translate.py

####################################################################################
#
# Distributed under MIT Licence
#   See https://github.com/house-of-abbey/GarminThermoNest/blob/main/LICENSE.
#
####################################################################################
#
# ThermoNest is a Garmin IQ application written in Monkey C and routinely tested on
# a Venu 2 device. The source code is provided at:
#            https://github.com/house-of-abbey/GarminThermoNest.
#
# J D Abbey & P A Abbey, 28 December 2022
#
#
# Description:
#
# Python script to automatically translate the strings.xml file to each supported
# language using Google Translate.
#
# Python installation:
#   pip install BeautifulSoup
#   pip install deep-translator
# NB. For XML formatting:
#   pip install lxml
#
# References:
#  * https://www.crummy.com/software/BeautifulSoup/bs4/doc/
#  * https://realpython.com/beautiful-soup-web-scraper-python/
#  * https://www.crummy.com/software/BeautifulSoup/bs4/doc/#parsing-xml
#  * https://www.crummy.com/software/BeautifulSoup/bs4/doc/#xml
#
####################################################################################

from deep_translator import GoogleTranslator
from bs4 import BeautifulSoup, Comment
import os

# List of tuples in the form os:
#  * Garmin IQ language three letter mnemonic,
#  * Google Translate language mnemonic,
#  * Language familiar name (mainly for reference)
languages: list[tuple[str, str, str]] = [
  ("ara", "ar", "Arabic"),
  ("bul", "bg", "Bulgarian"),
  ("zhs", "zh-CN", "Chinese (Simplified)"),
  ("zht", "zh-TW", "Chinese (Traditional)"),
  ("hrv", "hr", "Croatian"),
  ("ces", "cs", "Czech"),
  ("dan", "da", "Danish"),
  ("dut", "nl", "Dutch"),
  ("deu", "de", "German"),
  ("gre", "el", "Greek"),
  # ("eng", "en", "English"),
  ("est", "et", "Estonian"),
  ("fin", "fi", "Finnish"),
  ("fre", "fr", "French"),
  ("heb", "iw", "Hebrew"),
  ("hun", "hu", "Hungarian"),
  ("ind", "id", "Indonesian"),
  ("ita", "it", "Italian"),
  ("jpn", "ja", "Japanese"),
  ("kor", "ko", "Korean"),
  ("lav", "lv", "Latvian"),
  ("lit", "lt", "Lithuanian"),
  ("zsm", "ms", "Standard (Bahasa) Malay"),
  ("nob", "no", "Norwegian"),
  ("pol", "pl", "Polish"),
  ("por", "pt", "Portuguese"),
  ("ron", "ro", "Romanian"),
  # ("rus", "ru", "Russian"),
  ("slo", "sk", "Slovak"),
  ("slv", "sl", "Slovenian"),
  ("spa", "es", "Spanish"),
  ("swe", "sv", "Swedish"),
  ("tha", "th", "Thai"),
  ("tur", "tr", "Turkish"),
  ("ukr", "uk", "Ukrainian"),
  ("vie", "vi", "Vietnamese")
]

langLength = len(languages)

exceptionIds: list[str] = ["AppName", "AppVersionTitle"]
titleIds: list[str] = ["setMode", "tapIcon"]

i = 1
with open("./resources/strings/strings.xml", "r") as f:
  c = f.read().replace('\r', '')
  for l in languages:
    os.makedirs(f"./resources-{l[0]}/strings/", exist_ok=True)
    try:
      with open(f"./resources-{l[0]}/strings/corrections.xml", "r") as r:
        curr = BeautifulSoup(r.read().replace('\r', ''), features="xml")
    except FileNotFoundError:
      curr = BeautifulSoup("", features=["xml"])
    print(f"{i} of {langLength}: Translating English to {l[2]}")
    soup = BeautifulSoup(c, features="xml")
    soup.find(name="strings").insert_before("\n\n")
    soup.find(name="strings").insert_before(
      Comment(f"\n  Generated by Google Translate: English to {l[2]}\n  " +
      GoogleTranslator(source='en', target=l[1]).translate("Generated by Google Translate from English") +
      "\n")
    )
    soup.find(name="strings").insert_before("\n\n")

    for s in soup.find(name="strings").findAll(name="string"):
      s.insert_before("  ")
      if s["id"] in exceptionIds: continue
      
      s_curr = curr.find(name="string", attrs={ "id": s["id"] })
      if s_curr:
        s.string = s_curr.string
      else:
        a = GoogleTranslator(source='en', target=l[1]).translate(s.string)
        if s["id"] in titleIds:
          s.string = a.title()
        else:
          s.string = a
    for s in soup.find(name="strings").findAll(text=lambda text:isinstance(text, Comment)):
      s.insert_before("  ")
      s.replace_with(Comment(" " + GoogleTranslator(source='en', target=l[1]).translate(s) + " "))

    #print(str(soup))
    with open(f"./resources-{l[0]}/strings/strings.xml", "wb") as w:
      w.write(soup.encode("utf-8"))
    i += 1