markdown fix

gistfile1.py

src = '''\
Chapter 1
foo
- figure 1.1
bar
Section 1.1
- figure 1.2
- figure 1.3
baz
Section 1.2
- figure 1.4
Subsection 1.2.1
- figure 1.5
- figure 1.6

Chapter 2
- figure 2.1
Section 2.1
foo
Subsection 2.1.1
- figure 2.2
- figure 2.3
Subsection 2.1.2
- figure 2.4
Section 2.2
- figure 2.5
foobar
'''

import re

token_pattern = re.compile(r'''
        (?P<name>Chapter|Section|Subsection|-\ figure)
        \ (?P<num>([0-9]\.?)+)
        ''', re.VERBOSE)
def iterate(src, token_cb, text_cb):
    pos = 0
    while True:
        token = token_pattern.search(src, pos)
        if token is None: break
        text_cb(src[pos:token.start()])
        nums = [int(s) for s in token.group('num').split('.')]
        token_cb(token.group('name'), nums)
        pos = token.end()
    text_cb(src[pos:len(src)])

def fix(src):
    out = ''
    def token_cb(name, nums):
        nonlocal out
        if nums[0] >= 2: nums[0] += 1
        out += '%s %s' % (name, '.'.join(str(n) for n in nums))
    def text_cb(text):
        nonlocal out
        out += text
    iterate(src, token_cb, text_cb)
    return out

print(fix(src))

'''output:
Chapter 1
foo
- figure 1.1
bar
Section 1.1
- figure 1.2
- figure 1.3
baz
Section 1.2
- figure 1.4
Subsection 1.2.1
- figure 1.5
- figure 1.6

Chapter 3
- figure 3.1
Section 3.1
foo
Subsection 3.1.1
- figure 3.2
- figure 3.3
Subsection 3.1.2
- figure 3.4
Section 3.2
- figure 3.5
foobar

'''