Tab-Separated values in a text file

iso-3166-1.tsv

multidump.py

#-*- coding: utf-8 -*-
# multidump.py  (c)2022  Henrique Moreira

""" Dump multiple tables from single tab-separated value file 'iso-3166-1.tsv'
"""

# pylint: disable=missing-function-docstring

import unidecode
from zson.csv import CSV

IN_FILENAME = "iso-3166-1.tsv"
OUT_FILENAME = "iso-3166-1_byid.tsv"

def main():
    outs = [OUT_FILENAME]
    read_and_write(IN_FILENAME, outs)

def read_and_write(fin_name, outs:list):
    CSV.set_tab_separator()
    tsv = CSV()
    tsv.load(fin_name)
    for idx, out_name in enumerate(outs, 1):
        is_ok = write_to(tsv, idx, out_name)
        assert is_ok
    return True

def write_to(tsv, idx, out_name:str) -> bool:
    print("Writting to:", out_name)
    table = tsv.get()
    new, elems = [], []
    out = CSV()
    if idx == 1:
        new = sorted(table, key=lambda k: k[3])
        elems = [[dig, two, three, said(name, idx)] for name, two, three, dig in new]
        out.convert(elems)
    #print(f"::: (idx={idx}) first:", new[0], "; last:", new[-1])
    assert new, out_name
    assert elems
    dat = out.get()
    out.save(out_name)
    for elem in dat:
        print(elem)
    return True

def said(name:str, kind:int=0):
    """ Returns the approximate english name for the country, depending on 'kind'
	If kind is 0, no translation;
	if kind is -1, remove accents.
    """
    assert name == name.strip(), name
    if kind == 0:
        return name
    if kind <= -1:
        return unidecode.unidecode(name)
    if name.startswith("Virgin"):
        return name
    if "voire" in name:
        return "Ivory Coast"
    if "[" in name:
        name = name[:name.index("[")].strip()
    if name.endswith("(the)"):
        name = name[:-len("(the)")].strip()
    if name.endswith("(the Republic of)"):
        name = name[:-len("(the Republic of)")].strip()
    return said(name, -1)

# Main script
if __name__ == "__main__":
    main()

requirements.txt

unidecode