This is a contribution of by Benjamin William Barrett who is a PhD student at the Health and Biomedical Informatics track of Northwestern University Feinberg School of Medicine’s Health Sciences Integrated PhD (HSIP) program. He was geocoding healthcare encounters within a singular, privacy-preserving, environment via containerization and Docker but struggled to get the {appeears} R package authentication working.
I provided some guidance, mainly the use of a pre-created or spoofed keying, to avoid any user interaction in such a non-interactive setup. A similar approach is used in non-interactive unit testing of the package. A full description (taken from github of how to implement this in docker is provided below by Benjamin (many thanks go out to Benjamin for this write up). I hope this information might help some automate downloads of any data available on the AppEEARS API.
I was struggling for awhile to get the appeears package to function in a Docker container, and with Koen’s help finally got a working solution put together, so I’d like to document the solution here with a minimal working example in case it helps anyone else.
With Dockerfile
# Load R image
FROM rocker/r-ver:4.3.0
WORKDIR /app
# Install required packages
RUN apt-get update -y
RUN apt-get install libxml2-dev zlib1g-dev libfontconfig1-dev libssl-dev libcurl4-openssl-dev libharfbuzz-dev libfribidi-dev libfreetype6-dev libpng-dev libtiff5-dev libjpeg-dev libudunits2-dev cmake libnetcdf-dev libgdal-dev libgeos-dev libproj-dev libsqlite0-dev libsecret-1-dev libsodium23 -y
# Install R dependencies
RUN R -e "install.packages('appeears')"
COPY entrypoint.R .
WORKDIR /tmp
ENTRYPOINT ["/usr/local/bin/Rscript", "/app/entrypoint.R"]
And entrypoint.R
# Loading necessary packages
library(appeears)
# Creating a spoofed keyring
keyring::keyring_create("appeears", password = "spoof")
# Setting the key via username and password passed as command line environmental variables
rs_set_key(
user = Sys.getenv("USERNAME"),
password = Sys.getenv("PASSWORD")
)
# Pulling point-based Daymet data
df <- data.frame(
task = "Daymet",
subtask = "US-Ha1",
latitude = 42.5378,
longitude = -72.1715,
start = "2024-01-01",
end = "2024-12-31",
product = "DAYMET.004",
layer = c("dayl")
)
# Building the point-based request/task
task <- rs_build_task(df = df)
# Requesting the task to be executed
rs_request(
request = task,
user = Sys.getenv("USERNAME"),
transfer = TRUE,
path = getwd(),
verbose = TRUE
)
Activate Docker and on the command line navigate to the folder containing Dockerfile and entrypoint.R and run:
docker build -t my_container .
to build the Docker container.
Then, once the container is built, on the command line navigate to a working directory of interest and run:
docker run -e USERNAME="my_username" -e PASSWORD="my_password" --rm -v %cd%:/tmp my_container:latest
with my_username and my_password being your NASA EarthData username and password. This was run on Windows, but if using a different system you may have to sub %cd% out for $PWD or ${PWD}.
This will download selected data from AppEEARS into your working directory.
