is it possible to get one result from a map/reduce script getInputData to map phase

Yeah, just pass a single result from the getInput phase, not sure if that is what you mean.

that is what I am trying to do but getting this:

What am I looking at here?

logging from the map phase

what are you passing from the getInput?

let resultSet = fileSearch.run() resultSet.each(function (result){ seaResult = result; return false; }); return seaResult;

Your goal is to just pass the first result from that search, right?

yes, that is correct

It looks like its interpreting that properties from that object you are passing individually, try return fileSearch.run().getRange(0,1) instead, see if that gives you a better result in the map phase

much better

var values = JSON.parse(context.value).values;

, log that and should be good to go

if anyone looks back at this context.value contained {"recordType":"file","id":"3945945","values":{"name":"SC2020120106.XML"}} so I was able to get internal id with JSON.parse(context.value).id

a more generally useful way to reduce the result set is to add a filter that returns just one result.

I wanted to use M/R because of yeilding, the file is a very long XML of shipments from our 3PL, I write each shipment to a custom record, I want to be able to use the yeilding from M/R so that if soft governance is an issue then the M/R script will resume

The MR won't magically fix anything governance-wise, if you are saying you write each record down to its own reduce stage while interpreting the file in the map, then that makes perfect sense to do that.

yes, that is what I am doing

Actually, I think M/R has the ability to load the file directly

I'm not following?

I may be wrong on that but I thought I'd done that before

got it

wait, you say it's an XML file

yes sir

so the line-oriented behavior of that file support wouldn't apply

I'm loading the file in the map phase

I thought MR still has a 10,000 unit limit on

map()

?

1000 on map, 5000 on reduce per iteration

The map should be using liek no governance to convert to JSON though, right?

correct

yeah, design makes sense then to me

I'm only using 10 to load the file

and your

reduce

will fit in the 5000 limit in any iteration?

yes, creating only one custom record per reduce

ah, ok so you're emitting thousands of (unique) keys for reduce?

potentially yes

in effect, you're using reduce as another map (it's reducing a collection of 1 on each invocation)

I suppose yes, ideally our 3PL wouldn't send us so many shipment confirmations on one XML but we have to live with it I guess

I agree with Sandii that sounds sane enough. sorry to get off topic

good discussion

though I wouldn't be surprised if the underlying machinery were skewed towards having many

map

instances and relatively fewer

reduce

instances - you've just cleverly turned that around to the opposite, but if it works then it sounds like it's good leveraging of the governance.

To stalberts point, you could just convert the file into a JSON in the getInput, and pass the JSON down to the map where each invocation of the map is for each custom record entry.

wouldn't I hit the hard limit of 1000 governance on map?

Thats the beauty of MR scripts, if you write them correctly, you should never run out of governance and the only answer to "how do we make it faster" is (should be at least) to open the wallet and buy more suitecloud+ licenses.