nodbi provides a single user interface for interacting with many NoSQL databases.

So far we support the following DBs:

  • MongoDB
  • Redis (server based)
  • CouchDB
  • Elasticsearch
  • SQLite

Currently we have support for data.frame’s for the following operations

  • Create - all DBs
  • Exists - all DBs, except MongoDB
  • Get - all DBs
  • Query - all DBs, except Redis
  • Delete - all DBs
  • Update - just CouchDB

Install

cran version

dev version

install.packages("devtools")
devtools::install_github("ropensci/nodbi")
library("nodbi")

Initialize connections

Start CouchDB on the cli or with the app

Start Elasticsearch, e.g.:

If you want to use classic Redis server, we do that through the redux package, and you’ll need to start up Redis by e.g,. redis-server in your shell.

Start MongoDB: mongod (may need to do sudo mongod)

CouchDB

src <- src_couchdb()
docout <- docdb_create(src, key = "mtcars", value = mtcars)
head( docdb_get(src, "mtcars") )

Redis

src <- src_redis()
docdb_create(src, "mtcars", mtcars)
docdb_get(src, "mtcars")

MongoDB

library("ggplot2")
src <- src_mongo(verbose = FALSE)
ff <- docdb_create(src, "diamonds", diamonds)
docdb_get(src, "diamonds")

SQLite

src <- src_sqlite(dbname = ":memory:")
ff <- docdb_create(src, key = "mtcars", value = mtcars)
docdb_get(src, key = "mtcars")

Extension json1 is enabled in Package RSQLite (since version 1.1). This extension is used to emulate the behaviour of MongoDB with the methods for SQLite:

  • Parameter collection corresponds to the name of a table (parameter key) in the SQLite database dbname.
  • Json strings in parameters query and fields are translated into SQL commands, unless too complex.
  • Tables created by docdb_create() are defined as follows, with exactly two columns, an index column named _id and a column with json data named json:
CREATE TABLE mtcars ( _id TEXT PRIMARY_KEY NOT NULL, json JSON );
CREATE UNIQUE INDEX mtcars_index ON mtcars ( _id );

The following examples show the maximum level of complexity that can be used at this time with available json operaters (“\$eq”, “\$gt”, “\$gte”, “\$lt”, “\$lte”, “\$ne”); query implies AND of the comma separated expressions in the absence of a prefixed logical operator (available at this time: “\$and”, “\$or”).

Use with dplyr

library("dplyr")
src <- src_mongo(verbose = FALSE)
docdb_get(src, "diamonds") %>%
  group_by(cut) %>%
  summarise(mean_depth = mean(depth), mean_price = mean(price))

Meta

rofooter