The main Web of Things (WoT) concepts are described in the WoT Architecture document. The Web of Things is made of entities (Things) that can describe their capabilities in a machine-interpretable Thing Description (TD) and expose these capabilities through the WoT Interface, that is, network interactions modeled as Properties (for reading and writing values), Actions (to execute remote procedures with or without return values) and Events (for signaling notifications).

Scripting is an optional "convenience" building block in WoT and it is typically used in gateways that are able to run a WoT Runtime and script management, providing a convenient way to extend WoT support to new types of endpoints and implement WoT applications such as Thing Directory.

This specification describes a programming interface representing the WoT Interface that allows scripts to discover, operate Things and to expose locally defined Things characterized by WoT Interactions specified by a script.

The specification deliberately follows the WoT Thing Description specification closely. It is possible to implement simpler APIs on top of this API, or implementing directly the WoT network facing interface (i.e. the WoT Interface).

This specification is implemented at least by the Thingweb project also known as node-wot, which is considered the reference open source implementation at the moment. Check its source code, including examples. Other, closed source implementations have been made by WG member companies and tested against node-wot in plug-fests.

Implementers need to be aware that this specification is considered unstable. Vendors interested in implementing this specification before it eventually reaches the Candidate Recommendation phase should subscribe to the repository and take part in the discussions.

Please contribute to this draft using the GitHub Issue feature of the WoT Scripting API repository. For feedback on security and privacy considerations, please use the WoT Security and Privacy Issues.

Introduction

WoT provides layered interoperability based on how Things are used: "consumed" and "exposed", as defined in [[WOT-ARCHITECTURE]].

By consuming a TD, a client Thing creates a local runtime resource model that allows accessing the Properties, Actions and Events exposed by the server Thing on a remote device.

Exposing a Thing requires: This specification describes how to expose and consume Things by a script. Also, it defines a generic API for Thing discovery.

Typically scripts are meant to be used on bridges or gateways that expose and control simpler devices as WoT Things and have means to handle (e.g. install, uninstall, update etc.) and run scripts.

This specification does not make assumptions on how the WoT Runtime handles and runs scripts, including single or multiple tenancy, script deployment and lifecycle management. The API already supports the generic mechanisms that make it possible to implement script management, for instance by exposing a manager Thing whose Actions (action handlers) implement script lifecycle management operations.

Use Cases

The following scripting use cases are supported in this specification:

Consuming a Thing

Exposing a Thing

Discovery

This specification describes the conformance criteria for the following classes of user agent (UA).

Due to requirements of small embedded implementations, splitting WoT client and server interfaces was needed. Then, discovery is a distributed application, but typical scenarios have been covered by a generic discovery API in this specification. This resulted in using 3 conformance classes for a UA that implements this API, one for client, one for server, and one for discovery. An application that uses this API can introspect for the presence of the consume(), produce() and discover() methods on the WoT API object in order to determine which conformance class the UA implements.

WoT Consumer UA

Implementations of this conformance class MUST implement the {{ConsumedThing}} interface and the consume() method on the WoT API object.

WoT Producer UA

Implementations of this conformance class MUST implement {{ExposedThing}} interface and the produce() method on the WoT API object.

WoT Discovery UA

Implementations of this conformance class MUST implement the ThingDiscovery interface and the discover() method on the WoT API object.

These conformance classes MAY be implemented in a single UA.

This specification can be used for implementing the WoT Scripting API in multiple programming languages. The interface definitions are specified in [[!WEBIDL]].

The UA may be implemented in the browser, or in a separate runtime environment, such as Node.js or in small embedded runtimes.

Implementations that use ECMAScript executed in a browser to implement the APIs defined in this document MUST implement them in a manner consistent with the ECMAScript Bindings defined in the Web IDL specification [[!WEBIDL]].

Implementations that use TypeScript or ECMAScript in a runtime to implement the APIs defined in this document MUST implement them in a manner consistent with the TypeScript Bindings defined in the TypeScript specification [[!TYPESCRIPT]].

The ThingDescription type

      typedef object ThingDescription;
    

Represents a Thing Description (TD) as defined in [[!WOT-TD]]. It is expected to be a parsed JSON object that is validated using JSON schema validation.

Fetching a Thing Description

Fetching a TD given a URL should be done with an external method, such as the Fetch API or a HTTP client library, which offer already standardized options on specifying fetch details.

        try {
          let res = await fetch('https://tds.mythings.biz/sensor11');
          // ... additional checks possible on res.headers
          let td = await res.json();
          let thing = new ConsumedThing(td);
          console.log("Thing name: " + thing.getThingDescription().title);
        } catch (err) {
          console.log("Fetching TD failed", err.message);
        }
      

Expanding a Thing Description

Note that [[WOT-TD]] allows using a shortened Thing Description by the means of defaults and requiring clients to expand them with default values specified in [[WOT-TD]] for the properties that are not explicitly defined in a given TD.

To expand a TD given |td:ThingDescription|, run the following steps:
  1. For each item in the TD default values table from [[!WOT-TD]], if the term is not defined in |td|, add the term definition with the default value specified in [[!WOT-TD]].

Validating a Thing Description

The [[!WOT-TD]] specification defines how a TD should be validated. Therefore, this API expects the {{ThingDescription}} objects be validated before used as parameters. This specification defines a basic TD validation as follows.

To validate a TD given |td:ThingDescription|, run the following steps:
  1. If |td| is not an object, [= exception/throw =] a {{"TypeError"}} and abort these steps.
  2. If any of the mandatory properties defined in [[!WOT-TD]] for Thing that don't have default definitions are missing from |td|, [= exception/throw =] a {{"TypeError"}} and abort these steps.
  3. If JSON schema validation fails on |td|, [= exception/throw =] a {{"TypeError"}} and abort these steps.

The WoT API object

Defines the API entry point exposed as a singleton and contains the API methods.

The WOT interface

      [SecureContext, Exposed=(Window,Worker)]
      interface WOT {
        // methods defined in UA conformance classes
      };
    

Browser implementations should use a namespace object such as navigator.wot. Standalone runtimes may expose the API object through mechanisms like require() or import.

The consume() method

      partial interface WOT {
        Promise<ConsumedThing> consume(ThingDescription td);
      };
    
Belongs to the WoT Consumer conformance class. Expects an |td:ThingDescription| argument and returns a {{Promise}} that resolves with a {{ConsumedThing}} object that represents a client interface to operate with the Thing. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Run the validate a TD steps on |td|. If that [= exception/throws =], reject |promise| with that error and abort these steps.
  4. Let |thing:ConsumedThing| be a new {{ConsumedThing}} object constructed from |td|.
  5. Set up the WoT Interactions based on introspecting td as explained in [[!WOT-TD]] and [[!WOT-PROTOCOL-BINDINGS]]. Make a request to the underlying platform to initialize the Protocol Bindings. The details are private to the implementations and out of scope of this specification.
  6. Resolve |promise| with |thing|.

The produce() method

      partial interface WOT {
        Promise<ExposedThing> produce(ThingDescription td);
      };
    
Belongs to the WoT Producer conformance class. Expects a |td:ThingDescription| argument and returns a {{Promise}} that resolves with an {{ExposedThing}} object that extends {{ConsumedThing}} with a server interface, i.e. the ability to define request handlers. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Let |thing:ExposedThing| be a new {{ExposedThing}} object constructed with |td|.
  4. Resolve |promise| with |thing|.

The discover() method

      partial interface WOT {
        ThingDiscovery discover(optional ThingFilter filter = null);
      };
    
Belongs to the WoT Discovery conformance class. Starts the discovery process that will provide {{ThingDescription}} objects for Thing Descriptions that match an optional |filter:ThingFilter| argument of type {{ThingFilter}}. The method MUST run the following steps:
  1. If invoking this method is not allowed for the current scripting context for security reasons, [= exception/throw =] a {{"SecurityError"}} and abort these steps.
  2. Construct a ThingDiscovery object |discovery:ThingDiscovery| with |filter|.
  3. Invoke the discovery.start() method.
  4. Return |discovery|.

The ConsumedThing interface

Represents a client API to operate a Thing. Belongs to the WoT Consumer conformance class.

      [SecureContext, Exposed=(Window,Worker)]
      interface ConsumedThing {
        constructor(ThingDescription td);
        Promise<any> readProperty(DOMString propertyName,
                                  optional InteractionOptions options = null);
        Promise<PropertyMap> readAllProperties(optional InteractionOptions options = null);
        Promise<PropertyMap> readMultipleProperties(
                                  sequence<DOMString> propertyNames,
                                  optional InteractionOptions options = null);
        Promise<void> writeProperty(DOMString propertyName,
                                    any value,
                                    optional InteractionOptions options = null);
        Promise<void> writeMultipleProperties(PropertyMap valueMap,
                                    optional InteractionOptions options = null);
        Promise<any> invokeAction(DOMString actionName,
                                    optional any params = null,
                                    optional InteractionOptions options = null);
        Promise<void> observeProperty(DOMString name,
                                    WotListener listener,
                                    optional InteractionOptions options = null);
        Promise<void> unobserveProperty(DOMString name);
        Promise<void> subscribeEvent(DOMString name,
                                    WotListener listener,
                                    optional InteractionOptions options = null);
        Promise<void> unsubscribeEvent(DOMString name);
        ThingDescription getThingDescription();
      };
      dictionary InteractionOptions {
        unsigned long formIndex;
        object uriVariables;
      };
      typedef object PropertyMap;
      callback WotListener = void(any data);
    

Constructing ConsumedThing

After fetching a Thing Description as a JSON object, one can create a {{ConsumedThing}} object.

To create {{ConsumedThing}} with the {{ThingDescription}} |td:ThingDescription|, run the following steps:
  1. Run the expand a TD steps on |td|. If that fails, re-[= exception/throw =] the error and abort these steps.
  2. Let |thing:ConsumedThing| be a new {{ConsumedThing}} object.
  3. Let ||td|| be an internal slot of |thing| and let |td| be its value.
  4. Return |thing|.

The getThingDescription() method

Returns the internal slot ||td|| of the {{ConsumedThing}} object that represents the Thing Description of the {{ConsumedThing}}. Applications may consult the Thing metadata stored in ||td|| in order to introspect its capabilities before interacting with it.

The InteractionOptions dictionary

Holds the interaction options that need to be exposed for application scripts according to the Thing Description.

The formIndex property, if defined, represents an application hint for which Form definition, identified by this index, of the TD to use for the given WoT interaction. Implementations SHOULD use the Form with this index for making the interaction, but MAY override this value if the index is not found or not valid. If not defined, implementations SHOULD attempt to use the Form definitions in order of appearance as listed in the TD for the given Wot Interaction.

The uriVariables property if defined, represents the URI template variables to be used with the WoT Interaction that are represented as parsed JSON objects defined in [[!WOT-TD]].

The support for URI variables comes from the need exposed by [[WOT-TD]] to be able to describe existing TDs that use them, but it should be possible to write a Thing Description that would use Actions for representing the interactions that need URI variables and represent the URI variables as parameters to the Action and in that case that could be encapsulated by the implementations and the |options| parameter could be dismissed from the methods exposed by this API.

The PropertyMap type

Represents a map of Property names as strings to a value that the Property can take. It is used as a property bag for interactions that involve multiple Properties at once.

It could be defined in Web IDL (as well as {{ThingDescription}}) as a maplike interface from string to any.

The readProperty() method

Reads a Property value. Takes a string argument |propertyName| and and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns a Property value represented as `any` type. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform (via the Protocol Bindings) to retrieve the value of the Property given by |propertyName| with optional URI templates given in |options|' {{uriVariables}}.
  4. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  5. Let |value| be the result of the request.
  6. Run the validate Property value sub-steps on |value|:
    1. Based on the DataSchema definition, |value| MUST be a JSON value and comply to the data schema defined for the Property that is found in Thing Description. If this fails, [= exception/throw =] a {{"SyntaxError"}} and abort these sub-steps.
    2. Return |value|.
  7. If these above sub-steps failed, reject |promise| with {{SyntaxError}} and abort these steps.
  8. Otherwise resolve |promise| with |value|.

The readMultipleProperties() method

Reads multiple Property values with one or multiple requests. Takes the |propertyNames: string sequence| argument as a sequence of strings and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns an object that maps keys from |propertyNames| to values returned by this algorithm. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Let |result:object| be an object and for each string |name:string| in |propertyNames| add a property with key |name| and the value `null`.
  4. Make a request to the underlying platform (via the Protocol Bindings) to retrieve the Property values given by |propertyNames| with optional URI templates given in |options|' {{uriVariables}}.
  5. If this cannot be done with a single request with the Protocol Bindings of the Thing, then reject |promise| with a {{NotSupportedError}} and abort these steps.
  6. Process the reply and update all properties in |result| with the values obtained in the reply.
  7. If the above step fails at any point, reject |promise| with a {{SyntaxError}} and abort these steps.
  8. Resolve |promise| with |result|.

The readAllProperties() method

Reads all properties of the Thing with one or multiple requests. Takes an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns an object that maps keys from Property names to values returned by this algorithm. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform (via the Protocol Bindings) to retrieve the value of the all the Property definitions from the TD with optional URI templates given in |options|' {{uriVariables}}.
  4. If this cannot be done with a single request with the Protocol Bindings of the Thing, then reject |promise| with a {{NotSupportedError}} and abort these steps.
  5. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  6. Process the reply and let |result:object| be an object with the keys and values obtained in the reply.
  7. Resolve |promise| with |result|.

The writeProperty() method

Writes a single Property. Takes a string argument |propertyName:string|, a value argument |value:any| and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Run the validate Property value steps on |value|. If that fails, reject promise with {{SyntaxError}} and abort these steps.
  4. Make a request to the underlying platform (via the Protocol Bindings) to write |value| to the Property given by |propertyName| with optional URI templates given in |options.uriVariables|.
  5. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  6. Otherwise resolve |promise|.

As discussed in Issue #193, the design decision is that write interactions only return success or error, not the written value (optionally). TDs should capture the schema of the Property values, including precision and alternative formats. When a return value is expected from the interaction, an Action should be used instead of a Property.

The writeMultipleProperties() method

Writes a multiple Property values with one request. Takes a |properties:object| argument as an object with keys being Property names and values as Property values and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. For each key |name:string| in |properties|, take its value as |value| and run the validate Property value steps on |value|. If that fails for any |name|, reject promise with a {{SyntaxError}} and abort these steps.
  4. Make a single request to the underlying platform (via the Protocol Bindings) to write each Property provided in |properties| with optional URI templates given in |options| {{uriVariables}}.
  5. If this cannot be done with a single request with the Protocol Bindings of the Thing, then reject |promise| with a {{NotSupportedError}} and abort these steps.
  6. If the request fails, return the error received from the Protocol Bindings and abort these steps.
  7. Otherwise resolve |promise|.

The WotListener callback

User provided callback that takes `any` argument and is used for observing Property changes and handling Event notifications. Since subscribing to these are WoT interactions, they are not modelled with software events.

The observeProperty() method

Makes a request for Property value change notifications. Takes a string argument |propertyName:string|, a {{WotListener}} callback function |listener:WotListener| and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. If |listener| is not a {{Function}}, reject |promise| with a {{TypeError}} and abort these steps.
  4. Make a request to the underlying platform (via the Protocol Bindings) to observe Property identified by |propertyName| with optional URI templates given in |options|' {{uriVariables}}.
  5. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  6. Otherwise resolve |promise|.
  7. Whenever the underlying platform receives a notification for this subscription with new Property value |value|, run the following sub-steps:
    • If running the validate Property value steps on |value| fails, abort these steps.
    • Invoke |listener| with |value| as parameter.

The unobserveProperty() method

Makes a request for unsubscribing from Property value change notifications. Takes a string argument |propertyName:string| and returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform (via the Protocol Bindings) to stop observing the Property identified by |propertyName|.
  4. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  5. Otherwise resolve |promise|.

The invokeAction() method

Makes a request for invoking an Action and return the result. Takes a string argument |actionName:string|, an optional argument |params:any| and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns the result of the Action or an error. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform (via the Protocol Bindings) to invoke the Action identified by |actionName| with parameters provided in |params| with optional URI templates given in |options|'s {{uriVariables}}.
  4. If the request fails locally or returns an error over the network, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  5. Otherwise let |value| be the result returned in the reply and run the validate Property value steps on it. If that fails, reject |promise| with a {{SyntaxError}} and abort these steps.
  6. Reject |promise| with |value|.

The subscribeEvent() method

Makes a request for subscribing to Event notifications. Takes a string argument |eventName|, a {{WotListener}} callback function |listener:WoTListener| and an optional {{InteractionOptions}} |options:InteractionOptions| argument. It returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. If |listener| is not a {{Function}}, reject |promise| with a {{TypeError}} and abort these steps.
  4. Make a request to the underlying platform (via the Protocol Bindings) to subscribe to an Event identified by |eventName:string| with optional URI templates given in |options|' {{uriVariables}}.
  5. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  6. Otherwise resolve |promise|.
  7. Whenever the underlying platform receives a notification for this Event subscription, implementations SHOULD invoke |listener|, giving the data provided with the Event as parameter.

The unsubscribeEvent() method

Makes a request for unsubscribing from Event notifications. Takes a string argument |eventName:string| and returns success or failure. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform (via the Protocol Bindings) to unsubscribe from the Event identified by |eventName|.
  4. If the request fails, reject |promise| with the error received from the Protocol Bindings and abort these steps.
  5. Otherwise resolve |promise|.

ConsumedThing Examples

The next example illustrates how to fetch a TD by URL, create a {{ConsumedThing}}, read metadata (title), read property value, subscribe to property change, subscribe to a WoT event, unsubscribe.

        try {
          let res = await fetch("https://tds.mythings.org/sensor11");
          let td = res.json();
          let thing = new ConsumedThing(td);
          console.log("Thing " + thing.getThingDescription().title + " consumed.");
        } catch(e) {
          console.log("TD fetch error: " + e.message); },
        };
        try {
          // subscribe to property change for “temperature”
          await thing.observeProperty("temperature", value => {
            console.log("Temperature changed to: " + value);
          });
          // subscribe to the “ready” event defined in the TD
          await thing.subscribeEvent("ready", eventData => {
            console.log("Ready; index: " + eventData);
            // run the “startMeasurement” action defined by TD
            await thing.invokeAction("startMeasurement", { units: "Celsius" });
            console.log("Measurement started.");
          });
        } catch(e) {
          console.log("Error starting measurement.");
        }
        setTimeout( () => {
          console.log(“Temperature: “ + await thing.readProperty(“temperature”));
          await thing.unsubscribe(“ready”);
          console.log("Unsubscribed from the ‘ready’ event.");
        },
        10000);
      

The ExposedThing interface

The {{ExposedThing}} interface is the server API to operate the Thing that allows defining request handlers, Property, Action, and Event interactions.

      [SecureContext, Exposed=(Window,Worker)]
      interface ExposedThing: ConsumedThing {
        ExposedThing setPropertyReadHandler(DOMString name,
                PropertyReadHandler readHandler);
        ExposedThing setPropertyWriteHandler(DOMString name,
                PropertyWriteHandler writeHandler);
        ExposedThing setActionHandler(DOMString name, ActionHandler action);
        void emitEvent(DOMString name, any data);
        Promise<void> expose();
        Promise<void> destroy();
      };
      callback PropertyReadHandler = Promise<any>(
              optional InteractionOptions options = null);
      callback PropertyWriteHandler = Promise<void>(any value,
              optional InteractionOptions options = null);
      callback ActionHandler = Promise<any>(any params,
              optional InteractionOptions options = null);
    

Constructing {{ExposedThing}}

The {{ExposedThing}} interface extends {{ConsumedThing}}. It is constructed from a full or partial {{ThingDescription}} object.

Note that an existing {{ThingDescription}} object can be optionally modified (for instance by adding or removing elements on its |properties|, |actions| and |events| internal properties) and the resulting object can used for constructing an {{ExposedThing}} object. This is the current way of adding and removing Property, Action and Event definitions, as illustrated in the examples.

Before invoking expose(), the {{ExposedThing}} object does not serve any requests. This allows first constructing {{ExposedThing}} and then initialize its Properties and service handlers before starting serving requests.

To construct an {{ExposedThing}} with the {{ThingDescription}} |td:ThingDescription|, run the following steps:
  1. If invoking this method is not allowed for the current scripting context for security reasons, [= exception/throw =] a {{SecurityError}} and abort these steps.
  2. Run the expand a TD steps on |td|. If that fails, re-[= exception/throw =] the error and abort these steps.
  3. Let |thing:ExposedThing| be a new {{ExposedThing}} object.
  4. Let ||td|| be an internal slot of |thing| and let |td| be its value.
  5. Return |thing|.

Methods inherited from {{ConsumedThing}}

The readProperty(), readMultipleProperties(), readAllProperties(), writeProperty(), writeMultipleProperties(), writeAllProperties() methods have the same algorithmic steps as described in ConsumedThing, with the difference that making a request to the underlying platform MAY be implemented with local methods or libraries and don't necessarily need to involve network operations.

The implementation of {{ConsumedThing}} interface in an {{ExposedThing}} provide the default methods to interact with the {{ExposedThing}}.

After constructing an {{ExposedThing}}, a script can initialize its Properties and can set up the optional read, write and action request handlers (the default ones are provided by the implementation). The script provided handlers MAY use the default handlers, thereby extending the default behavior, but they can also bypass them, overriding the default behavior. Finally, the script would call expose() on the {{ExposedThing}} in order to start serving external requests.

The PropertyReadHandler callback

A function that is called when an external request for reading a Property is received and defines what to do with such requests. It returns a {{Promise}} and resolves it when the value of the Property matching the |name:string| argument is obtained, or rejects with an error if the property is not found or the value cannot be retrieved.

The setPropertyReadHandler() method

Takes |name:string| as string argument and |readHandler:PropertyReadHandler| as argument of type PropertyReadHandler. Sets the service handler for reading the specified Property matched by |name|. Throws on error. Returns a reference to |this| object for supporting chaining.

The |readHandler| callback function should implement reading a Property and SHOULD be called by implementations when a request for reading a Property is received from the underlying platform.

There MUST be at most one handler for any given Property, so newly added handlers MUST replace the previous handlers. If no handler is initialized for any given Property, implementations SHOULD implement a default property read handler based on the Thing Description.

Handling Property read requests

When a network request for reading Property |propertyName:string| is received by the implementation, run the following steps:
  1. If a Property with |propertyName| does not exist, return a {{ReferenceError}} in the reply and abort these steps.
  2. If there is a user provided read handler registered with setPropertyReadHandler(), invoke that given |propertyName|, return the obtained value in the reply and abort these steps.
  3. Otherwise, if there is a default read handler provided by the implementation, invoke it with |propertyName|, include the returned |value| with the reply and abort these steps.
  4. if there is no default handler defined by the implementation, return a {{NotSupportedError}} with the reply and abort these steps.

Handling Property observe requests

When a network request for observing a Property |propertyName:string| is received by the implementation, run the following steps:
  1. If a Property with |propertyName| does not exist, return an error in the reply (as defined in the Thing Description) and abort these steps.
  2. Save the request sender information to the Property's internal observer list in order to be able to notify about Property value changes.

The PropertyWriteHandler callback

A function that is called when an external request for writing a Property is received and defines what to do with such requests. It expects the requested new |value| as argument and returns a {{Promise}} which is resolved when the value of the Property that matches the |name:string| argument has been updated with |value|, or rejects with an error if the property is not found or the value cannot be updated.

Note that the code in this callback function can read the property before updating it in order to find out the old value, if needed. Therefore the old value is not provided to this function.

The setPropertyWriteHandler() method

Takes |name:string| as string argument and |writeHandler:PropertyWriteHandler| as argument of type PropertyWriteHandler. Sets the service handler that will be used for serving write requests for writing the specified Property matched by |name|. Throws on error. Returns a reference to |this| object for supporting chaining.

Note that even for readonly Properties it is possible to specify a write handler, as explained in Issue 199. In this case, the write handler may define in an application-specific way to fail the request.

There MUST be at most one write handler for any given Property, so newly added handlers MUST replace the previous handlers. If no write handler is initialized for any given Property, implementations SHOULD implement default property update if the Property is writeable and notifying observers on change if the Property is observeable, based on the Thing Description.

Handling Property write requests

When a network request for writing a Property |propertyName:string| with a new value |value| is received, implementations SHOULD run the following update property steps, given |propertyName|, |value| and |mode| set to `"single"`:
  1. If a Property with |propertyName| does not exist, return a {{ReferenceError}} in the reply and abort these steps.
  2. If there is a user provided write handler registered with setPropertyWriteHandler(), or if there is a default write handler,
    1. Invoke the handler with |propertyName|. If it fails, return the error in the reply and abort these steps.
    2. Otherwise, if |mode| is `"single"`, reply to the request with the new value, following to the Protocol Bindings.
    3. For each item stored in the internal observer list of the Property with |propertyName|, send an observe reply with the new value attached.
  3. If there is no handler to handle the request, return a {{NotSupportedError}} in the reply and abort these steps.
When a network request for writing multiple Properties given in an object |propertyNames| is received, run the following steps:
  1. For each property with key |name| and value |value| defined in |propertyNames|, run the update property steps with |name|, |value| and |mode| set to `"multiple"`.
  2. Reply to the request (by sending a single or multiple replies) according to the Protocol Bindings defined for the Property.

The ActionHandler callback

A function that is called when an external request for invoking an Action is received and defines what to do with such requests. It is invoked with a |params:object| dictionary argument. It returns a {{Promise}} that rejects with an error or resolves if the action is successful.

The setActionHandler() method

Takes |name:string| as string argument and |action:ActionHandler| as argument of type ActionHandler. Sets the handler function for the specified Action matched by |name|. May throw an error. Returns a reference to |this| object for supporting chaining.

The |action| callback function will implement an Action and SHOULD be called by implementations when a request for invoking the Action is received from the underlying platform.

There MUST be at most one handler for any given Action, so newly added handlers MUST replace the previous handlers.

Handling Action requests

When a network request for invoking the Action identified by |name:string| is received, the runtime SHOULD execute the following steps:
  1. If an Action identified by |name| does not exist, return a {{ReferenceError}} in the reply and abort these steps.
  2. If there is a user provided action handler registered with setActionHandler(), invoke that wih |name|, return the resulting |value| with the reply and abort these steps.
  3. Otherwise return a {{NotSupportedError}} with the reply and abort these steps.

The emitEvent() method

Takes a |name:string| argument denoting an Event name, and a |data:any| argument of `any` type. The method MUST run the following steps:
  1. If invoking this method is not allowed for the current scripting context for security reasons, [= exception/throw =] a {{SecurityError}} and abort these steps.
  2. If an Event with the name |name| is not found, [= exception/throw =] a {{NotFoundError}} and abort these steps.
  3. Make a request to the underlying platform to send an Event with |data| attached as property, using the Protocol Bindings, then abort these steps.

The expose() method

Start serving external requests for the Thing, so that WoT Interactions using Properties, Actions and Events will be possible. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Run the expand a TD steps on the internal slot ||td||.
  4. Run the validate a TD on ||td||. If that fails, reject |promise| with a {{TypeError}} and abort these steps.
  5. For each Property definition in ||td||, initialize an |internal observer list| internal slot in order to store observe request data needed to notify the observers on value changes.
  6. Set up the WoT Interactions based on introspecting ||td|| as explained in [[!WOT-TD]] and [[!WOT-PROTOCOL-BINDINGS]]. Make a request to the underlying platform to initialize the Protocol Bindings and then start serving external requests for WoT Interactions (read, write and observe Properties, invoke Actions and manage Event subscriptions), based on the Protocol Bindings. The details are private to the implementations and out of scope of this specification.
  7. If there was an error during the request, reject |promise| with an {{Error}} object |error| with |error|'s |message| set to the error code seen by the Protocol Bindings and abort these steps.
  8. Otherwise resolve |promise| and abort these steps.

The destroy() method

Stop serving external requests for the Thing and destroy the object. Note that eventual unregistering should be done before invoking this method. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If invoking this method is not allowed for the current scripting context for security reasons, reject |promise| with a {{SecurityError}} and abort these steps.
  3. Make a request to the underlying platform to stop serving external requests for WoT Interactions, based on the Protocol Bindings.
  4. If there was an error during the request, reject |promise| with an {{Error}} object |error| with its |message| set to the error code seen by the Protocol Bindings and abort these steps.
  5. Otherwise resolve |promise| and abort these steps.

ExposedThing Examples

The next example illustrates how to create an {{ExposedThing}} based on a partial TD object constructed beforehands.

        try {
          let temperaturePropertyDefinition = {
            type: "number",
            minimum: -50,
            maximum: 10000
          };
          let tdFragment = {
            properties: {
              temperature: temperaturePropertyDefinition
            },
            actions: {
              reset: {
                description: "Reset the temperature sensor",
                input: {
                  temperature: temperatureValueDefinition
                },
                output: null,
                forms: []
              },
            },
            events: {
              onchange: temperatureValueDefinition
            }
          };
          let thing1 = await WOT.produce(tdFragment);
          // initialize Properties
          await thing1.writeProperty("temperature", 0);
          // add service handlers
          thing1.setPropertyReadHandler("temperature", () => {
             return readLocalTemperatureSensor();  // Promise
          });
          // start serving requests
          await thing1.expose();
        } catch (err) {
           console.log("Error creating ExposedThing: " + err);
        }
      

The next example illustrates how to add or modify a Property definition on an existing {{ExposedThing}}: take its |td| property, add or modify it, then create another {{ExposedThing}} with that.

        try {
          // create a deep copy of thing1's TD
          let instance = JSON.parse(JSON.stringify(thing1.td));
          const statusValueDefinition = {
            type: "object",
            properties: {
              brightness: {
                type: "number",
                minimum: 0.0,
                maximum: 100.0,
                required: true
              },
              rgb: {
                type: "array",
                "minItems": 3,
                "maxItems": 3,
                items : {
                    "type" : "number",
                    "minimum": 0,
                    "maximum": 255
                }
              }
          };
          instance["name"] = "mySensor";
          instance.properties["brightness"] = {
            type: "number",
            minimum: 0.0,
            maximum: 100.0,
            required: true,
          };
          instance.properties["status"] = statusValueDefinition;
          instance.actions["getStatus"] = {
            description: "Get status object",
            input: null,
            output: {
              status : statusValueDefinition;
            },
            forms: [...]
          };
          instance.events["onstatuschange"] = statusValueDefinition;
          instance.forms = [...];  // update
          var thing2 = new ExposedThing(instance);
          // TODO: add service handlers
          await thing2.expose();
          });
        } catch (err) {
           console.log("Error creating ExposedThing: " + err);
        }
      

The ThingDiscovery interface

Discovery is a distributed application that requires provisioning and support from participating network nodes (clients, servers, directory services). This API models the client side of typical discovery schemes supported by various IoT deployments.

The {{ThingDiscovery}} object is constructed given a filter and provides the properties and methods controlling the discovery process.

      [SecureContext, Exposed=(Window,Worker)]
      interface ThingDiscovery {
        constructor(optional ThingFilter filter = null);
        readonly attribute ThingFilter? filter;
        readonly attribute boolean active;
        readonly attribute boolean done;
        readonly attribute Error? error;
        void start();
        Promise<ThingDescription> next();
        void stop();
      };
    

The {{ThingDiscovery}} interface has a next() method and a done property, but it is not an Iterable. Look into Issue 177 for rationale.

The discovery results internal slot is an internal queue for temporarily storing the found {{ThingDescription}} objects until they are consumed by the application using the next() method. Implementations MAY optimize the size of this queue based on e.g. the available resources and the frequency of invoking the next() method.

The filter property represents the discovery filter of type ThingFilter specified for the discovery.

The active property is `true` when the discovery is actively ongoing on protocol level (i.e. new TDs may still arrive) and `false` otherwise.

The done property is `true` if the discovery has been completed with no more results to report and discovery results is also empty.

The error property represents the last error that occured during the discovery process. Typically used for critical errors that stop discovery.

Constructing {{ThingDiscovery}}

To create {{ThingDiscovery}} with a |filter:ThingFilter| or type {{ThingFilter}}, run the following steps:
  1. If |filter| is not an object or `null`, [= exception/throw =] a {{TypeError}} and abort these steps.
  2. Let |discovery:ThingDiscovery| be a new {{ThingDiscovery}} object.
  3. Set the filter property to |filter|.
  4. Set the active and done properties to `false`. Set the error property to `null`.
  5. Return |discovery|.

The start() method sets active to `true`. The stop() method sets the active property to |false|, but done may be still `false` if there are {{ThingDescription}} objects in the discovery results not yet consumed with next().

During successive calls of next(), the active property may be `true` or `false`, but the done property is set to `false` by next() only when both the active property is `false` and discovery results is empty.

The DiscoveryMethod enumeration

        typedef DOMString DiscoveryMethod;
      

Represents the discovery type to be used:

The ThingFilter dictionary

Represents an object containing the constraints for discovering Things as key-value pairs.

        dictionary ThingFilter {
          (DiscoveryMethod or DOMString) method = "any";
          USVString? url;
          USVString? query;
          object? fragment;
        };
      

The method property represents the discovery type that should be used in the discovery process. The possible values are defined by the DiscoveryMethod enumeration that MAY be extended by string values defined by solutions (with no guarantee of interoperability).

The url property represents additional information for the discovery method, such as the URL of the target entity serving the discovery request, for instance the URL of a Thing Directory (if method is `"directory"`), or otherwise the URL of a directly targeted Thing.

The query property represents a query string accepted by the implementation, for instance a SPARQL or JSON query. Support may be implemented locally in the WoT Runtime or remotely as a service in a Thing Directory.

The fragment property represents a template object used for matching property by property against discovered Things.

The start() method

Starts the discovery process. The method MUST run the following steps:
  1. If invoking this method is not allowed for the current scripting context for security reasons, set the error property to a {{SecurityError}} and abort these steps.
  2. If discovery is not supported by the implementation, set the error property to {{NotSupportedError}} and abort these steps.
  3. Let |filter| denote the filter property.
  4. If the |filter| is defined,
    • If |filter|'s |query| is defined, pass it as an opaque string to the underlying implementation to be matched against discovered items. The underlying implementation is responsible to parse it e.g. as a SPARQL or JSON query and match it against the Thing Descriptions found during the discovery process. If queries are not supported, set |this.error| to {{NotSupportedError}} and abort these steps.
  5. Create the discovery results internal slot for storing discovered {{ThingDescription}} objects.
  6. Request the underlying platform to start the discovery process, with the following parameters:
    • If |filter|s |method| is not defined or the value is `"any"`, use the widest discovery method supported by the underlying platform.
    • Otherwise if |filter|s |method| is `"local"`, use the local Thing Directory for discovery. Usually that defines Things deployed in the same device, or connected to the device in slave mode (e.g. sensors connected via Bluetooth or a serial connection).
    • Otherwise if |filter|s |method| is `"directory"`, use the remote Thing Directory specified in |filter.url|.
    • Otherwise if |filter|s |method| is `"multicast"`, use all the multicast discovery protocols supported by the underlying platform.
  7. When the underlying platform has started the discovery process, set the active property to `true`.
  8. Whenever a new Thing Description |td:ThingDescription| is discovered by the underlying platform, run the following sub-steps:
    1. Fetch |td| as a JSON object |json|. If that fails, set the error property to {{SyntaxError}}, discard |td| and continue the discovery process.
    2. If |filter|'s |query| is defined, check if |json| is a match for the query. The matching algorithm is encapsulated by implementations. If that returns `false`, discard |td| and continue the discovery process.
    3. If |filter|'s |fragment| is defined, for each property defined in it, check if that property exists in |json|'s properties and has the same value. If this is `false` in any checks, discard |td| and continue the discovery process.
    4. Otherwise add |td| to the discovery results.
    5. At this point implementations MAY control the flow of the discovery process (depending on memory constraints, for instance temporarily stop discovery if the queue is getting too large, or resume discovery when the queue is emptied sufficiently).
  9. Whenever an error occurs during the discovery process,
    1. Let |error| be a new {{Error}} object. Set |error|'s |name| to `"DiscoveryError"`.
    2. If there was an error code or message provided by the Protocol Bindings, set |error|'s |message| to that value as string.
    3. Set error property to |error|.
    4. If the error is irrecoverable and discovery has been stopped by the underlying platform, set the active property to `false`.
  10. When the underlying platform reports the discovery process has completed, set the active property to `false`.

The next() method

Provides the next discovered {{ThingDescription}} object. The method MUST run the following steps:
  1. Return a {{Promise}} |promise:Promise| and execute the next steps in parallel.
  2. If the active property is `true`, wait until the discovery results internal slot is not empty.
  3. If discovery results is empty and the active property is `false`, set the done property to `true` and reject |promise|.
  4. Remove the first {{ThingDescription}} object |td| from discovery results.
  5. Resolve |promise| with |td| and abort these steps.

The stop() method

Stops or suppresses the discovery process. It might not be supported by all discovery methods and endpoints, however, any further discovery results or errors will be discarded and the discovery is marked inactive. The method MUST run the following steps:
  1. Request the underlying platform to stop the discovery process. If this returns an error, or if it is not possible, for instance when discovery is based on open ended multicast requests, the implementation SHOULD discard subsequent discovered items.
  2. Set the active property to `false`.

Discovery Examples

The following example finds {{ThingDescription}} objects of Things that are exposed by local hardware, regardless how many instances of WoT Runtime it is running. Note that the discovery can end (become inactive) before the internal discovery results queue is emptied, so we need to continue reading {{ThingDescription}} objects until done. This is typical with local and directory type discoveries.

        let discovery = new ThingDiscovery({ method: "local" });
        do {
          let td = await discovery.next();
          console.log("Found Thing Description for " + td.title);
          let thing = new ConsumedThing(td);
          console.log("Thing name: " + thing.getThingDescription().title);
        } while (!discovery.done);
      

The next example finds {{ThingDescription}} objects of Things listed in a Thing Directory service. We set a timeout for safety.

        let discoveryFilter = {
          method: "directory",
          url: "http://directory.wotservice.org"
        };
        let discovery = new ThingDiscovery(discoveryFilter);
        setTimeout( () => {
            discovery.stop();
            console.log("Discovery stopped after timeout.");
          },
          3000);
        do {
          let td = await discovery.next();
          console.log("Found Thing Description for " + td.title);
          let thing = new ConsumedThing(td);
          console.log("Thing name: " + thing.getThingDescription().title);
        } while (!discovery.done);
        if (discovery.error) {
          console.log("Discovery stopped because of an error: " + error.message);
        }
      

The next example is for an open-ended multicast discovery, which likely won't complete soon (depending on the underlying protocol), so stopping it with a timeout is a good idea. It will likely deliver results one by one.

        let discovery = new ThingDiscovery({ method: "multicast" });
        setTimeout( () => {
            discovery.stop();
            console.log("Stopped open-ended discovery");
          },
          10000);
        do {
          let td = await discovery.next();
          let thing = new ConsumedThing(td);
          console.log("Thing name: " + thing.getThingDescription().title);
        } while (!discovery.done);
      

Security and Privacy

A detailed discussion of security and privacy considerations for the Web of Things, including a threat model that can be adapted to various circumstances, is presented in the informative document [[!WOT-SECURITY-GUIDELINES]]. This section discusses only security and privacy risks and possible mitigations directly relevant to the scripts and WoT Scripting API.

A suggested set of best practices to improve security for WoT devices and services has been documented in [[!WOT-SECURITY-BEST-PRACTICES]]. That document may be updated as security measures evolve. Following these practices does not guarantee security, but it might help avoid common known vulnerabilities.

The WoT security risks and possible mitigations are concerning the following groups:

Scripting Runtime Security and Privacy Risks

This section is normative and contains specific risks relevant for the WoT Scripting Runtime.

Corrupted Input Security and Privacy Risk

A typical way to compromise any process is to send it a corrupted input via one of the exposed interfaces. This can be done to a script instance using WoT interface it exposes.

Mitigation:
Implementors of this API SHOULD perform validation on all script inputs. In addition to input validation, fuzzing should be used to verify that the input processing is done correctly. There are many tools and techniques in existence to do such validation. More details can be found in [[!WOT-SECURITY-TESTING]].

Physical Device Direct Access Security and Privacy Risk

In case a script is compromised or misbehaving, the underlying physical device (and potentially surrounded environment) can be damaged if a script can use directly exposed native device interfaces. If such interfaces lack safety checks on their inputs, they might bring the underlying physical device (or environment) to an unsafe state (i.e. device overheats and explodes).

Mitigation:
The WoT Scripting Runtime SHOULD avoid directly exposing the native device interfaces to the script developers. Instead, a WoT Scripting Runtime should provide a hardware abstraction layer for accessing the native device interfaces. Such hardware abstraction layer should refuse to execute commands that might put the device (or environment) to an unsafe state. Additionally, in order to reduce the damage to a physical WoT device in cases a script gets compromised, it is important to minimize the number of interfaces that are exposed or accessible to a particular script based on its functionality.

Provisioning and Update Security Risk

If the WoT Scripting Runtime supports post-manufacturing provisioning or updates of scripts, WoT Scripting Runtime or any related data (including security credentials), it can be a major attack vector. An attacker can try to modify any above described element during the update or provisioning process or simply provision attacker's code and data directly.

Mitigation:
Post-manufacturing provisioning or update of scripts, WoT Scripting Runtime or any related data should be done in a secure fashion. A set of recommendations for secure update and post-manufacturing provisioning can be found in [[!WOT-SECURITY-GUIDELINES]].

Security Credentials Storage Security and Privacy Risk

Typically the WoT Scripting Runtime needs to store the security credentials that are provisioned to a WoT device to operate in WoT network. If an attacker can compromise the confidentiality or integrity of these credentials, then it can obtain access to the WoT assets, impersonate WoT things or devices or create Denial-Of-Service (DoS) attacks.

Mitigation:
The WoT Scripting Runtime should securely store the provisioned security credentials, guaranteeing their integrity and confidentiality. In case there are more than one tenant on a single WoT-enabled device, a WoT Scripting Runtime should guarantee isolation of each tenant provisioned security credentials. Additionally, in order to minimize a risk that provisioned security credentials get compromised, the WoT Scripting Runtime should not expose any API for scripts to query the provisioned security credentials.

Script Security and Privacy Risks

This section describes specific risks relevant for script developers.

Corrupted Script Input Security and Privacy Risk

A script instance may receive data formats defined by the TD, or data formats defined by the applications. While the WoT Scripting Runtime SHOULD perform validation on all input fields defined by the TD, scripts may be still exploited by input data.

Mitigation:
Script developers should perform validation on all application defined script inputs. In addition to input validation, fuzzing could be used to verify that the input processing is done correctly. There are many tools and techniques in existence to do such validation. More details can be found in [[!WOT-SECURITY-TESTING]].

Denial Of Service Security Risk

If a script performs a heavy functional processing on received requests before the request is authenticated, it presents a great risk for Denial-Of-Service (DOS) attacks.

Mitigation:
Scripts should avoid heavy functional processing without prior successful authentication of requestor. The set of recommended authentication mechanisms can be found in [[!WOT-SECURITY-BEST-PRACTICES]].

Stale TD Security Risk

During the lifetime of a WoT network, a content of a TD can change. This includes its identifier, which might not be an immutable one and might be updated periodically.

Mitigation:
Scripts should use this API to subscribe for notifications on TD changes and do not rely on TD values to remain persistent.

While stale TDs can present a potential problem for WoT network operation, it might not be a security risk.

Terminology and conventions

The generic WoT terminology is defined in [[!WOT-ARCHITECTURE]]: Thing, Thing Description (in short TD), Web of Things (in short WoT), WoT Interface (same as WoT network interface), Protocol Bindings, WoT Runtime, Consuming a Thing Description, Thing Directory, WoT Interactions, Property, Action, Event etc.

JSON-LD is defined in [[!JSON-LD]] as a JSON document that is augmented with support for Linked Data.

The terms URL, URL scheme, URL host, URL path, URL record, parse a URL, absolute-URL string, path-absolute-URL string, basic URL parser are defined in [[!URL]].

The terms MIME type, Parsing a MIME type, Serializing a MIME type, valid MIME type string, JSON MIME type are defined in [[!MIMESNIFF]].

The terms UTF-8 encoding, UTF-8 decode, encode, decode are defined in [[!ENCODING]].

string, parse JSON from bytes and serialize JSON to bytes, are defined in [[!INFRA]].

{{Promise}}, Error, JSON, JSON.stringify, JSON.parse, internal method and internal slot are defined in [[!ECMASCRIPT]].

The terms browsing context, top-level browsing context, global object, current settings object, executing algorithms in parallel are defined in [[!HTML5]] and are used in the context of browser implementations.

The term secure context is defined in [[!WEBAPPSEC]].

IANA media types (formerly known as MIME types) are defined in RFC2046.

The terms hyperlink reference and relation type are defined in [[!HTML5]] and RFC8288.

API design rationale

API rationale usually belongs to a separate document, but in the WoT case the complexity of the context justifies including basic rationale here.

Approaches to WoT application development

The WoT Interest Group and Working Group have explored different approaches to application development for WoT that have been all implemented and tested.

No Scripting API

It is possible to develop WoT applications that only use the WoT network interface, typically exposed by a WoT gateway that presents a REST-ful API towards clients and implements IoT protocol plugins that communicate with supported IoT deployments. One such implementation is the Mozilla WebThings platform.

Simple Scripting API

WoT Things show good synergy with software objects, so a Thing can be represented as a software object, with Properties represented as object properties, Actions as methods, and Events as events. In addition, metadata is stored in special properties. Consuming and exposing is done with factory methods that produce a software object that directly represents a remote Thing and its interactions. One such implementation is the Arena Web Hub project.

In the next example, a Thing that represents interactions with a lock would look like the following: the |status| property and the open() method are directly exposed on the object.

          let lock = await WoT.consume(‘https://td.my.com/lock-00123’);
          console.log(lock.status);
          lock.open('withThisKey');
        

This API, aligned with [[WOT-TD]]

Since the direct mapping of Things to software objects have had some challenges, this specification takes another approach that exposes software objects to represent the Thing metadata as data property and the WoT interactions as methods. One implementation is node-wot in the the Eclipse ThingWeb project, which is the current reference implementation of the API specified in this document.

The same example now would look like the following: the |status| property and the open() method are represented indirectly.

          let res = await fetch(‘https://td.my.com/lock-00123’);
          let td = await res.json();
          let lock = new ConsumedThing(td);
          console.log(lock.readProperty(‘status’));
          lock.invokeAction(‘open’, 'withThisKey');
        

In conclusion, the WoT WG decided to explore the third option that closely follows the [[WOT-TD]] specification. Based on this, a simple API can also be implemented. Since Scripting is an optional module in WoT, this leaves room for applications that only use the WoT network interface. Therefore all three approaches above are supported by [[WOT-TD]].

Moreover, the WoT network interface can be implemented in many languages and runtimes. Consider this API an example for what needs to be taken into consideration when designing a Scripting API for WoT.

Fetching and validating a TD

The fetch(url) method has been part of this API in earlier versions. However, now fetching a TD given a URL should be done with an external method, such as the Fetch API or a HTTP client library, which offer already standardized options on specifying fetch details. The reason is that while simple fetch operations (covering most use cases) could be done in this API, when various fetch options were needed, there was no point in duplicating existing work to re-expose those options in this API.

Since fetching a TD has been scoped out, and TD validation is defined externally in [[WOT-TD]], that is scoped out, too. This specification expects a TD as parsed JSON object that has been validated according to the [[WOT-TD]] specification.

Factory vs constructors

The factory methods for consuming and exposing Things are asynchronous and fully validate the input TD. In addition, one can also construct {{ConsumedThing}} and {{ExposedThing}} by providing a parsed and validated TD. Platform initialization is then done when needed during the WoT interactions. So applications that prefer validating a TD themselves, may use the constructors, whereas applications that leave validation to implementations and prefer interactions initialized up front SHOULD use the factory methods on the WoT API object.

Observers

Earlier drafts used the Observer construct, but since it has not become standard, a new design was needed that was light enough for embedded implementations. Therefore observing Property changes and handling WoT Events is done with callback registrations.

Using Events

This API ended up not using software events at all, for the following reasons:
  • Subscription to WoT Events may be different from handling software events (subscription might need parameters, might involve security tokens etc).
  • Most implementations are for Node.js and browser implementations will likely be libraries (because possible dependency management issues in native implementations), using Events has been challenging.
  • Observing Property changes and handling WoT Events is done with the solution above.

Polymorphic functions

The reason to use function names like readProperty(), readMultipleProperties() etc. instead of a generic polymorphic read() function is that the current names map exactly to the "op" vocabulary from the Form definition in [[WOT-TD]].

Changes

The following is a list of major changes to the document. Major versions of this specification are the following:

For a complete list of changes, see the github change log. You can also view the recently closed issues.

Open issues

The following problems are being discussed and need most attention:

Full Web IDL

Acknowledgements

Special thanks to former editor Johannes Hund (until August 2017, when at Siemens AG) and Kazuaki Nimura (until December 2018) for developing this specification. Also, the editors would like to thank Dave Raggett, Matthias Kovatsch, Michael Koster, Elena Reshetova, Michael McCool as well as the other WoT WG members for their comments, contributions and guidance.