Class ModelBuilder<In, Out>

Model builder to construct and build a Model consisting of graph shapes and services.

Usage

Models can be created using the ModelBuilder. Once you have added all services and constructed the graph, you can build the model using the build() function. A promise will be returned with the created model.

import { ModelBuilder } from '@openhps/core';

ModelBuilder.create()
    .build().then(model => {
        // ...
    });
Copy

The graph shape of a model is immutable and can not be altered after building.

Shape Builder

Shapes can be created by starting with the from() function. This function takes an optional parameter of one or multiple source nodes.

In order to end a shape, the to() function needs to be called with one or more optional sink nodes.

import { ModelBuilder } from '@openhps/core';

ModelBuilder.create()
    .from()
    .to()
    .build().then(model => {
        // ...
    });
Copy

Alternatively for readability with multiple shapes, the shapes can individually be created using the addShape() function as shown below.

import { ModelBuilder, GraphBuilder } from '@openhps/core';

ModelBuilder.create()
    .addShape(
      GraphBuilder.create()
        .from()
        .to())
    .build().then(model => {
        // ...
    });
Copy

Building Source Processors

It is possible to have multiple processing nodes between the source and sink. These processing nodes can manipulate the data frame when it traverses from node to node.

import { ModelBuilder } from '@openhps/core';

ModelBuilder.create()
    .from(...)
    .via(new ComputingNode())
    .via(new AnotherComputingNode())
    .to(...)
    .build().then(model => {
        // ...
    });
Copy

Helper Functions

Helper functions can replace the via() function. Commonly used nodes such as frame filters, merging of data frames from multiple sources, ... can be replaced with simple functions as filter() or merge() respectively.

import { ModelBuilder } from '@openhps/core';
import { CSVSourceNode, CSVSinkNode } from '@openhps/csv';

ModelBuilder.create()
    .from(
        new CSVSourceNode('scanner1.csv', ...),
        new CSVSourceNode('scanner2.csv', ...),
        new CSVSourceNode('scanner3.csv', ...)
    )
    .filter((frame: DataFrame) => true)
    .merge((frame: DataFrame) => frame.source.uid)
    .via(new ComputingNode())
    .via(new AnotherComputingNode())
    .to(new CSVSinkNode('output.csv', ...))
    .build().then(model => {
        // ...
    });
Copy

Debug Logging

When building the model, you can provide a logger callback that has two arguments. An error level complying with normal log levels and a log object that represents an object.

import { ModelBuilder } from '@openhps/core';

ModelBuilder.create()
    // Set the logger that will be used by all nodes and services
    .withLogger((level: string, log: any) => {
        console.log(log);
    })
    // ...
    .build().then(model => {
     // ...
    });
Copy

Adding Services

Adding services can be done using the addService() function in the model builder.

import { ModelBuilder } from '@openhps/core';

ModelBuilder.create()
    .addService(...)
    // ...
    .build().then(model => {

    });
Copy