Getting a script to respond to a word within a phrase ...

[Extraneousnomenclature]

OK, this script works fine when someone says "hello." It says hello right back.
But what if someone says "hello there" or "I just wanted to say hello" or something like that.
How do I change the script so it responds to "hello" embedded in a phrase?

default
{
    state_entry()
    {
        llListen(0,"","","");
    }
    listen(integer channel,string name, key id, string msg)
    {
        if (msg == "hello")
        {
            llSay(0, "hello");
        }
    }
}

 

[Rolig Loon]

Use llSubStringIndex, which is designed to tell you where a particular substring is within a longer string. For example,

integer i = llSubStringIndex(message, "hello");

will tell you where the substring "hello" lies in message.   If it isn't there at all, i will be equal to -1.  So, a test for

if (llSubStringIndex(message,"hello") != -1)

will succeed if "hello" is found and will fail if it isn't.  Remember, of course, that "hello" is not the same as "Hello" or "HELLO", so you'll have to do some filtering to start using the function wisely.  Still, that's the basic idea.

[Extraneousnomenclature]

Thanks!That worked, and I can finally faintly glimpse the logic.I had puzzled over llSubStringIndex in the LSL wiki without being able to figure out how to apply it.And yes, now on to figure out llToLower or whatever that good stuff is that will let me say "Hello "default{    state_entry()    {        llListen(0,"","","");    }    listen(integer channel,string name, key id, string msg)    {       if (llSubStringIndex(msg,"hello") != -1)        {            llSay(0, "hello");        }    }}        

 

[Rolig Loon]

Be daring.  Give it a try. You're doing fine so far.

[steph Arnott]

I not know if this what you intend, but anyone or any object in the sim that says hello on the public_channel will output hello from your script. ADDED: Well depending on the braodcast range that being used. Also you will have to run three passes to cover different upper case and lower case possibilities, else the whole message will have to be converted.

[steph Arnott]

Just out of curiousity is this way better?

default{    state_entry()    {        llListen(0,"","","");    }    listen(integer channel,string name, key id, string msg)    {       if (~llSubStringIndex(llToLower(msg),"hello"))        {            llSay(0, "hello");        }    }}

 

 

[Rolig Loon]

Yes, of course, but I was letting the OP discover that and a number of other things for herself.  She'll probably learn better that way.  She's doing OK so far.  :smileywink:

[Madelaine McMasters]

---

Rolig Loon wrote:

Yes, of course, but I was letting the OP discover that and a number of other things for herself.  She'll probably learn better that way.  She's doing OK so far.  :smileywink:

---

Will Rogers said, "There are three kinds of men. The one that learns by reading. The few who learn by observation. The rest of them have to pee on the electric fence for themselves."

What Will Rogers neglected to mention is that peeing on the electric fence is still the fastest way for readers and observers to learn. And that, with a suitable substitution (in my case it was touching an electric fence with a buttered piece of bread), this applies to women as well.

;-)

[steph Arnott]

I lost that in translation, so have no clue as what that means. As for my version of the code I was just curious.

[Madelaine McMasters]

Steph,

Rolig was suggesting that people learn better by actually trying things, rather than being shown the way. And that's why she was holding back and not showing the OP how to do it.

The famous American humorist Will Rogers made the wry observation that some people may learn only by doing, and that you probably can't show or tell them anything. That was not what Rolig was suggesting, though I think we've all met people who are a little like that, or we have been someone who is a little like that.

You wouldn't have to search very hard to find someone who thinks that trying to teach me is pointless.

;-)

[ChaosRaine]

Can I change the word the script is looking for to a veriable? Like by adding

 

list greeting = hello,hi,hey,morning;

then changing the listen line like this

(~llSubStringIndex(llToLower(msg),greeting))

[Rolig Loon]

No.  You'd need to test each one of your possible greeting words separately, which is going to be slow.  Also, because you are leaving the llListen open, the script will continue to do all of that work whenever anyone says anything within chat range, thus adding a significant chat lag to the sim.  You can lose a lot of friends for doing this.  The only time I could justify using something like this script would be, for example, if you had a password-protected doorway that expected a chat keyword within a few seconds after you step on the doormat.  So, something like ....

 

list gGreeting = ["hello","hi","hey","morning"];integer gLsn;default{    collision_start(integer num)    {        llSetTimerEvent(10.0);        gLsn = llListen(0,"",llDetectedKey(0),"");    }    listen(integer channel,string name, key id, string msg)    {       llListenRemove(gLsn);       llSetTimerEvent(0.0);       integer i;       while(i < llGetListLength(gGreeting))       {           if (~llSubStringIndex(llToLower(msg),llList2String(gGreeting,i)))          {            llSay(0, "hello");          }          ++i;       }    }    timer()    {        llSetTimerEvent(0.0);        llListenRemove(gLsn);    }}

 

 

 

 

[Helium Loon]

There are a few ways to speed up things....

 

All the greetings have different starts.  Do your comparisons on a substring (in this case, the first 2 characters differ in every greeting in the list) so the comparisons execute faster.  Not a lot in this particular case, but for more extensive lists it can add up quickly.

 

Open listens on the public channel are potentially lag monsters.  Do like Rolig says, and only open it when someone gets close by (use a sensor or a collision volume or collision object) then close it after a timeout.  There are potential cases where it will then not work correctly (the avi enters, but doesn't say anything before the timeout.)  A repeat sensor with a long time between refreshes (the same as the timeout of the timer, or a little less) that keeps the listen from being removed if someone is still in range.

 

Or just put a keypad and require entering a code.  Or have a listen on a non-public channel that the users you want to be able to enter are given and only listen on that (and maybe an object they can use that will echo what they type onto that channel when active.)

 

Lots of potential workarounds to avoid lag and poor responsiveness.