Handling row-major order observations

All observations returned by the extended API in response to requests for tabulations are returned in row-major order.

The main challenge in working with these observations is to compute, for each observation, the indices of the corresponding category within each table dimension.

A simple way to achieve this is shown in a snippet of Python code below, written in a generic style to make it easy to translate to other languages.

dim_lengths = [3, 4, 5]  # example dimension lengths
dim_indices = [0, 0, 0]  # indices into dimensions to obtain categories
values = range(3 * 4 * 5)  # values of correct length, actual values not relevant

for v in values:
  print(dim_indices) # process the row here: use dim_indices to find categories
  i = len(dim_indices) - 1
  while i >= 0:
    dim_indices[i] += 1
    if dim_indices[i] < dim_lengths[i]:
        break
    dim_indices[i] = 0
    i -= 1

Worked examples of how to translate GraphQL outputs to other data structures in different languages, are available in our Cantabular examples repository on GitHub.