How to write an aggregate query to figure the count in Mongodb - mongodb

Question

I have the following JSON context and I have a requirement to aggregate the distinct keywords under conversion_token group and the count it is repeated; for instance:
"conversion_token": [
{
"keyword": "DBMS",
"count":4,
"classify":2
}
the keyword DBMS is used multiple times under different constructs in the json provided, the aggregate should display the
"conversion_token": [
{
"keyword": "DBMS",
"count":6,
"classify":2
}
{
"keyword": "NVL",
"count":2,`enter code here`
"classify":2
}
etc..
How can I do this?
{
"select_emp": {
"specification": {
"input": [
"p_empno"
],
"declare_stmt": {
"anchorvariable": [
"V_ENAME",
"V_HIREDATE",
"V_TITLE",
"V_REPORTSTO",
"V_DISP_DATE",
"V_INS_COUNT",
"CITY_FROM"
],
"tablename_variable": [
"EMPLOYEE.V_ENAME",
"EMPLOYEE.V_HIREDATE",
"EMPLOYEE.V_TITLE",
"EMPLOYEE.V_REPORTSTO",
"EMPLOYEE.V_DISP_DATE",
"EMPLOYEE.V_INS_COUNT",
"EMPLOYEE.CITY_FROM"
]
}
},
"body": {
"select_stmt1": {
"columns": [
"FIRSNAME",
"HIREDATE",
"TITLE",
"REPORTSTO"
],
"tablename": [
"EMPLOYEE"
],
"conversion_token": [
{
"keyword": "NVL",
"count": 1,
"classify": 2
}
]
},
"select_stmt2": {
"columns": [
"CITY"
],
"tablename": [
"EMPLOYEE"
],
"conversion_token": [
{
"keyword": "DECODE",
"count": 1,
"classify": 3
}
]
},
"dbms_stmt1": {
"dbms_putline": [
"P_EMPNO",
"V_ENAME",
"V_DISP_DATE",
"V_REPORTSTO"
],
"conversion_token": [
{
"keyword": "DBMS",
"count": 1,
"classify": 2
}
]
},
"forloop1": {
"select_stmt": {
"columns": [
"EMPLOYEEID",
"ROWID"
],
"tablename": [
"EMPLOYEE"
],
"conversion_token": [
{
"keyword": "DBMS",
"count": 1,
"classify": 2
}
]
}
},
"merge_stmt1": {
"merge_into": "EMPLOYEE",
"merge_using": {
"columns": [
"EMPLOYEEID",
"LASTNAME",
"TITLE",
"BIRTHDATE",
"HIREDATE",
"ADDRESS",
"CITY",
"STATE",
"COUNTRY",
"POSTALCODE",
"PHONE",
"FAX",
"EMAIL",
"BONUS"
],
"tablename": [
"EMPLOYEE"
]
},
"merge_update": {
"columns": [
"BONUS"
],
"tablename": [
"EMPLOYEE"
]
},
"merge_delete": {
"columns": [
"BONUS"
],
"tablename": [
"EMPLOYEE"
]
},
"merge_insert": {
"columns": [
"EMPLOYEEID",
"LASTNAME",
"FIRSTNAME",
"TITLE",
"BIRTHDATE",
"HIREDATE",
"ADDRESS",
"CITY",
"STATE",
"COUNTRY",
"POSTALCODE",
"PHONE",
"FAX",
"EMAIL",
"BONUS"
],
"tablename": [
"EMPLOYEE"
]
},
"conversion_token": [
{
"keyword": "Merge",
"count": 1,
"classify": 4
}
]
},
"exception_handling1": {
"dbms_putline": [
"P_EMPNO"
],
"conversion_token": [
{
"keyword": "DBMS",
"count": 1,
"classify": 2
}
]
}
}
}
}

Answer 1

First you need to concat conversion_token arrays into one using project and $concatArrays. Then using $unwind, you can separate each conversion_token object to make them ready for aggregation. After $unwind, you can group them by keyword using $group and you can take sum of count.
db.collection.aggregate([
{
$project: {
conversion_tokens: {
$concatArrays:
[
"$select_emp.body.select_stmt1.conversion_token",
"$select_emp.body.select_stmt2.conversion_token",
"$select_emp.body.dbms_stmt1.conversion_token",
"$select_emp.body.forloop1.conversion_token",
"$select_emp.body.merge_stmt1.conversion_token",
"$select_emp.body.exception_handling1.conversion_token"
]
}
}
}, {
$unwind: "$conversion_tokens"
}, {
$group: {
_id: "$conversion_tokens.keyword",
count: {
$sum: "$conversion_tokens.count"
}
}
}
])
This will produce an array like:
{
"_id": "DBMS",
"count": 3
}, {
"_id": "NVL",
"count": 1
}, {
"_id": "DECODE",
"count": 1
}
If you want to change the _id key with keyword, you can use $project again.