Skip to content

An R package for storing and retrieving data.frames in git repositories.

License

Notifications You must be signed in to change notification settings

ropensci/git2rdata

Repository files navigation

git2rdata git2rdata logo

CRAN status Project Status: Active – The project has reached a stable, usable state and is being actively developed. lifecycle ROpenSci review Licence minimal R version DOI codecov GitHub forks GitHub stars GitHub code size in bytes GitHub repo size

Please visit the git2rdata website at https://ropensci.github.io/git2rdata/. The vignette code on the website link to a rendered version of the vignette. Functions have a link to their help file.

Rationale

The git2rdata package is an R package for writing and reading dataframes as plain text files. A metadata file stores important information.

  1. Storing metadata allows to maintain the classes of variables. By default, git2rdata optimizes the data for file storage. The optimization is most effective on data containing factors. The optimization makes the data less human readable. The user can turn this off when they prefer a human readable format over smaller files. Details on the implementation are available in vignette("plain_text", package = "git2rdata").
  2. Storing metadata also allows smaller row based diffs between two consecutive commits. This is a useful feature when storing data as plain text files under version control. Details on this part of the implementation are available in vignette("version_control", package = "git2rdata"). Although we envisioned git2rdata with a git workflow in mind, you can use it in combination with other version control systems like subversion or mercurial.
  3. git2rdata is a useful tool in a reproducible and traceable workflow. vignette("workflow", package = "git2rdata") gives a toy example.
  4. vignette("efficiency", package = "git2rdata") provides some insight into the efficiency of file storage, git repository size and speed for writing and reading.

Why Use Git2rdata?

  • You can store dataframes as plain text files.
  • The dataframe you read identical information content as the one you wrote.
    • No changes in data type.
    • Factors keep their original levels, including their order.
    • Date and date-time format are unambiguous, documented in the metadata.
  • The data and the metadata are in a standard and open format, making it readable by other software.
  • git2rdata checks the data and metadata during the reading. read_vc() informs the user if there is tampering with the data or metadata.
  • Git2rdata integrates with the git2r package for working with git repository from R.
    • Another option is using git2rdata solely for writing to disk and handle the plain text files with your favourite version control system outside of R.
  • The optimization reduces the required disk space by about 30% for both the working directory and the git history.
  • Reading data from a HDD is 30% faster than read.table(), writing to a HDD takes about 70% more time than write.table().
  • Git2rdata is useful as a tool in a reproducible and traceable workflow. See vignette("workflow", package = "git2rdata").
  • You can detect when a file was last modified in the git history. Use this to check whether an existing analysis is obsolete due to new data. This allows to not rerun up to date analyses, saving resources.

Talk About git2rdata at

useR!2019 in Toulouse, France

<iframe width="560" height="315" src="https://www.youtube-nocookie.com/embed/sbRPmakBFqo" frameborder="0" allow="accelerometer; autoplay; encrypted-media; gyroscope; picture-in-picture" allowfullscreen></iframe>

Installation

Install from CRAN

install.packages("git2rdata")

Install the development version from GitHub

# installation requires the "remotes" package
# install.package("remotes")

# install with vignettes (recommended)
remotes::install_github(
  "ropensci/git2rdata", 
  build = TRUE, 
  dependencies = TRUE, 
  build_opts = c("--no-resave-data", "--no-manual")
)
# install without vignettes
remotes::install_github("ropensci/git2rdata"))

Usage in Brief

The user stores dataframes with write_vc() and retrieves them with read_vc(). Both functions share the arguments root and file. root refers to a base location where to store the dataframe. It can either point to a local directory or a local git repository. file is the file name to use and can include a path relative to root. Make sure the relative path stays within root.

# using a local directory
library(git2rdata)
root <- "~/myproject" 
write_vc(my_data, file = "rel_path/filename", root = root)
read_vc(file = "rel_path/filename", root = root)
root <- git2r::repository("~/my_git_repo") # git repository

More details on store dataframes as plain text files in vignette("plain_text", package = "git2rdata").

# using a git repository
library(git2rdata)
repo <- repository("~/my_git_repo")
pull(repo)
write_vc(my_data, file = "rel_path/filename", root = repo, stage = TRUE)
commit(repo, "My message")
push(repo)
read_vc(file = "rel_path/filename", root = repo)

Please read vignette("version_control", package = "git2rdata") for more details on using git2rdata in combination with version control.

What Data Sizes Can Git2rdata Handle?

The recommendation for git repositories is to use files smaller than 100 MiB, a repository size less than 1 GiB and less than 25k files. The individual file size is the limiting factor. Storing the airbag dataset (DAAG::nassCDS) with write_vc() requires on average 68 (optimized) or 97 (verbose) byte per record. The file reaches the 100 MiB limit for this data after about 1.5 million (optimized) or 1 million (verbose) observations.

Storing a 90% random subset of the airbag dataset requires 370 kiB (optimized) or 400 kiB (verbose) storage in the git history. Updating the dataset with other 90% random subsets requires on average 60 kiB (optimized) to 100 kiB (verbose) per commit. The git history reaches the limit of 1 GiB after 17k (optimized) to 10k (verbose) commits.

Your mileage might vary.

Citation

Please use the output of citation("git2rdata")

Folder Structure

  • R: The source scripts of the R functions with documentation in Roxygen format
  • man: The help files in Rd format
  • inst/efficiency: pre-calculated data to speed up vignette("efficiency", package = "git2rdata")
  • testthat: R scripts with unit tests using the testthat framework
  • vignettes: source code for the vignettes describing the package
  • man-roxygen: templates for documentation in Roxygen format
  • pkgdown: source files for the git2rdata website
  • .github: guidelines and templates for contributors
git2rdata
├── .github 
├─┬ inst
│ └── efficiency
├── man 
├── man-roxygen 
├── pkgdown
├── R
├─┬ tests
│ └── testthat
└── vignettes

Contributions

git2rdata welcomes contributions. Please read our Contributing guidelines first. The git2rdata project has a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.

rOpenSci footer