Merge

Interleave several streams into one index-sorted stream, tagging each event with the integer index of the source it came from. This is the causal, order-preserving way to fan several streams into a single timeline.

Feeding Merge lazy iterators of (value, index, source) events returns a lazy iterator (each source is numbered by a per-source arrival counter); feeding arrays or (values, index) tuples returns the eager (values, sources, index) 3-tuple.

Example

Two streams are merged into one stream ordered by index. sources records which stream each event came from.

import numpy as np
from screamer import Merge

a_v = np.array([1.0, 3.0])
a_k = np.array([1, 3])
b_v = np.array([2.0, 4.0])
b_k = np.array([2, 4])

values, sources, idx = Merge()(a_v, b_v, index=[a_k, b_k])
print(list(zip(idx.tolist(), sources.tolist())))

[(1, 0), (2, 1), (3, 0), (4, 1)]

split is the inverse: it restores the original per-source streams.