Ramda - how to use multiple functions on the same data structure - ramda.js

Question

I am trying to use multiple ramda functions on this example:
const data = {
"tableItems": [
{
"id": 1,
"name": "1",
"startingPoint": true,
"pageNumber": 15,
"nodes": [
100,
200
]
},
{
"id": 2,
"name": "2",
"startingPoint": true,
"pageNumber": 14,
"nodes": [
300,
400
]
}
],
"nodes": [
{
"id": 100,
"tableItemId": 1,
"content": "test"
},
{
"id": 200,
"tableItemId": 1,
"content": "test"
},
{
"id": 300,
"tableItemId": 2,
"content": "test"
},
{
"id": 400,
"tableItemId": 2,
"content": "test"
}
]
}
I am trying to create new JSON which should look like this where nodes array should be filled with another ramda function:
const newJSON = [
{
"id": "chapter-1",
"name": "2",
"nodes": []
},
{
"id": "chapter-2",
"name": "1",
"nodes": []
}
]
I started with:
let chapters = [];
let chapter;
const getChapters = R.pipe(
R.path(['tableItems']),
R.sortBy(R.prop('pageNumber')),
R.map((tableItem) => {
if(tableItem.startingPoint) {
chapter = {
id: `chapter-${chapters.length+1}`,
name: tableItem.name,
nodes: []
}
chapters.push(chapter);
}
return tableItem
})
)
But how to combine getNodes which needs access to the whole scope of data?
I tried pipe but something is not working.
Example:
const getNodes = R.pipe(
R.path(['nodes']),
R.map((node) => {
console.log(node)
})
)
R.pipe(
getChapters,
getNodes
)(data)
Any help would be appreciated.

Answer 1

We could write something like this, using Ramda:
const {pipe, sortBy, prop, filter, map, applySpec, identity, propEq, find, __, addIndex, assoc} = R
const transform = ({tableItems, nodes}) => pipe (
filter (prop ('startingPoint')),
sortBy (prop ('pageNumber')),
map (applySpec ({
name: prop('name'),
nodes: pipe (prop('nodes'), map (pipe (propEq ('id'), find (__, nodes))), filter (Boolean))
})),
addIndex (map) ((o, i) => assoc ('id', `chapter-${i + 1}`, o))
) (tableItems)
const data = {tableItems: [{id: 1, name: "1", startingPoint: true, pageNumber: 15, nodes: [100, 200]}, {id: 2, name: "2", startingPoint: true, pageNumber: 14, nodes: [300, 400]}], nodes: [{id: 100, tableItemId: 1, content: "test"}, {id: 200, tableItemId: 1, content: "test"}, {id: 300, tableItemId: 2, content: "test"}, {id: 400, tableItemId: 2, content: "test"}]}
console .log (transform (data))
.as-console-wrapper {max-height: 100% !important; top: 0}
<script src="//cdnjs.cloudflare.com/ajax/libs/ramda/0.27.1/ramda.min.js"></script>
First we filter the tableItems to include only those with startingPoint of true, then we sort the result by pageNumber. Then for each, we create name and nodes elements, based on the original data and on a function that maps the node values to the element in the initial nodes property. Finally, for each one, we add the chapter-# id element using addIndex (map).
This works, and is not horrible. It would take a fair bit of work to make this entirely point-free, I believe. And I don't find it worthwhile... especially because this Ramda version doesn't add anything to a simpler vanilla implementation:
const transform = ({tableItems, nodes}) =>
tableItems
.filter (x => x .startingPoint)
.sort (({pageNumber: a}, {pageNumber: b}) => a - b)
.map (({name, nodes: ns}, i) => ({
id: `chapter-${i + 1}`,
name,
nodes: ns .map (n => nodes .find (node => node .id == n)) .filter (Boolean)
}))
const data = {tableItems: [{id: 1, name: "1", startingPoint: true, pageNumber: 15, nodes: [100, 200]}, {id: 2, name: "2", startingPoint: true, pageNumber: 14, nodes: [300, 400]}], nodes: [{id: 100, tableItemId: 1, content: "test"}, {id: 200, tableItemId: 1, content: "test"}, {id: 300, tableItemId: 2, content: "test"}, {id: 400, tableItemId: 2, content: "test"}]}
console .log (transform (data))
.as-console-wrapper {max-height: 100% !important; top: 0}
This works similarly to the above except that it assigns the id at the same time as name and nodes.
I'm a founder of Ramda and remain a big fan. But it doesn't always add anything to vanilla modern JS.

Answer 2

You can use a curried function. Because the pipe will always pipe the result of the previous function call into the next function. You can use R.tap if you want to step over.
However, I guess you want to have the data object and the output of the previous function call both in your getNodes function. In that case you can use a curried function, where you pass the response of the previous function as last parameter.
const getNodes = R.curryN(2, function(data, tableItemList){
console.log(tableItemList) // result of previous function call
return R.pipe(
R.path(['nodes']),
R.map((node) => {
console.log('node:', node);
})
)(data)
})
And use it like:
R.pipe(
getChapters,
getNodes(data)
)(data)

Answer 3

I would split the solution into two steps:
Prepare the tableItems and nodes to the required end state using R.evolve - filter, sort, and then use R.toPairs the tableItems to get an array that includes the index and the object. Group the nodes by id so you can pick the relevant nodes by id in the combine step.
Combine both properties to create the end result by mapping the new tableItems, and using R.applySpec to create the properties.
const {pipe, evolve, filter, prop, sortBy, toPairs, groupBy, map, applySpec, path, flip, pick} = R
const transform = pipe(
evolve({ // prepare
tableItems: pipe(
filter(prop('startingPoint')),
sortBy(prop('pageNumber')),
toPairs
),
nodes: groupBy(prop('id'))
}),
({ tableItems, nodes }) => // combine
map(applySpec({
id: ([i]) => `chapter-${+i + 1}`,
name: path([1, 'name']),
nodes: pipe(path([1, 'nodes']), flip(pick)(nodes)),
}))(tableItems)
)
const data = {tableItems: [{id: 1, name: "1", startingPoint: true, pageNumber: 15, nodes: [100, 200]}, {id: 2, name: "2", startingPoint: true, pageNumber: 14, nodes: [300, 400]}], nodes: [{id: 100, tableItemId: 1, content: "test"}, {id: 200, tableItemId: 1, content: "test"}, {id: 300, tableItemId: 2, content: "test"}, {id: 400, tableItemId: 2, content: "test"}]}
console.log(transform(data))
.as-console-wrapper {max-height: 100% !important; top: 0}
<script src="//cdnjs.cloudflare.com/ajax/libs/ramda/0.27.1/ramda.min.js"></script>