#!/usr/bin/python import collections import os import string import sys CHAPTER_MARKER = '## ' STATUS_MARKER = '[status]: # ' ACT_MARKER = '[act]: # ' def count_words(line): count = 0 for word in line.strip().split(' '): if not word.strip() or word == '*' or word.startswith('#'): continue count += 1 return count def main(): arguments = sys.argv[1:] filename = arguments[0] chapter_number = None act_number = None total_word_count = 0 word_count_by_chapter = collections.defaultdict(int) word_count_by_status = collections.defaultdict(int) word_count_by_act = collections.defaultdict(int) status_by_chapter = {} for line in open(filename).readlines(): if line.startswith(CHAPTER_MARKER): word_count_by_act[act_number] += word_count_by_chapter[chapter_number] total_word_count += word_count_by_chapter[chapter_number] if chapter_number in status_by_chapter: word_count_by_status[status_by_chapter[chapter_number]] += 1 chapter_number = int(line[len(CHAPTER_MARKER):]) word_count_by_chapter[chapter_number] = 1 # Start at one, because the chapter number itself counts as a word. if chapter_number in status_by_chapter: word_count_by_status[chapter_status] += 1 elif line.startswith(STATUS_MARKER): status_by_chapter[chapter_number] = line[len(STATUS_MARKER):].strip('()\n') elif line.startswith(ACT_MARKER): act_number = int(line[len(ACT_MARKER):].strip('()\n')) word_count_by_act[act_number] = 1 else: line_word_count = count_words(line) word_count_by_chapter[chapter_number] += line_word_count if chapter_number in status_by_chapter: word_count_by_status[status_by_chapter[chapter_number]] += line_word_count # Do some final accounting after the last chapter. word_count_by_act[act_number] += word_count_by_chapter[chapter_number] total_word_count += word_count_by_chapter[chapter_number] if chapter_number in status_by_chapter: word_count_by_status[status_by_chapter[chapter_number]] += 1 # Print out word counts. for chapter_number, chapter_word_count in word_count_by_chapter.items(): if chapter_number is None: continue chapter_status = status_by_chapter.get(chapter_number) print( 'chapter {}: {:,} words{}'.format( chapter_number, chapter_word_count, ' ({})'.format(chapter_status) if chapter_status else '', ) ) print() for act_number, act_word_count in word_count_by_act.items(): if act_number is None: continue print('act {}: {:,} words (~{}%)'.format(act_number, act_word_count, act_word_count * 100 // total_word_count)) for status, status_word_count in word_count_by_status.items(): print('{}: {:,} words (~{}%)'.format(status, status_word_count, status_word_count * 100 // total_word_count)) print('total: {:,} words'.format(total_word_count)) if __name__ == '__main__': main()