testing error with reticulate

[joshualerickson]

I'm getting this error below when running the tests within the testthat framework. For instance, I can run the code outside of the function test_that and the error doesn't occur. It only occurs when running the test... I also get there error in Base R. I'll just block it out for now but keep an eye out for when summarise has year, month and multiple years for landsat.

library(testthat)
library(tidyrgee)
#> 
#> Attaching package: 'tidyrgee'
#> The following object is masked from 'package:stats':
#> 
#>     filter
library(rgee)
ee_Initialize()
#> -- rgee 1.1.2 --------------------------------------- earthengine-api 0.1.312 -- 
#>  v user: not_defined
#>  v Initializing Google Earth Engine: v Initializing Google Earth Engine:  DONE!
#>  v Earth Engine account: users/joshualerickson 
#> --------------------------------------------------------------------------------

test_that("working with summarise by filter and grouping", {
  
  ## Landsat 8
  roi <- ee$Geometry$Point(-114.275, 45.891)
  
  
  ld_ic = ee$ImageCollection("LANDSAT/LC08/C01/T1_SR")$filterBounds(roi)
  
  # with == and year
  filter_year_month <- ld_ic %>%
    as_tidyee() %>%
    filter(year == 2018) %>%
    group_by(year, month) %>%
    summarise(stat = c('mean', 'median'))
  
  meta <- filter_year_month$ee_ob$getInfo()
  
  expect_equal(length(meta[["features"]][[1]][["bands"]]), 24)
  
  expect_equal(filter_year_month$vrt$band_names[[1]], c("B1_mean", "B2_mean",
                                                        "B3_mean", "B4_mean",
                                                        "B5_mean", "B6_mean",
                                                        "B7_mean", "B10_mean",
                                                        "B11_mean","sr_aerosol_mean",
                                                        "pixel_qa_mean", "radsat_qa_mean",
                                                        "B1_median", "B2_median",
                                                        "B3_median", "B4_median",
                                                        "B5_median", "B6_median",
                                                        "B7_median", "B10_median",
                                                        "B11_median","sr_aerosol_median",
                                                        "pixel_qa_median","radsat_qa_median" ))
  
  filter_year_month <- ld_ic %>%
    as_tidyee() %>%
    filter(year %in% c(2016:2019)) %>%
    group_by(year, month) %>%
    summarise(stat = c('mean', 'median'))
  
  meta <- filter_year_month$ee_ob$getInfo()
  
  expect_equal(length(meta[["features"]][[1]][["bands"]]), 24)
  
})
#> -- Error (<text>:38:3): working with summarise by filter and grouping ----------
#> <Rcpp::exception/C++Error/error/condition>
#> Error in `py_get_attr_impl(x, name, silent)`: OverflowError: Python int too large to convert to C long
#> Backtrace:
#>   1. ... %>% summarise(stat = c("mean", "median"))
#>   5. tidyrgee::as_tidyee(.)
#>   7. reticulate:::`$.python.builtin.object`(x, "first")
#>   8. reticulate:::py_get_attr_or_item(x, name, TRUE)
#>   9. reticulate::py_get_attr(x, name)
#>  10. reticulate:::py_get_attr_impl(x, name, silent)
#> Error:
#> ! Test failed

^{Created on 2022-06-29 by the reprex package (v2.0.1)}

[zackarno]

yeah I ran this code as well and experienced the same as you. Everything works correctly until you put it inside test_that.

As it does not seem to effect the functionality leaving the test out for now seems correct. It looks like another tricky one to understand.

seems likely that it might have some thing to do with large numeric dates:

r-spatial/rgee#79 (comment)

rstudio/reticulate#154 (comment)