improve leftJoin nested mapObject when equivalence do and do not match

user_6621346 0
Viewed 100

I am using the following dataweave function, and it does works.

%dw 2.0
import * from dw::core::Arrays
output application/json

var mysqlInvoices = [
    {
        "id": 1,
        "owner": "Joseph"
    },
    {
        "id": 2,
        "owner": "Maria"
    }
]

var sapInvoices = [
    {
        "number": 3,
        "issuedBy": "XYZ"
    },
    {
        "number": 4,
        "issuedBy": "ABC"
    }
]
---
leftJoin(mysqlInvoices, sapInvoices, (m) -> m.id, (s) -> s.number) map (item, index) -> 
  (item.l mapObject (sItem, sKey) -> 
        (if ((sKey) as String == "id") "identifier" 
        else if ((sKey) as String == "owner") "ownerName" 
        else (sKey)): sItem)
  ++ 
  (if (item.r != null) 
    item.r mapObject (sItem, sKey) -> 
        (sKey): sItem
  else 
    sapInvoices[0] mapObject 
        (sItem, sKey) -> (sKey): "")

However, I am thinking if I can improve this function at two points:

  1. change the key conditions:

I dont think that is the best practice to check every key match an if condition to change it:

(if ((sKey) as String == "id") "identifier" 
else if ((sKey) as String == "owner") "ownerName" 
else (sKey)): sItem
  1. Use the original object to map it as an empty string when the leftJoin do not match keys:
sapInvoices[0] mapObject (sItem, sKey) -> 
(sKey): ""

I am uncomfortable with these two points, and I believe that there are ways to improve this code, I just dont know how.

If there is a very different way of doing the same task, I also appreciate that kind of suggestion.

dataweave mule optimization performance
2 Answers

Give the following a try, if anything the code seems a bit simpler:

%dw 2.0
import * from dw::core::Arrays
output application/json

var mysqlInvoices = [
    {
        "id": 1,
        "owner": "Joseph"
    },
    {
        "id": 2,
        "owner": "Maria"
    }
]

var sapInvoices = [
    {
        "number": 3,
        "issuedBy": "XYZ"
    },
    {
        "number": 4,
        "issuedBy": "ABC"
    }
]

var fs2rn = {
    id: "identifier",
    owner: "ownerName"
}

var rightEmpty= {number:"",issuedBy:""}
---
leftJoin(
    // Do the field renaming at the very begining
    mysqlInvoices map ($ mapObject {(fs2rn[$$] default $$): $}), 
    sapInvoices, 
    (m) -> m.identifier, 
    (s) -> s.number
)
// Iterate over the results
// Get just the values, and colapse the objects into a single object
map (
    {($ pluck $)}
)
// Iterate over the results and use pattern-matching to 
// 
map (
    $ match {
        // Check if you have an id but not a number fields
        // In which case add the rightEmpty object
        case o if (o.identifier? and not (o.number?)) -> o ++ rightEmpty
        // Or give the object because you now have both an id and a number
        else o -> o
    }
)

The features and functions I used are:

If I was to give you an advice, it would be to better indent your code. Nonetheless, pretty good job!

user_30789860

Based on George's answer, you can remove pluck and match and directly combine left and right table. See below:

%dw 2.0
import * from dw::core::Arrays
output application/json

var mysqlInvoices = [
    {
        "id": 1,
        "owner": "Joseph"
    },
    {
        "id": 2,
        "owner": "Maria"
    }
]

var sapInvoices = [
    {
        "number": 3,
        "issuedBy": "XYZ"
    },
    {
        "number": 4,
        "issuedBy": "ABC"
    }
]

var fs2rn = {
    id: "identifier",
    owner: "ownerName"
}

var rightEmpty= {number:"",issuedBy:""}
---
leftJoin(
    // Do the field renaming at the very begining
    mysqlInvoices map ($ mapObject {(fs2rn[$$] default $$): $}), 
    sapInvoices, 
    (m) -> m.identifier, 
    (s) -> s.number
) map (item) -> item.l ++ (item.r default rightEmpty)
user_85054950
Related