alpha_centrality() crashes when `weights` argument is a string other than "weight"

[davidskalinder]

Describe the bug
The docs for alpha_centrality() say that the weights argument is

A character scalar that gives the name of the edge attribute to use in the adjacency matrix. If it is NULL, then the ‘weight’ edge attribute of the graph is used, if there is one. Otherwise, or if it is NA, then the calculation uses the standard adjacency matrix.

The function seems to work properly when the weights argument is "weight", NULL, or NA, but crashes when it's anything else.

The problem seems to be with one or more conditions in the block of ifs in both the dense and sparse functions:

rigraph/R/centrality.R

Line 1260 in 2810da8

if (is.null(weights) && "weight" %in% edge_attr_names(graph)) {

rigraph/R/centrality.R

Line 1301 in 2810da8

if (is.null(weights) && "weight" %in% edge_attr_names(graph)) {

It looks like the lines that go wrong were introduced in 81e82d8 and/or 4dda4fd.

#910 might obviate the need for this in the long term, so any fix might be temporary (but this logic shouldn't be reused without edits in the meantime, e.g., in #905).

To reproduce

library(igraph)
#> 
#> Attaching package: 'igraph'
#> The following objects are masked from 'package:stats':
#> 
#>     decompose, spectrum
#> The following object is masked from 'package:base':
#> 
#>     union

star <- make_star(10)
E(star)$myweight <- sample(ecount(star))
alpha_centrality(star, weights = "myweight", sparse = TRUE)
#> Error in h(simpleError(msg, call)): error in evaluating the argument 'x' in selecting a method for function 't': no such edge attribute

^{Created on 2023-10-21 with reprex v2.0.2}

Version information

• R/igraph version: 1.5.1
• R version: 4.1.2 (2021-11-01)
• Operating system: Ubuntu 20.04.6 LTS

> sessionInfo()
R version 4.1.2 (2021-11-01)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Ubuntu 20.04.6 LTS

Matrix products: default
BLAS:   /usr/lib/x86_64-linux-gnu/openblas-pthread/libblas.so.3
LAPACK: /usr/lib/x86_64-linux-gnu/openblas-pthread/liblapack.so.3

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] igraph_1.5.1

loaded via a namespace (and not attached):
 [1] viridis_0.6.4      pkgload_1.3.3      tidyr_1.3.0        tidygraph_1.2.3   
 [5] viridisLite_0.4.2  R.utils_2.12.2     ggraph_2.1.0       shiny_1.7.5.1     
 [9] yaml_2.3.7         remotes_2.4.2.1    ggrepel_0.9.4      sessioninfo_1.2.2 
[13] lattice_0.20-45    pillar_1.9.0       glue_1.6.2         digest_0.6.33     
[17] promises_1.2.1     polyclip_1.10-6    colorspace_2.1-0   Matrix_1.6-1.1    
[21] R.oo_1.25.0        htmltools_0.5.6.1  httpuv_1.6.11      clipr_0.8.0       
[25] pkgconfig_2.0.3    devtools_2.4.5     haven_2.5.3        unix_1.5.5        
[29] purrr_1.0.2        xtable_1.8-4       scales_1.2.1       processx_3.8.2    
[33] tweenr_2.0.2       later_1.3.1        ggforce_0.4.1      tibble_3.2.1      
[37] styler_1.10.2      generics_0.1.3     farver_2.1.1       ggplot2_3.4.4     
[41] usethis_2.2.2      ellipsis_0.3.2     cachem_1.0.8       withr_2.5.1       
[45] cli_3.6.1          magrittr_2.0.3     crayon_1.5.2       mime_0.12         
[49] memoise_2.0.1      evaluate_0.22      ps_1.7.5           R.methodsS3_1.8.2 
[53] fs_1.6.3           fansi_1.0.5        R.cache_0.16.0     MASS_7.3-55       
[57] forcats_1.0.0      pkgbuild_1.4.2     profvis_0.3.8      tools_4.1.2       
[61] prettyunits_1.2.0  hms_1.1.3          lifecycle_1.0.3    stringr_1.5.0     
[65] munsell_0.5.0      reprex_2.0.2       callr_3.7.3        compiler_4.1.2    
[69] rlang_1.1.1        grid_4.1.2         rstudioapi_0.15.0  htmlwidgets_1.6.2 
[73] miniUI_0.1.1.1     corrr_0.4.4        rmarkdown_2.25     gtable_0.3.4      
[77] graphlayouts_1.0.1 R6_2.5.1           gridExtra_2.3      knitr_1.44        
[81] dplyr_1.1.3        fastmap_1.1.1      utf8_1.2.3         stringi_1.7.12    
[85] Rcpp_1.0.11        vctrs_0.6.4        tidyselect_1.2.0   xfun_0.40         
[89] urlchecker_1.0.1

[maelle]

@szhorvat in the future shouldn't this error? If one wants to provide another weights vector, one should pass it directly?

my_weights <- edge_attr(star, "myweight")
alpha_centrality(star, weights = my_weights, sparse = TRUE)

do I understand this correctly?

[szhorvat]

No, why do you think so? It should be possible to pass any vector of weights. But we won't be able to pass the name of an attribute containing weights.

[maelle]

But we won't be able to pass the name of an attribute containing weights.

Right, I didn't write it correctly but I understood it correctly 🙈 So, the weights argument will not be a string other than weight, as it'll be a vector.

[rbutleriii]

Also came across this, and it would be convenient if it could be both, i.e. you could pass it the string of an edge attribute or just pass it a vector of numeric weights. It looks like it could be just:

  if (is.null(weights) && "weight" %in% edge_attr_names(graph)) {
    ## weights == NULL and there is a "weight" edge attribute
    attr <- "weight"
  } else if (is.null(weights)) {
    ## weights == NULL, but there is no "weight" edge attribute
    attr <- NULL
  } else if (is.character(weights) && length(weights) == 1) {
    ## name of an edge attribute, nothing to do
    attr <- weights #### <------------- formerly "weight"
  } else if (any(!is.na(weights))) {
    ## weights != NULL and weights != rep(NA, x)
    graph <- set_edge_attr(graph, "weight", value = as.numeric(weights))
    attr <- "weight"
  } else {
    ## weights != NULL, but weights == rep(NA, x)
    attr <- NULL
  }