_stats builtin reduction for min/max/mean/variance

git-svn-id: https://svn.apache.org/repos/asf/couchdb/trunk@921592 13f79535-47bb-0310-9956-ffa450edef68

2 files changed, 36 insertions, 1 deletions

diff --git a/share/www/script/test/reduce_builtin.js b/share/www/script/test/reduce_builtin.js
index 7938a0cf..0db6bc09 100644
--- a/share/www/script/test/reduce_builtin.js
+++ b/share/www/script/test/reduce_builtin.js
@@ -22,6 +22,14 @@ couchTests.reduce_builtin = function(debug) {
 
   var summate = function(N) {return (N+1)*N/2;};
 
+  var sumsqr = function(N) { 
+    var acc = 0;
+    for (var i=1; i<=N; ++i) {
+      acc += i*i;
+    }
+    return acc;
+  }
+
   // this is the same test as the reduce.js test
   // only we'll let CouchDB run reduce in Erlang
   var map = function (doc) {
@@ -32,6 +40,12 @@ couchTests.reduce_builtin = function(debug) {
   T(result.rows[0].value == 2*summate(numDocs));
   result = db.query(map, "_count");
   T(result.rows[0].value == 1000);
+  result = db.query(map, "_stats");
+  T(result.rows[0].value.sum == 2*summate(numDocs));
+  T(result.rows[0].value.count == 1000);
+  T(result.rows[0].value.min == 1);
+  T(result.rows[0].value.max == 500);
+  T(result.rows[0].value.sumsqr == 2*sumsqr(numDocs));
 
   result = db.query(map, "_sum", {startkey: 4, endkey: 4});
   T(result.rows[0].value == 8);
diff --git a/src/couchdb/couch_query_servers.erl b/src/couchdb/couch_query_servers.erl
index 3095b199..ff89a67a 100644
--- a/src/couchdb/couch_query_servers.erl
+++ b/src/couchdb/couch_query_servers.erl
@@ -154,7 +154,10 @@ builtin_reduce(reduce, [<<"_count">>|BuiltinReds], KVs, Acc) ->
     builtin_reduce(reduce, BuiltinReds, KVs, [Count|Acc]);
 builtin_reduce(rereduce, [<<"_count">>|BuiltinReds], KVs, Acc) ->
     Count = builtin_sum_rows(KVs),
-    builtin_reduce(rereduce, BuiltinReds, KVs, [Count|Acc]).
+    builtin_reduce(rereduce, BuiltinReds, KVs, [Count|Acc]);
+builtin_reduce(Re, [<<"_stats">>|BuiltinReds], KVs, Acc) ->
+    Stats = builtin_stats(Re, KVs),
+    builtin_reduce(Re, BuiltinReds, KVs, [Stats|Acc]).
 
 builtin_sum_rows(KVs) ->
     lists:foldl(fun
@@ -164,6 +167,24 @@ builtin_sum_rows(KVs) ->
             throw({invalid_value, <<"builtin _sum function requires map values to be numbers">>})
     end, 0, KVs).
 
+builtin_stats(reduce, [[_,First]|Rest]) when is_number(First) ->
+    Stats = lists:foldl(fun([_K,V], {S,C,Mi,Ma,Sq}) when is_number(V) ->
+        {S+V, C+1, erlang:min(Mi,V), erlang:max(Ma,V), Sq+(V*V)};
+    (_, _) ->
+        throw({invalid_value,
+            <<"builtin _stats function requires map values to be numbers">>})
+    end, {First,1,First,First,First*First}, Rest),
+    {Sum, Cnt, Min, Max, Sqr} = Stats,
+    {[{sum,Sum}, {count,Cnt}, {min,Min}, {max,Max}, {sumsqr,Sqr}]};
+
+builtin_stats(rereduce, [[_,First]|Rest]) ->
+    {[{sum,Sum0}, {count,Cnt0}, {min,Min0}, {max,Max0}, {sumsqr,Sqr0}]} = First,
+    Stats = lists:foldl(fun([_K,Red], {S,C,Mi,Ma,Sq}) ->
+        {[{sum,Sum}, {count,Cnt}, {min,Min}, {max,Max}, {sumsqr,Sqr}]} = Red,
+        {Sum+S, Cnt+C, erlang:min(Min,Mi), erlang:max(Max,Ma), Sqr+Sq}
+    end, {Sum0,Cnt0,Min0,Max0,Sqr0}, Rest),
+    {Sum, Cnt, Min, Max, Sqr} = Stats,
+    {[{sum,Sum}, {count,Cnt}, {min,Min}, {max,Max}, {sumsqr,Sqr}]}.
 
 % use the function stored in ddoc.validate_doc_update to test an update.
 validate_doc_update(DDoc, EditDoc, DiskDoc, Ctx, SecObj) ->