`make test` to generate initialisms for Japanese place names, e.g. 池袋 → E. K. Bukuro

initialjp.py

from __future__ import print_function
import romkan

with open('vendor/Noun.place.csv', encoding='utf-8') as in_file:
    data = [line.split(',') for line in in_file]

kana2romaji = {
    'エ': 'A',
    'ビ': 'B',
    'シ': 'C',
    'シー': 'C',
    'イ': 'E',
    'イー': 'E',
    'イイ': 'E',
    'ジ': 'G',
    'アイ': 'I',
    'ケ': 'K',
    'エン': 'N',
    'オ': 'O',
    'オオ': 'O',
    'オウ': 'O',
    'ピ': 'P',
    'キュ': 'Q',
    'キュウ': 'Q',
    'ユ': 'U',
    'ユウ': 'U',
}

forbidden = 'ャュョッン'


def get_initialism(reading):
    out = []
    initials = 0
    while reading:
        for kana in sorted(kana2romaji, key=len, reverse=True):
            if reading.startswith(kana) and (reading == kana or
                                             reading[len(kana)] not in forbidden):
                out.append(kana2romaji[kana])
                out.append('. ')
                initials += 1
                reading = reading[len(kana):]
                break
        else:
            roma = romkan.to_roma(reading)
            out.append(roma.capitalize())
            reading = ''
    return ''.join(out), initials


def main():
    for item in data:
        name, reading = item[0], item[11]
        initialism, count = get_initialism(reading)
        if count > 1:
            print(name, initialism)


if __name__ == '__main__':
    main()

Makefile

vendor:
	mkdir -p $(@)

NOUN_DATA := vendor/Noun.place.csv
NOUN_DATA_URL := https://raw.githubusercontent.com/taku910/mecab/master/mecab-ipadic/Noun.place.csv

$(NOUN_DATA): | vendor
	curl $(NOUN_DATA_URL) | iconv -f EUC-JP -t UTF-8 > $(@)

ENV := .env

$(ENV):
	virtualenv $(@)
	$(@)/bin/pip install romkan

.PHONY: test
test: | $(NOUN_DATA) $(ENV)
	$(ENV)/bin/python initialjp.py

Output (from a slightly older version) here:
https://pastebin.com/VAfDAF5u