[SOLVED] RSS Guard can localize feeds

[martinrotter]

@StealerCook - if you are interested, ping me in this discussion.

[StealerCook]

Hey Hi, Is this about Feed translation? Sry i don't remember starting thread here...

[martinrotter]

Sure, no prob. It is relatively advanced functionality. Some knowledge is needed - installing Python, adding "exe" to "PATH" variable, launching executables on command line (to test the script) etc., but it is all needed because localization of feeds is not directly implemented as GUI option in any feed reader I know and only some feed readers offer reasonable "work-around" - the only I know are RSS Guard (of course), Liferea and probably newsboat.

Thing is that once you get the grasp of the "scraping" feature, you can do crazy magic like filtering out advertisements, CSS/HTML formatting or you can even generate feeds for any remote or local data source.

[martinrotter]

Here is a screenshot of localized feed I tested right now. Feed is localized from english to portuguese - notice the "Post-process script" box where the script is called with parameters "en" (english) "pt_BR" (portuguese):

[martinrotter]

[martinrotter]

Also, it is probably best if you use latest "development build" - https://github.com/martinrotter/rssguard/releases/tag/devbuild

[StealerCook]

Thankyou. I can manage with python settings and thanks for your guide. It really has cool features.

[StealerCook]

@martinrotter just tried this out. But I get this error. Did I miss anything?

[martinrotter]

@StealerCook what is py -3 in the beginning of the post-process script line? it should be something like C:\path\to\python.exe or just python if it is in your PATH. Also, it is good to launch rssguard with --log switch (see --help for more info) and generate debug log and investigate the log as it may contain more info

[StealerCook]

Hey Thankyou. I thought it would work. I tried with the python path, now it isn't providing any error. But i don't see the translated feed. In the left side you can still see feeds in English.

[martinrotter]

what happens when you push "Fetch it now" button? do you get any error message?

[StealerCook]

No Error. Just shows "All metadata fetched successfully."

[martinrotter]

Just tested with your feed. I had to increase feed download timeout in settings, then:

Notice that title of feed is now translated after "Fetch it now". Also, do you have all dependecies installed? They are listed in the script file - read it.

[martinrotter]

Also, you can test the script from command line to get better info if it actually works.

[StealerCook]

Yes, downloaded all the dependencies. How do you change download timeout, do you mean that feed connection timeout?
I tried with command line too and that provided me this.

[martinrotter]

btw, your screenshot from rssguard is promising, it failed at the end of script,

also, you use powershell? if so then your command is wrong, it must be:
curl https://www.reddit.com/r/TOR.rss | Select-Object -ExpandProperty Content | python.exe .\translate-feed.py en pt true

[martinrotter]

one more discovered thing, if you try many times, reddit returns HTML (not XML) saying that you are a robot etc

[martinrotter]

i uploaded new version of the script! please try it.

https://github.com/martinrotter/rssguard/blob/master/resources/scripts/scrapers/translate-feed.py

[StealerCook]

Thankyou for updating the script. I just ran below command in PowerShell. No Error. But doesn't translate.

curl https://www.reddit.com/r/TOR.rss | Select-Object -ExpandProperty Content | python feed_translate.py en pt true

[martinrotter]

ok, there is a function "translate_string" in

https://github.com/martinrotter/rssguard/blob/master/resources/scripts/scrapers/translate-feed.py#L58

tweak it so that it does not enclose its inner calls in try block:

def translate_string(to_translate):
    if to_translate is None:
      return to_translate

    translated_text = translator.translate(to_translate, src = lang_from, dest = lang_to)

    if not parallel:
      time.sleep(0.2)

    return translated_text.text

then try again and you should see the actual errors

I will repeat one thing. Those scripts in the repository are semi-DIY and I completely know that they will not work out-of-box for everyone and adjustments might have to be made.