Browse Source

Added emdash 0x2014 as a word separator

deroshkin_branch
deroshkin 2 weeks ago
parent
commit
357eceabad
  1. 4
      novel_stats/novel_stats.py

4
novel_stats/novel_stats.py

@ -4,6 +4,7 @@
import argparse
import collections
import tempfile
import re
CHAPTER_MARKER = '## '
STATUS_MARKER = '[status]: # '
@ -11,12 +12,13 @@ ACT_MARKER = '[act]: # '
# Standard markdown comment marker, supported by Pandoc and Calibre's ebook-convert.
COMMENT_MARKER = '[//]: # '
TITLE_MARKER = '# '
WORD_SEPS = [' ','']
def count_words(line):
count = 0
for word in line.strip().split(' '):
for word in re.split('|'.join(WORD_SEPS), line.strip()):
if not word.strip() or word == '*' or word.startswith('#'):
continue

Loading…
Cancel
Save