs_and_cols(append)
def extend_multipoint_cpu(
sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, missing, offsets, eligible_inds, *aggs_and_cols
):
for i in eligible_inds:
if missing[i] is True:
continue
start = offsets[i]
stop = offsets[i + 1]
for j in range(start, stop, 2):
_perform_extend_points(
i, j, sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, *aggs_and_cols
)
def extend(aggs, df, vt, bounds):
from spatialpandas.geometry import PointArray
aggs_and_cols = aggs + info(df, aggs[0].shape[:2])
sx, tx, sy, ty = vt
xmin, xmax, ymin, ymax = bounds
geometry = df[geometry_name].array
if geometry._sindex is not None:
# Compute indices of potentially intersecting polygons using
# geometry's R-tree if there is one
eligible_inds = geometry.sindex.intersects((xmin, ymin, xmax, ymax))
else:
# Otherwise, process all indices
eligible_inds = np.arange(0, len(geometry), dtype='uint32')
missing = geometry.isna()
if isinstance(geometry, PointArray):
values = geometry.flat_values
extend_point_cpu(
sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, missing, eligible_inds, *aggs_and_cols
)
else:
values = geometry.buffer_values
offsets = geometry.buffer_offsets[0]
extend_multipoint_cpu(
sx, tx, sy, ty, xmin, xmax, ymin, ymax,
values, missing, offsets, eligible_inds, *aggs_and_cols
)
return extend
