======  Quick project: translating text [python] ======

{{tag>dev python wordpress}}

Lately, I've been considering multilanguage support on my [[https://blog.nervtech.org/|family blogging site]]. And my my my... 😳 This is still so expensive in the end! All the extensions/plugins you can find for wordpress will request you to pay at some point to perform translations, which is fair I guess. But I don't like paying for things I can do myself. So question is, can I do that ? Translate large texts easily ? lol 😅

And thus, this project at hand here: let's see if I can do that in python conviniently! I found the [[https://pypi.org/project/deep-translator|deep-translator]] package already, so let's give that a try.


====== ======

===== The command line arguments =====

As usual let's start with the command line arguments that I could use here:

  * **input_text**: the input text to translate, should come as a first mandatory argument.
  * **source_language**: The source language of the input text, should default to **fr**
  * **target_language**: The target language of the translation, should default to **en**
  * **inverse**: should allow swapping of the source/target languages for quick reverse translations.

And that should be a good start already I believe. Let's now setup the component skeleton, with a simple command ''translate'':

<sxh python; highlight: []>"""TextTranslator module"""
import logging
import sys

from deep_translator import GoogleTranslator

from nvp.nvp_component import NVPComponent
from nvp.nvp_context import NVPContext

logger = logging.getLogger(__name__)


class TextTranslator(NVPComponent):
    """TextTranslator component class"""

    def __init__(self, ctx: NVPContext, proj=None):
        """Component constructor"""
        NVPComponent.__init__(self, ctx)

        desc = {
            "translate": None,
        }
        ctx.define_subparsers("main", desc)

        psr = ctx.get_parser('main.translate')
        psr.add_argument("input_text", type=str,
                         help="Input text to translate")
        psr.add_argument("-s", "--src", dest="source_language", type=str, default="fr",
                         help="Source language to use")
        psr.add_argument("-t", "--tgt", dest="target_language", type=str, default="en",
                         help="Target language to use")
        psr.add_argument("-i", "--inverse", dest="inverse", action='store_true',
                         help="Swap src/tgt languages")

    def process_command(self, cmd0):
        """Re-implementation of process_command"""

        if cmd0 == 'translate':
            return self.translate()

        return False</sxh>

===== Translate method implementation =====

And here is the translate method I quickly wrote on a first pass: <sxh python; highlight: []>    def translate(self):
        """Translate the input text here."""

        text = self.get_param("input_text")
        # logger.info("Should translate the input text '%s' here", text)

        src = self.get_param("source_language")
        tgt = self.get_param("target_language")

        translator = GoogleTranslator(source=src, target=tgt)
        result = translator.translate(text)

        # logger.info("Translation to %s:\n\"%s\"", tgt, result)
        sys.stdout.write(f"=> {result}\n")
        sys.stdout.flush()

        return True</sxh>

That is working great already: <sxh bash; highlight: []>kenshin@Saturn /cygdrive/d/Projects/NervHome
$ nvp translate "Salut Manu. Comment ça va?"
=> Hi Manu. How's it going?
</sxh>

===== Adding support for language swapping =====

Now adding support to swap source/target languages: <sxh python; highlight: []>    def translate(self):
        """Translate the input text here."""

        text = self.get_param("input_text")
        # logger.info("Should translate the input text '%s' here", text)

        src = self.get_param("source_language")
        tgt = self.get_param("target_language")
        swap = self.get_param("inverse")
        if swap:
            src, tgt = tgt, src

        translator = GoogleTranslator(source=src, target=tgt)
        result = translator.translate(text)

        # logger.info("Translation to %s:\n\"%s\"", tgt, result)
        sys.stdout.write(f"=> {result}\n")
        sys.stdout.flush()

        return True</sxh>

And the result this time is: <sxh bash; highlight: []>kenshin@Saturn /cygdrive/d/Projects/NervHome
$ nvp translate -i "Hi Manu. How's it going?"
=> Salut Manou. Comment ça va?
</sxh>

Nobody called me "Manou" since a loooonnng time now, but appart from that we are all good already! So that's great 👍!

<note>from there I thought about using the Linguee or Pons translators for single words, but Linguee gives me an exception, and PONS give me a large chunk of mixed english/french text for a single word, so it's a mess from my perspective lol</note>

**Conclusion**: let's just stop here for this first implementation, that's good enough for me and with that I can start testing translating my blog content myself (and then I'll see if I need to extend this tool further or not ;-)