Wrote a script for localization.

src/localize.py

+#!/usr/bin/python3
+
+import copy
+import json
+import os
+import sys
+
+import langid
+
+
+def detect_language(text):
+    return langid.classify(text)
+
+
+def load(path):
+    with open(path, encoding="utf-8") as file:
+        notebook = json.load(file)
+    return notebook
+
+
+def human_intervention(language, probability, text):
+    print(
+        f"I classified this text:\n`{text}`\nas {language} with a non normalized probability of {probability}.\n"
+    )
+    while True:
+        print(
+            "Can you help me identifying it? write one among 'en', 'de', 'eq' (for equations), then press enter"
+        )
+        language = input()
+        if language in ["en", "de", "eq"]:
+            print("\n\n")
+            return language
+
+
+def write(path, notebook):
+    with open(path, "w", encoding="utf-8") as file:
+        json.dump(notebook, file, indent=4)
+
+
+def assign_languages(notebook):
+    cells = notebook["cells"]
+
+    for i, cell in enumerate(cells):
+        if cell["cell_type"] == "markdown":
+            text = str(cell["source"]).strip("\n# []'")
+            language, probability = detect_language(text)
+
+            if probability >= -10 or language not in ["en", "de"]:
+                # langid is doing something strange, we need to intervene
+                language = human_intervention(language, probability, text)
+
+            notebook["cells"][i]["metadata"]["natural_language"] = language
+
+
+def select_language(notebook, language):
+    cells = notebook["cells"]
+
+    selected_cells = []
+    for cell in cells:
+        if cell["cell_type"] == "markdown":
+            if cell["metadata"]["natural_language"] == language:
+                selected_cells.append(cell)
+            elif cell["metadata"]["natural_language"] == "eq":
+                selected_cells.append(cell)
+        else:
+            selected_cells.append(cell)
+
+    new_notebook = copy.deepcopy(notebook)
+    new_notebook["cells"] = selected_cells
+
+    return new_notebook
+
+
+if __name__ == "__main__":
+    PATH = sys.argv[1]
+    BASENAME = os.path.splitext(os.path.basename(PATH))[0]
+
+    NOTEBOOK = load(PATH)
+    assign_languages(NOTEBOOK)
+
+    NOTEBOOK_EN = select_language(NOTEBOOK, "en")
+    NOTEBOOK_DE = select_language(NOTEBOOK, "de")
+
+    write(f"{BASENAME}_en.ipynb", NOTEBOOK_EN)
+    write(f"{BASENAME}_de.ipynb", NOTEBOOK_DE)