_Icicle-Cube_ wrote:

PGBFLITE6373 wrote:

banana439monkey wrote:

medians wrote:

(#20)
It's because of the DjangoBB filter, they can't easily change it.

…yes they can

Banana

Why haven't they uncensored crap and damn, or censored discord? ☠️ (or remove filerbypasses like I just did)

Ok, maybe. I think the problem is that this isn't commonly used as a filter bypass. I, therefore, suggest that the filter bot would instead convert it to regular letters and then censor it. That was if it was used in an ok manner then they wouldn't be muted. That way, I could use them in a regular manner.

the forum filter has a different filtering system.

and it would be a huge pain to add that to the forums

OP: We don't want to hijack the topic, but if this were to be the main point, there would be far more support since this could benefit more people.

ajskateboarder wrote:

banana439monkey wrote:

(#42)

ajskateboarder wrote:

roofogato wrote:

(#36)
i feel this would cause troubles with languages that have the lil symbols above their letters, like vietnamese

That's what I also said, but the filter doesn't even deal with languages besides English, so there's no harm in removing these symbols to filter-check.

you can't just ban those symbols!
the point is to make scratch ACCESSIBLE TO ALL

I don't want the symbols to be banned (banning a ton of langs == bad), just involve another step where the filterbot checks for curses in the glitchy text.

As PBGFLITE said, the “glitchy” Unicode symbols are to be stripped from the content, then checked with the filterbot assuming it's English. It would try regardless of what language the text is written in.

This is relatively easy to check, just convert the content to Unicode:

ä̸̛̰̙́͝ -> "\x61\xcc\xb8\xcc\x9b\xcd\x84\xcd\x9d\xcc\xb0\xcc\x99"

Check for any sequences starting with \x{0-9} and remove everything except for them. This should leave you with plain ASCII to check with the filterbot for English swearing.

ok but then in some languages, accents can be used to denote other words, sometimes somewhat inappropriate for scratch

Banana

banana439monkey wrote:

ajskateboarder wrote:

banana439monkey wrote:

(#42)

ajskateboarder wrote:

roofogato wrote:

(#36)
i feel this would cause troubles with languages that have the lil symbols above their letters, like vietnamese

That's what I also said, but the filter doesn't even deal with languages besides English, so there's no harm in removing these symbols to filter-check.

you can't just ban those symbols!
the point is to make scratch ACCESSIBLE TO ALL

I don't want the symbols to be banned (banning a ton of langs == bad), just involve another step where the filterbot checks for curses in the glitchy text.

As PBGFLITE said, the “glitchy” Unicode symbols are to be stripped from the content, then checked with the filterbot assuming it's English. It would try regardless of what language the text is written in.

This is relatively easy to check, just convert the content to Unicode:

ä̸̛̰̙́͝ -> "\x61\xcc\xb8\xcc\x9b\xcd\x84\xcd\x9d\xcc\xb0\xcc\x99"

Check for any sequences starting with \x{0-9} and remove everything except for them. This should leave you with plain ASCII to check with the filterbot for English swearing.

ok but then in some languages, accents can be used to denote other words, sometimes somewhat inappropriate for scratch

Banana

yea, in gaelic, there are many words that seem offensive in English, but because of the accent mark, are no longer offensive.

roofogato wrote:

yea, in gaelic, there are many words that seem offensive in English, but because of the accent mark, are no longer offensive.

The filterbot is not equipped to handle languages besides English (and should go in its own suggestion), so what differences would be made anyway?

ajskateboarder wrote:

(#50)

roofogato wrote:

yea, in gaelic, there are many words that seem offensive in English, but because of the accent mark, are no longer offensive.

The filterbot is not equipped to handle languages besides English (and should go in its own suggestion), so what differences would be made anyway?

that's not nearly the point!
the filter is very equipped to handle other languages than english (it's literally documented in cleanspeak's documentation), and even if it wasn't, the nature of scratch is that it's meant to be free and accessible to all
regardless of who you are, the languages you speak or don't speak, race, sexual orientation, how computer-literate you are, your financial status
in scratch, A11Y MATTERS, as does making sure that languages do not get unintentionally blocked off!

Banana

Those characters are actual combining characters that can be used for non-zalgo purposes. Not allowing them could censor a lot of fine things, for example in the conlanging studio.

okey dokey thanks to @ajskateboarder the post is now more agreeable(i hope): check glitch/zenith font for bad words

among_us1w2 wrote:

okey dokey thanks to @ajskateboarder the post is now more agreeable(i hope): check glitch/zenith font for bad words

for just bad words? too many combinations of symbols to add to the filter

roofogato wrote:

among_us1w2 wrote:

okey dokey thanks to @ajskateboarder the post is now more agreeable(i hope): check glitch/zenith font for bad words

for just bad words? too many combinations of symbols to add to the filter

No, the filter removes the symbols (not visually, just when filtering) and checks for bad words hidden behind the symbols.
It would take trillions of years to filter every combination of symbols.

An0therRand0mC0der wrote:

roofogato wrote:

among_us1w2 wrote:

okey dokey thanks to @ajskateboarder the post is now more agreeable(i hope): check glitch/zenith font for bad words

for just bad words? too many combinations of symbols to add to the filter

No, the filter removes the symbols (not visually, just when filtering) and checks for bad words hidden behind the symbols.
It would take trillions of years to filter every combination of symbols.

again, other languages.

how would it tell the difference between just an accent and a “glitch” font?

like the term for “leave a” in gaelic has what LOOKS like a slur in english, but isn't in gaelic.

so you couldn't say “leave a comment on my page!” in gaelic if this was added

banana439monkey wrote:

(#51)

ajskateboarder wrote:

(#50)

roofogato wrote:

yea, in gaelic, there are many words that seem offensive in English, but because of the accent mark, are no longer offensive.

The filterbot is not equipped to handle languages besides English (and should go in its own suggestion), so what differences would be made anyway?

that's not nearly the point!
the filter is very equipped to handle other languages than english (it's literally documented in cleanspeak's documentation), and even if it wasn't, the nature of scratch is that it's meant to be free and accessible to all

Is that true? Huh I guess I change my mind (again).

If the filter checks different languages, this would lead to false-negatives when dealing with swearing in other languages because removing accents (and entire non-ASCII letters) from text changes its meaning. Swearing in German could easily be done since the accents would be removed, changing its meaning, thus not making it swearing.

Sorry among_us, but no support.

ajskateboarder wrote:

(#57)

banana439monkey wrote:

(#51)

ajskateboarder wrote:

(#50)

roofogato wrote:

yea, in gaelic, there are many words that seem offensive in English, but because of the accent mark, are no longer offensive.

The filterbot is not equipped to handle languages besides English (and should go in its own suggestion), so what differences would be made anyway?

that's not nearly the point!
the filter is very equipped to handle other languages than english (it's literally documented in cleanspeak's documentation), and even if it wasn't, the nature of scratch is that it's meant to be free and accessible to all

Is that true? Huh I guess I change my mind (again).

If the filter checks different languages, this would lead to false-negatives when dealing with swearing in other languages because removing accents (and entire non-ASCII letters) from text changes its meaning. Swearing in German could easily be done since the accents would be removed, changing its meaning, thus not making it swearing.

Sorry among_us, but no support.

so
i think it is configured to check locales but the cleanspeak documentation definitely does document that you can check based on locales: https://cleanspeak.com/docs/3.x/tech/tutorials/filtering-phrases

Banana

Here is a solution, what if we just add ok words to the filter, as we do for fine words that have bad words in them (such as skyscraper)?

I think this will do more harm. The issue is that if mobile users will have problems commenting using glitch text or anything. And possibly the glitches filtering out. The solution was to detect the language we are using.

PGBFLITE6373 wrote:

Here is a solution, what if we just add ok words to the filter, as we do for fine words that have bad words in them (such as skyscraper)?

that's janky and that is not at all how cleanspeak works

Banana

cs3868895 wrote:

I think this will do more harm. The issue is that if mobile users will have problems commenting using glitch text or anything. And possibly the glitches filtering out. The solution was to detect the language we are using.

Ḭ̸͌͛̓͑͂͜͠͝ͅ ̵̧͎͙̞̝͊͘h̵̡̙̤̗̳̯͖̊͂̈́̏̾͠o̵̧͇̜̽̅̔̇̇͂̏̈́̍̌̉̅̌̚͘ṕ̷͖̩̥e̵̢̘̗͒̐͋̎̕ ̶̨̰̠̱̩͔͈̦͕̫̔̐̿͊̚͘Ĭ̶̦̫͎̣͉͋̎̅̆̉͘͠͠͝ ̶̧̨̡̪̥̯͖̰̮̟͔͈͒̆̇͂͗͋̇̃̄͂̿̚͝ͅh̷̖̹̓̾̈́̊͋̆̾̉̈̂͘̕͠͠è̷̡̟̼͎̜͗̎͆̽͒͒͠͝l̷̢̧̢̛̛̖̘̺̰̱̞̙̬̱̯̹̜̭̎͂̒̑̂̓̃͊̊́͋̈́̈̌͘ͅp̵̛̣̫̮̱̻͆̐̃̃̂̌̋̋̂ē̴͎̩̺̗͊͌̇d̴͔̰̭̒̾̽̊̍̽͒̊̐̉ ̶̧̛͚͚̠͕͉̟̩̖̜͋̒̃̈́́̆̓͜ͅy̸̢̜̮̳̣̙̹̦͈̱͎͚͔͈̻̱͊̈̽̋̄͒́̃̏͘o̸̢̦̯̼͕͔͓̟̫̘̟̙͎̙̠͆̈̌̊̑̍͒̿͛͊͑̎̀̕̚͝͝ų̶̧̧̳̙̤͎̭͙̂̐̓̒̉͗͘.̶̢̩̱̺̼̲̖̮̜̣̩͌͜͝


if you want to, you can quote to see if there's hidden swears idk

agreed.

banana439monkey wrote:

PGBFLITE6373 wrote:

Here is a solution, what if we just add ok words to the filter, as we do for fine words that have bad words in them (such as skyscraper)?

that's janky and that is not at all how cleanspeak works

Banana

then how does ST fix filter issues?

Cleanspeak is already janky. We could just make levels for it. If it is a bad word? mute unless its the lanagues fault. if it looks like a bad word. no mute.

PGBFLITE6373 wrote:

banana439monkey wrote:

PGBFLITE6373 wrote:

Here is a solution, what if we just add ok words to the filter, as we do for fine words that have bad words in them (such as skyscraper)?

that's janky and that is not at all how cleanspeak works

Banana

then how does ST fix filter issues?

you can go ask the questions about scratch about that if you want to yk

PGBFLITE6373 wrote:

(#63)

banana439monkey wrote:

PGBFLITE6373 wrote:

Here is a solution, what if we just add ok words to the filter, as we do for fine words that have bad words in them (such as skyscraper)?

that's janky and that is not at all how cleanspeak works

Banana

then how does ST fix filter issues?

They add words to a black list based on what people report content for and also complaints from its dedicated suggestion (can't find the link for it)