Jupyter notebook translator

README.md

Usage

run python jupyter_translate.py, and you will be asked for 3 inputs:

• path to notebook
• source language
• target language

A new notebook will be created in target language in the same path as notebook.
It will preserve original language cells too, for easier human debugging of the translation.

You need a DeepL API

jupyter_translate.py

import json
import os
import sys
import deepl
import readline

my_key = open("./my_key.txt", "r").read().strip()
translator = deepl.Translator(my_key)


def needs_translate(cell):
    if is_toc_cell(cell):
        return False
    if is_code_cell(cell):
        return False
    if is_html_cell(cell):
        return False

    return True


def get_cell_text(cell):
    return "\n".join(cell["source"])


def is_latex_cell(cell):
    cell_text = get_cell_text(cell)
    return cell_text.startswith("$")


def is_html_cell(cell):
    cell_text = get_cell_text(cell)
    return cell_text.startswith("<")


def is_toc_cell(cell):
    return cell["metadata"].get("toc")


def is_code_cell(cell):
    return cell["cell_type"] == "code"


def deepl_translate(text, from_lang, to_lang):
    result = translator.translate_text(text, source_lang=from_lang, target_lang=to_lang)
    translated_text = result.text

    return translated_text


def generate_translated_notebook(source_lang, target_lang, filename):
    with open(filename, "rb") as f:
        notebook = json.load(f)

    # clone old notebook (maintain metadata, indent, etc) and reset cells
    new_notebook = notebook.copy()
    new_notebook["cells"] = []
    n_translations = 0

    for i, original_cell in enumerate(notebook["cells"]):
        if i % 10 == 0:
            print(f"Evaluating cell #{i}")

        new_notebook["cells"].append(original_cell)

        if not needs_translate(original_cell):
            continue

        # else: needs translation
        n_translations += 1
        new_cell = original_cell.copy()
        original_text = get_cell_text(original_cell)
        translated_text = deepl_translate(original_text, source_lang, target_lang)

        new_cell['source'] = translated_text
        
        new_notebook["cells"].append(new_cell)

    print(f"Translated {n_translations}/{i} cells")

    filepath_translated = f"{filename.split('.ipynb')[0]}_{target_lang}.ipynb"
    with open(filepath_translated, "w") as f:
        print(f"You can find translated notebook here: {filepath_translated}")
        json.dump(new_notebook, f, indent=2)


if __name__ == "__main__":
    readline.set_completer_delims(' \t\n=')
    readline.parse_and_bind("tab: complete")

    filename = input("File path: ")

    print(f"You entered: {filename}")
    if not os.path.exists(filename):
        print("That file path does not exist")
        sys.exit()

    source_lang = input("Source language: ")
    target_lang = input("Target language: ")


    generate_translated_notebook(source_lang, target_lang, filename)