Prompt class
Base class for built-in prompts and can be used to build new custom prompts. The Prompt class
provides the basic logic to prompt/re-prompt a user and provides a set of extensible hooks to
customize the prompts recognition of the users reply as well as the output sent to the user.
Prompts should always have at least one onRecognize() handler registered and
they support adding any number of matches() or matchesAny() handlers
which can be used to add special user initiated commands to the prompt. For instance, the
built-in prompts add a matches('BotBuilder.RepeatIntent')
to listen for a user to ask to
have a prompt repeated, causing the prompt to send its initial prompt again.
- Extends
Constructors
Prompt(IPrompt |
Creates a new customizable instance of the prompt. Your new prompt should be added as a dialog to either a bot or library. |
Properties
features | The prompts current configured set of features. |
Methods
add |
Called once for each dialog within a library to give the dialog a chance to add its
|
begin<T>(Session, T) | Called when a new dialog session is being started. |
begin |
Binds an action to the dialog that will start another dialog anytime it's triggered. The new dialog will be pushed onto the stack so it does not automatically end the current task. The current task will be continued once the new dialog ends. The built-in prompts will automatically re-prompt the user once this happens but that behaviour can be disabled by setting the promptAfterAction flag when calling a built-in prompt. |
cancel |
Binds an action to the dialog that will cancel the dialog anytime it's triggered. When canceled, the dialogs parent will be resumed with a resumed code indicating that it was canceled. |
clone(Action |
Returns a clone of an existing ActionSet. |
custom |
Binds a custom action to the dialog that will call the passed in onSelectAction handler when triggered. |
dialog |
Called when a root dialog is being interrupted by another dialog. This gives the dialog that's being interrupted a chance to run custom logic before it's removed from the stack. The dialog itself is responsible for clearing the dialog stack and starting the new dialog. |
dialog |
A child dialog has ended and the current one is being resumed. |
end |
Binds an action that will end the conversation with the user when triggered. |
find |
Called during the Library.findRoutes() call for each dialog on the stack to determine if any of the dialogs actions are triggered by the users utterance. |
format |
Creates the message to send for the prompt. This is called automatically by sendPrompt()
so in most cases you'll want to register a onFormatMessage() handler to
customize the message sent for a prompt. You should only need to call this method if you're
implementing your own |
gettext(Session, Text |
Returns the text for a prompt that's been localized using the namespace of the prompts caller. |
matches(Reg |
Invokes a handler when a given intent is detected in the users utterance. For
|
matches |
Invokes a handler when any of the given intents are detected in the users utterance. For
|
on |
Registers a handler that will be called to create the outgoing IMessage
that will be sent for the prompt. This handler is only called when the current
prompt/retryPrompt is of type |
on |
Registers a handler that will be called every time the prompt is about to send a message to
the user. You can use this hook to implement your own custom prompt sending logic.
Multiple handlers can be registered and calling |
on |
Registers a handler that will be called everytime the prompt receives a reply from the user.
The handlers Multiple handlers can be registered and unlike the other handler types, all of the registered will be called and handler providing the highest confidence score will be chosen as the winner. When customizing one of the built-in prompt types you'll often want to disable the prompts default recognizer logic. This can be achieved by setting the features of the prompt when you create it. Just keep in mind that if you completely disable the prompts default recognizer logic, you'll need to do all of the recognition yourself. |
recognize(IRecognize |
Parses the users utterance and assigns a score from 0.0 - 1.0 indicating how confident the dialog is that it understood the users utterance. This method is always called for the active dialog on the stack. A score of 1.0 will indicate a perfect match and terminate any further recognition. When the score is less than 1.0, every dialog on the stack will have its recognizeAction() method called as well to see if there are any named actions bound to the dialog that better matches the users utterance. Global actions registered at the bot level will also be evaluated. If the dialog has a score higher then any bound actions, the dialogs replyReceived() method will be called with the result object returned from the recognize() call. This lets the dialog pass additional data collected during the recognize phase to the replyReceived() method for handling. Should there be an action with a higher score then the dialog the action will be invoked instead of the dialogs replyReceived() method. The dialog will stay on the stack and may be resumed at some point should the action invoke a new dialog so dialogs should prepare for unexpected calls to dialogResumed(). |
recognizer(IIntent |
Adds a new recognizer plugin to the Prompt which will be run everytime the user replies to the prompt. |
reload |
Binds an action to the dialog that will cause the dialog to be reloaded anytime it's triggered. This is useful to implement logic that handle user utterances like "start over". |
reply |
Processes messages received from the user. Called by the dialog system. |
select |
Selects the route that had the highest confidence score for the utterance. |
send |
Sends a prompt to the user for the current turn. This can be called from a [matches()][#matches]
handler to manually send a prompt/reprompt to the user. To force sending of the initial prompt
you would need to set |
trigger |
Binds an action to the dialog that will make it the active dialog anytime it's triggered. The default behaviour is to interupt any existing dialog by clearing the stack and starting the dialog at the root of the stack. The dialog being interrupted can intercept this interruption by adding a custom onInterrupted handler to their trigger action options. Additionally, you can customize the way the triggered dialog is started by providing a custom onSelectAction handler to your trigger action options. |
Constructor Details
Prompt(IPromptFeatures)
Creates a new customizable instance of the prompt. Your new prompt should be added as a dialog to either a bot or library.
new Prompt(features?: IPromptFeatures)
Parameters
- features
- IPromptFeatures
(Optional) features used to customize the prompts behaviour.
Property Details
features
The prompts current configured set of features.
public features: T
Property Value
T
Method Details
addDialogTrigger(ActionSet, string)
Called once for each dialog within a library to give the dialog a chance to add its
triggerAction()
to the libraries global action set. These triggers get mapped to
a beginDialogAction()
that starts the dialog when the trigger condition is met.
function addDialogTrigger(actions: ActionSet, dialogId: string)
Parameters
- actions
- ActionSet
Libraries global action set.
- dialogId
-
string
The fully qualified ID of the dialog to trigger.
begin<T>(Session, T)
Called when a new dialog session is being started.
function begin<T>(session: Session, args?: T)
Parameters
- session
- Session
Session object for the current conversation.
- args
-
T
(Optional) arguments passed to the dialog by its parent.
beginDialogAction(string, string, IBeginDialogActionOptions)
Binds an action to the dialog that will start another dialog anytime it's triggered. The new dialog will be pushed onto the stack so it does not automatically end the current task. The current task will be continued once the new dialog ends. The built-in prompts will automatically re-prompt the user once this happens but that behaviour can be disabled by setting the promptAfterAction flag when calling a built-in prompt.
function beginDialogAction(name: string, id: string, options?: IBeginDialogActionOptions)
Parameters
- name
-
string
Unique name to assign the action.
- id
-
string
ID of the dialog to start.
- options
- IBeginDialogActionOptions
(Optional) options used to configure the action. If matches is specified the action will listen for the user to say a word or phrase that triggers the action, otherwise the action needs to be bound to a button using CardAction.dialogAction() to trigger the action. You can also use dialogArgs to pass additional params to the dialog being started.
Returns
cancelAction(string, TextOrMessageType, ICancelActionOptions)
Binds an action to the dialog that will cancel the dialog anytime it's triggered. When canceled, the dialogs parent will be resumed with a resumed code indicating that it was canceled.
function cancelAction(name: string, msg?: TextOrMessageType, options?: ICancelActionOptions)
Parameters
- name
-
string
Unique name to assign the action.
(Optional) message to send the user prior to canceling the dialog.
- options
- ICancelActionOptions
(Optional) options used to configure the action. If matches is specified the action will listen for the user to say a word or phrase that triggers the action, otherwise the action needs to be bound to a button using CardAction.dialogAction() to trigger the action.
Returns
clone(ActionSet)
Returns a clone of an existing ActionSet.
function clone(copyTo?: ActionSet)
Parameters
- copyTo
- ActionSet
(Optional) instance to copy the current object to. If missing a new instance will be created.
Returns
customAction(IDialogActionOptions)
Binds a custom action to the dialog that will call the passed in onSelectAction handler when triggered.
function customAction(options: IDialogActionOptions)
Parameters
- options
- IDialogActionOptions
The options used to configure the action. If matches is specified the action will listen for the user to say a word or phrase that triggers the action. Custom matching logic can be provided using onFindAction.
Returns
dialogInterrupted(Session, string, any)
Called when a root dialog is being interrupted by another dialog. This gives the dialog that's being interrupted a chance to run custom logic before it's removed from the stack. The dialog itself is responsible for clearing the dialog stack and starting the new dialog.
function dialogInterrupted(session: Session, dialogId: string, dialogArgs: any)
Parameters
- session
- Session
Session object for the current conversation.
- dialogId
-
string
ID of the dialog that should be started.
- dialogArgs
-
any
Arguments that should be passed to the new dialog.
dialogResumed<T>(Session, IDialogResult<T>)
A child dialog has ended and the current one is being resumed.
function dialogResumed<T>(session: Session, result: IDialogResult<T>)
Parameters
- session
- Session
Session object for the current conversation.
- result
Result returned by the child dialog.
endConversationAction(string, TextOrMessageType, ICancelActionOptions)
Binds an action that will end the conversation with the user when triggered.
function endConversationAction(name: string, msg?: TextOrMessageType, options?: ICancelActionOptions)
Parameters
- name
-
string
Unique name to assign the action.
(Optional) message to send the user prior to ending the conversation.
- options
- ICancelActionOptions
(Optional) options used to configure the action. If matches is specified the action will listen for the user to say a word or phrase that triggers the action, otherwise the action needs to be bound to a button using CardAction.dialogAction() to trigger the action.
Returns
findActionRoutes(IRecognizeDialogContext, (err: Error, results: IRouteResult[]) => void)
Called during the Library.findRoutes() call for each dialog on the stack to determine if any of the dialogs actions are triggered by the users utterance.
function findActionRoutes(context: IRecognizeDialogContext, callback: (err: Error, results: IRouteResult[]) => void)
Parameters
- context
- IRecognizeDialogContext
The context of the incoming message as well as the dialogData for the evaluated dialog.
- callback
-
(err: Error, results: IRouteResult[]) => void
Function to invoke with the top candidate route(s).
formatMessage(Session, TextType, TextType, (err: Error, msg: IMessage) => void)
Creates the message to send for the prompt. This is called automatically by sendPrompt()
so in most cases you'll want to register a onFormatMessage() handler to
customize the message sent for a prompt. You should only need to call this method if you're
implementing your own sendPrompt()
logic.
function formatMessage(session: Session, text: TextType, speak: TextType, callback: (err: Error, msg: IMessage) => void)
Parameters
- session
- Session
Session object for the current conversation.
- text
- TextType
Current prompt/retryPrompt text.
- speak
- TextType
Current speak/retrySpeak SSML. This value may be null.
- callback
-
(err: Error, msg: IMessage) => void
Function to receive the created message.
gettext(Session, TextType, string)
Returns the text for a prompt that's been localized using the namespace of the prompts caller.
static function gettext(session: Session, text: TextType, namespace?: string)
Parameters
- session
- Session
Current session for the conversation.
- text
- TextType
Prompt to localize.
- namespace
-
string
(Optional) library namespace to use for localizing the prompt. By default the namespace of the prompts caller will be used.
Returns
string
matches(RegExp | string, string | IDialogWaterfallStep[] | IDialogWaterfallStep, any)
Invokes a handler when a given intent is detected in the users utterance. For string
based
intents, the intent can either be an intent returned by a recognizer() registered
for the prompt or it can be an intent that flows in from a global recognizer().
NOTE: The full details of the match, including the list of intents & entities detected, will be passed to the args of the first waterfall step or dialog that's started.
function matches(intent: RegExp | string, dialogId: string | IDialogWaterfallStep[] | IDialogWaterfallStep, dialogArgs?: any)
Parameters
- intent
-
RegExp | string
- intent: {RegExp} - A regular expression that will be evaluated to detect the users intent.
- intent: {string} - A named intent returned by an IIntentRecognizer plugin that will be used to match the users intent.
- dialogId
-
string | IDialogWaterfallStep[] | IDialogWaterfallStep
- dialogId: _{string} - The ID of a dialog to begin when the intent is matched.
- dialogId: {IDialogWaterfallStep[]} - Waterfall of steps to execute when the intent is matched.
- dialogId: {IDialogWaterfallStep} - Single step waterfall to execute when the intent is matched. Calling a built-in prompt or starting a new dialog will result in the current dialog ending upon completion of the child prompt/dialog.
- dialogArgs
-
any
(Optional) arguments to pass the dialog that started when dialogId
is a {string}.
Returns
matchesAny(RegExp[] | string[], string | IDialogWaterfallStep[] | IDialogWaterfallStep, any)
Invokes a handler when any of the given intents are detected in the users utterance. For string
based
intents, the intent can either be an intent returned by a recognizer() registered
for the prompt or it can be an intent that flows in from a global recognizer().
NOTE: The full details of the match, including the list of intents & entities detected, will be passed to the args of the first waterfall step or dialog that's started.
function matchesAny(intent: RegExp[] | string[], dialogId: string | IDialogWaterfallStep[] | IDialogWaterfallStep, dialogArgs?: any)
Parameters
- intent
-
RegExp[] | string[]
- intent: {RegExp[]} - Array of regular expressions that will be evaluated to detect the users intent.
- intent: {string[]} - Array of named intents returned by an IIntentRecognizer plugin that will be used to match the users intent.
- dialogId
-
string | IDialogWaterfallStep[] | IDialogWaterfallStep
- dialogId: _{string} - The ID of a dialog to begin when the intent is matched.
- dialogId: {IDialogWaterfallStep[]} - Waterfall of steps to execute when the intent is matched.
- dialogId: {IDialogWaterfallStep} - Single step waterfall to execute when the intent is matched. Calling a built-in prompt or starting a new dialog will result in the current dialog ending upon completion of the child prompt/dialog.
- dialogArgs
-
any
(Optional) arguments to pass the dialog that started when dialogId
is a {string}.
Returns
onFormatMessage((session: Session, text: TextType, speak: TextType, callback: (err: Error, message?: IMessage) => void) => void)
Registers a handler that will be called to create the outgoing IMessage
that will be sent for the prompt. This handler is only called when the current
prompt/retryPrompt is of type string|string[]
. Anytime the prompt/retryPrompt is an
IMessage|IIsMessage
the configured message is used so your handler will not be called.
Multiple handlers can be registered and the first handler to call callback()
with a message
will be used. Calling callback(null, null)
will cause processing to move to the next handler
in the chain.
function onFormatMessage(handler: (session: Session, text: TextType, speak: TextType, callback: (err: Error, message?: IMessage) => void) => void)
Parameters
- handler
-
(session: Session, text: TextType, speak: TextType, callback: (err: Error, message?: IMessage) => void) => void
Function that will be called to create an IMessage
for the current prompt. Call callback()
with either a message or null
to continue processing.
Returns
Prompt<any>
onPrompt((session: Session, next: Function) => void)
Registers a handler that will be called every time the prompt is about to send a message to
the user. You can use this hook to implement your own custom prompt sending logic.
Multiple handlers can be registered and calling next()
will invoke the next handler in
the chain. The final handler performs the prompts default logic which is to create a new
message using formatMessage() and then send it.
function onPrompt(handler: (session: Session, next: Function) => void)
Parameters
- handler
-
(session: Session, next: Function) => void
Function that will be called anytime sendPrompt() is called.
Returns
Prompt<any>
onRecognize((context: IRecognizeDialogContext, callback: (err: Error, score: number, response?: any) => void) => void)
Registers a handler that will be called everytime the prompt receives a reply from the user.
The handlers callback()
can be used to return a confidence score that it understood the
users input as well as the value that should be returned to the caller of the prompt.
Calling callback(null, 1.0, true);
would indicate a high confidence that the user answered
the prompt and would return a boolean
true as the response from the prompt. Any response
type is possible, including objects. Calling callback(null, 0.0);
indicates that the users
input was not understood at all and that they should be re-prompted.
Multiple handlers can be registered and unlike the other handler types, all of the registered will be called and handler providing the highest confidence score will be chosen as the winner. When customizing one of the built-in prompt types you'll often want to disable the prompts default recognizer logic. This can be achieved by setting the features of the prompt when you create it. Just keep in mind that if you completely disable the prompts default recognizer logic, you'll need to do all of the recognition yourself.
function onRecognize(handler: (context: IRecognizeDialogContext, callback: (err: Error, score: number, response?: any) => void) => void)
Parameters
- handler
-
(context: IRecognizeDialogContext, callback: (err: Error, score: number, response?: any) => void) => void
Function that will be called to recognize the users reply to a prompt.
Returns
Prompt<any>
recognize(IRecognizeDialogContext, (err: Error, result: IRecognizeResult) => void)
Parses the users utterance and assigns a score from 0.0 - 1.0 indicating how confident the dialog is that it understood the users utterance. This method is always called for the active dialog on the stack. A score of 1.0 will indicate a perfect match and terminate any further recognition. When the score is less than 1.0, every dialog on the stack will have its recognizeAction() method called as well to see if there are any named actions bound to the dialog that better matches the users utterance. Global actions registered at the bot level will also be evaluated. If the dialog has a score higher then any bound actions, the dialogs replyReceived() method will be called with the result object returned from the recognize() call. This lets the dialog pass additional data collected during the recognize phase to the replyReceived() method for handling.
Should there be an action with a higher score then the dialog the action will be invoked instead of the dialogs replyReceived() method. The dialog will stay on the stack and may be resumed at some point should the action invoke a new dialog so dialogs should prepare for unexpected calls to dialogResumed().
function recognize(context: IRecognizeDialogContext, callback: (err: Error, result: IRecognizeResult) => void)
Parameters
- context
- IRecognizeDialogContext
The context of the request.
- callback
-
(err: Error, result: IRecognizeResult) => void
Function to invoke with the recognition results.
recognizer(IIntentRecognizer)
Adds a new recognizer plugin to the Prompt which will be run everytime the user replies to the prompt.
function recognizer(plugin: IIntentRecognizer)
Parameters
- plugin
- IIntentRecognizer
The recognizer to add.
Returns
reloadAction(string, TextOrMessageType, IBeginDialogActionOptions)
Binds an action to the dialog that will cause the dialog to be reloaded anytime it's triggered. This is useful to implement logic that handle user utterances like "start over".
function reloadAction(name: string, msg?: TextOrMessageType, options?: IBeginDialogActionOptions)
Parameters
- name
-
string
Unique name to assign the action.
(Optional) message to send the user prior to reloading the dialog.
- options
- IBeginDialogActionOptions
(Optional) options used to configure the action. If matches is specified the action will listen for the user to say a word or phrase that triggers the action, otherwise the action needs to be bound to a button using CardAction.dialogAction() to trigger the action. You can also use dialogArgs to pass additional params to the dialog when reloaded.
Returns
replyReceived(Session, IRecognizeResult)
Processes messages received from the user. Called by the dialog system.
function replyReceived(session: Session, recognizeResult?: IRecognizeResult)
Parameters
- session
- Session
Session object for the current conversation.
- recognizeResult
- IRecognizeResult
selectActionRoute(Session, IRouteResult)
Selects the route that had the highest confidence score for the utterance.
function selectActionRoute(session: Session, route: IRouteResult)
Parameters
- session
- Session
Session object for the current conversation.
- route
- IRouteResult
Results returned from the call to findActionRoutes().
sendPrompt(Session)
Sends a prompt to the user for the current turn. This can be called from a [matches()][#matches]
handler to manually send a prompt/reprompt to the user. To force sending of the initial prompt
you would need to set session.dialogData.turns = 0;
before calling sendPrompt()
.
function sendPrompt(session: Session)
Parameters
- session
- Session
Session object for the current conversation.
triggerAction(ITriggerActionOptions)
Binds an action to the dialog that will make it the active dialog anytime it's triggered. The default behaviour is to interupt any existing dialog by clearing the stack and starting the dialog at the root of the stack. The dialog being interrupted can intercept this interruption by adding a custom onInterrupted handler to their trigger action options. Additionally, you can customize the way the triggered dialog is started by providing a custom onSelectAction handler to your trigger action options.
function triggerAction(options: ITriggerActionOptions)
Parameters
- options
- ITriggerActionOptions
Options used to configure the action.