MongoDB Aggregation group and count strings -

i have problem counting different logstatusses collection. i'd following result query:

month | importername | nroferror | nrofdebug | nrofinfo | nrofwarning

so includes grouping month , importername , counting number of documents different statusses.

my mongodb collection:

{     "_id" : "8ec84cb7-5099-4a9d-be00-a40200a67c5a",     "messages" : [         {             "logstatus" : "error",             "message" : "my test message"         },         {             "logstatus" : "error",             "message" : "my test message"         },         {             "logstatus" : "error",             "message" : "my test message"         },         {             "logstatus" : "error",             "message" : "my test message"         },         {             "logstatus" : "error",             "message" : "my test message"         }     ],     "starttime" : new date("2014-12-15t10:06:09.00z"),     "endtime" : new date("2014-12-15t13:06:09.00z"),     "haserrors" : true,     "haswarnings" : false,     "importername" : "myimporter" } 

i have following query's:

db.sessionlogitems.aggregate ([ {     $project:     {         month :{$month : "$starttime"},         name: "$importername",         status: "$messages.logstatus",          _id: 0     } } ]) 

result:

month: 12, "name" : "importername", status: ["error", "error", "info"]

and

db.sessionlogitems.aggregate ([ {      $unwind: "$messages"  },  {     $group: { _id: "$messages", number : {$sum : 1 }} },  {     $sort: {number : -1 } }  ]) 

result: "_id" : { "logstatus" : "warning", "message" : "my test warning" }, "number" :5 "_id" : { "logstatus" : "error", "message" : "my test message" }, "number" : 5

but can't seem figure out correct query. appreciated!

edit:

my example above 1 out of many documents. have several importers have starttime , endtime. importers have several logmessages , 4 possible logstatusses: "error", "info", "debug", "warning". i'd have overview per month , per importer how many errors, infos, debugs , warnings produced.

assuming there no overlap in "month" between starttime , endtime values can use starttime value basis grouping key. of magic other "fields" comes $cond operator decides whether count value or not:

db.sessionlogitems.aggregate([      // unwind array de-normalize documents contained     { "$unwind": "$messages" },      // month , importer form grouping key     { "$group": {         "_id": {              "month": { "$month": "$starttime" },             "importername": "$importername"         },         "nroferror": {            "$sum": {                "$cond": [                    { "$eq": [ "$messages.logstatus", "error" ] },                    1,                    0                 ]             }         },         "nrofdebug": {            "$sum": {                "$cond": [                    { "$eq": [ "$messages.logstatus", "debug" ] },                    1,                    0                 ]             }         },         "nrofinfo": {            "$sum": {                "$cond": [                    { "$eq": [ "$messages.logstatus", "info" ] },                    1,                    0                 ]             }         },         "nrofwarning": {            "$sum": {                "$cond": [                    { "$eq": [ "$messages.logstatus", "warning" ] },                    1,                    0                 ]             }         }     }} ]) 

so "status" value tested , matched or not appropriate count value added appropriate field.