Tests on travis failing

[zkamvar]

Epicontacts currently has a test failing on travis that doesn't look like a fluke:

https://travis-ci.org/reconhub/epicontacts/jobs/540650748#L1345-L1357

tests in ‘tests/testthat.R’ failed.
Last 13 lines of output:
  4: eval_bare(get_expr(quo), get_env(quo))
  5: zz[k = nocoords]
  6: `[.epicontacts`(zz, k = nocoords)
  7: data.frame(c(x$linelist[1], x$linelist[-1][k]), stringsAsFactors = FALSE)
  8: x$linelist[-1][k]
  9: `[.data.frame`(x$linelist[-1], k)
  10: stop("undefined columns selected")
  
  ══ testthat results  ═══════════════════════════════════════════════════════════
  OK: 141 SKIPPED: 3 WARNINGS: 1 FAILED: 2
  1. Failure: Returns epicontacts object subsetted correctly (@test_subset.epicontacts.R#115) 
  2. Error: Returns epicontacts object subsetted correctly (@test_subset.epicontacts.R#119) 
  
  Error: testthat unit tests failed

[zkamvar]

I believe the culprit was 4492006 since subset.epicontacts() relies on subset_clusters_by_id()

[zkamvar]

I think the issue here is that the subset data contains a row with missing values in the linelist

[finlaycampbell]

This looks like previously buggy behaviour that was fixed in the update. The command z[k = nocoords] should only be subsetting columns, but the saved, subsetted reference linelist has 12 rows whereas the original linelist has 13. In the current version both the pre-subsetted and post-subsetted linelists have 13 rows. I'd suggest overwriting rds/z.rds with the new output, as this gives the expect output.

[zkamvar]

As I mentioned in #105 (comment), I don't think this is an issue where we can simply update the output.

@nistara, I believe 4492006 may have caused some tests to fail on travis (see #104). They fail because the contacts list now has 13 rows, an extra column, and the dates have changed to numeric values (see below). Can you write a test for what you should expect to happen from the subset_clusters_by_id() function and then fix the issue?

library(epicontacts)
ebola_sim <- outbreaks::ebola_sim
x <- make_epicontacts(ebola_sim$linelist,
                      ebola_sim$contacts,
                      id = "case_id",
                      to = "case_id", from = "infector",
                      directed = FALSE)

dates <- as.Date(c("2014-04-08","2015-03-28"))


id <- names(which.max(get_degree(x, "out")))
z <- thin(subset(x, cluster_id = id), 2)
nocoords <- grep("(lat|lon)", names(z$linelist), perl = TRUE, invert = TRUE) - 1
z
#> 
#> /// Epidemiological Contacts //
#> 
#>   // class: epicontacts
#>   // 13 cases in linelist; 12 contacts;  non directed 
#> 
#>   // linelist
#> 
#> # A tibble: 13 x 12
#>    id    generation date_of_infecti… date_of_onset date_of_hospita…
#>    <chr>      <int>            <dbl>         <dbl>            <dbl>
#>  1 a6c6…          5            16226         16237            16239
#>  2 e4be…          6            16238         16243            16246
#>  3 e278…          7            16245         16252            16261
#>  4 916d…          6            16238         16264            16269
#>  5 4c10…          7            16268         16271            16271
#>  6 1f16…          7            16266         16277            16277
#>  7 c259…          7            16265         16275            16281
#>  8 8188…          7            16267         16284            16285
#>  9 ee65…          7            16268         16282            16288
#> 10 9f0c…          8            16276         16289            16290
#> 11 8863…          9            16290         16300            16302
#> 12 6ca9…          7            16267         16297            16306
#> 13 3ff1…         NA               NA            NA               NA
#> # … with 7 more variables: date_of_outcome <dbl>, outcome <chr>,
#> #   gender <chr>, hospital <chr>, lon <dbl>, lat <dbl>, id.1 <chr>
#> 
#>   // contacts
#> 
#> # A tibble: 12 x 3
#>    from   to     source 
#>    <chr>  <chr>  <chr>  
#>  1 a6c614 3ff1bc funeral
#>  2 a6c614 e4bee8 other  
#>  3 e4bee8 e27816 other  
#>  4 a6c614 916d0a other  
#>  5 916d0a 4c1040 other  
#>  6 916d0a 1f1620 other  
#>  7 916d0a c2595d other  
#>  8 916d0a 8188cd other  
#>  9 916d0a ee65e4 other  
#> 10 c2595d 9f0c37 other  
#> 11 9f0c37 88632e other  
#> 12 916d0a 6ca927 other
z$linelist
#>        id generation date_of_infection date_of_onset
#> 1  a6c614          5             16226         16237
#> 2  e4bee8          6             16238         16243
#> 3  e27816          7             16245         16252
#> 4  916d0a          6             16238         16264
#> 5  4c1040          7             16268         16271
#> 6  1f1620          7             16266         16277
#> 7  c2595d          7             16265         16275
#> 8  8188cd          7             16267         16284
#> 9  ee65e4          7             16268         16282
#> 10 9f0c37          8             16276         16289
#> 11 88632e          9             16290         16300
#> 12 6ca927          7             16267         16297
#> 13 3ff1bc         NA                NA            NA
#>    date_of_hospitalisation date_of_outcome outcome gender
#> 1                    16239           16247 Recover      f
#> 2                    16246              NA    <NA>      f
#> 3                    16261           16264 Recover      m
#> 4                    16269              NA    <NA>      f
#> 5                    16271              NA Recover      f
#> 6                    16277           16284   Death      f
#> 7                    16281           16322 Recover      m
#> 8                    16285           16311 Recover      f
#> 9                    16288           16289   Death      m
#> 10                   16290           16295   Death      f
#> 11                   16302           16307 Recover      m
#> 12                   16306           16313 Recover      m
#> 13                      NA              NA    <NA>   <NA>
#>              hospital       lon      lat   id.1
#> 1               other -13.21907 8.477376 a6c614
#> 2               other -13.22992 8.478471 e4bee8
#> 3  Connaught Hospital -13.24715 8.466236 e27816
#> 4               other -13.21633 8.480811 916d0a
#> 5               other -13.23596 8.477045 4c1040
#> 6               other -13.21373 8.468914 1f1620
#> 7   Military Hospital -13.26345 8.464106 c2595d
#> 8  Connaught Hospital -13.26120 8.463342 8188cd
#> 9  Connaught Hospital -13.23375 8.470376 ee65e4
#> 10              other -13.22357 8.483618 9f0c37
#> 11               <NA> -13.25422 8.460642 88632e
#> 12  Military Hospital -13.23645 8.468757 6ca927
#> 13               <NA>        NA       NA 3ff1bc

^{Created on 2019-06-06 by the reprex package (v0.3.0)}

[zkamvar]

Note that the extra row here has the ID "3ff1bc"... which doesn't appear to exist in the linelist:

grep("3ff1bc", outbreaks::ebola_sim$linelist$case_id)
#> integer(0)

^{Created on 2019-06-06 by the reprex package (v0.3.0)}

[finlaycampbell]

Ah soz you're right, good catch.
OK I've just pushed a potential fix to 7f5f401 and added a bunch of tests that test the function is actually doing what it's supposed to, i.e. returning all linelist and contact elements associated with the cluster a given ID belongs to.

[finlaycampbell]

Fixed as of f3463b6