Single View Creator

Microservice initialization

Each Single View needs a dedicated Microservice. This service will listen on the Projections changes that affect the Single View and consequently updates its data. This service is the Single View Creator.

To start configuring your own Single View Creator, you can begin from the Single View Creator Template.

info

If you are searching for a ready-to-use plugin that only requires some configuration files to update your single views, please visit this link.

Single View Creator Template

Initialize the service

The service starts in index.js file. First of all, the template uses the Custom Plugin Lib to instantiate a service. Inside its callback, the single-view-creator-lib is initialized to deal with the complexity of the Fast Data components.

const singleViewCreator = getSingleViewCreator(log, config, customMetrics)

await singleViewCreator.initEnvironment() // connect Mongo, Kafka and create the patient instance
service.decorate('patient', singleViewCreator.k8sPatient)

Where config is an object whose fields represent the Microservice environment variables.

Some environment variables will be pre-compiled when you create the service from template, others won't, but they will still have a placeholder as value. Replace it with the correct value.

Here some tips:

• TYPE: should be the name of the single view which your single view creator is responsible for
• SINGLE_VIEWS_COLLECTION: should be the name of the single view which your single view creator is responsible for
• PROJECTIONS_CHANGES_COLLECTION: if you have set a custom projection change collection name from advanced, then set its name. Otherwise it is fd-pc-SYSTEM_ID where SYSTEM_ID is the id of the System of Records this single view creator is responsible for.
• SINGLE_VIEWS_PORTFOLIO_ORIGIN: should be equals to the SYSTEM_ID you have set in PROJECTIONS_CHANGES_COLLECTION
• SINGLE_VIEWS_ERRORS_COLLECTION: it is the name of a MongoDB CRUD you want to use as collection for single view errors.
• KAFKA_BA_TOPIC: topic where to send the before-after, which is the single view document before and after a change
• SEND_BA_TO_KAFKA: true if you want to send to Kafka the before-after information about the update changes of the single view
• KAFKA_SVC_EVENTS_TOPIC: topic used to queue Single View Creator state changes (e.g. single view creation)

Now, we start the single-view-creator:

const resolvedOnStop = singleViewCreator.startCustom({
  strategy: aggregatorBuilder(projectionsDB),
  mapper,
  validator,
  singleViewKeyGetter: singleViewKey,
  upsertSingleView: upsertSV(),
  deleteSingleView: deleteSV(),
})

• strategy is the function that performs the aggregation over the projections
• mapper is the function that takes as input the raw aggregation result and maps the data to the final Single View
• validator is the validation function which determines if the Single View is valid (and thus inserted or updated in Mongo) or not (and thus deleted)
• singleViewKeyGetter is the function that, given the projections changes identifier, returns the data used as selector to find the single view document on mongo to update or delete
• upsertSingleView is the function that updates or inserts the Single View to the Single Views collection on Mongo
• deleteSingleView is the function that deletes the Single View from the Single Views collection on Mongo

upsertSV and deleteSV are two utility functions that the library exports that handle the upsert and the delete of the single view.

note

The deleteSV function makes a real delete of the document on MongoDb. So, unlike the projections deletion, it does not make a virtual delete.

The Single View creator needs to be stopped when the process is stopping. To do that, we use the onClose hook:

service.addHook('onClose', async() => {
  log.fatal({ type: 'END' }, 'Single View Creator is stopping...')
  await singleViewCreator.stop()

  // this is a promise resolved when the infinite loop which processes the single views ends.
  // Here we wait for the resolving of the promise. You don't need to call it.
  await resolvedOnStop
  log.fatal({ type: 'END' }, 'Single View Creator stopped')
  await mongoClient.close()
})

info

You can use the template and all the Mia-Platform libraries only under license. For further information contact your Mia Platform referent

note

This documentation refers to the @mia-platform-internal/single-view-creator-lib ^9.x.x.

The core of your work on this service are the files inside the src folder.

singleViewKey.js: It takes as input the identifier of the projection change and returns the key object used to select the document of the Single View collection that needs to be updated. This key corresponds to the query object fed to mongodb, therefore you can return any legal mongo query.

In the example below, we expect to have the field myId as primary key of the Single View collection.

const get = require('lodash.get')

module.exports = function singleViewKeyGenerator(logger, projectionChangeIdentifier) {
  const IDENTIFIER_KEY = 'UNIQUE_ID'
  
  // get the single view id from the identifier
  const myId = projectionChangeIdentifier[IDENTIFIER_KEY]

  return {
    myId,
  }
}

pipeline.js: It takes as input a MongoDB instance and returns a function. This function takes as input the projection change identifier and returns an array.

If it is empty, a delete operation will be executed on the single view identified by the singleViewKeyGenerator result. If it is not empty, an upsert operation will be executed on the single view identified by the singleViewKeyGenerator result.

note

If the pipeline returns an array with more than one element, only the first element will be used for the upsert.

module.exports = (mongoDb) => {
  return async(logger, projectionChangeIdentifier) => {

    const uniqueId = projectionChangeIdentifier.UNIQUE_ID
    const MY_PROJECTION = 'projection-name'

    // retrieve data from all projections you need for your single view
    const projectionCollection = mongoDb.collection(MY_PROJECTION)
    const projectionDataById = await projectionCollection.findOne({
        UNIQUE_ID: uniqueId,
        __STATE__: 'PUBLIC'
    })

    if (!projectionDataById) {
      // it's expected to be a delete
      logger.debug({ UNIQUE_ID: uniqueId }, 'single view public data not found')
      return []
    }
    const singleViewData = projectionDataById
    logger.debug({ singleViewData }, 'single view retrieved data')

    return [
      singleViewData,
    ]
  }
}

mapper.js: It is a function that takes as argument the first element (if defined) of the result of the pipeline, and returns an object containing the value updated for the single view. The object returned should match the schema of the single view.

module.exports = (logger, singleViewData) => {
  const { UNIQUE_ID, NAME } = singleViewData

  return {
    myId: UNIQUE_ID,
    name: NAME,
  }
}

Inside the mapper a renaming and repositioning of the fields can be applied.

note

We suggest to implement inside the mapper all the aggregation logic that can be reused for all the clients that will read the Single Views, they should be as generic as possible. It is a good practice to have some calculation and aggregation logic inside Single View Creator as far as it is reusable. If you have to apply some custom logic try to do it inside and API Adapter specific for the client.

Validate a Single View

The startCustom function accepts a function in the configuration object called validator, which is the validation function.

The validation of a Single View determines what to do with the current update. If the single view is determined as "non-valid", the delete function will be called. Otherwise, if the result of the validation is positive, it will be updated or inserted in the Single Views collection, through the upsert function. Delete function and upsert function will be explained in the next paragraph.

For this reason, the validation procedure should not be too strict, since a Single View declared as "invalid" would not be updated or inserted to the database. Rather, the validation is a check operation to determine if the current Single View should be handled with the upsert or delete functions.

By default, in this template we set as validator a function that returns always true. So we accept all kinds of single views, but, if you need it, you can replace that function with your own custom validator.

The input fields of the validation function are the logger and the Single View, while the output is a boolean containing the result of the validation.

function singleViewValidator(logger, singleView) {
  ... checks on singleView

  // returns a boolean
  return validationResult
}

Customize Upsert and Delete functions

If you want, you can replace both upsertSV and deleteSV with your own custom functions to perform those operations.

These functions represents the last step of the creation (or deletion) of a Single View, in which the Single View collection is actually modified.

In case the validation is succeeded, the upsert function will be called with the following arguments:

• logger is the logger
• singleViewCollection is the the Mongo collection object
• singleView is the result of the mapping operation
• singleViewKey is the Single View key

On the other hand, if the validation has a negative result, the delete function will be called with the same arguments, except for the singleView, which will not be handled by the delete function.

In both cases, some operation should be done on singleViewCollection in order to modify the Single View with the current singleViewKey, with the idea of "merging" the current result with the one already present in the database.

For example, we have a Customer Single View with a list of products the customer bought from different e-commerce and we receive an update for a new object on a specific shop, in that case we don't want to replace the list of bought products with the last one arrived but we want to push the product in the list in order to have the complete history of purchases.

For both functions, the output is composed of an object containing two fields:

• old which contains the old Single View
• new which contains the new Single View

These values will be the respectively the before and after of the message sent to the KAFKA_BA_TOPIC topic, that is the topic responsible for tracking any result of the Single View creator.

async function upsertSingleViewFunction(
  logger,
  singleViewCollection,
  singleView,
  singleViewKey)
{
  logger.trace('Upserting Single View...')
  const oldSingleView = await singleViewCollection.findOne(singleViewKey)

  await singleViewCollection.replaceOne(
    singleViewKey,
    singleView,
    { upsert: true }
  )

  logger.trace({ isNew: Boolean(oldSingleView) }, 'Updated Single View')
  return {
    old: oldSingleView,
    new: singleView,
  }
}

async function deleteSingleViewFunction(
  logger,
  singleViewCollection,
  singleViewKey)
{
  logger.trace('Deleting Single View...')
  const oldSingleView = await singleViewCollection.findOne(singleViewKey)

  if (oldSingleView !== null) {
    try {
      await singleViewCollection.deleteOne(singleViewKey)
    } catch (ex) {
      logger.error(`Error during Single View delete: ${ex}`)
    }
  }

  logger.trace('Single view deletion procedure terminated')
  return {
    old: oldSingleView,
    new: null,
  }
}

Error handling

When generating a Single View, every error that occurs is saved in MongoDb, with a format that satisfies the schema requirements of the CRUD service, so that you can handle those errors using the Console. The fields of the error messages when they are first created are:

• _id: a unique identifier of the record, automatically generated
• portfolioOrigin: a value concerning the origin of the error, defaults to UNKNOWN_PORTFOLIO_ORIGIN
• type: the Single View type
• identifier: id of the projection changes
• errorType: the error details
• createdAt: time of creation
• creatorId: set to single-view-creator
• __STATE__: set to PUBLIC
• updaterId: set to single-view-creator
• updatedAt: time of creation

It is highly recommended to use a TTL index to enable the automatic deletion of older messages, which can be done directly using the Console, as explained here.