latex2eoa.py

#!/usr/bin/env python3
# -*- coding: utf-8; mode: python -*-

__version__ = "1.0"
__date__ = "20170621"
__author__ = "kthoden@mpiwg-berlin.mpg.de"

import sys
import re

# to be dealt with: EOAfigure

regex_replacements = {
    r'\$\$(.*?)\$\$' : r"\\begin{EOAequationnonumber}\g<1>\\end{EOAequationnonumber}",
    r'\$(.*?)\$' : r"\\EOAineq{\g<1>}",
    r'\\section' : r"\\EOAsection",
    r'{\\em'  : r"\\EOAemph{",
    r'\\begin{equation}' : r"\\begin{EOAequation}{equation_label}",
    r'\\end{equation}' : r"\\end{EOAequation}",
    r'\\(begin|end){quote}' : r"\\\g<1>{EOAquote}",
    r'\\(begin|end){itemize}' : r"\\\g<1>{EOAitems}",
    r'\\label' : r"\\EOAlabel",
    r'\\ref' : r"\\EOAref",
    r'\\citet' : r"\\EOAciteauthoryear",
    r'{\\bf' : r"\\EOAbold{",
    r'\\footnote' : r"\\EOAfn",
    r'\\bibliography{' : r"\\EOAbibliographydatabase{"
    }

def find_latex_commands(string):
    """Finds latex commands in a text, return a list of unique
    occurrences.
    """

    command_pattern = re.compile(r'\\[a-zA-Z0-9]+', re.UNICODE)
    found_commands = set(re.findall(command_pattern, string))

    return found_commands
# def find_latex_commands ends here

def main(infile):
    """The main bit"""

    with open(infile, 'r') as input_file:
        data = input_file.read()

    # list_of_commands = find_latex_commands(data)
    # print(list_of_commands)

    for instance in regex_replacements.keys():
        data = re.sub(instance, regex_replacements[instance], data)
    print(data)
# def main ends here

if __name__ == '__main__':
    infile = sys.argv[-1]
    main(infile)
# finis