Extract-Transform-Load Framework for Medium Data

A predictable and pipeable framework for performing ETL (extract-transform-load) operations on publicly-accessible medium-sized data set. This package sets up the method structure and implements generic functions. Packages that depend on this package download specific data sets from the Internet, clean them up, and import them into a local or remote relational database management system.


etl is an R package to facilitate Extract - Transform - Load (ETL) operations for medium data. The end result is generally a populated SQL database, but the user interaction takes place solely within R.

etl is now on CRAN, so you can install it in the usual way, then load it.

install.packages("etl")
library(etl)

Instantiate an etl object using a string that determines the class of the resulting object, and the package that provides access to that data. The trivial mtcars database is built into etl.

cars <- etl("mtcars")
## /tmp/RtmpjA2zLQ/file7b654f607c84.sqlite3
class(cars)
## [1] "etl_mtcars" "etl"        "src_sqlite" "src_sql"    "src"

etl works with a local or remote database to store your data. Every etl object extends a dplyr::src_sql object. If, as in the example above, you do not specify a SQL source, a local RSQLite database will be created for you. However, you can also specify any source that inherits from dplyr::src_sql.

library(RPostgreSQL)
db <- src_postgres(dbname = "mtcars", user = "postgres", host = "localhost")
library(RMySQL)
db <- src_mysql(dbname = "mtcars", user = "r-user", password = "mypass", host = "localhost")
cars <- etl("mtcars", db)

At the heart of etl are three functions: etl_extract(), etl_transform(), and etl_load().

The first step is to acquire data from an online source.

cars %>%
  etl_extract()
## Extracting raw data...

This creates a local store of raw data.

These data may need to be transformed from their raw form to files suitable for importing into SQL (usually CSVs).

cars %>%
  etl_transform()
## Transforming raw data...

Populate the SQL database with the transformed data.

cars %>%
  etl_load()
## Loading processed data...

## Data was successfully written to database.

To populate the whole database from scratch, use etl_create.

cars %>%
  etl_create()
## Loading SQL script at /home/bbaumer/R/x86_64-pc-linux-gnu-library/3.3/etl/sql/init.sqlite

## Extracting raw data...

## Transforming raw data...

## Loading processed data...

## Data was successfully written to database.

You can also update an existing database without re-initializing, but watch out for primary key collisions.

cars %>%
  etl_update()

Under the hood, there are four functions that etl_update chains together:

getS3method("etl_update", "default")
## function(obj, ...) {
##   obj <- obj %>%
##     etl_extract(...) %>%
##     etl_transform(...) %>%
##     etl_load(...)
##   invisible(obj)
## }
## <environment: namespace:etl>

etl_create is simply a call to etl_update that forces the SQL database to be written from scratch.

getS3method("etl_create", "default")
## function(obj, ...) {
##   obj <- obj %>%
##     etl_init(...) %>%
##     etl_update(...) %>%
##     etl_cleanup(...)
##   invisible(obj)
## }
## <environment: namespace:etl>

Now that your database is populated, you can work with it as a src data table just like any other dplyr source.

cars %>%
  tbl("mtcars") %>%
  group_by(cyl) %>%
  summarise(N = n(), mean_mpg = mean(mpg))
## Source:   query [?? x 3]
## Database: sqlite 3.8.6 [/tmp/RtmpjA2zLQ/file7b654f607c84.sqlite3]
## 
##      cyl     N mean_mpg
##    <int> <int>    <dbl>
## 1      4    11 26.66364
## 2      6     7 19.74286
## 3      8    14 15.10000
## ..   ...   ...      ...

Suppose you want to create your own ETL package called pkgname. All you have to do is write a package that requires etl, and then you have to write two S3 methods:

etl_extract.etl_pkgname()
etl_load.etl_pkgname()

You may also wish to write

etl_transform.etl_pkgname()
etl_cleanup.etl_pkgname()

All of these functions must take and return an object of class etl_pkgname that inherits from etl. Please see the packages listed below for examples.

Packages that use the etl framework:

tools::dependsOnPkgs("etl")
## [1] "airlines" "fec"      "macleish" "nyc311"

News

etl 0.3.5 (2016-11-28)

  • Fixed CRAN failures on Solaris (thanks Brian Ripley)

etl 0.3.4 (2016-11-07)

  • Added src_mysql_cnf as shorthand for connecting to MySQL
  • Fixed CRAN failures on Solaris (thanks Brian Ripley)
  • Moved to file.path uniformly (#7)
  • Moved smart_download to downloader for HTTPS

etl 0.3.3 (2016-07-27)

  • Moved is.etl to main documentation for etl (30dee378)
  • Fixed typo in DESCRIPTION (4e77fba2)
  • Fixed bug in etl_load.etl_mtcars by making etl_transform safer
  • Made verify_con messages easier to read
  • Added new functions for help with computing dates and matching filenames to dates
  • Added several tests
  • Added new_filenames argument to smart_download
  • Re-implemented etl_init (#7)
  • Renamed get_schema to find_schema (1c0a4e3)

etl 0.3.1 (2016-06-07)

  • released to CRAN
  • Added a NEWS.md file to track changes to the package.

Reference manual

It appears you don't have a PDF plugin for this browser. You can click here to download the reference manual.

install.packages("etl")

0.3.7 by Ben Baumer, 2 months ago


http://github.com/beanumber/etl


Report a bug at https://github.com/beanumber/etl/issues


Browse source code at https://github.com/cran/etl


Authors: Ben Baumer [aut, cre], Carson Sievert [ctb]


Documentation:   PDF Manual  


CC0 license


Imports DBI, datasets, downloader, lubridate, methods, stringr, readr, rlang, rvest, tibble, utils, xml2

Depends on dplyr

Suggests devtools, dbplyr, knitr, macleish, RSQLite, RPostgreSQL, RMySQL, MonetDBLite, ggplot2, testthat, rmarkdown


Depended on by macleish, nyctaxi.

Suggested by mdsr.


See at CRAN