replace_post_tex.py

import re

def replace_dollar_tex(s):
    l = len(s)
    i, j, stack = 0, 0, 0
    new_txt = ''
    while i < l:
        if s[i] == "\\" and (i + 1) < l:
            if s[i + 1] == '$':
                # skip if it is escaped dollar
                new_txt += '$'
                i += 1
            elif stack == 0:
                # otherwise just copy it
                new_txt += s[i]
        elif s[i] == '$':
            if stack == 0: # first open dollar
                stack = 1
                j = i + 1
            elif stack == 1: # second dollar
                if i == j:
                    # consecutive dollar
                    # (second open dollar)
                    stack = 2
                    j = i + 1
                else:
                    # non-consecutive dollar
                    # (close dollar)
                    stack = 0
                    # print('single: %s' % s[j:i])
                    new_txt += '[imath]%s[/imath]' % s[j:i]
            else: # stack == 2
                # first close dollar
                stack = 0
                # print('double: %s' % s[j:i])
                new_txt += '[imath]%s[/imath]' % s[j:i]
                # skip the second close dollar
                i += 1
        elif stack == 0:
            # non-escaped and non enclosed characters
            new_txt += s[i]
        i += 1
    return new_txt

def replace_display_tex(s):
    # replace '\[ * \]'
    regex = re.compile(r'\\\[(.+?)\\\]', re.DOTALL)
    return re.sub(regex, r"[imath]\1[/imath]", s)

def replace_inline_tex(s):
    # replace '\\( * \\)'
    regex = re.compile(r'\\\\\((.+?)\\\\\)', re.DOTALL)
    return re.sub(regex, r"[imath]\1[/imath]", s)

def unwrap_isolated_tex_group(text, group_name):
    regex = re.compile(
        r"\\begin{" + group_name +
        r"\*?}(.+?)\\end{" + group_name +
        r"\*?}(?!\s+\[/imath\])", re.DOTALL) # negative lookahead
    return re.sub(regex, r"[imath]\1[/imath]", text)

def unwrap_isolated_tex_groups(text,
    groups=['align', 'alignat', 'equation', 'gather']):
    for grp in groups:
        text = unwrap_isolated_tex_group(text, grp)
    return text

if __name__ == '__main__':
    # curl http://math.stackexchange.com/questions/1886701
    # to test real-world MSE parsing.
    test = r'''
    In this paper, the authors investigate the double-pole Nahm-type sums
    \[
    \mathcal D_{t,s}(w,q) = \sum_{n_1,\ldots,n_t\geq 0} (-w;q){n_1} q^{n_s} \prod{r=1}^t \frac{q^{n_rn_{r+1}+n_r}}{(q;q){n_r}^2},
    \]
    where \\(t\geq 1\\), \\(1\leq s \leq t+1\\) and \\(n{t+1}=0\\).
    '''
    print(test)
    print(replace_display_tex(replace_inline_tex(test)))