I need some help with translating the following CouchDB views from javascript to erlang. I need them in erlang, because in javascript the view uses all of the available stack memory and crashes couchjs (see this bugreport ).
The current map functions I have in javascript are:
sync/transaction_keys
function(doc) {
if(doc.doc_type == "Device") {
for(key in doc.transactions)
emit(key, null);
}
}
and sync/transcation
function(doc) {
if(doc.doc_type == "Device") {
for(key in doc.transactions) {
t = doc.transactions[key];
t.device = doc.device;
emit(key, t);
}
}
}
An example document would be:
{
"_id": "fcef7b5c-cbe6-31af-8363-2b446a7e4cf2",
"_rev": "3-c90abd075404a75744fd3e5e4f04ebad",
"device": "fcef7b5c-cbe6-31af-8363-2b446a7e4cf2",
"doc_type": "Device",
"transactions": {
"79fe8630-c0c0-30c6-9913-79b2f93e3e6e": {
"timestamp": 1309489169533,
"version": 10008,
"some_more_data" : "more_data"
}
"e4678930-c465-76a6-8821-75a3e888765a": {
"timestamp": 1309489169533,
"version": 10008,
"some_more_data" : "more_data"
}
}
}
Basically sync/transaction_keys emits all keys of the transaction dictionary and sync/transaction does emit all entries in the transaction dictionary.
Unfortunately I never used Erlang before and I need to rewrite that code pretty soon, so any help is very weled.
Thanks in advance.
I need some help with translating the following CouchDB views from javascript to erlang. I need them in erlang, because in javascript the view uses all of the available stack memory and crashes couchjs (see this bugreport https://issues.apache/jira/browse/COUCHDB-893).
The current map functions I have in javascript are:
sync/transaction_keys
function(doc) {
if(doc.doc_type == "Device") {
for(key in doc.transactions)
emit(key, null);
}
}
and sync/transcation
function(doc) {
if(doc.doc_type == "Device") {
for(key in doc.transactions) {
t = doc.transactions[key];
t.device = doc.device;
emit(key, t);
}
}
}
An example document would be:
{
"_id": "fcef7b5c-cbe6-31af-8363-2b446a7e4cf2",
"_rev": "3-c90abd075404a75744fd3e5e4f04ebad",
"device": "fcef7b5c-cbe6-31af-8363-2b446a7e4cf2",
"doc_type": "Device",
"transactions": {
"79fe8630-c0c0-30c6-9913-79b2f93e3e6e": {
"timestamp": 1309489169533,
"version": 10008,
"some_more_data" : "more_data"
}
"e4678930-c465-76a6-8821-75a3e888765a": {
"timestamp": 1309489169533,
"version": 10008,
"some_more_data" : "more_data"
}
}
}
Basically sync/transaction_keys emits all keys of the transaction dictionary and sync/transaction does emit all entries in the transaction dictionary.
Unfortunately I never used Erlang before and I need to rewrite that code pretty soon, so any help is very weled.
Thanks in advance.
Share Improve this question asked Jul 23, 2011 at 13:28 SimonSimon 1,6813 gold badges14 silver badges20 bronze badges 4- How many transactions do you have in your larger documents? That should be OK, though I'd avoid global variables (I don't think that really matters in this case). – Dustin Commented Jul 23, 2011 at 19:04
- about 594 transactions/document. But it keeps growing because each 15min another transaction will be added. – Simon Commented Jul 24, 2011 at 13:11
- Perhaps you'd be better off representing these transactions as new documents. Is there any reason not to do that? It seems like it'd be easier to model this that way. – Dustin Commented Jul 25, 2011 at 2:15
- Yeah, that's what I did before, but I was told to change it to 1 document/device. – Simon Commented Jul 27, 2011 at 12:33
1 Answer
Reset to default 15I just did your second one (the more plicated one). The first can easily be extrapolated from there:
fun({Doc}) ->
%% Helper function to get a toplevel value from this doc.
F = fun(B) -> proplists:get_value(B, Doc) end,
%% switch on doc type
case F(<<"doc_type">>) of
<<"Device">> ->
%% Grab the transactions from this document
{Txns} = F(<<"transactions">>),
lists:foreach(fun({K,V}) ->
%% Emit the key and the value as
%% the transaction + the device
%% id
{T} = proplists:get_value(K, Txns),
Emit(K, {[{<<"device">>, F(<<"device">>)} | T]})
end,
Txns);
_ -> false %% Not a device -- ignoring this document
end
end.