is it possible to write regular expression in $cond in MongoDB - regex

Question

I need to use $cond to combine differenet column, and one $cond I need to write is as following:
create_widget: {
$sum:{
$cond:[{$and: [ {$eq: ['$Method', 'POST']},
{Url:{$regex: /.*\/widgets$/}} ]}, 1, 0]
}
}
and this code is not right, it seems, regular expression can not be put here.Is there any other way to do this? I want to match Url and regular expression and put the code under $cond.
A sample data looks as
{"BrandId":"a","SessionId":"a1","Method":"POST","Url":"/sample/widgets"}
{"BrandId":"a","SessionId":"a2","Method":"POST","Url":"/sample/blog"}
{"BrandId":"b","SessionId":"b1","Method":"PUT","Url":"/sample/widgets"}
The whole code I wrote is as following:
db.tmpAll.aggregate([
{$group: {
_id: {BrandId:'$BrandId'},
SessionId: {$addToSet: '$SessionId'},
create_widget: {
$sum:{
$cond:[{$and: [ {$eq: ['$Method', 'POST']},
{} ]}, 1, 0]
}
}
}},
{$group: {
_id: '$_id.BrandId',
distinct_session: {$sum: {$size: '$SessionId'}},
create_widget: {$sum: '$create_widget'}
}}
]);
The expected result of sample code is
{ "_id" : "a", "distinct_session" : 2, "create_widget" : 1 }
{ "_id" : "b", "distinct_session" : 1, "create_widget" : 0 }

Answer 1

For MongoDB 4.2 and newer production releases, and in the 4.1.11 and newer development versions, use $regexMatch which is a syntactic sugar on top of $regexFind which can be used for regex matching and capturing.
db.tmpAll.aggregate([
{ "$group": {
"_id": {
"BrandId": "$BrandId",
"SessionId": "$SessionId"
},
"widget_count": {
"$sum": {
"$cond": [
{
"$and": [
{ "$eq": ["$Method", "POST"] },
{ "$regexMatch": {
"input": "$Url",
"regex": /widget/
} }
]
}, 1, 0
]
}
},
"session_count": { "$sum": 1 }
} },
{ "$group": {
"_id": "$_id.BrandId",
"create_widget": { "$sum": "$widget_count" },
"distinct_session": { "$sum": "$session_count" }
} }
]);
There is an open JIRA issue for this SERVER-8892 - Use $regex as the expression in a $cond. However, as a workaround, For older MongoDB versions which do not have the above features, use the following workaround in your aggregation pipeline.
It uses the $substr operator in the $project operator stage to extract the part of the URL and acts as a workaround for the regex. :
db.tmpAll.aggregate([
{ "$group": {
"_id": {
"BrandId": "$BrandId",
"SessionId": "$SessionId"
},
"widget_count": {
"$sum": {
"$cond": [
{
"$and": [
{ "$eq": ["$Method", "POST"] },
{ "$eq": [ { "$substr": [ "$Url", 8, -1 ] }, "widget"] }
]
}, 1, 0
]
}
},
"session_count": { "$sum": 1 }
} },
{ "$group": {
"_id": "$_id.BrandId",
"create_widget": { "$sum": "$widget_count" },
"distinct_session": { "$sum": "$session_count" }
} }
]);
Output
/* 1 */
{
"result" : [
{
"_id" : "a",
"create_widget" : 1,
"distinct_session" : 2
},
{
"_id" : "b",
"create_widget" : 0,
"distinct_session" : 1
}
],
"ok" : 1
}