@grafana/faro-core

Warning: currently pre-release and subject to frequent breaking changes. Use at your own risk.

Core package of Faro.

The entire architecture of the library is contained within this package. Out of the box, it doesn't collect any signals, but it provides the API to do it as well as multiple helper methods.

Installation

import { initializeFaro } from '@grafana/faro-core';

initializeFaro({
  // ...
});

Options

Faro Web SDK requires a configuration object parameter with the following properties:

Property	Description	Type	Default Value Variable
app	Application metadata	App
dedupe	A flag for toggling deduplication filter	boolean
globalObjectKey	String that should be used when defining the Faro instance property on the global object	string	defaultGlobalObjectKey = 'faro'
instrumentations	Array of instrumentations that should be ran	Instrumentation[]
internalLoggerLevel	The level of information printed to console for internal messages	InternalLoggerLevel	defaultInternalLoggerLevel = InternalLoggerLevel.ERROR
isolate	A flag that will create an isolated Faro instance	boolean
metas	Array of metas that should be logged	MetaItem[]
parseStacktrace	A function used to parse stack traces	StacktraceParser
paused	Flag for initializing the Faro instance as paused	boolean
preventGlobalExposure	Flag for toggling the definition on the global object	boolean
transports	Array of transports that should be used	Transport[]
batching.enabled	Enable batch sending of events	boolean	true
batching.sendTimeout	The interval in which to send event batches	number	250
batching.itemLimit	The size of the signal buffer before the batch is sent (overrides interval)	number	50

Besides the mandatory properties, Faro configuration also supports the following optional properties:

Property	Description	Type	Default Value
beforeSend	Hook invoked before pushing event to transport. Can be used to modify or filter events	BeforeSendHook	undefined
eventDomain	event.domain attribute of an event, to be set on every event item as default	string	undefined
ignoreErrors	Error message patterns for errors that should be ignored	Patterns	[]
session	Session metadata	Session	undefined
user	User metadata	User	undefined
view	View metadata	View	undefined

Faro instance

Faro instance is an object which can be accessed by either importing it from the package or by referencing it from the global object (window in browsers and global in Node.js).

// Browser/Node.js
import { faro } from '@grafana/faro-core';

faro.api.pushLog(/* ... */);

// Browser
window.faro.api.pushLog(/* ... */);

// Node.js
global.faro.api.pushLog(/* ... */);

API

The api property on the Faro instance contains all the necessary methods to push new events.

Errors

• pushError - is a method to push an error/exception to the Faro instance. It accepts a mandatory message parameter and an optional one where you can set:

  • skipDedupe - a flag for enforcing error push even if the error is identical to the previous one.
  • stackFrames - an array of stack frames. Defaults to parsing error.stack if present.
  • type - the type of exception. Default value: error.name or "error".

  faro.api.pushError(new Error('This is an error'));
  
  faro.api.pushError(new Error('This is an unhandled exception'), { skipDedupe: true });
  
  faro.api.pushError(new Error('This is an unhandled exception'), { type: 'unhandledException' });
  
  faro.api.pushError(new Error('This is an error with stack frames'), {
    stackFrames: [
      {
        filename: 'file.js',
        function: 'myFunction',
        colno: 120,
        lineno: 80,
      },
    ],
  });
  
  faro.api.pushError(new Error('This is a log with context'), {
    context: {
      message: 'React error boundary',
      componentStackTrace: {...}
    },
  });

Events

• pushEvent - is a method to push an event to the Faro instance. It accepts a mandatory name parameter and a couple of optional parameters.

  The first optional parameter is called attributes and it is a key-value object where you can set additional properties for your event.

  The second optional parameter is called domain and it is a string that's used to group events together.

  The third optional parameter is an optional options object where you can set:

  • skipDedupe - a flag for enforcing event push even if the event is identical to the previous one.

  faro.api.pushEvent('user-logged-in');
  
  faro.api.pushEvent('user-logged-in', { username: 'the-user-id' });
  
  faro.api.pushEvent('user-logged-in', { username: 'the-user-id' }, 'auth');
  
  faro.api.pushEvent('user-logged-in', { username: 'the-user-id' }, 'auth', { skipDedupe: true });

Logs

• pushLog - is a method to register a log event. The method accepts a mandatory args parameter which is an array of arguments that will be stringified and sent to the transports and an optional one where you can set:

  • skipDedupe - a flag for enforcing log push even if the log is identical to the previous one.
  • logLevel - the type of message that we register.
  • context - a plain object containing primitive values that will be recorded along with the message.

  faro.api.pushLog(['This is a log', 'With another message']);
  
  faro.api.pushLog(['This is a log'], { skipDedupe: true });
  
  faro.api.pushLog(['This is a warning'], { level: LogLevel.WARN });
  
  faro.api.pushLog(['This is a log with context'], {
    context: {
      randomNumber: Math.random(),
    },
  });

Measurements

• pushMeasurement - is a method for registering metrics. The method accepts a mandatory payload parameter and an optional parameter for passing additional options:

  • skipDedupe - a flag for enforcing measurement push even if the measurement is identical to the previous one.

  faro.api.pushMeasurement({
    type: 'custom',
    values: {
      my_custom_metric: Math.random(),
    },
  });
  
  faro.api.pushMeasurement(
    {
      type: 'custom',
      values: {
        my_custom_metric: Math.random(),
      },
    },
    { skipDedupe: true }
  );
  
  faro.api.pushMeasurement(
    {
      type: 'custom',
      values: {
        my_custom_metric: Math.random(),
      },
    },
    { context: { hello: 'world' } }
  );

Traces

• pushTraces - is a method for manually sending traces through the transports. The method only accepts a mandatory payload.

  faro.api.pushTraces({
    resourceSpans: [
      /* OTel Traces */
    ],
  });

• getOTEL - is a method for getting the OpenTelemetry instance used by Faro.

  const otel = faro.api.getOTEL();

• isOTELInitialized - is a method for checking if OpenTelemetry package was correctly initialized and there is an OpenTelemetry instance used by Faro.

  const isInitialized = faro.api.isOTELInitialized();

Instrumentations

Instrumentations are packages that leverage the Faro Web SDK API to provide automatic mechanisms for collecting data. They are just simple functions that are executed when the agent is initialized.

Please note that the core package does not contain any instrumentations out of the box and they should be provided by platform specific packages like @grafana/faro-web-sdk.

You can also write your own instrumentations:

import { initializeFaro, BaseInstrumentation } from '@grafana/faro-core';

export class MyInstrumentation extends BaseInstrumentation {
  readonly version = '1.0.0';
  readonly name = 'my-instrumentation';

  initialize(): void {
    this.faro.api.pushLog(['hello from my instrumentation']);
  }
}

initializeFaro({
  // ...
  instrumentations: [new MyInstrumentation()],
});

Metas

Metas are objects that will be attached to every event that is triggered by the API.

Out of the box, only one meta is provided: sdk which contains information about the Faro instance and its version.

Additional metas may be provided by platform packages like @grafana/faro-web-sdk.

You can also define your own metas:

import { initializeFaro } from '@grafana/faro-core';

initializeFaro({
  metas: [
    // Define a static meta
    {
      app: {
        name: 'my-app',
        version: '1.0.0',
      },
    },

    // Define a meta which caches some values on initialization
    () => {
      return {
        user: {
          username: getUser().name,
        },
      };
    },
  ],
});

Transports

Transports are functions that will be called for every event that is triggered by the API. They are used to do something with the data after collecting it.

Out of the box, no transports are provided in the core package and they should be provided by platform specific packages like @grafana/faro-web-sdk.

You can also define your own transports:

import { initializeFaro, BaseTransport, TransportItem } from '@grafana/faro-core';

class MyTransport extends BaseTransport {
  send(item: TransportItem) {
    // do something with paylaod
  }
}

initializeFaro({
  transports: [new MyTransport()],
});

The transports also support batch processing wiht the batchEnabled flag set to true and a positive, non zero batchSendTimeout. The batch executor will group items that share the same metas together and call the send function for each group of signals.

Batched transports

Currently the Transport interface supports implementations of the send function that accept either a TransportItem or an array of them. The latter is useful when you want to batch process signals with your custom transport. In order to achieve this, you should override the isBatched method of the BaseTransport.

export class MyTransport extends BaseTransport {
  async send(item: TransportItem): Promise<void> {
    // do something with the item
  }
}

/* should be */

export class MyBatchedTransport extends BaseTransport {
  async send(items: TransportItem[]): Promise<void> {
    // do something with the array of items
    // all of them share the same metas
  }

  // This is required for the transports layer to differentiate
  // between the two types of Transport implementations
  override isBatched(): boolean {
    return true;
  }
}

Unpatched console

Some instrumentations might override the default console methods but Faro instance provides a way to access the unmodified console methods.

faro.unpatchedConsole.log('This is a log');
faro.unpatchedConsole.warn('This is a warning');

Pause / unpause

Faro instance can be paused by invoking faro.pause(). This will prevent events from being sent to transports. Call faro.unpause() to resume capturing events.

Isolated Faro instances

Sometimes you may want to create one or more isolated Faro instances. For example:

• you want to bundle Faro in a reusable library and report certain events only for it while the project where the library is used has its own Faro instance
• you want to log certain events in one system while other events in other systems
• E2E libraries that may create multiple Faro instances without refreshing the page

In order to achieve this, you can use the isolate flag when initializing the agent:

// faro 1 will be isolated
const faro1 = initializeFaro({
  // ...
  isolate: true,
});

// globalFaro will be available globally
const globalFaro = initializeFaro({
  // ...
  isolate: false,
});

// another isolated Faro instance
const faro2 = initializeFaro({
  // ...
  isolate: true,
});

Although an isolated agent may sound like a great idea, there are some limitations which apply to them:

• some instrumentations will still register globally (i.e. exceptions instrumentation or console instrumentation)
• an isolated Faro instance will not be available on the global object
• the Faro reference should be stored by the project as it won't be available via import { faro } from '@grafana/faro-core';