Reduce multiple field values in CouchDB - mapreduce

Question

For a notification system I am storing user messages in CouchDB. My document structure looks like this:
{
"_id": "bc325c2f6a99194ecab6e7bbae81b609",
"_rev": "1-9745ddb21cefe3dbc8cc1f7f7bd8d11d",
"uid": "bc325c2f6a99194ecab6e7bbae80eb29",
"msg": "Hi there, here is a message for you!",
"level": "warning",
"status": "new",
"created": 1467180077,
"read": null
}
The field status can have the two values new or read - I am trying to write a Map/Reduce view to return a count for a certain user with how many messages they have with the status new and how many with the status read.
The Map Function is pretty straight forward:
function(doc) {
emit(doc.uid, doc.status);
}
and it returns something like this:
| key | value
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | "read"
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | "read"
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | "read"
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | "new"
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | "new"
I am trying now to figure out how to write a reduce function that would produce this:
| key | value
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | ["read":3, "new":2]
+------------------------------------+------------------------------
If I use this reduce function
function (key, values, rereduce) {
return values;
}
I get:
| key | value
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | ["read","read","read","new","new"]
+------------------------------------+------------------------------
I tried to use return count(values) but it returns null.
It seems I can't wrap my head around, how to approach this. Anyone who could put me on the right track here?

Answer 1

My recommendation would be to emit the uid and status as your key, then you will get the counts as your reduced value.
First, we'll start with the map function:
function (doc) {
emit([ doc.uid, doc.status ]);
}
Notice we're emitting a key only, and we're not bothering with a value. (it turns out we don't need it in this case)
Then, use the following built-in reduce function
_count
Your view output will have keys like: (remember, we don't care about value)
[ "bc325c2f6a99194ecab6e7bbae80eb29", "read" ]
[ "bc325c2f6a99194ecab6e7bbae80eb29", "read" ]
[ "bc325c2f6a99194ecab6e7bbae80eb29", "read" ]
[ "bc325c2f6a99194ecab6e7bbae80eb29", "new" ]
[ "bc325c2f6a99194ecab6e7bbae80eb29", "new" ]
When you query the view, make sure to include group=true (the reduce=true is implied)
You'll notice that your view has the following key/value pairs now:
[ "bc325c2f6a99194ecab6e7bbae80eb29", "read" ]: 3
[ "bc325c2f6a99194ecab6e7bbae80eb29", "new" ]: 2
If your database has more documents in it, you'll see all the other user ids as well. To filter down to the user you care about, simply use:
startkey=["bc325c2f6a99194ecab6e7bbae80eb29"]
endkey=["bc325c2f6a99194ecab6e7bbae80eb29",{}]
These arrays look kinda weird, I'm sure, but those just ensure that any values for the "status" will be matched. (check out the documentation on Views Collation for more information)
This approach will scale up rather nicely, allowing any other status values to work without any code changes.

Answer 2

I found a way of doing this, but I am not sure if it is the right approach:
The reduce function:
function (key, values, rereduce) {
var result = new Object;
result.read = 0;
result.new = 0;
for(var i = 0; i < values.length; ++i){
if(values[i] == 'new'){
result.new++;
}
if(values[i] == 'read'){
result.read++;
}
};
return result;
}
will produce this result:
| key | value
+------------------------------------+------------------------------
| bc325c2f6a99194ecab6e7bbae80eb29 | {read: 2, new: 1}
+------------------------------------+------------------------------