Build continuous PuD code ranges based on ZIP:distance mapping. Writes to stdout.

pudz.py

#! /usr/bin/env python
# -*- coding: utf-8 -*-

from bisect import bisect_left
import begin

@begin.start(auto_convert=True, short_args=False)
def main(airport: """3 letter airport code""",
         infile:  """input file with whitespace separted (code, dist) pairs""",
         breaks:  """whitespace separated string containing sorted list with
                  "more than" distance breaks""",
         cutoff:  """value up to which distances will be included in zone
                  def (will be ignored if 0)""" = 0):
    """
    Build continuous PuD code ranges based on ZIP: distance mapping. Writes to
    stdout.
    """

    with open(infile, 'r') as fp:
        code_dist = (line.strip().split() for line in fp)
        code_dist = [(code, int(dist)) for code, dist in code_dist if
                dist.isdecimal() and (cutoff == 0 or int(dist) <= cutoff)]

    breaks = [int(b) for b in breaks.split()]

    # first sort by distance to assign zones:
    code_dist.sort(key=lambda k: k[1])
    zones = (breaks[bisect_left(breaks, d) - 1] for _, d in code_dist)
    code_zone = [(code, zone) for (code, _), zone in zip(code_dist, zones)]

    # now sort by code to assemble continuous ranges with same zone:
    code_zone.sort(key=lambda k: k[0])
    code_from = code_zone[0][0]
    for (current_code, current_zone), (prev_code, prev_zone) in zip(
            code_zone[1:], code_zone[:-1]):
        if current_zone == prev_zone:
            continue
        else:
            print('{0}\t{1}\t{2}\t{3}'.format(airport, code_from, prev_code,
                                              prev_zone))
            code_from = current_code
    print('{0}\t{1}\t{2}\t{3}'.format(airport, code_from, current_code,
          current_zone))