arq2tex.py

#!/usr/bin/env python


import re


ARQ_COL = '|'
ARQ_HLINE = '-'
ARQ_HHLINE = '='

TEX_HLINE = '\\hline'
TEX_COL = '&'
TEX_EOL = '\\\\'


def get_ncols(lines):
    for line in (l.strip() for l in lines):
        ncols = line.strip(ARQ_COL).count(ARQ_COL)
        if ncols != 0:
            return ncols + 1
    return 1


def hline():
    print TEX_HLINE


def is_hline(line):
    return line == ARQ_HLINE * len(line)


def hhline():
    print TEX_HLINE, TEX_HLINE


def is_hhline(line):
    return line == ARQ_HHLINE * len(line)


def row(line, translate=[('"', '')]):
    def camelCase(s):
        chars = []
        for i, char in enumerate(s):
            if char.isspace():
                chars.append(char)
            else:
                chars.append(char.upper())
                chars.extend(s[i+1:])
                break
        return ''.join(chars)

    def snake2camel(s):
        tokens = (tok for tok in s.split('_') if tok)
        return ' '.join(map(camelCase, tokens))

    for source, target in translate:
        line = re.sub(re.escape(source), target, line)
    line = re.sub('  *', ' ', line)
    line = re.sub('&', '\&', line)
    line = ARQ_COL.join(map(snake2camel, line.split(ARQ_COL)))
    print re.sub(re.escape(ARQ_COL), TEX_COL, line.strip(ARQ_COL)) + TEX_EOL


def start_table(ncols, colstyle='c|'):
    print '\\begin{table}'
    print '\\centering'
    print '\\begin{tabular}{|l|%s}' % (colstyle * (ncols - 1))


def end_table():
    print '\\end{tabular}'
    print '\\caption{}'
    print '\\label{}'
    print '\\end{table}'


def convert_lines(lines):
    lines = list(lines)
    start_table(ncols=get_ncols(lines))
    for line in (l.strip() for l in lines):
        if is_hline(line):
            hline()
        elif is_hhline(line):
            hhline()
        else:
            row(line)
    end_table()


if __name__ == '__main__':
    import fileinput
    convert_lines(fileinput.input())