On 22 Nov 2008, at 18:19, Jedediah Smith wrote:
Let's say I have a couchdb full of documents representing
transactions e.g.
{
type: "transaction",
account: "bob",
date: "2008-11-22 21:01:00",
amount: 123.45
}
where amount can be positive or negative, representing a credit or
debit respectively
I want a view of balance by account and date i.e. given an account
and date, get the sum of all transaction amounts for that account,
up to that date.
map: function(doc) {
emit(doc.date, doc.amount);
}
reduce function(key, values) {
return sum(values):
}
If you query `_view/viewname`, you get the total sum of all
`doc.amount` instances.
If you are using [2008, 11, 22, 21, 01, 00] for the date you can do:
- query `_view/viewname?group_level=1` to get all amounts per year
- query `_view/viewname?group_level=2` to get all amounts per months
- etc...
I assume though you want to be able to get these values per
`doc.account`. For this, do
map: function(doc) {
emit([doc.account, doc.date], doc.amount);
}
reduce: same as above.
Query with `_view/viewname?startkey=["bob", null]&endkey=["bob", {}]`
to get all data for bob. Use `group_level` as above. Substitute `null`
and `{}` with dates to get amounts in a key-range.
Cheers
Jan
--
Can this be done with only a couchdb view while meeting these
requirements:
- no client logic
- no storing intermediate data in documents
- optimal use of incremental view updates (adding/updating a
transaction only invalidates view rows with same or later date)
If not, what's the best compromise that minimizes client logic?
