« Back to index

Experimental

Table of Contents:

• ChatBot
• FirebaseDB
• GraphQL
• ImageBot

ChatBot

The ChatBot component is a non-visible component for chatting with an AI chatbot. This version uses a proxy run by MIT that in turn uses the ChatGPT generative large language model.

Properties

ApiKey

An ApiKey for ChatGPT. User supplied. If provided, we will use it instead of the API key embedded in the chat proxy service.

Note: We do not provide this as a DesignerProperty, it should be stored using the blocks, preferably using the Obfuscated Text block to provide some protection (not perfect protection) of the key embedded in a packaged app.

Model

Set the name of the model to use. See https://appinv.us/chatbot for the current list of supported models. Leaving this blank will result in the default model set by the provider being used

Provider

Set the name of the provider to use. See https://appinv.us/chatbot for the current list of supported providers.

System

The “System” value given to ChatGPT. It is used to set the tone of a conversation. For example: “You are a funny person.”

Token

The MIT Access token to use. MIT App Inventor will automatically fill this value in. You should not need to change it.

Events

ErrorOccurred(responseCode,responseText)

The ErrorOccurred event will be run when an error occurs during processing, such as if your you are over usage quota, or some other error signaled by ChatGPT or PaLM. See https://appinv.us/chatbot for current information.

GotResponse(responseText)

Event indicating that a request has finished and has returned data (output from ChatBot).

Methods

Converse(question)

Ask a question of the Chat Bot. Successive calls will remember information from earlier in the conversation. Use the “ResetConversation” function to reset for a new conversation.

ConverseWithImage(question,source)

Ask a question of the Chat Bot using an Image. Successive calls will remember information from earlier in the conversation. Use the “ResetConversation” function to reset for a new conversation.

ResetConversation()

Reset the current conversation, Chat bot will forget any previous conversation when responding in the future.

FirebaseDB

The Firebase component communicates with a Web service to store and retrieve information. The component has methods to store a value under a tag and to retrieve the value associated with the tag. It also possesses a listener to fire events when stored values are changed.

Additional Information

Properties

DeveloperBucket

Getter for the DeveloperBucket.

FirebaseToken

Getter for the FirebaseToken.

FirebaseURL

Specifies the URL for the Firebase.

The default value is currently my private Firebase URL, but this will eventually changed once the App Inventor Candle plan is activated.

Persist

If true, variables will retain their values when off-line and the App exits. Values will be uploaded to Firebase the next time the App is run while connected to the network. This is useful for applications which will gather data while not connected to the network. Note: AppendValue and RemoveFirst will not work correctly when off-line, they require a network connection.

Note: If you set Persist on any Firebase component, on any screen, it makes all Firebase components on all screens persistent. This is a limitation of the low level Firebase library. Also be aware that if you want to set persist to true, you should do so before connecting the Companion for incremental development.

ProjectBucket

Getter for the ProjectBucket.

Events

DataChanged(tag,value)

Indicates that the data in the Firebase has changed. Launches an event with the tag and value that have been updated.

FirebaseError(message)

Indicates that the communication with the Firebase signaled an error.

FirstRemoved(value)

Event triggered by the “RemoveFirst” function. The argument “value” is the object that was the first in the list, and which is now removed.

GotValue(tag,value)

Indicates that a GetValue request has succeeded.

TagList(value)

Event triggered when we have received the list of known tags. Used with the “GetTagList” Function.

Methods

AppendValue(tag,valueToAdd)

Append a value to the end of a list atomically. If two devices use this function simultaneously, both will be appended and no data lost.

ClearTag(tag)

Asks Firebase to forget (delete or set to “null”) a given tag.

GetTagList()

Get the list of tags for this application. When complete a “TagList” event will be triggered with the list of known tags.

GetValue(tag,valueIfTagNotThere)

GetValue asks Firebase to get the value stored under the given tag. It will pass valueIfTagNotThere to GotValue if there is no value stored under the tag.

RemoveFirst(tag)

Return the first element of a list and atomically remove it. If two devices use this function simultaneously, one will get the first element and the the other will get the second element, or an error if there is no available element. When the element is available, the “FirstRemoved” event will be triggered.

StoreValue(tag,valueToStore)

Asks Firebase to store the given value under the given tag.

Unauthenticate()

Unauthenticate from Firebase.

Firebase keeps track of credentials in a cache in shared_prefs It will re-use these credentials as long as they are valid. Given That we retrieve a FirebaseToken with a version long life, this will effectively be forever. Shared_prefs survive an application update and depending on how backup is configured on a device, it might survive an application removal and reinstallation.

Normally this is not a problem, however if we change the credentials used, for example the App author is switching from one Firebase account to another, or invalided their firebase.secret, this cached credential is invalid, but will continue to be used, which results in errors.

This function permits us to unauthenticate, which tosses the cached credentials. The next time authentication is needed we will use our current FirebaseToken and get fresh credentials.

GraphQL

The GraphQL component communicates with a GraphQL endpoint to execute queries and mutations. It represents returned data as a dictionary. All queries have an associated operation name and are executed asynchronously. Completed queries trigger different events depending on whether there the queries had any errors.

Properties

GqlEndpointUrl

Specifies the URL for this GraphQL component.

GqlHttpHeaders

Specifies the HTTP headers for this GraphQL component as a JSON dictionary.

Events

GqlGotError(gqlQueryName,gqlError)

Triggers an event indicating that there were one or more errors when executing the query. This method should be executed in the application’s main thread.

GqlGotResponse(gqlQueryName,gqlResponse)

Triggers an event indicating that the given operation has successfully executed and returned data. This method should be executed in the application’s main thread.

Methods

GqlQuery(gqlQueryName,gqlQuery)

Executes an arbitrary query against the GraphQL endpoint.

ImageBot

The ImageBot is a non-visible component that uses DALL-E 2 to create and edit images. You must supply your own OpenAI API key for this component by setting its ApiKey property in the blocks.

Properties

ApiKey

Specifies the ApiKey used to authenticate with the ImageBot.

InvertMask

Specifies whether the mask used for editing should have its alpha channel inverted.

Size

Specifies the size of the generated image. Can be one of 256, 512, or 1024.

Token

The MIT Access token to use. MIT App Inventor will automatically fill this value in. You should not need to change it.

Events

ErrorOccurred(responseCode,responseText)

The ErrorOccurred event will be run when an error occurs during processing, such as if you forget to provide an API key or the server is overloaded.

ImageCreated(fileName)

The ImageCreated event will be run when the ImageBot successfully creates an image.

ImageEdited(fileName)

The ImageCreated event will be run when the ImageBot successfully edits an image.

Methods

CreateImage(description)

Create an image using the given description.

EditImageWithMask(imageSource,maskSource,prompt)

Edit the imageSource using the given description. The editable area of the image should be indicated by the maskSource. The sources can be a Canvas, an Image, or a string representing the path to a file.