MTW is not translating

[rohankatyal29]

I can't seem to get MTW to translate. I am using yandex for the translations.
Here are the relevant screenshots:

[ceilican]

You must must select the translation configuration that you want to use. Your third screenshot shows that you selected "do not trans" late.That is why it is not translating.

[ceilican]

Anyway, the GUI could be improved to make it clear that a selection must be made.

[rohankatyal29]

@ceilican: I have activated the translation, but I still can't see any change on the page

[ceilican]

The latest stable version from the Chrome Web Store works for me.

[ceilican]

I won't have time to investigate this issue deeply now.

Note that, if the page is long, MTW takes a long time to display the translations.

[alexklibisz]

I've noticed this as an inconsistency between sites. For example, Wikipedia appears to work fine with a Yandex key, but Hacker News (news.ycombinator.com) does not.

Working on Wikipedia: http://i.imgur.com/5W14c8i.gif
Non-working on Hacker News: http://i.imgur.com/lkb2Wyj.gif

I started debugging by looking at the function requestTranslations on line 21 in MindTheWord.js. After some debugging, it looks like Hacker News takes about 10 seconds to hit that line, and when it does, the sourceWords object is empty. In the image below I've added a debugger statement on line 24 in MindTheWord.js. Notice the console statements showing that sourceWords is empty:

I'm a potential GSOC student, so I'm going to take some time now to look at the GSOC puzzles before moving back to this issue, but hopefully this information can get someone moving in the right direction.

[ankit-m]

@alexklibisz The source words are empty because there are no <p> tags in the hacker news page. If you look at line 217 of MindTheWord.js you'll see that getAllWords() is searching by p tags.

On the other hand, if you see the comments page of YC, you will see translations

But we can definitely look for other ways to implement it.

[ankit-m]

A quick search got me this. You can get all the text in a document with the following code

var body = document.body;
var textContent = body.textContent || body.innerText;
console.log(textContent); 

I tested it now, it does work. But speed is still an issue.

[alexklibisz]

@ankit-m I noticed that only p tags are considered from another issue I worked on just now. Good catch. Is this a considerable speed issue? Is it taking noticeably longer to handle?

[ceilican]

@ankit-m @rohankatyal29, Was this issue solved?