Archivo para traducir notebooks [Translation file for notebook]

translate.py

from googletrans import Translator
import nbformat
import glob
import sys
import os

# References:
# http://nbformat.readthedocs.io/en/latest/api.html
# https://pypi.python.org/pypi/googletrans

def translate_string(source_text, lang_dest, lang_source):
    """
    Translation function, powered by google translate and fixed for
    custom markdown errors.
    """
    # General translation
    translator = Translator()
    t = translator.translate(source_text, src=lang_source, dest=lang_dest)
    translated_text = t.text
    # Fix the easy markdown errors
    fixed_markdown = translated_text.replace(r"] (", r"](")
    fixed_markdown = fixed_markdown.replace(r"</ ", r"</")
    fixed_markdown = fixed_markdown.replace(r"\ frac", r"\frac")
    fixed_markdown = fixed_markdown.replace(r"\ pi", r"\pi")
    fixed_markdown = fixed_markdown.replace(r"\ begin", r"\begin")
    fixed_markdown = fixed_markdown.replace(r"\ end", r"\end")
    fixed_markdown = fixed_markdown.replace(r"\ partial", r"\partial")
    fixed_markdown = fixed_markdown.replace(r"\ sum", r"\sum")
    fixed_markdown = fixed_markdown.replace(r"\ quad", r"\quad")
    fixed_markdown = fixed_markdown.replace(r"help ()", r"help()")
    fixed_markdown = fixed_markdown.replace(r"print ()", r"print()")
    fixed_markdown = fixed_markdown.replace(r"\ $", r"\$")
    fixed_markdown = fixed_markdown.replace(r"`` `", r"```")
    fixed_markdown = fixed_markdown.replace(r" / ", r"/")
    fixed_markdown = fixed_markdown.replace(r"—", r"-")

    # Fix opening italics
    for mark in [r'"', r'¿', r'¡', r'`']:
        s_old = '_ {}'.format(mark)
        s_new = '_{}'.format(mark)
        fixed_markdown = fixed_markdown.replace(s_old, s_new)
    # Fix closing italics
    for mark in [r'"', r'?', r'!', r'`']:
        s_old = '{} _'.format(mark)
        s_new = '{}_'.format(mark)
        fixed_markdown = fixed_markdown.replace(s_old, s_new)
    print(fixed_markdown)
    return fixed_markdown

def translate_notebook(ipynb_folder, format_version, lang_dest, lang_source="eng"):
    """
    Translate the notebook cells, but only for those cells that are
    markdown, the others might be code or something else.
    """
    # Translate all the markdown cells
    my_translated_notebook_path = my_notebook_path.replace(".ipynb", "_"+lang_dest+".ipynb")
    print("Original:\t"+my_notebook_path)
    print("Translated:\t"+my_translated_notebook_path)
    nb = nbformat.read(my_notebook_path, as_version=format_version)
    for i, cell in enumerate(nb["cells"]):
        print("\tTranslating cell {} from {}".format(i, len(nb["cells"])))
        if cell["cell_type"]=="markdown":
            cell["source"] = translate_string(cell["source"], lang_dest, lang_source)
        else:
            pass
    print("Saving the translated notebook")
    nbformat.write(nb, fp=my_translated_notebook_path)

if __name__=="__main__":
    # Parameters
    lang_dest = "en"
    format_version = 4
    if len(sys.argv)==2:
        ipynb_path = sys.argv[1]
        all_files = sorted(glob.glob(ipynb_path))
        # Translate all the notebooks not translated already
        for my_notebook_path in all_files:
            if "_"+lang_dest+".ipynb" in my_notebook_path:
                pass # Skip if there's no need to translate (already translated, so will be overwritten)
            else:
                translate_notebook(my_notebook_path, format_version, lang_dest, "es")
    else:
        print("Run as:")
        print("\t$ python3 translate.py EngComp/modules/1_offtheground/1_Interacting_with_Python.ipynb")
        print("or more than one with wildmarks")
        print('\t$ python3 translate.py "EngComp/modules/1_offtheground/*.ipynb"')