I am currently evaluating Mongo and CouchDB for storing data points (analytics).
During my earlier interaction with CouchDB, I loved its JSONP based interface. I could perform all CRUD operations from purely JavaScript. Just run CouchDB and write some JavaScript - no server side component needed.
When comparing this to Mongo, is there any REST interface available? Is it possible to do CRUD purely from javascript in Mongo?
Thanks!
There is no full-blown REST interface to MongoDB, mainly because the server uses native binary protocol for efficiency. You can find few REST wrappers in official documentation (edit: MongoDB inc has now deleted this information):
- RESTHeart (Java 8) is a the data REST API server part of the MongoDB ecosystem. RESTHeart uses a standard representation format based on HAL with full native mongodb data support via the strict mode representation of BSON. It provides API for CRUD and data model operations, built-in authentication and authorization and it supports CORS. It is easy to setup and a docker container is available. RESTHeart is also fast and lightweight (~7 Mb footprint and ~200 Mb RAM peek usage).
- Sleepy Mongoose (Python) is a full featured REST interface for MongoDB which is available as a separate project.
- Rest on Mongo for node.js. The older MongoDB Rest is no longer maintained.
- Simple REST Interface The mongod process includes a simple read-only REST interface for convenience. For full REST capabilities we recommend using an external tool such as Sleepy.Mongoose.
The MongoDB Atlas Data API in Preview was also released in November 2021 to use with a hosted MongoDB instance through the company's Atlas offering. It lets you send complex queries and aggregations to MongoDB over a standard HTTPS interface, though it isn't currently recommended for direct client-side access.
For instance, once a cluster is created and the Data API is enabled for it, the following request can be used to insert a document -
curl --request POST \
'https://data.mongodb-api.com/app/<Unique ID>/endpoint/data/beta/action/insertOne' \
--header 'Content-Type: application/json' \
--header 'Access-Control-Request-Headers: *' \
--header 'api-key: <Data API Key>' \
--data-raw '{
"dataSource": "Cluster0",
"database": "todo",
"collection": "tasks",
"document": {
"status": "open",
"text": "Do the dishes"
}
}'
and the following to do an aggregation -
curl --location --request POST 'https://data.mongodb-api.com/app/<Unique ID>/endpoint/v1/beta/action/aggregate' \
--header 'Content-Type: application/json' \
--header 'Access-Control-Request-Headers: *' \
--header 'api-key:<Data API Key>' \
--data-raw '{
"collection":"movies",
"database":"sample_mflix",
"dataSource": "Cluster0",
"pipeline": [
{
"$search": {
"index": "default",
"text": {
"query": "Brad Pitt",
"path": {
"wilcard": "*"
}
}
}
}
]
}
Both the API and Atlas offer free tiers and only take a few minutes to spin up.
Full disclosure - I work for MongoDB, Inc.
Another option (shameless plug) is DrowsyDromedary.
We developed Drowsy out of frustration with the other REST options for Mongo. Namely, we found that:
Sleepy Mongoose is not really RESTful. It's a hacky HTTP interface that does not follow REST conventions; it doesn't use the standard REST HTTP methods, it doesn't use URLs to properly identify resources... We also found the options for limiting the maximum number of results (which is limited by default) rather confusing.
mongodb-rest offers a proper REST interface, but we found it to be a pain in the ass to run. It would die regularly, and drove our sysadmin insane (who admittedly has little experience running node.js services).
The built-in REST interface would have been great, but it being read-only means it's mostly useless for our needs.
DrowsyDromedary, was developed to address many of the above issues:
bundle, and then rackup, and you're running). It can also be easily deployed under Apache or nginx.
Edit
Python Eve is a fantastic solution for this. I have found it very easy to use and highly configurable.
Eve is an open source Python REST API framework designed for human beings. It allows to effortlessly build and deploy highly customizable, fully featured RESTful Web Services.
Eve is powered by Flask, Redis, Cerberus, Events and offers support for both MongoDB and SQL backends [*].
The codebase is thoroughly tested under Python 2.6, 2.7, 3.3, 3.4 and PyPy.
It is as simple to get started as:
from eve import Eve
app = Eve()
app.run()
and to use as
$ curl -i http://example.com/people
HTTP/1.1 200 OK
REST endpoint schemas are easily defined in a settings file.
Original
I haven't used it yet, but RESTHeart looks like an answer to this.
From the documentation it looks easy to set up and use.
mongodb-rest's fork by ActibeUnits (github repo) is one of the fastest and most feature rich rest interface to mongodb that I've seen.
I would highly recommend it.
RESTHeart is a the new data REST API server part of the MongoDB ecosystem.
It provides API for CRUD and data model operations, built-in authentication and authorization and supports CORS.
It uses a standard representation format based on HAL+json with full support of the native mongodb data format via the strict mode representation of BSON.
It is easy to setup and a docker container is available.
RESTHeart is fast (see performance test result) and lightweight (~7 Mb footprint and ~200 Mb RAM peek usage).
Not really. The mongod process includes a simple read-only REST interface for convenience. For full REST capabilities you need to use an external tool such as Sleepy.Mongoose.
