Perspective API

tankerkiller125

Update 0.1.5

Try/Catch block surrounds the API call, should prevent posting comments/topics from failing if misconfiguration or error

Changelog

Updated Copyright info in files
API Calls now in Try/Catch block
Enabling experimental models now disables the regular counterparts (resolves issues with some languages supported in experimental but not in regular models.

tankerkiller125

Update 0.1.6

Changelog

Fixes issue introduced in 0.1.5
Removes automatic dis-approvals (very glitch) automatic flagging is now the only part active

OrdinaryJellyfish

I'm trying this extension out on a forum, but it doesn't seem to be working.

I have my threshold set as 10 (for testing), and every model and use max value enabled. I put the content in perspectiveapi.com and it gave a score of 0.99, so shouldn't it be flagged? When I post it then log into the admin account, there aren't any flags. No errors are being thrown either...I can't figure it out.

EDIT: now I'm getting errors...they're all complaining about parameters not being arrays.

First line:
PHP Notice

tankerkiller125

OrdinaryJellyfish I don't have the try/catch block extended far enough it would seem. This means that the API request is failing and in doing so is failing to set the $scores variable.

I've just tagged a new release that resolves this issue.

OrdinaryJellyfish

tankerkiller125 now i'm just confused 😂

PHP Error

tankerkiller125

Update 0.1.7

Changelog

Resolves issue with Try/Catch block
Now logs errors properly

tankerkiller125

OrdinaryJellyfish That would be expected, the experimental models are only there for the languages that aren't supported by the non-experimental versions. https://github.com/conversationai/perspectiveapi/blob/master/2-api/models.md#languages-in-production has a full listing of what languages are supported by what models. Generally speaking if your an English forum use the non-experimental.

OrdinaryJellyfish

tankerkiller125 yeah I tried disabling experimental but it doesn't flag anymore with no errors. Not being violent enough in my testing?

tankerkiller125

OrdinaryJellyfish so long as use Max is enabled you should be able to just drop the f bomb and it should be flagged. (If the profanity filter is on)

OrdinaryJellyfish

tankerkiller125 max is enabled, along with profanity.

tankerkiller125

Update 0.1.8

Changelog

Allow installation on beta 12 installs and upgrades
Translation strings have moved to be prefixed to tank-perspective, all language packs will need to be updated.

Littlegolden

tankerkiller125 I think this is not the language pack's fault?

rob006

Littlegolden Are you sure you have 0.1.8 version installed?

tankerkiller125

Update 0.1.9

Changelog

Fixed translation string (thanks Littlegolden)

Littlegolden

rob006 Yeah, it's the neweast version. Not only Chinese interface, but also English interface.

tankerkiller125

Littlegolden Clear cache, translation strings were working on dev environment.

Littlegolden

tankerkiller125 I did clear cache.... but nothing changed... and another forum has the same situation with me.

tankerkiller125

Littlegolden I'll take a look later today, it's possible I forgot to compile the JS for the extension

Littlegolden

tankerkiller125 Thank you 🎁

danielunited

Great work on the extension! Is there an option to automatically hide flagged posts, and not show them to other community members (until there's a moderator approval)?

tankerkiller125

danielunited Originally this extension did exactly that, however there were major bugs in the approval process (notably you couldn't approve them) so i removed that feature. Maybe when I have time I'll add it back in and properly diagnose the issue.

« Previous Page Next Page »