zeppelin-mongodb-interpreter

MongoDB interpreter for Apache Zeppelin.

Supported versions of MongoDB: >= 3.0

If you are interested, there is a Docker image for Zeppelin with MongoDB interpreter: https://hub.docker.com/r/cthiebault/zeppelin-mongodb/

Build

Requirement: Zeppelin must be in your local repo.

mvn clean package

Download

If you cannot build the jar, you can download it in the release page

Deployment

Update $ZEPPELIN_HOME/conf/zeppeln-site.xml

<property>
  <name>zeppelin.interpreters</name>
  <value>...,org.apache.zeppelin.mongodb.MongoDbInterpreter</value>
</property>

Create $ZEPPELIN_HOME/interpreter/mongodb
Copy interpreter jar in $ZEPPELIN_HOME/interpreter/mongodb

In some cases, mongodb is not visible in the list of the available interpreters. In this case, after the previous steps, you can create a new interpreter and in the interpreter group selection, you should be able to select mongodb.

Configuration

Parameter	Default value	Description
mongo.shell.path	mongo	Mongo shell path
mongo.shell.command.timeout	60000	Mongo command timeout
mongo.shell.command.table.limit	1000	Limit of documents displayed in a table
mongo.server.database	test	MongDB database name
mongo.server.host	localhost	Host of the MongDB server
mongo.server.port	27017	Port of the MongDB server
mongo.server.username		Username for authentication
mongo.server.password		Password for authentication
mongo.server.authentdatabase		Database used for authentication

How to use

In Zeppelin, use %mongodb in a paragraph. After that, you can type the same Javascript code you use when you write scripts for the Mongo shell. For more information, please consult: https://docs.mongodb.com/manual/tutorial/write-scripts-for-the-mongo-shell/

There are several functions that have been added to help you in Zeppelin:

printTable(cursor, fields, flattenArray): to print a table (i.e. it uses %table). Arguments:
- cursor: a DBQuery or DBCommandCursor instance
- fields: an array of field names to put in the table (can be null)
- flattenArray: if true, the arrays in the documents will also be flatten (false by default)
DBQuery.prototype.table: to print a table (it invokes the previous function)
DBCommandCursor.prototype.table: same as above

Examples:

%mongodb

// Display a table
db.zipcodes.find({ "city":"CHICAGO", "state": "IL" }).table()

%mongodb

var states = db.zipcodes.aggregate( [
   { $group: { _id: "$state", totalPop: { $sum: "$pop" } } },
   { $match: { totalPop: { $lt: 1000*1000 } } },
   { $sort: { totalPop: 1 } }
] )

// Build a 'table'
print("%table state\ttotalPop")
states.forEach(state => { print(state._id + "\t" + state.totalPop) })

Examples

Configuration:
Queries (these examples come from: https://docs.mongodb.com/manual/tutorial/aggregation-zip-code-data-set/)

ssabzzz / zeppelin-mongodb-interpreter