Skip to content

Latest commit

 

History

History

resolve-map

Folders and files

NameName
Last commit message
Last commit date

parent directory

..
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

resolve-map

npm version npm downloads Twitter Follow

This project is part of the @thi.ng/umbrella monorepo.

About

DAG resolution of vanilla objects & arrays with internally linked values.

This is useful for expressing complex configurations with derived values or computing interrelated values without having to specify the order of computations.

It's common practice to use nested JS objects for configuration purposes. Frequently some values in the object are copies or derivatives of other values, which can lead to mistakes during refactoring and / or duplication of effort.

To avoid these issues, this library provides the ability to define single sources of truth, create references (links) to these values and provide a resolution mechanism to recursively expand their real values and / or compute derived values. Both absolute & relative references are supported.

Status

STABLE - used in production

Installation

yarn add @thi.ng/resolve-map
// ES module
<script type="module" src="https://unpkg.com/@thi.ng/resolve-map?module" crossorigin></script>

// UMD
<script src="https://unpkg.com/@thi.ng/resolve-map/lib/index.umd.js" crossorigin></script>

Package sizes (gzipped, pre-treeshake): ESM: 872 bytes / CJS: 921 bytes / UMD: 981 bytes

Dependencies

Usage examples

Several demos in this repo's /examples directory are using this package.

A selection:

Screenshot Description Live demo Source
Filterable commit log UI w/ minimal server to provide commit history Demo Source
Basic crypto-currency candle chart with multiple moving averages plots Demo Source

Statistical analysis

In this example we construct a graph to compute a number of statistical properties for some numeric input array. The graph is a plain object of possibly dependent functions, which can be specified in any order. Each function uses ES6 object destructuring to look up and execute other computations in the graph. Each computation is only executed once.

import { resolve } from "@thi.ng/resolve-map";
import * as tx from "@thi.ng/transducers";

// define object of interrelated computations to be executed later
// the `src` key used by most functions is still missing here and
// will be injected later as well
const stats = {
    // sequence average
    mean: ({ src }) => tx.mean(src),
    // sequence range
    range: ({ min, max }) => max - min,
    // computes sequence min val
    min: ({ src }) => tx.min(src),
    // computes sequence max val
    max: ({ src }) => tx.max(src),
    // sorted copy
    sorted: ({ src }) => [...src].sort((a, b) => a - b),
    // standard deviation
    sd: ({ src, mean })=>
        Math.sqrt(
            tx.transduce(tx.map((x) => Math.pow(x - mean, 2)), tx.add(), src) /
            (src.length - 1)),
    // compute 10th - 90th percentiles
    percentiles: ({ sorted }) => {
        return tx.transduce(
            tx.map((x) => sorted[Math.floor(x / 100 * sorted.length)]),
            tx.push(),
            tx.range(10, 100, 5)
        );
    }
};

// inject some source data to analyze

// Note: we wrap the data as function to avoid `resolve`
// attempting to resolve each array item as well. this is
// purely for performance reasons and would also work without
// wrapping.

// Note 2: If the `stats` graph is meant to be re-usable in
// the future you MUST use the spread operator to create a
// shallow copy, because `resolve` mutates the given object
resolve({...stats, src: () => [ 1, 6, 7, 2, 4, 11, -3 ]})
// {
//     mean: 4,
//     range: 14,
//     min: -3,
//     max: 11,
//     sorted: [ -3, 1, 2, 4, 6, 7, 11 ],
//     sd: 4.546060565661952,
//     percentiles: [ -3, 1, 2, 2, 4, 6, 6, 7, 11 ],
//     src: [ 1, 6, 7, 2, 4, 11, -3 ]
// }

Theme configuration

import { resolve } from "@thi.ng/resolve-map";

resolve({
    colors: {
        bg: "white",
        text: "black",
        selected: "red",
    },
    main: {
        fontsizes: [12, 16, 20]
    },
    button: {
        bg: "@/colors/text",
        label: "@/colors/bg",
        // resolve with abs path inside fn
        fontsize: ($) => `${$("/main/fontsizes/0")}px`,
    },
    buttonPrimary: {
        bg: "@/colors/selected",
        label: "@/button/label",
        // resolve with relative path inside fn
        fontsize: ($) => `${$("../main/fontsizes/2")}px`,
    }
});
// {
//     colors: {
//         bg: "white",
//         text: "black",
//         selected: "red"
//     },
//     main: {
//         fontsizes: [ 12, 16, 20 ]
//     },
//     button: {
//         "bg": "black",
//         "label": "white",
//         "fontsize": "12px"
//     },
//     buttonPrimary: {
//         bg: "red",
//         label: "black",
//         fontsize: "20px"
//     }
// }

API

Generated API docs

resolve(obj)

Visits all key-value pairs or array items in depth-first order, expands any reference values, mutates the original object and returns it. Cyclic references are not allowed and will throw an error. However, refs pointing to other refs are recursively resolved (again, provided there are no cycles). Reference values are special strings representing lookup paths of other values in the object and are prefixed with @ for relative refs or @/ for absolute refs and both using / as path separator (Note: trailing slashes are NOT allowed!). Relative refs are resolved from the currently visited object and support "../" prefixes to access any parent levels. Absolute refs are always resolved from the root level (the original object passed to this function).

// `c` references sibling `d`
// `d` references parent `a`
resolve({a: 1, b: {c: "@d", d: "@/a"} })
// { a: 1, b: { c: 1, d: 1 } }

Any function values are called using two possible conventions:

  1. If the user function uses ES6 object destructuring for its first argument, the given object keys are resolved prior to calling the function and the resolved values provided as first argument (object) and a general resolve function as second argument.
  2. If no de-structure form is found in the function's arguments, the function is only called with resolve as argument.

Important: ES6 destructuring can only be used for ES6 compile targets and will fail when transpiling to ES5. If you're not sure, use the 2nd (legacy) form. Also, since ES6 var names can't contain special characters, destructured keys can ALWAYS only be looked up as siblings of the currently processed key.

The resolve function provided as arg to the user function accepts a path (without @ prefix) to look up any other values in the root object.

// `c` uses ES6 destructuring form to look up `a` & `b` values
// `d` uses provided resolve fn arg `$` to look up `c`
resolve({a: 1, b: 2, c: ({a,b}) => a + b, d: ($) => $("c") })
// { a: 1, b: 2, c: 3, d: 3 }

// last item references item @ index = 2
resolve([1,2, ($) => $("0") + $("1"), "@2"])
// [1, 2, 3, 3]

The return value of the user provided function is used as final value for that key in the object. This mechanism can be used to compute derived values of other values stored anywhere in the root object. Function values will always be called only once. Therefore, in order to associate a function as final value to a key, it MUST be wrapped with an additional function, as shown for the e key in the example below. Similarly, if an actual string value should happen to start with @, it needs to be wrapped in a function (see f key below).

// `a` is derived from 1st array element in `b.d`
// `b.c` is looked up from `b.d[0]`
// `b.d[1]` is derived from calling `e(2)`
// `e` is a wrapped function
// `f` is wrapped to ignore `@` prefix
res = resolve({
  a: ($) => $("b/c") * 100,
  b: { c: "@d/0", d: [2, ($) => $("../../e")(2) ] },
  e: () => (x) => x * 10,
  f: () => "@foo",
})
// { a: 200, b: { c: 2, d: [ 2, 20 ] }, e: [Function], f: "@foo" }

res.e(2);
// 20

Authors

Karsten Schmidt

License

© 2018 - 2020 Karsten Schmidt // Apache Software License 2.0