def query_knn(self, location: np.ndarray, k: int) -> (np.ndarray, np.ndarray):
idx = h3.geo_to_h3(location[0], location[1], self.h3res)
i = 0
indices = np.zeros(0, dtype=np.uint64)
ring = np.zeros(0, dtype=np.uint64)
while indices.shape[0] < k:
i += 2
k_ring = h3.k_ring(idx, i)
ring = np.setdiff1d(k_ring, ring, assume_unique=True)
i0 = np.searchsorted(self.h3arr, ring, side='left', sorter=self.h3idx)
i1 = np.searchsorted(self.h3arr, ring, side='right', sorter=self.h3idx)
indices = np.hstack((indices,
np.hstack([np.arange(i, j, dtype=np.uint64)
for i, j in zip(i0, i1) if i != j])))
idx = self.h3idx[indices]
dist = gm.vec_haversine(self.locations[idx, 0],
self.locations[idx, 1],
location[0], location[1])
dist_idx = np.argsort(dist)
return idx[dist_idx[:k]], dist[dist_idx[:k]]
def test5():
expected = {
617700169957507071,
617700169957769215,
617700169958031359,
617700169958293503,
617700169961177087,
617700169964847103,
617700169965109247,
}
out = h3.k_ring(617700169958293503, 1)
assert isinstance(out, np.ndarray)
assert set(out) == expected
def query_radius(self,
location: np.ndarray,
r: float) -> np.ndarray:
edge_len = h3.edge_length(self.h3res, unit="m")
idx = h3.geo_to_h3(location[0], location[1], self.h3res)
ring = h3.k_ring(idx, 1 + int(round(r / edge_len)))
i0 = np.searchsorted(self.h3arr, ring, side='left', sorter=self.h3idx)
i1 = np.searchsorted(self.h3arr, ring, side='right', sorter=self.h3idx)
indices = np.hstack([np.arange(i, j) for i, j in zip(i0, i1) if i != j])
idx = self.h3idx[indices]
dist = gm.vec_haversine(self.locations[idx, 0], self.locations[idx, 1],
location[0], location[1])
return self.h3idx[indices[np.argwhere(dist <= r).ravel()]]
def neighbours(self) -> HexIdSet:
return set(h3.k_ring(self.id, k=1))