Method __pairwise_step

tests/common/test_vector.py:318–345

Simple algorithm to select a subset of 'vectors' (each with N items) to cover every parameter pair. With 'min_cover_count'=1 all pairs will be covered (if possible), for bigger values only those vectors will be selected that cover at least 'min_cover_count' new pairs

(self, vectors, results, pairs_covered, min_cover_count)

    return True

  def __pairwise_step(self, vectors, results, pairs_covered, min_cover_count):
    """ Simple algorithm to select a subset of 'vectors' (each with N items) to cover
        every parameter pair.
        With 'min_cover_count'=1 all pairs will be covered (if possible), for bigger
        values only those vectors will be selected that cover at least 'min_cover_count'
        new pairs and the unused vectors will be returned in the result of the function.
        'pairs_covered' is a set of the already covered pairs. The key is a tuple of 4
        elements (i, v[i], j, v[j]) where i<j and both are < N. Each vector covers
        N * (N - 1) pairs.
    """
    remaining = []
    for v in vectors:
      cnt = 0
      for i, x in enumerate(v):
        for j, y in enumerate(v[i + 1:], i + 1):
          t = (i, x, j, y)
          if t not in pairs_covered:
            cnt += 1
      if cnt == 0: continue
      if cnt < min_cover_count:
        remaining.append(v)
        continue
      results.append(v)
      for i, x in enumerate(v):
        for j, y in enumerate(v[i + 1:], i + 1):
          t = (i, x, j, y)
          pairs_covered.add(t)
    return remaining