is it possible to do random reads of .tif or .nc files hosted on OSF Storage?

[mikoontz-vp]

Hi all,

This is such a great package and its existence makes it so much easier to recommend OSF as a hosting platform for projects.

I'm trying to figure out whether it is possible to read data directly from the OSF storage, specifically raster layers. For instance, if I host a .tif on AWS S3, I can use /vsis3/ and the {terra} package to read the raster metadata without first downloading the whole file. I can then crop it and just download a smaller portion: r = terra::rast(glue::glue('s3://{bucket_name}/{raster_basename})). Is something like this possible with the OSF API, which might then make it a potential hugely valuable feature for {osfr}?

Here's an OSF page for an OSF Storage-hosted .tif in case that's handy for anyone: https://osf.io/kwreq

[mikoontz]

An update here-- I recently learned that I could use the /vsicurl/ approach on raster that has a direct http link. I also learned about how to get a direct download link for an OSF id from CenterForOpenScience/osf.io#7020 and CenterForOpenScience/osf.io#8256.

So I'm able to do what I want using:

r <- terra::rast("/vsicurl/https://osf.io/download/kwreq")
r

Gives me:

class       : SpatRaster 
dimensions  : 104, 93, 2005  (nrow, ncol, nlyr)
resolution  : 11132, 11132  (x, y)
extent      : -434148, 601128, -656788, 500940  (xmin, xmax, ymin, ymax)
coord. ref. : NAD83 / California Albers (EPSG:3310) 
source      : kwreq 
names       : kwreq_1, kwreq_2, kwreq_3, kwreq_4, kwreq_5, kwreq_6, ... 

Note that this also works for vector files, because it is GDAL that implements the /vsicurl/ method:

(this one is pretty big, so might take a bit of time; probably a good example for walking the line between when it makes more sense to download the file locally then read it versus using the vsi approach)

gpkg <- sf::read_sf("/vsicurl/https://osf.io/download/74xcw")

And for completeness, we can also do this with .csv files by passing the https link directly to our file reading function:

csv <- readr::read_csv("https://osf.io/download/rgtvs")
csv_base <- utils::read.csv("https://osf.io/download/rgtvs")