Mobile client application

1. Introduction

The Sponge mobile client application is a Flutter application that provides a generic GUI to call remote Sponge actions. It can be run on both Android and iOS. It could be used as the one app to rule them all for Sponge services that publish actions via the Sponge REST API. All business logic has to be placed in Sponge actions, so the only requirement to have the application working is to create your own Sponge service by writing a knowledge base that define actions that will be visible and ready to run in the application or to connect to an existing Sponge service.

The application is currently under development. It will be released as an open source.

The application is especially useful when data and functionality are more important than visual aspects of a GUI, e.g. for prototyping, rapid application development, low cost solutions, etc. The reason is that the application provides only a generic and opinionated GUI whose customization is limited.

Figure 1. The mobile application architecture

One of many use cases of the application is to connect to IoT devices that have Sponge installed and execute available actions. Different types of such devices provide different sets of actions that are specific to a device features. For example one device could have a camera and provide an action to take a picture. Another device could have a temperature sensor and provide its readings or have a machine learning model and provide predictions.

The mobile application uses several Sponge concepts such as actions and action arguments metadata, data types metadata, action and data type features, annotated values, provided action arguments and categories. It supports the convention over configuration paradigm.

You could build a customized GUI using your own Flutter code that will call the same Sponge actions. In that case the generic Sponge mobile client would be used as a Flutter library in your Flutter project.

There is also a draft of an additional, experimental functionality that will allow users to subscribe to Sponge events in the mobile application. In case of a new event a notification will be displayed. The subscribed events will be shown in the list. By tapping on an event the user will be directed to a screen presenting an action associated with this event. The subscription uses a Sponge gRPC service published on a default port (i.e. the REST API port plus 1).

2. Functionalities

The following chapters show the key functionalities of the mobile application.

2.1. Connections

Figure 2. Selecting a connection to the Sponge instance in the action list screen

You may configure many connections to Sponge REST API services. The application allows you to connect to a single service at the same time.

Figure 3. List of connections to Sponge instances

You may add, edit and remove connections to Sponge instances as well as activate a connection. To remove a connection swipe the corresponding element.

Figure 4. Editing a connection to a Sponge instance

A Sponge address is the URL of the Sponge instance.

2.2. Action list

Figure 5. The action list

The main screen shows the list of actions defined in the connected Sponge engine. Only actions that have argument and result metadata are available. This is because the application uses a generic access to the actions utilizing their data types, labels, descriptions, features and so on. The number in the action icon is the number of action arguments.

To call an action or set action attributes tap the triangular icon on the right side of the action label.

The floating button allows to refresh the action list from the server. The refresh button clears all entered action arguments and received results.

The application currently doesn’t supports all Sponge data types.

Figure 6. The navigation drawer

The navigation drawer allows switching between the available main views.

2.4. Action call

Figure 7. The action call that manages the Raspberry Pi LCD display

Actions may have read only, provided arguments only to show a data from the server (see the Current LCD text attribute). The REFRESH button retrieves the current values of read only, provided arguments from the server.

The definition of the action that manages the Raspberry Pi LCD display

class ManageLcd(Action):
    def onConfigure(self):
        self.withLabel("Manage the LCD text and color")
        self.withDescription("Provides management of the LCD properties (display text and color). A null value doesn't change an LCD property.")
        self.withArgs([
            StringType("currentText").withMaxLength(256).withNullable(True).withFeatures({"maxLines":2})
                .withLabel("Current LCD text").withDescription("The currently displayed LCD text.").withProvided(ProvidedMeta().withValue().withReadOnly()),
            StringType("text").withMaxLength(256).withNullable(True).withFeatures({"maxLines":2})
                .withLabel("Text to display").withDescription("The text that will be displayed in the LCD.").withProvided(ProvidedMeta().withValue()),
            StringType("color").withMaxLength(6).withNullable(True).withFeatures({"characteristic":"color"})
                .withLabel("LCD color").withDescription("The LCD color.").withProvided(ProvidedMeta().withValue().withOverwrite()),
            BooleanType("clearText").withNullable(True).withDefaultValue(False)
                .withLabel("Clear text").withDescription("The text the LCD will be cleared.")
        ]).withNoResult()
        self.withFeature("icon", "monitor")
    def onCall(self, currentText, text, color, clearText = None):
        sponge.call("SetLcd", [text, color, clearText])
    def onProvideArgs(self, context):
        grovePiDevice = sponge.getVariable("grovePiDevice")
        if "currentText" in context.names:
            context.provided["currentText"] = ProvidedValue().withValue(grovePiDevice.getLcdText())
        if "text" in context.names:
            context.provided["text"] = ProvidedValue().withValue(grovePiDevice.getLcdText())
        if "color" in context.names:
            context.provided["color"] = ProvidedValue().withValue(grovePiDevice.getLcdColor())

class SetLcd(Action):
    def onCall(self, text, color, clearText = None):
        sponge.getVariable("grovePiDevice").setLcd("" if (clearText or text is None) else text, color)

Figure 8. The action call that manages the Grove Pi sensors and actuators

The action call screen allows editing the action arguments.

The definition of the action that manages the Grove Pi sensors and actuators

class ManageSensorActuatorValues(Action):
    def onConfigure(self):
        self.withLabel("Manage the sensor and actuator values").withDescription("Provides management of the sensor and actuator values.")
        self.withArgs([
            NumberType("temperatureSensor").withNullable().withLabel(u"Temperature sensor (°C)").withProvided(ProvidedMeta().withValue().withReadOnly()),
            NumberType("humiditySensor").withNullable().withLabel(u"Humidity sensor (%)").withProvided(ProvidedMeta().withValue().withReadOnly()),
            NumberType("lightSensor").withNullable().withLabel(u"Light sensor").withProvided(ProvidedMeta().withValue().withReadOnly()),
            NumberType("rotarySensor").withNullable().withLabel(u"Rotary sensor").withProvided(ProvidedMeta().withValue().withReadOnly()),
            NumberType("soundSensor").withNullable().withLabel(u"Sound sensor").withProvided(ProvidedMeta().withValue().withReadOnly()),
            BooleanType("redLed").withLabel("Red LED").withProvided(ProvidedMeta().withValue().withOverwrite()),
            IntegerType("blueLed").withMinValue(0).withMaxValue(255).withLabel("Blue LED").withProvided(ProvidedMeta().withValue().withOverwrite()),
            BooleanType("buzzer").withLabel("Buzzer").withProvided(ProvidedMeta().withValue().withOverwrite())
        ]).withNoResult()
        self.withFeature("icon", "thermometer")
    def onCall(self, temperatureSensor, humiditySensor, lightSensor, rotarySensor, soundSensor, redLed, blueLed, buzzer):
        grovePiDevice = sponge.getVariable("grovePiDevice")
        grovePiDevice.setRedLed(redLed)
        grovePiDevice.setBlueLed(blueLed)
        grovePiDevice.setBuzzer(buzzer)
    def onProvideArgs(self, context):
        values = sponge.call("GetSensorActuatorValues", [context.names])
        for name, value in values.iteritems():
            context.provided[name] = ProvidedValue().withValue(value)

class GetSensorActuatorValues(Action):
    def onCall(self, names):
        values = {}
        grovePiDevice = sponge.getVariable("grovePiDevice")
        if "temperatureSensor" or "humiditySensor" in names:
            th = grovePiDevice.getTemperatureHumiditySensor()
            if "temperatureSensor" in names:
                values["temperatureSensor"] = th.temperature if th else None
            if "humiditySensor" in names:
                values["humiditySensor"] = th.humidity if th else None
        if "lightSensor" in names:
            values["lightSensor"] = grovePiDevice.getLightSensor()
        if "rotarySensor" in names:
            values["rotarySensor"] = grovePiDevice.getRotarySensor().factor
        if "soundSensor" in names:
            values["soundSensor"] = grovePiDevice.getSoundSensor()
        if "redLed" in names:
            values["redLed"] = grovePiDevice.getRedLed()
        if "blueLed" in names:
            values["blueLed"] = grovePiDevice.getBlueLed()
        if "buzzer" in names:
            values["buzzer"] = grovePiDevice.getBuzzer()
        return values

Figure 9. The action call that sends an SMS from the Raspberry Pi

Actions arguments may be edited in multiline text fields.

The definition of the action that sends an SMS from the Raspberry Pi

class SendSms(Action):
    def onConfigure(self):
        self.withLabel("Send an SMS").withDescription("Sends a new SMS.")
        self.withArgs([
            StringType("recipient").withFormat("phone").withLabel("Recipient").withDescription("The SMS recipient."),
            StringType("message").withMaxLength(160).withFeatures({"maxLines":5}).withLabel("Message").withDescription("The SMS message.")
        ]).withNoResult()
        self.withFeature("icon", "cellphone-text")
    def onCall(self, recipient, message):
        gsm.sendSms(recipient, message)

Figure 10. The action call argument editor for a color type

The color picker widget allows a user to choose a color as an argument value.

The definition of the action that takes a color argument

class ChooseColor(Action):
    def onConfigure(self):
        self.withLabel("Choose a color").withDescription("Shows a color argument.")
        self.withArg(
            StringType("color").withMaxLength(6).withNullable(True).withFeatures({"characteristic":"color"})
                .withLabel("Color").withDescription("The color.")
        ).withResult(StringType())
        self.withFeatures({"icon":"format-color-fill"})
    def onCall(self, color):
        return ("The chosen color is " + color) if color else "No color chosen"

Figure 11. The action call argument editor for a digit drawing

The drawing panel allows a user to paint an image that will be set as an argument value in an action call.

The definition of the action that recognizes a handwritten digit

class DigitsPredict(Action):
    def onConfigure(self):
        self.withLabel("Recognize a digit").withDescription("Recognizes a handwritten digit")
        self.withArg(createImageType("image")).withResult(IntegerType().withLabel("Recognized digit"))
        self.withFeature("icon", "brain")
    def onCall(self, image):
        predictions = py4j.facade.predict(image)
        prediction = max(predictions, key=predictions.get)
        probability = predictions[prediction]

        # Handle the optional predictionThreshold Sponge variable.
        predictionThreshold = sponge.getVariable("predictionThreshold", None)
        if predictionThreshold and probability < float(predictionThreshold):
            self.logger.debug("The prediction {} probability {} is lower than the threshold {}.", prediction, probability, predictionThreshold)
            return None
        else:
            self.logger.debug("Prediction: {}, probability: {}", prediction, probability)
            return int(prediction)

def imageClassifierServiceInit(py4jPlugin):
    SpongeUtils.awaitUntil(lambda: py4jPlugin.facade.isReady())

Figure 12. The action call for an attribute of type drawing

The action call screen shows all action arguments.

Figure 13. The action call result for a digit recognition

If the action has been called, the result is shown below the action label. If the result can’t be fully shown in the action list, you may tap the result to see the details.

Figure 14. The action call argument editor for a doodle drawing

Drawing panels can be configured in a corresponding action definition, where a color, a background color etc. could be specified.

The definitions of actions that represent drawing and viewing a doodle

from java.lang import System
from os import listdir
from os.path import isfile, join, isdir

class DrawAndUploadDoodle(Action):
    def onConfigure(self):
        self.withLabel("Draw and upload a doodle").withDescription("Shows a canvas to draw a doodle and uploads it to the server")
        self.withArg(
              BinaryType("image").withLabel("Doodle").withMimeType("image/png")
                     .withFeatures({"characteristic":"drawing", "width":300, "height":250, "background":"FFFFFF", "color":"000000", "strokeWidth":2})
        )
        self.withResult(StringType().withLabel("Status"))
        self.withFeatures({"icon":"brush"})
    def onCall(self, image):
        filename = str(System.currentTimeMillis()) + ".png"
        SpongeUtils.writeByteArrayToFile(image, sponge.getProperty("doodlesDir") + "/" + filename)
        return "Uploaded as " + filename

class ListDoodles(Action):
    def onConfigure(self):
        self.withLabel("List doodles").withDescription("Returns a list of doodle filenames").withFeatures({"visible":False})
        self.withNoArgs().withResult(ListType(StringType()).withLabel("Doodles"))
    def onCall(self):
        dir = sponge.getProperty("doodlesDir")
        doodles = [f for f in listdir(dir) if isfile(join(dir, f)) and f.endswith(".png")] if isdir(dir) else []
        return sorted(doodles, reverse=True)

class ViewDoodle(Action):
    def onConfigure(self):
        self.withLabel("View a doodle").withDescription("Views a doodle")
        self.withArg(StringType("image").withLabel("Doodle name").withProvided(ProvidedMeta().withValue().withValueSet().withOverwrite()))
        self.withResult(BinaryType().withAnnotated().withMimeType("image/png").withLabel("Doodle image"))
        self.withFeature("icon", "drawing")
    def onCall(self, name):
        return AnnotatedValue(SpongeUtils.readFileToByteArray(sponge.getProperty("doodlesDir") + "/" + name)).withFeatures({"filename":"doodle_" + name})
    def onProvideArgs(self, context):
        if "image" in context.names:
            doodles = sponge.call("ListDoodles")
            context.provided["image"] = ProvidedValue().withValue(doodles[0] if doodles else None).withValueSet(doodles)

def onStartup():
    sponge.logger.info(str(sponge.call("ListDoodles")))

Figure 15. The action call for a doodle drawing as an argument

The action call screen shows all action arguments, for example a drawing.

Figure 16. The action call that shows argument dependencies

Action arguments may depend on each other. Argument dependencies are supported in the action call panel and allow creating simple, interactive forms where some arguments are provided by the server, some entered by the user, some read only and some depend on the values of others. The important thing is that all that configuration is defined in an action in a knowledge base placed on the server side, not in the mobile application.

The definition of the action that provides arguments with dependencies

class DependingArgumentsAction(Action):
    def onConfigure(self):
        self.withLabel("Action with depending arguments")
        self.withArgs([
            StringType("continent").withLabel("Continent").withProvided(ProvidedMeta().withValueSet()),
            StringType("country").withLabel("Country").withProvided(ProvidedMeta().withValueSet().withDependency("continent")),
            StringType("city").withLabel("City").withProvided(ProvidedMeta().withValueSet().withDependency("country")),
            StringType("river").withLabel("River").withProvided(ProvidedMeta().withValueSet().withDependency("continent")),
            StringType("weather").withLabel("Weather").withProvided(ProvidedMeta().withValueSet())
        ]).withResult(StringType().withLabel("Sentences"))
        self.withFeature("icon", "flag")
    def onCall(self, continent, country, city, river, weather):
        return "There is a city {} in {} in {}. The river {} flows in {}. It's {}.".format(city, country, continent, river, continent, weather.lower())
    def onProvideArgs(self, context):
        if "continent" in context.names:
            context.provided["continent"] = ProvidedValue().withValueSet(["Africa", "Asia", "Europe"])
        if "country" in context.names:
            continent = context.current["continent"]
            if continent == "Africa":
                countries = ["Nigeria", "Ethiopia", "Egypt"]
            elif continent == "Asia":
                countries = ["China", "India", "Indonesia"]
            elif continent == "Europe":
                countries = ["Russia", "Germany", "Turkey"]
            else:
                countries = []
            context.provided["country"] = ProvidedValue().withValueSet(countries)
        if "city" in context.names:
            country = context.current["country"]
            if country == "Nigeria":
                cities = ["Lagos", "Kano", "Ibadan"]
            elif country == "Ethiopia":
                cities = ["Addis Ababa", "Gondar", "Mek'ele"]
            elif country == "Egypt":
                cities = ["Cairo", "Alexandria", "Giza"]
            elif country == "China":
                cities = ["Guangzhou", "Shanghai", "Chongqing"]
            elif country == "India":
                cities = ["Mumbai", "Delhi", "Bangalore"]
            elif country == "Indonesia":
                cities = ["Jakarta", "Surabaya", "Medan"]
            elif country == "Russia":
                cities = ["Moscow", "Saint Petersburg", "Novosibirsk"]
            elif country == "Germany":
                cities = ["Berlin", "Hamburg", "Munich"]
            elif country == "Turkey":
                cities = ["Istanbul", "Ankara", "Izmir"]
            else:
                cities = []
            context.provided["city"] = ProvidedValue().withValueSet(cities)
        if "river" in context.names:
            continent = context.current["continent"]
            if continent == "Africa":
                rivers = ["Nile", "Chambeshi", "Niger"]
            elif continent == "Asia":
                rivers = ["Yangtze", "Yellow River", "Mekong"]
            elif continent == "Europe":
                rivers = ["Volga", "Danube", "Dnepr"]
            else:
                rivers = []
            context.provided["river"] = ProvidedValue().withValueSet(rivers)
        if "weather" in context.names:
            context.provided["weather"] = ProvidedValue().withValueSet(["Sunny", "Cloudy", "Raining", "Snowing"])

Figure 17. The action call that shows argument dependencies and value sets

Allowed argument values can be defined in an action and provided from the server every time the action call screen is shown or an argument dependency value changes.

2.5. Action result

Figure 18. The action binary result

Actions may return contents that can be viewed for example as a HTML or a PDF file using the mobile OS viewers.

The definitions of the actions that return a HTML and a PDF file respectively

class HtmlFileOutput(Action):
    def onConfigure(self):
        self.withLabel("HTML file output").withDescription("Returns the HTML file.")
        self.withNoArgs().withResult(BinaryType().withMimeType("text/html").withLabel("HTML file"))
        self.withFeatures({"icon":"web"})
    def onCall(self):
        return String("""
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01//EN">
<html>
    <head>
      <title>HTML page</title>
    </head>
    <body>
        <!-- Main content -->
        <h1>Header</h1>
        <p>Some text
    </body>
</html>
""").getBytes("UTF-8")

class PdfFileOutput(Action):
    def onConfigure(self):
        self.withLabel("PDF file output").withDescription("Returns the PDF file.")
        self.withNoArgs().withResult(BinaryType().withMimeType("application/pdf").withLabel("PDF file"))
        self.withFeatures({"icon":"file-pdf"})
    def onCall(self):
        return sponge.process(ProcessConfiguration.builder("curl", "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf")
                              .outputAsBinary()).run().outputBinary

Figure 19. The action console formatted result

Actions may return a console output, for example the result of running the df -h command on the server.

The definitions of the actions that returns OS commands output

from org.openksavi.sponge.util.process import ProcessConfiguration

class OsGetDiskSpaceInfo(Action):
    def onConfigure(self):
        self.withLabel("Get disk space info").withDescription("Returns the disk space info.")
        self.withNoArgs().withResult(StringType().withFormat("console").withLabel("Disk space info"))
        self.withFeature("icon", "console")
    def onCall(self):
        return sponge.process(ProcessConfiguration.builder("df", "-h").outputAsString()).run().outputString

class OsDmesg(Action):
    def onConfigure(self):
        self.withLabel("Run dmesg").withDescription("Returns the dmesg output.")
        self.withNoArgs().withResult(StringType().withFormat("console").withLabel("The dmesg output"))
        self.withFeature("icon", "console")
    def onCall(self):
        return sponge.process(ProcessConfiguration.builder("dmesg").outputAsString()).run().outputString

Figure 20. The action Markdown formatted result

Actions may return a Markdown formatted text.

3. Advanced use cases

3.1. Context actions

Context actions can be specified for actions, record arguments and list elements (see the list-details) to provide related, customized sub-actions. Context actions should be specified as the contextActions feature statically for a type or an action or dynamically for an annotated value. The latter option takes precedence.

Figure 21. The action with context actions

The definition of the action with context actions

class ActionWithContextActions(Action):
    def onConfigure(self):
        self.withLabel("Action with context actions").withArgs([
            StringType("arg1").withLabel("Argument 1"),
            StringType("arg2").withLabel("Argument 2")
        ]).withNoResult().withFeature("contextActions", [
            "ActionWithContextActionsContextAction1", "ActionWithContextActionsContextAction2(arg2)", "ActionWithContextActionsContextAction3(arg2=arg2)",
            "ActionWithContextActionsContextAction4(arg1)", "ActionWithContextActionsContextAction5", "MarkdownText()"
        ])
        self.withFeature("icon", "attachment")
    def onCall(self, arg1, arg2):
        pass

class ActionWithContextActionsContextAction1(Action):
    def onConfigure(self):
        self.withLabel("Context action 1").withArgs([
            RecordType("arg").withFields([
                StringType("arg1").withLabel("Argument 1"),
                StringType("arg2").withLabel("Argument 2")
            ])
        ]).withResult(StringType())
        self.withFeatures({"visible":False, "icon":"tortoise"})
    def onCall(self, arg):
        return arg["arg1"]

class ActionWithContextActionsContextAction2(Action):
    def onConfigure(self):
        self.withLabel("Context action 2").withArgs([
            StringType("arg").withLabel("Argument"),
            StringType("additionalText").withLabel("Additional text"),
        ]).withResult(StringType())
        self.withFeatures({"visible":False, "icon":"tortoise"})
    def onCall(self, arg, additionalText):
        return arg + " " + additionalText

class ActionWithContextActionsContextAction3(Action):
    def onConfigure(self):
        self.withLabel("Context action 3").withArgs([
            StringType("arg1").withLabel("Argument 1"),
            StringType("arg2").withLabel("Argument 2"),
            StringType("additionalText").withLabel("Additional text"),
        ]).withResult(StringType())
        self.withFeatures({"visible":False, "icon":"tortoise"})
    def onCall(self, arg1, arg2, additionalText):
        return arg1 + " " + arg2 + " " + additionalText

class ActionWithContextActionsContextAction4(Action):
    def onConfigure(self):
        self.withLabel("Context action 4").withArgs([
            StringType("arg1NotVisible").withLabel("Argument 1 not visible").withFeatures({"visible":False}),
            StringType("arg2").withLabel("Argument 2"),
        ]).withResult(StringType())
        self.withFeatures({"visible":False, "icon":"tortoise"})
    def onCall(self, arg1NotVisible, arg2):
        return arg1NotVisible + " " + arg2

class ActionWithContextActionsContextAction5(Action):
    def onConfigure(self):
        self.withLabel("Context action 5").withArgs([
            RecordType("arg").withFields([
                StringType("arg1").withLabel("Argument 1"),
                StringType("arg2").withLabel("Argument 2")
            ]).withFeatures({"visible":False}),
            StringType("additionalText").withLabel("Additional text")
        ]).withResult(StringType())
        self.withFeatures({"visible":False, "icon":"tortoise"})
    def onCall(self, arg, additionalText):
        return arg["arg1"] + " " + additionalText

Modes of passing arguments to a context action:

Default (e.g. "ContextAction"). In case of a context action for an action, all action arguments as a record (represented as a map) will be passed as the first argument of the context action. In case of a record argument or a list element, the record or the list element value will be passed as the first argument respectively.
Explicit substitution (format "ContextAction(targetArgName=sourceArgName,…)", e.g. "ContextAction(contextArg1=arg1,contextArg2=arg2)"). In case of a context action for an action, the argument contextArg1 of the context action will be set to a value of the argument arg1 of the parent action and the argument contextArg2 of the context action will be set to a value of the argument arg2.
Implicit substitution (format "ContextAction(sourceArgName,…)", e.g. "ContextAction(arg1)"). In case of a context action for an action, the first argument of the context action will be set to a value of the argument arg1 of the parent action.
No arguments passed (e.g. "ContextAction()").

The target argument must be have the same type as the source value.

A source argument name could be a path if a source value is a record, e.g. "ContextAction(contextArg1: arg1.field1)". The this keyword can be used as the source argument name. In case of a context action for an action it is a record of all parent action arguments. In case of a record argument or a list element it is the value itself.

If a context action requires more arguments than passed from a parent action, a new action call screen will be shown. If a context value is an annotated value, the screen will show a header containing a label of the annotated value.

The context actions feature is not propagated in an annotated value feature to sub-actions.

3.2. List-details

Figure 22. The list-details action in the action list

Figure 23. The list-details main action screen

Figure 24. The CRUD and context actions

action form list details subaction create

Figure 25. The create/add sub-action

Figure 26. The read/view sub-action

action form list details subaction update

Figure 27. The update/modify sub-action

action form list details subaction delete

Figure 28. The delete/remove sub-action

action form list details subaction context binary

Figure 29. The context action returning a binary result

action form list details subaction context no result

Figure 30. The context action with no result

action form list details subaction context args

Figure 31. The context action with an additional argument

action form list details subaction context args result

Figure 32. The result of the context action with an additional argument

The definitions of actions implementing the list-details

from org.openksavi.sponge.util.process import ProcessConfiguration

def createBookRecordType(name):
    """ Creates a book record type.
    """
    return RecordType(name).withFields([
        IntegerType("id").withLabel("ID").withNullable().withFeature("visible", False),
        StringType("author").withLabel("Author"),
        StringType("title").withLabel("Title")
])

class RecordLibraryForm(Action):
    def onConfigure(self):
        self.withLabel("Library (books as records)")
        self.withArgs([
            StringType("search").withNullable().withLabel("Search").withFeature("responsive", True),
            StringType("order").withLabel("Sort by").withProvided(ProvidedMeta().withValue().withValueSet()),
            ListType("books").withLabel("Books").withFeatures({
                    "createAction":"RecordCreateBook", "readAction":"RecordReadBook", "updateAction":"RecordUpdateBook", "deleteAction":"RecordDeleteBook",
                # Provided with overwrite to allow GUI refresh.
                }).withProvided(ProvidedMeta().withValue().withOverwrite().withDependencies(["search", "order"])).withElement(
                        createBookRecordType("book").withAnnotated()
                )
        ]).withNoResult().withCallable(False)
        self.withFeatures({
            "refreshLabel":None, "clearLabel":None, "cancelLabel":None,
        })
        self.withFeature("icon", "library-books")
    def onProvideArgs(self, context):
        global LIBRARY
        if "order" in context.names:
            context.provided["order"] = ProvidedValue().withValue("author").withAnnotatedValueSet([
                AnnotatedValue("author").withLabel("Author"), AnnotatedValue("title").withLabel("Title")])
        if "books" in context.names:
            context.provided["books"] = ProvidedValue().withValue(
                # Context actions are provided dynamically in an annotated value.
                map(lambda book: AnnotatedValue(book.toMap()).withLabel("{} - {}".format(book.author, book.title)).withFeature("contextActions", [
                        "RecordBookContextBinaryResult", "RecordBookContextNoResult", "RecordBookContextAdditionalArgs"]),
                    sorted(LIBRARY.findBooks(context.current["search"]), key = lambda book: book.author.lower() if context.current["order"] == "author" else book.title.lower())))

class RecordCreateBook(Action):
    def onConfigure(self):
        self.withLabel("Add a new book")
        self.withArg(
            createBookRecordType("book").withLabel("Book").withProvided(ProvidedMeta().withValue()).withFields([
                # Overwrite the author field.
                StringType("author").withLabel("Author").withProvided(ProvidedMeta().withValueSet(ValueSetMeta().withNotLimited())),
            ])
        ).withNoResult()
        self.withFeatures({"visible":False, "callLabel":"Save", "clearLabel":None, "cancelLabel":"Cancel", "icon":"plus-box"})

    def onCall(self, book):
        global LIBRARY
        LIBRARY.addBook(book["author"], book["title"])

    def onProvideArgs(self, context):
        global LIBRARY
        if "book" in context.names:
            # Create an initial, blank instance of a book and provide it to GUI.
            context.provided["book"] = ProvidedValue().withValue({})
        if "book.author" in context.names:
            context.provided["book.author"] = ProvidedValue().withValueSet(LIBRARY.getAuthors())

class RecordReadBook(Action):
    def onConfigure(self):
        self.withLabel("View the book")
        # Must set withOverwrite to replace with the current value.
        self.withArg(createBookRecordType("book").withAnnotated().withLabel("Book").withProvided(
            ProvidedMeta().withValue().withOverwrite().withDependency("book.id")))
        self.withNoResult().withCallable(False)
        self.withFeatures({"visible":False, "clearLabel":None, "callLabel":None, "cancelLabel":"Close", "icon":"book-open"})
    def onProvideArgs(self, context):
        global LIBRARY
        if "book" in context.names:
            context.provided["book"] = ProvidedValue().withValue(AnnotatedValue(LIBRARY.getBook(context.current["book.id"]).toMap()))

class RecordUpdateBook(Action):
    def onConfigure(self):
        self.withLabel("Modify the book")
        self.withArg(
            # Must set withOverwrite to replace with the current value.
            createBookRecordType("book").withAnnotated().withLabel("Book").withProvided(
                    ProvidedMeta().withValue().withOverwrite().withDependency("book.id")).withFields([
                StringType("author").withLabel("Author").withProvided(ProvidedMeta().withValueSet(ValueSetMeta().withNotLimited())),
            ])
        ).withNoResult()
        self.withFeatures({"visible":False, "clearLabel":None, "callLabel":"Save", "cancelLabel":"Cancel", "icon":"square-edit-outline"})
    def onCall(self, book):
        global LIBRARY
        LIBRARY.updateBook(book.value["id"], book.value["author"], book.value["title"])
    def onProvideArgs(self, context):
        global LIBRARY
        if "book" in context.names:
            context.provided["book"] = ProvidedValue().withValue(AnnotatedValue(LIBRARY.getBook(context.current["book.id"]).toMap()))
        if "book.author" in context.names:
            context.provided["book.author"] = ProvidedValue().withValueSet(LIBRARY.getAuthors())

class RecordDeleteBook(Action):
    def onConfigure(self):
        self.withLabel("Remove the book")
        self.withArg(createBookRecordType("book").withAnnotated()).withNoResult()
        self.withFeatures({"visible":False, "callLabel":"Save", "clearLabel":None, "cancelLabel":"Cancel", "icon":"delete", "confirmation":True})

    def onCall(self, book):
        global LIBRARY
        self.logger.info("Deleting book id: {}", book.value["id"])
        LIBRARY.removeBook(book.value["id"])

class RecordBookContextBinaryResult(Action):
    def onConfigure(self):
        self.withLabel("Text sample as PDF")
        self.withArg(
            createBookRecordType("book").withAnnotated().withFeature("visible", False)
        ).withResult(BinaryType().withAnnotated().withMimeType("application/pdf").withLabel("PDF"))
        self.withFeatures({"visible":False, "icon":"file-pdf"})
    def onCall(self, book):
        return AnnotatedValue(sponge.process(ProcessConfiguration.builder("curl", "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf")
                              .outputAsBinary()).run().outputBinary)

class RecordBookContextNoResult(Action):
    def onConfigure(self):
        self.withLabel("Return the book")
        self.withArg(
            createBookRecordType("book").withAnnotated().withFeature("visible", False)
        ).withNoResult().withFeatures({"visible":False, "icon":"arrow-left-bold"})
    def onCall(self, book):
        pass

class RecordBookContextAdditionalArgs(Action):
    def onConfigure(self):
        self.withLabel("Add book comment")
        self.withArgs([
            createBookRecordType("book").withAnnotated().withFeature("visible", False),
            StringType("comment").withLabel("Comment").withFeatures({"multiline":True, "maxLines":2})
        ]).withResult(StringType().withLabel("Added comment"))
        self.withFeatures({"visible":False, "icon":"comment-outline"})
    def onCall(self, book, message):
        return message

The main action should have a list argument that is provided with the overwrite option. The action shouldn’t be callable. The list argument type can be annotated and the provided list elements may have labels (AnnotatedValue().withLabel()) and descriptions. The list argument may have the following features: createAction, readAction, updateAction, deleteAction. Their values are the sub-action names that will be called to perform the CRUD operations.

There are two types of sub-actions: CRUD actions and context actions. CRUD actions implement create, read, update and delete operations. Context actions implement customized operations related to a list element.

The CRUD actions should not be visible in the actions list so they should have the visible feature set to False.

In the default scenario read, update and delete actions should have the first argument corresponding to the value of the list element. In most cases the argument visible feature should be set to False to hide it. Its type should be the same as the list element’s type. The value of the list element will be passed as this argument. In the case of a create action, no argument corresponding to any list element is necessary.

The result of a create, an update and a delete action is ignored and should be set to withNoResult.

After calling a CRUD action the main action arguments are refreshed.

4. User experience

Figure 33. The application dart theme

The application may be switched to the dark theme in the settings.

5. Supported Sponge concepts

5.1. Data types

Table 1. Data types
Type	Description
`AnyType`	Not supported.
`BinaryType`	Editing (as an action attribute) is supported only for `image/png` mime type with `drawing` characteristic feature. Viewing is supported for image formats supported by Flutter and other binary content supported by the `open_file` Flutter plugin that is used by the application.
`BooleanType`	Supported.
`DateTimeType`	Viewing supported. Editing is currently limited to the `DATE_TIME` date kind.
`DynamicType`	A limited support. This functionality is experimental.
`IntegerType`	Supported.
`ListType`	Editing is supported with limitations (see the Advanced use cases chapter). A unique list with a provided element value set is represented as a multichoice widget. Viewing is not supported. This functionality is experimental.
`MapType`	Not supported.
`NumberType`	Supported.
`ObjectType`	Not supported.
`RecordType`	Editing supported. Viewing support is limited to fields that have simple data types. This functionality is experimental.
`StreamType`	Not supported.
`StringType`	Supported.
`TypeType`	Not supported.
`VoidType`	Only viewing supported.

5.2. Data type formats

Table 2. Supported data type formats
Format	Description
`phone`	A phone number format. Applicable for `StringType`.
`email`	An email format. Applicable for `StringType`.
`url`	A URL format. Applicable for `StringType`.
`console`	A console format. Text is presented using a monospaced font. Applicable for `StringType`.
`markdown`	A Markdown format. Applicable for `StringType`.

5.3. Features

Table 3. Supported features
Feature	Applies to	Description
`visible`	Action	If `True`, an action will be visible in the main list of actions in the application. Defaults to `true`.
`visible`	Type	If `True`, an action argument or a record field is visible in the action call screen. Defaults to `true`.
`icon`	Action	An action icon name. The supported set of icons is limited to the material design icons (currently v3.6.95).
`confirmation`	Action	If `True` then before calling the action a confirmation dialog will be shown. Defaults to `false`.
`characteristic`	Type	A value of this feature indicates a special meaning of the type.
`characteristic: drawing`	`BinaryType`	A `BinaryType` that represents a drawing. Currently supported only for mime type `image/png`.
`characteristic: color`	`StringType`	A `StringType` that represents a color as a hexadecimal RGB string.
`filename`	`BinaryType`	A filename associated with a binary value.
`intent`	Action	An action with an intent is handled by the application in a specific way.
`intent: login`	Action	Should be set in an action that represents a user login in the user management functionality. See the user management example project.
`intent: logout`	Action	Should be set in an action that represents a user logout in the user management functionality.
`intent: signUp`	Action	Should be set in an action that implements a user sign up in the user management functionality.
`intent`	Type	A type with an intent is handled by the application in a specific way.
`intent: username`	Action argument type	Indicates that the action argument represents a username. Applies only for actions that implement the user management functionality. This intent may be omitted if an action argument name is `username`.
`intent: password`	Action argument type	Indicates that the action argument represents a password. Applies only for actions that implement the user management functionality. This intent may be omitted if an action argument name is `password`.
`multiline`	`StringType`	If `True`, a string will be multilined in the GUI. Defaults to `false`.
`maxLines`	`StringType`	A maximum number of lines in the GUI.
`obscure`	`StringType`	If `True`, a string will be obscured in the GUI, e.g. in case of passwords. Defaults to `false`.
`callLabel`	Action	An action call button label in the action call screen. Defaults to `RUN` if the feature is not set. If set to `None`, the button will not be shown.
`refreshLabel`	Action	An action refresh button label in the action call screen. Defaults to `REFRESH` if the feature is not set. If set to `None`, the button will not be shown. The refresh button fetches provided action arguments from the server. Only arguments provided with read only or overwrite flags will be refreshed.
`clearLabel`	Action	An action clear button label in the action call screen. Defaults to `CLEAR` if the feature is not set. If set to `None`, the button will not be shown. The clear button resets the action argument values.
`cancelLabel`	Action	An action cancel button label in the action call screen. Defaults to `CANCEL` if the feature is not set. If set to `None`, the button will not be shown.
`contextActions`	Action, `RecordType`, `ListType` action argument	Context actions. For more information on context actions and sub-actions see the Advanced use cases chapter.
`createAction`	`ListType` action argument	A create sub-action for a list element.
`readAction`	`ListType` action argument	A read sub-action for a list element.
`updateAction`	`ListType` action argument	A update sub-action for a list element.
`deleteAction`	`ListType` action argument	A delete sub-action for a list element.
`width`	`BinaryType` drawing	An image width.
`height`	`BinaryType` drawing	An image height.
`strokeWidth`	`BinaryType` drawing	A drawing stroke width.
`color`	`BinaryType` drawing	A drawing pen color.
`background`	`BinaryType` drawing	A drawing background color.

A data type property can be dynamically overwritten by a corresponding feature in an AnnotatedValue. The feature name has to be exactly the same as the data type property name. The overwrite is handled by the Sponge mobile client application.

6. Included demos

The access to actions in the mobile application is generic. However the application may include demos that use a customized UI.

6.1. Handwritten digit recognition

Figure 34. The navigation drawer if connected to a Sponge instance that supports a digit recognition

If the current connection points to a Sponge instance that has the required action that performs a handwritten digit recognition, this demo is enabled in the navigation drawer.

Figure 35. The digit recognition demo - the information dialog

Figure 36. The digit recognition demo - drawing a digit

The digit recognition demo screen allows drawing a digit that will be recognized by the Sponge action. After each stroke the remote action call is made and the result is shown in the circle.