Edge class, less looping

[lmmx]

generate_candidate_edge in spans.py should be a class CandidateEdge, and rather than returning None as the silent negative case if a negative if statement (awful double negative!) it should set a property is_valid based on the three test conditions

• in fact, is_valid can be a method which hides the 3 conditions out of the control flow of the init method

...and then the init method should end with a simple if self.is_valid(): before setting value (if not valid then the value was initialised as None)

• set value as None at the class level to make it default
• check generate_candidate_edge is only used in spans.assemble_spans
• rather than if edge is not None (which itself could just be if edge), use a listcomp
• in fact the whole block could be passed to the method at once rather than repeatedly looping in and passing the same parameters, so the listcomp will probably have a couple of layers

The syntax for this is [x for x in top_level_list for sublist in top_level_list] so here it’d be:

candidate_edges = [CandidateEdge(cinfo_i, cinfo_list[j]) for i, cinfo_i in enumerate(cinfo_list) for j in range(i) if CandidateEdge(cinfo_i, cinfo_list[j]).is_valid]

However this instantiates the Edge class twice! This could in fact be a good use for the walrus operator, though it’d technically become a generator expression within a listcomp (the listcomp is just the is_valid filter part)

candidate_edges = [
  edge for edge in [
    CandidateEdge(cinfo_i, cinfo_list[j])
    for i, cinfo_i in enumerate(cinfo_list)
    for j in range(i)
  ]
  if edge.is_valid()
]

• Also think about renaming cinfo_list to contours? Double check what’s actually in the variable (it’s been named to emphasise it’s a list but type annotations could do that)