ArcGIS Enterprise Portal

[roimartins]

Hi, thanks for the package creation . this will provide capabilities to promote ArcGIS Enterprise as a spatial data server for any data science solution and not only linked to users with an ArcGIS Pro installation ( as happens with arcgisbinding) .

At the moment I managed to obtain n httr2 token using auth_user() fuinction with my built-in Portal user with user name and password. However this token is not working when used with with arc_open() function:

> arcgislayers::arc_open( "https://name_of_the_server/server/rest/services/Hosted/name_of_the_feature/FeatureServer/0", token = user_token ) 
Error in `arcgislayers::arc_open()`:
! Status code: 498
Error : Invalid Token

Then I tried to use the other authentication methods with my user with OAuth approach but I didnt find the instructions to follow this approach up . Unfortunaley in the README file this link looks it is broken : see the {arcgislayers} authorization article.

I have tried with auth_client() function but this requires a ArcGIS Developer accoutn linked to the AGOL. However, it is not possible to link ArcGIS Developer account to an ArcGIS Enterprise Portal , only gives the option to use a AGOL URL .

Unfortunately , so far the only method to connect to not public layers in AGOL / AGE is arcgisbinding , with the limitation above mentioned.

[JosiahParry]

Thanks, @roimartins. I will work on removing that link from the README. Please see this article: https://r.esri.com/r-bridge-site/location-services/connecting-to-a-portal.html

Per the doc, the recommendation is to use auth_code(). You should avoid using auth_user() as that requires a username and password which is a security risk.

[JosiahParry]

This is probably some rate limiting that is being enforced by the server that you are calling.

Is there any reason why you need to perform this in a loop?

For example instead do:

years <- 2016:2023
max_year <- max(years)

arc_select(
  sar_f,
  fields = fields_s,
  where =  paste0("year >= 2016 and year <= ", max_year)
)

[roimartins]

@JosiahParry may I am using in the wrong way the arc_open function but currently does not allow to use the function without a token . So if we want to use it with public services still require a token ?

Even using the examples in the documentation rise the error with the missing token :

furl <- paste0(
"https://services3.arcgis.com/ZvidGQkLaDJxRSJ2/arcgis/rest/services/",
"PLACES_LocalData_for_BetterHealth/FeatureServer/0"
)

arc_open(furl)

Error in arc_open():
! Status code: 498
Error : Invalid token.
Run rlang::last_trace() to see where the error occurred.

[JosiahParry]

You probably have an expired token. You should restart your R session. If you provide a token it will always be checked first, even for public services. It's a little annoying. One thing we can do is to check if your token has expired and then remove it from your query with a warning.

But, my guess is that you reopened a stale R session and tried to run this code. Try restarting your R session or removing your token with unset_arc_token()

[roimartins]

Thanks again, yes that was the issue. The unset_arc_token() worked.

In addition to your idea also may the error messsage could suggest potential solutions. e.g. Get a new token if the wervice is protectted or use the 'unset_arc_token()' if the web service is open ?

[roimartins]

I have been using also teh aptial predicates with arc_select 9) to get intersecgtionand works perfectly fine . Just I needed a bit of research to create the poitn as st_sfc () with the right projectio to be passed into the 'filter_geom' parameter.

So , I sorted with : haul_sel_geom = st_point(c ( lon, lat ) ) |> st_sfc() |> st_set_crs(4326)

I like the idea that you dont have to pass the other parameters that are needed in the API GUI such : type of geometry and CRS of teh geometry . all is taken form teh sfc object.

Great stuff!