CouchDB view reduce one doc per key - mapreduce

Question

I'm trying to solve what seems like a fairly simple problem with a couchDb view, but I'm not even getting close to the target with my result set.
Rather than updating documents, I'm creating a new document every time as my versioning strategy, and tying those documents together with a versioning field called ver. The very first document in a version chain will see the ver field and the _id field having the same value. All subsequent documents in the chain will have the same ver field as previous docs in the chain, but will have a unique _id field. These documents also have a createdTime field which is my way of knowing which document is the latest.
Here's my documents:
{
"_id": "abcd-1234-efgh-9876",
"ver": "abcd-1234-efgh-9876",
"createdTime": "2020-01-12 01:15:00 PM -0600",
...
},
{
"_id": "uopa-3849-pmdi-1935",
"ver": "abcd-1234-efgh-9876",
"createdTime": "2020-02-16 02:39:00 PM -0600",
...
}
Here's my map function:
function (doc) {
emit(doc.ver, doc);
}
Here's my reduce function:
function(keys, values, rereduce) {
var latestVersions = {};
for (var i = 0; i < keys.length; i++) {
var found = latestVersions[keys[i][0]];
if (!found || found.createdTime < values[i].createdTime) {
latestVersions[keys[i][0]] = values[i];
}
}
return latestVersions;
}
And finally, here's my desired output from the view (just the doc that I want):
{
"_id": "uopa-3849-pmdi-1935",
"ver": "abcd-1234-efgh-9876",
"createdTime": "2020-02-16 02:39:00 PM -0600",
...
}
What am I missing here? The reduce function is returning both records, which is not what I want. Is what I'm trying to achieve possible or is there a better way to go about this?
Update
I was able to get this to work when a single key is used to access the view, which is one of my use cases.
function (keys, values, rereduce) {
var toReturn = values[0];
for (var i = 1; i < values.length; i++) {
if (values[i].createdTime > toReturn.createdTime) {
toReturn = values[i];
}
}
return toReturn;
}
I have another use case that will be returning all of the data in the view, however. The desired result there is the same as above, but the function I'm using for single keys will only ever return one result. How do I filter multiple values with a shared key such that 1 "shared" key:n values -> 1 key:1 value.

Answer 1

I was finally able to resolve this when I stumbled upon this couchbase article. It was much more articulate than some of the other dry computer-science documentation.
I still do not understand why certain items are grouped in a reduce method and other ones are not. For example, reduce was called 5 times for 6 items that shared an identical key; only one of the keys had actually grouped anything -- an array of two documents. It probably has something to do with those dry computer-science B-tree documents I glossed over.
Anyway, I was able to determine that all I needed to do was group the values by the ver field in both scenarios (the only difference being that rereduce had a 2 dimensional array). Here's what my reduce function ended up looking like:
function (keys, values, rereduce) {
var toValues = function(myMap) {
return Object.keys(myMap).map(function(key) {
return myMap[key];
});
}
if (rereduce) {
// values should look like [[{...}, {...}], [{...}]]
var outputMap = {};
for (var i = 0; i < values.length; i++) {
for (var j = 0; j < values[i].length; j++) {
var currentEl = values[i][j];
var found = outputMap[currentEl.ver];
if ((found && found.createdDate < currentEl.createdDate) || !found) {
outputMap[currentEl.ver] = currentEl;
}
}
}
return toValues(outputMap);
} else {
var outputMap = {};
for (var i = 0; i < values.length; i++) {
var found = outputMap[values[i].ver];
if ((found && found.createdDate < values[i].createdDate) || !found) {
outputMap[values[i].ver] = values[i];
}
}
return toValues(outputMap);
}
}