Most effective way to match commands when developing a chatbot?

Question

I'm creating a simple chatbot through the Messenger Platform API and I'm pretty much stuck on how to effectively recognise a set of commands that the bot can react to. I'm currently using a switch statement to detect commands that begin with an exclamation mark (e.g. !showlist; !additem <single/set of parameter(s)>).

Here is what I currently have:

switch(true){
            case stristr($receivedMsg,'!additem'):
....
}

At any matching stage the code, either execute a set of statements or extrapolate the eventual parameters first and then executes some statements with them.

The issues I'm having with the above setup are the following:

• in case of no parameters commands it is possible to get the related code to execute even if the command is misspelled. E.g. !additem#$%% will still invoke the actual command's code in the switch statement.

• in case of commands that take parameters, when retrieving those parameters with say this statement:

  $item=str_replace('!additem', '', $receivedMsg);
  

  it is very easy to include unwanted text in the parameters; you may deal with spaces with trim() or imply there will always be a space and edit the above statement to include it in the function. E.g.

  $item=str_replace('!additem ', '', $receivedMsg);
  

  but this makes other problem arise when trying to separate the command from the params.

I'm aware that a solution could be hardcoding with systematic string manipulation functions but that doesn't seem correct to me. What do people do in this situation? Isn't there a specific way to exactly match commands and securely spot eventual users' typos?

Answer 1

You didn't work with Regular Expressions in your own solution but tagged it correctly. By stristr() function I found you are not looking for more coming commands so I applied the same logic onto RegEx:

$msg = 'Gonna add it !additem param1,param2';
preg_match('~(!\S+)\s*(.*)~', $msg, $match);
$command = $match[1];
$parameters = preg_split('~\s*,\s*~', $match[2]);

I tried to do it a one-liner but later thought this would be much more cleaner. BTW I wonder about the switch statement.

RegEx Breakdown:

~   # regex delimiter
    (   # Start of Capturing Group (1)
        !\S+    # Match non-space characters that start with !
    )   # End of CG1
    \s* # Any number of white-sapce characters
    (   # Start of CG2
        .*  # Match up to end
    )   # End of CG2
~   # regex delimiter

preg_split too receives a regex as its first argument and tries to split on it, almost a explode with regex. \s*,\s* means a comma that may be enclosed in any number of spaces.

Answer 2

if ($receivedMsg[0] == '!')
    switch (strtolower(substr(reset(explode(' ', $receivedMsg)), 1)))
    {
        case 'additem':
            // do this
            break;
        case 'delitem':
            // do that
            break;
        default:
            echo 'Command not recognized.';
    }

Well that's one way to do it. You can also declare an array with the functions that handle each command, example:

$handles = [
    'additem' = function ($args) { /* add something */ },
    'delitem' = function ($args) { /* del something */ },
    // ...
];

if ($receivedMsg[0] == '!')
{
    $args = explode(' ', $receivedMsg);
    $cmd  = strtolower(substr($args[0], 1));
    if (isset($handles[$cmd]))
        $handles[$cmd]($args);
    else
        echo 'Command not recognized.';
}

Answer 3

Basing my solution on the answers that @Havenhard and @revo provided, I've written the following solution that perfectly works for me:

$this->senderId = $messaging['sender']['id'];
$command_handlers = [
        'additem' => "addItemCommand",
        'showlist' => "showListCommand",
        'rngroup' => "renameGroupCommand"
    ];

    $actionCompletedOrh = new OutRequestHandler($this->senderId);

    if(!empty($messaging['message']) && empty($messaging['message']['quick_reply'])){
        $receivedMsg = $messaging['message']['text'];
        $replyMsg = "";
        $this->performSenderAction(0);
        //isCommand uses this regex to perform the evaluation 
        //(^!\w+\s+([\w,\s]*$)|^!\w+$)"
        if($this->isCommand($receivedMsg)){
            //regex matching to get params in raw form
            preg_match("~(^!\w+\s+([\w,\s]*$)|^!\w+$)~",$receivedMsg,$match);
            //regex matching to get the command
            preg_match("~^!\w+~",$match[0],$_match); 
            $command = strtolower($_match[0]);
            $params = null; 
            if(count($match)>2){
                //the function below uses preg_split as in @revo's example
                $params = $this->getCommandParams($match[2]);
            }
            if(array_key_exists(substr($command,1), $command_handlers)){
                $func = $command_handlers[substr($command,1)];
                $replyMsg=$this->$func($params,$connection);
            }
            else{
                $replyMsg=$this->getPromptMessage("cmer1");
            }
        }
        else{
            //All other messages - possibly processed with NPL
        }
        $this->performSenderAction(2);
        $replyMsg = json_encode($replyMsg);
        $actionCompletedOrh->sendJustTextMessage($replyMsg,$access_token);
    }

Do you see anything I could improve? Please let me know what and why in the comments!