Fork and clone this repository.
Please use git clone --recursive
to ensure you retrieve the sample-data
submodule.
By the end of the lesson, students should be able to:
- Use the MongoDB shell to interact with MongoDB databases and collections
- Create, Read, Update, and Delete documents in MongoDB collections using the MongoDB shell.
Basic JavaScript.
Required readings:
- Completely unbiased explanation of noSQL
- Mongo DB Is Web Scale
- Getting Started with MongoDB (MongoDB Shell Edition)
We'll run brew install mongodb
then execute the first two commands suggested.
Run sudo apt-get install mongodb
Relational databases are good at modeling data that fits into tables. What do you use if your data isn't that structured?
Perhaps a noSQL data-store. An important consideration is ACID versus BASE and the CAP theorem.
MongoDB is a schema-less document-store that organizes documents in collections. What does this mean?
Relational Database Term | MongoDB Term |
---|---|
database | database |
table | collection |
row | document |
column | field |
We'll use mongo-crud
as the database to hold our tables and mongo to interact with it. mongo
is MongoDB's command line client which lets us execute commands interactively and from scripts.
First we start the mongo shell:
$ mongo mongo-crud
MongoDB shell version: 3.0.7
connecting to: mongo-crud
>
The command to list databases is show databases
(or show dbs
):
> show databases
local 0.078GB
>
Unlike PostgreSQL, MongoDB lets us select a database that hasn't been created. When we add a collection, the database will be created.
If we didn't specify the database on the command line we can connect to a database with use <database name>
:
MongoDB shell version: 3.0.7
connecting to: test
> use mongo-crud
switched to db mongo-crud
> show databases
local 0.078GB
>
Our first collection will be called people
. It has no entries.
> show collections
> db.people.count();
0
This is a common pattern in MongoDB: you can refer to things that don't yet exist, and it will cooperate. MongoDB won't create them until you give it something to remember.
- Inserting data - Overview of adding documents to a collection.
- db..insert() - detailed documentation of MongoDB's
insert
collection method. - Importing data - overview of MongoDB's
mongoimport
command line utility. - mongoimport - detailed documentation of MongoDB's
mongoimport
command line utility.
MongoDB's mongoimport
command will let us load bulk data from a JSON
or CSV
file.
First we'll load data in bulk from sample-data/csv/people.csv
. We'll save the command in scripts/import/people.sh
.
mongoimport --db=mongo-crud --collection=people --type=csv --headerline --file=sample-data/csv/people.csv
If we want to clear the collection before the import we pass the --drop
flag.
Now that we've inserted data into it, the mongo-crud
database and the people
collection both exist.
$ mongo mongo-crud
MongoDB shell version: 3.0.7
connecting to: mongo-crud
> show dbs
local 0.078GB
mongo-crud 0.078GB
> show collections
people
system.indexes
> db.people.count();
2438
Next we'll use the insert
collection method to add a few more people. We'll save our invocations in insert/people.js
. We'll execute that script using the mongo
load
method. Let's give these people a middle_initial or a nick_name. Note that the attributes we choose for these people need not match those from the data we loaded in bulk.
> load('scripts/insert/people.js');
MongoDB uses JSON natively (technically BSON), which makes it well suited for JavaScript applications. Conveniently, MongoDB lets us specify the JSON as a JavaScript object.
Together we'll add a few cities then we'll bulk load sample-data/csv/cities.csv
.
Add a pet to the pets
collection using insert
then bulk load sample-data/csv/pets.csv
.
Next add a person to the people
collection using insert
then bulk load sample-data/csv/people.csv
.
- Querying - Overview of retrieving data from MongoDB.
- Queries - More detailed overview on retrieving data.
- find - Detailed documentation on the
find
collection method. - findOne - Detailed documentation on the
findOne
collection method. - Data aggregation - Overview of summarizing documents.
- aggregate - Detailed documentation on the
aggregate
collection method.
Let's see some what we can learn about the people in the database.
Note: In a MongoDB shell script we'll use .forEach(printjson)
to display results. When using the REPL the .pretty()
method can be quite helpful.
What do we see?
-
MongoDB gave each of our documents a unique ID field, called _id.
-
MongoDB doesn't care that some documents have fewer or more attributes.
Let's look at the SQL to MongoDB Mapping Chart.
Together we'll build a query to get the count of cities by country.
Write a query to get the count of animals by kind born before 2010.
Then write a query to count people by height.
- Updating Data - overview of changing documents
- update - detailed documentation of MongoDB's
update
collection method. - Update Operators - The different modifications we can make during an update.
MongoDB makes it easy to add an array of items to a document. We'll update some people and give them some pets. Then we'll remove everyone's nick_name.
Let's update some cities population.
Update weight for pets then people.
- Removing Data - Overview of removing documents from a collection.
- remove - detailed documentation of MongoDB's
remove
collection method.
If we want to clean up, db.<collection>.drop();
drops the specified collection and db.dropDatabase();
drops the current database.
We'll remove a few people from the data-store.
Let's remove the cities that don't have a region.
Remove pets born before 1996 then people taller than 5'11".