[Question] Allow to change Language

[qin2dim]

[Question] Allow to change Language

[QIN2DIM]

I am trying to edit the tag mapping.I didn't think anyone would pay attention to the project.(/▽＼)

[QIN2DIM]

However, from the conclusion of my research, this i18n compatibility is not an easy job.

On the PC, hCaptcha determines what language to use based on the browser's startup parameters, and on Linux it is based on the session window's locale environment variable.

---

In addition, the tags given by hCaptcha may not be in the standard encoding format.You can run the following code in your favorite Python IDE, which is a very clever trick I found while researching how hCaptcha works.

print("AWESΟME" == "AWESOME")
print("bіcycle" == "bicycle")

So, we need to write at least one adaptor function manually to convert these unusual characters.

[saitishmukhametov]

just change label_aliases and get_label function, since in my (russian) language we have different word endings u have to consider it into label_aliases.
worked perfectly fine for me

        try:
            _label = label_obj.text.split('с ')[1]
        except (AttributeError, IndexError):

        self.label_alias = {
            "велосипедами": "bicycle",
            "поездами": "train",
            "грузовиками": "truck",
            "автобусами": "bus",
            "автобусами": "bus",
            "самолетами": "aeroplane",
            "лодками": "boat",
            "лодками": "boat",
            "машинами": "car",
            "мотоциклами": "motorbike",
        }

[QIN2DIM]

Are you running on Linux? 👀

[saitishmukhametov]

nope

[QIN2DIM]

Yes, this does solve the problem.

But I'm looking for a way to automatically adapt to the language and match the corresponding splitting function to the locale environment variable.This doesn't look easy

[QIN2DIM]

nope

Did you comment out this line of code? 👀

hcaptcha-challenger/src/services/utils/toolbox/toolbox.py

Line 58 in 1bc20e2

options.add_argument("--lang=zh-CN") # 可能仅在 Windows 生效

[saitishmukhametov]

the problem is, that some sites force language on hcaptcha, about the lang stuff: i use only antihcaptcha class
greasyfork hcaptcha solver script solves this problem by uploading target image to wolframimage or something like this, where they get keywords and then compare them with long list of built-in keywords

[QIN2DIM]

stuff

The solution you propose seems to be quite time consuming. Note that the hCaptcha challenge has a time limit of roughly 130 seconds, and the page element will reset after that timeout.

force language

Yes, like EPIC login, they remove the checkbox and lock the element language.

regular

But if it's a regular hCaptcha it will have a section to switch the element language.

From the picture below you should understand why I wanted to unify the language in the first place. If we want to do adaptation, I think we need to rely on other modules, and the complexity of this split word is too terrible.

My idea was to find a way to use elements of a particular language in a variety of contexts. Once and for all.

[QIN2DIM]

I've standardized the challenge language in the test cases and it works on all major operating systems. d6aa5ed

that some sites force language on hcaptcha

Could you provide some application scenarios or links about this situation?

[skrrppop]

Hey what i meant is when changing Language the label doesnt get recognized.
And if u try to translate it with using self.label_alias not all aliases are included or correct
Do you understand what i mean?
And thanks for youre project i learned alot with it

[QIN2DIM]

Hey what i meant is when changing Language the label doesnt get recognized.

So won't it be possible to identify accurately after the language is unified? 👀

And if u try to translate it with using self.label_alias not all aliases are included or correct

I don't quite understand the exact meaning of the phrase.

The purpose of self.label_alias is not to translate, but to correct miscoding.

I did a lot of testing and I found that the number of image categories in hCaptcha is constant and that the tags are not completely random and "infinite".So, with this simple mapping, you can cover all the cases.

At certain fixed times of the day, the challenge labels encountered are the same.

[skrrppop]

I mean when ur getting the label when its in english and than using self.label_alias to get the correspondend chinese word so u can pass it to the image classificator .
It turned out that aeroplane in english is airplane on the label text

[QIN2DIM]

I still don't understand what you mean...🤦‍♂️

I think you may not be reading the purpose of this variable. I have mentioned above. The purpose of self.label_alias is not to translate, but to correct miscoding.

[QIN2DIM]

Oh, my God, there are three people talking in here! Now I know exactly what you need.

[skrrppop]

I still don't understand what you mean...🤦‍♂️

I think you may not be reading the purpose of this variable. I have mentioned above. The purpose of self.label_alias is not to translate, but to correct miscoding.

Ah im an idiot . But when im trying to use an english label for example like 'train' for _label it doesnt work .

Image

)

[QIN2DIM]

WHAT! What did you do with the source code. 😂😂😂😂

[QIN2DIM]

Ultimately, I think it's a pseudo-need. I don't think it should matter what language is used to open the challenge.

If you do have to start a challenge in a particular language for some indescribable reason, you can refer to this friend's approach and write your own splitting methods as well as label alias.

A final reminder. The purpose of self.label_alias is not to translate, but to correct miscoding. So, this layer of mapping must be present even if you open the challenge in English.

[skrrppop]

if i use _label = label_obj.text.split()[-1] i get the correct label but the programm doesnt work and the tactical_retreat function retuns True

[QIN2DIM]

That's because you didn't add the cleansed noun to the self.label_alias.