I would like to make an RMB bot

[Kethania]

Hi there!

I have a question regarding the script rules. I want to make a bot that would react to certain RMB posts containing a command for the bot and post a response, much like bots on Discord.

As an example of what I have in mind, one of the commands I would like to implement is to compare the policies of two nations. This would look something like this:

To clarify:

• the bot would only respond to commands directed directly at it, I don't want to make it possible to trigger it on accident;
• it would not mention any nations other than the one that activated it;
• it would only respond with a single message per command;
• it would only operate with the region's administration's approval;
• it would comply with all the rate limits (no more than 10 requests per minute, no simultaneous requests);
• the content of the messages would not break the site rules.

I have read the Script Rules and the relevant 'case law' on the forums, and I understand that posting on the RMB is a Restricted Action and as such having a script post anything on the RMB completely automatically is illegal. However, in this case, the bot would be responding to user input -- the message that triggered it! The Script Rules state

We consider a tool to be working "automatically" if it executes a restricted action in any way other than by immediately responding to a user's mouse click (or similar input) at the ratio of one click to one action.

and an RMB message with a command is kind of similar to a mouse click, right? Right?

I don't really feel that I'm doing anything nefarious or going against the spirit of the rules, and I also don't think I'd put much load on NS's servers (I would only send a single POST request per, well, post), but I want extra clarification before doing anything that could angry the mods. I would not mind any additional requirements on the messages the bot would make, such as having all of them have a disclaimer that they were made semi-automatically, or having the nation say 'bot' in the name.

As this was never requested on this forum before (everyone seems to only want to post messages completely automatically), I don't know if anyone other than me wants and would make use of this feature, but it would be pretty cool to have this added to the API as well.

Hoping for a reply from the admin team. Cheers!

[Frisbeeteria]

and an RMB message with a command is kind of similar to a mouse click, right? Right?

The way I interpret that aspect of the rules is that the input is managed by the person running the script. You're handing off script operations to any random player who may or may not want that implementation. That may be one interpretation, but I don't think it was intended as such.

it would comply with all the rate limits (no more than 60 requests per minute, no simultaneous requests);

This tells me that you haven't scrupulously read the script rules, where the limit is clearly set at "no more than 10 requests per minute".

I'm not an admin, but I'd provisionally rule this illegal.

~ Frisbeeteria
Senior Game Moderator

[All Wild Things]

That would be one cool bot!

I think you breach the "one click" rule though. By clicking "Lodge message", you'd be performing two actions. The first action is the usual "Lodge message" action performed by the player. The second action is an additional "Lodge message" action performed by the bot. So although I agree you're within the spirit of the rules, you wouldn't be within the letter of the rules.

If you got past that issue, you'd hit some technical concerns. Imagine a scenario where one bot triggers another bot triggers the first bot etc. (Because other people will want their own bots too, and you have no control over how they are coded). At ten posts per minute, that could get very messy before one of the mods deleted the bot nation.

And you just know that some smart arse is going to create a nation called "Bot Nation not really though compare policies" to see what happens when you compare policies with "Bot Nation not really though"

Do Discord bots have some sort of mechanism to prevent loops like that?

[United Calanworie]

That would be one cool bot!

I think you breach the "one click" rule though. By clicking "Lodge message", you'd be performing two actions. The first action is the usual "Lodge message" action performed by the player. The second action is an additional "Lodge message" action performed by the bot. So although I agree you're within the spirit of the rules, you wouldn't be within the letter of the rules.

If you got past that issue, you'd hit some technical concerns. Imagine a scenario where one bot triggers another bot triggers the first bot etc. (Because other people will want their own bots too, and you have no control over how they are coded). At ten posts per minute, that could get very messy before one of the mods deleted the bot nation.

And you just know that some smart arse is going to create a nation called "Bot Nation not really though compare policies" to see what happens when you compare policies with "Bot Nation not really though"

Do Discord bots have some sort of mechanism to prevent loops like that?

Yes. Per discord.py:

Code: Select all

async def process_commands(self, message: Message) -> None:
        """|coro|
        This function processes the commands that have been registered
        to the bot and other groups. Without this coroutine, none of the
        commands will be triggered.
        By default, this coroutine is called inside the :func:`.on_message`
        event. If you choose to override the :func:`.on_message` event, then
        you should invoke this coroutine as well.
        This is built using other low level tools, and is equivalent to a
        call to :meth:`~.Bot.get_context` followed by a call to :meth:`~.Bot.invoke`.
        This also checks if the message's author is a bot and doesn't
        call :meth:`~.Bot.get_context` or :meth:`~.Bot.invoke` if so.
        Parameters
        -----------
        message: :class:`discord.Message`
            The message to process commands for.
        """
        if message.author.bot:
            return

        ctx = await self.get_context(message)
        await self.invoke(ctx)


If they're a bot, the command processor just ignores it.

[Kethania]

and an RMB message with a command is kind of similar to a mouse click, right? Right?

The way I interpret that aspect of the rules is that the input is managed by the person running the script. You're handing off script operations to any random player who may or may not want that implementation. That may be one interpretation, but I don't think it was intended as such.

Ah, I see, that wasn't my impression on reading the rules, but that could be implied I guess. Thank you for your input, but I still hope for an admin to weigh in. Perhaps additional rules could be put in place to prevent abuse while allowing non-malicious usage.

it would comply with all the rate limits (no more than 60 requests per minute, no simultaneous requests);

This tells me that you haven't scrupulously read the script rules, where the limit is clearly set at "no more than 10 requests per minute".

This was a typo in the post, apologies! Do not worry, the ratelimit is set up correctly in my code.

Imagine a scenario where one bot triggers another bot triggers the first bot etc. (Because other people will want their own bots too, and you have no control over how they are coded). At ten posts per minute, that could get very messy before one of the mods deleted the bot nation.

Yes, I am aware that this could be an issue, this is why I said in the original post

I would not mind any additional requirements on the messages the bot would make, such as having all of them have a disclaimer that they were made semi-automatically, or having the nation say 'bot' in the name.

with the implication that, as NS doesn't have a way to separate bot accounts from user-controlled nations the way Discord does (unless the admins wish to implement it), it could be emulated by, for example

• requiring all bot generated messages to contain a prescribed disclaimer;
• requiring all bots to ignore messages with that disclaimer.

While this is certainly an issue to be aware of, it's not an existential problem.

Also, the way I see it is that these sorts of bots should be operated by the region's administration, or at least with their consent. I'd personally ban anyone doing these sort of shenanigans on my region's RMB without asking me first :p

[United Calanworie]

I would definitely like to see this be made possible/get an admin ruling on it, because I can see so many good ways to benefit regional integration with automatic welcomes on the RMB, or a “dispatch watch,” etc.

Lots of potential for good, but I also can see potential for abuse.

[Flanderlion]

I think if bots were able to post on the RMB, it should be both via an API and only from nations with comms authority in the region. I personally am still leaning against, as like, I don't want to see dead regions with only bot posts on their RMBs.

[[violet]]

Interesting! I think this is worth pursuing.

There are a few challenges, which you've highlighted. A concern for me is also that we don't currently offer a way for you to subscribe to events -- you can't just ask to be pinged whenever there's a new message on an RMB you're tracking. Instead, you'd need to keep polling the RMB to see if there's anything new. You'll no doubt want your bot to be responsive, so I wonder whether you're going to wind up polling a large number of regions every second or two, 99% of which is pulling the same messages over and over, just to listen for new messages. We might be able to offer a listener-based API to solve this, though.

I think I'd also want to create an RMB posting API, just to keep automated messaging separate. This would let us do things like automatic throttling and tagging of all automated messages, which would be better than hoping bot authors remember to do it and chasing after them when they don't.

In general, though, I think it would be a worthy experiment, and we could at least try it out for a while and see how it goes.

[Kethania]

Interesting! I think this is worth pursuing.
...
In general, though, I think it would be a worthy experiment, and we could at least try it out for a while and see how it goes.

Thank you! I shall make the first version shortly.

There are a few challenges, which you've highlighted. A concern for me is also that we don't currently offer a way for you to subscribe to events -- you can't just ask to be pinged whenever there's a new message on an RMB you're tracking. Instead, you'd need to keep polling the RMB to see if there's anything new. You'll no doubt want your bot to be responsive, so I wonder whether you're going to wind up polling a large number of regions every second or two, 99% of which is pulling the same messages over and over, just to listen for new messages. We might be able to offer a listener-based API to solve this, though.

At the moment, I'm actually polling the happenings (with a period of 30 seconds). This avoids the issue of having to poll several regions separately (in fact, I only have a single polling loop in my application for many different features), but is still far from ideal. A subscription (websocket-based, perhaps) API would definitely be greatly appreciated, though I assume implementing it would be a lot of work for you, with all the legacy code and all.

I think I'd also want to create an RMB posting API, just to keep automated messaging separate. This would let us do things like automatic throttling and tagging of all automated messages, which would be better than hoping bot authors remember to do it and chasing after them when they don't.

Thank you for considering this, I will track the forums and switch my code to use the API as soon as it's available.

[Flanderlion]

Interesting! I think this is worth pursuing.

There are a few challenges, which you've highlighted. A concern for me is also that we don't currently offer a way for you to subscribe to events -- you can't just ask to be pinged whenever there's a new message on an RMB you're tracking. Instead, you'd need to keep polling the RMB to see if there's anything new. You'll no doubt want your bot to be responsive, so I wonder whether you're going to wind up polling a large number of regions every second or two, 99% of which is pulling the same messages over and over, just to listen for new messages. We might be able to offer a listener-based API to solve this, though.

I think I'd also want to create an RMB posting API, just to keep automated messaging separate. This would let us do things like automatic throttling and tagging of all automated messages, which would be better than hoping bot authors remember to do it and chasing after them when they don't.

In general, though, I think it would be a worthy experiment, and we could at least try it out for a while and see how it goes.

In OPs case, they have a private shard for the nation that they could be polling if the user is going to tag the nation each time.

So posting by bot in any form would still be prohibited, just there would be an API created for it which would make it legal if done through that?

Would the nation require comms authority in the region they were in? Would they be able to post on embassy regions? Would any member of a GCR/UCR be able to post automatic posts on their own RMB? - I'm hoping for yes, no and no.

[Kethania]

The provisional version of the bot is live! Currently it only has a single command, but I do plan on developing it further.

[Merni]

I have never used NS RSS feeds, but wouldn't it be possible to use the private RSS feed (notices) (https://www.nationstates.net/page=subscribe) to find when the bot is mentioned in an RMB message? To my knowledge (/notamod) this wouldn't be subject to the API rate limit.

Edit: If it wasn't clear enough, this was in response to:

There are a few challenges, which you've highlighted. A concern for me is also that we don't currently offer a way for you to subscribe to events -- you can't just ask to be pinged whenever there's a new message on an RMB you're tracking. Instead, you'd need to keep polling the RMB to see if there's anything new. You'll no doubt want your bot to be responsive, so I wonder whether you're going to wind up polling a large number of regions every second or two, 99% of which is pulling the same messages over and over, just to listen for new messages. We might be able to offer a listener-based API to solve this, though.

Of course, RSS is not exactly this -- it still requires regularly polling the feed -- but I would hope that would be easier on the server than an API query.

[All Wild Things]

The provisional version of the bot is live! Currently it only has a single command, but I do plan on developing it further.

Fantastic! Well done!

[Homyland]

Does it work on any region, or just in TCB?

[Luna Amore]

Does it work on any region, or just in TCB?

Looks like it’s just TCB right now.

[[violet]]

I have never used NS RSS feeds, but wouldn't it be possible to use the private RSS feed (notices) (https://www.nationstates.net/page=subscribe) to find when the bot is mentioned in an RMB message?

This sounds like a great idea. The bot should use the "notices" Private Shard, which is more efficient that RSS but offers access to the same info.

[Kethania]

In OPs case, they have a private shard for the nation that they could be polling if the user is going to tag the nation each time.

This sounds like a great idea. The bot should use the "notices" Private Shard, which is more efficient that RSS but offers access to the same info.

Ah, yes, thank you. For now I will continue to rely on happenings (as I need them for other parts of my application anyway), but will keep this in mind for the future.

Does it work on any region, or just in TCB?

Looks like it’s just TCB right now.

Well, I can enable it for some of the regions TCB has embassies with, so if one.. were to... construct embassies with TCB .. aha ha, just kidding.. unless..?

[[violet]]

As a trial, I've added a Nation Private Command API for making RMB posts.

It's currently undocumented, but follow the same process as for lodging a dispatch:

https://www.nationstates.net/pages/api. ... tecommands

You'll need to send "c=rmbpost", plus these parameters: nation, region, text, mode

E.g.

Code: Select all

curl -H "X-Password: hunter2" -A "UserAgent Example" "https://www.nationstates.net/cgi-bin/api.cgi" --data "nation=testlandia&region=testregionia&c=rmbpost&text=Hello%20there.&mode=prepare"

curl -H "X-Pin: 1234567890" -A "UserAgent Example" "https://www.nationstates.net/cgi-bin/api.cgi" --data "nation=testlandia&region=testregionia&c=rmbpost&text=Hello%20there.&mode=execute&token=1234567890abcdefg"

This will automatically add a signature that says "This is an automated message," and apply some stricter flood control.

Can you please see if you can get this working? I would like all bots to operate via the API.

[August]

-snip-

Not the OP, but I just tested this and it works like a charm.

A couple notes for the curious:
1. The "This is an automated message" tagline is part of the message and can be manually edited out afterward--not that it matters, because that that point it is no longer an automated message.
2. It works through embassies, assuming the destination region has enabled embassy posting.
3. Character limits still apply.
4. Upon a success, you can look for the following XML in the response if you want the link to the new post.

Code: Select all

<NATION id="the_moose">
<SUCCESS>Your message has been lodged! &lt;a href="/region=august/page=display_region_rmb?postid=44913308#p44913308"&gt;&lt;span class="smalltext"&gt;View your post.&lt;/span&gt;&lt;/a&gt;</SUCCESS>
</NATION>

5. Upon a failure, you can look for the following XML in the response.

Code: Select all

<NATION id="the_moose">
<ERROR>Oops! The server experienced an error lodging your message.</ERROR>
</NATION>

Will edits be permitted in the future?

[United Calanworie]

As a trial, I've added a Nation Private Command API for making RMB posts.

It's currently undocumented, but follow the same process as for lodging a dispatch:

https://www.nationstates.net/pages/api. ... tecommands

You'll need to send "c=rmbpost", plus these parameters: nation, region, text, mode

E.g.

Code: Select all

curl -H "X-Password: hunter2" -A "UserAgent Example" "https://www.nationstates.net/cgi-bin/api.cgi" --data "nation=testlandia&region=testregionia&c=rmbpost&text=Hello%20there.&mode=prepare"

curl -H "X-Pin: 1234567890" -A "UserAgent Example" "https://www.nationstates.net/cgi-bin/api.cgi" --data "nation=testlandia&region=testregionia&c=rmbpost&text=Hello%20there.&mode=execute&token=1234567890abcdefg"

This will automatically add a signature that says "This is an automated message," and apply some stricter flood control.

Can you please see if you can get this working? I would like all bots to operate via the API.

Just tested this over in Karma, it works well! Small note on the response, I'm not sure the success message is super useful? Getting

Code: Select all

<NATION id="united_calanworie">
<SUCCESS>Your message has been lodged! &lt;a href="/region=karma/page=display_region_rmb?postid=44913378#p44913378"&gt;&lt;span class="smalltext"&gt;View your post.&lt;/span&gt;&lt;/a&gt;</SUCCESS>
</NATION>


as a response is usable, but to be honest I would prefer a 0/1 response like the verify shard.

[Flanderlion]

Looks like you both are a lot faster than me (First time using tokens). I also tried it, is cool, same results as above. So it is intended that nations without comms authority can post via bots on the RMB? I guess any region bar TRR can just ban nations that they don't want.

[United Calanworie]

So it is intended that nations without comms authority can post via bots on the RMB? I guess any region bar TRR can just ban nations that they don't want.

I'd assume it's intended behavior, yes.

[Flanderlion]

So it is intended that nations without comms authority can post via bots on the RMB? I guess any region bar TRR can just ban nations that they don't want.

I'd assume it's intended behavior, yes.

I think regions should be allowed to choose whether they want to accept bot posts via embassy partners, bot posts from residents and ideally whitelist - essentially like how they can do with recruiters. Normal RMB posts and bot posts are quite different, and while we've been discussing offsite and really like the feature (going to be great for retention programmes), it feels like it is ripe for abuse in the current state.

Edit: Like the recruiter settings - just changing some of the options to embassy regions, embassy region comms officers - as API/regional admin and probably even officers could be dropped from the list.

[Homyland]

In OPs case, they have a private shard for the nation that they could be polling if the user is going to tag the nation each time.

This sounds like a great idea. The bot should use the "notices" Private Shard, which is more efficient that RSS but offers access to the same info.

Ah, yes, thank you. For now I will continue to rely on happenings (as I need them for other parts of my application anyway), but will keep this in mind for the future.

Looks like it’s just TCB right now.

Well, I can enable it for some of the regions TCB has embassies with, so if one.. were to... construct embassies with TCB .. aha ha, just kidding.. unless..?

You guys should apply for embassies with TNP.

[Doge Land]

I always get this error during the execute phase:

Code: Select all

<NATION id="doge_land">
<ERROR>Invalid token in execute mode. Please try again.</ERROR>
</NATION>

I don't know why. I've made sure I've copied and pasted the right token into the token field but still, nothing is working and I keep getting this error.