renv chokes scanning dependencies erroring ungracefully

[MilesMcBain]

Error was initially reported to me in MilesMcBain/capsule#9

Running renv::init(bare = TRUE) in this project (https://github.com/njtierney/njtcv/tree/b002f11b78aac3f9c239dfba6d71895c06ca0408) gives:

[nav] r$> renv::init(bare = TRUE)
WARNING: One or more problems were discovered while enumerating dependencies.

Error in split.default(x = seq_len(nrow(x)), f = f, drop = drop, ...) :
  group length is 0 but data length > 0
Traceback (most recent calls last):
11: renv::init(bare = TRUE)
10: renv_dependencies_scope(project, action = "init")
 9: withCallingHandlers(dependencies(path, progress = FALSE, errors = errors,
        dev = TRUE), renv.dependencies.error = renv_dependencies_error_handler(message,
        errors))
 8: dependencies(path, progress = FALSE, errors = errors, dev = TRUE)
 7: renv_dependencies_impl(path = path, root = root, progress = progress,
        errors = errors, dev = dev, ...)
 6: renv_dependencies_report(errors)
 5: split(bound, bound$file)
 4: split.data.frame(bound, bound$file)
 3: lapply(split(x = seq_len(nrow(x)), f = f, drop = drop, ...),
        function(ind) x[ind, , drop = FALSE])
 2: split(x = seq_len(nrow(x)), f = f, drop = drop, ...)
 1: split.default(x = seq_len(nrow(x)), f = f, drop = drop, ...)

I spelunked into the innards of init and found the problems object you pull out of global state in renv_dependencies_report appears to have an unexpected shape:

[ins] Browse[8]> problems
[[1]]
[[1]]$file
NULL

[[1]]$error
<simpleError in parse(text = text, ...): <text>:1:1: unexpected '<'
1: <
    ^>

Although the file was NULL I grepped for < and found some potentially edge-case-smelling usage in a file containing this code:

embolden_name <- function(string,
                          name = "Nicholas J. Tierney"){
  str_replace_all(string = string,
                  pattern = name,
                  replacement = glue::glue("\\\\textbf{<name>}",
                                           .open = "<",
                                           .close = ">"))
}

Renaming the file (embolden_name.R) to have a .txt extension and rerunning init(bare = TRUE) worked.

So I think it's a parsing edge case, but I don't think renv should be performing parsing anyway due to bare = TRUE (see: #771)

Unfortunately, I have to tap out debugging there because the data is being maintained in global state via side effects in ways that I can't come to grips with quickly.

- Session info ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------   
 setting  value
 version  R version 4.0.5 (2021-03-31)
 os       Windows 10 x64
 system   x86_64, mingw32
 ui       RTerm
 language (EN)
 collate  English_Australia.1252
 ctype    English_Australia.1252
 tz       Australia/Brisbane
 date     2021-06-09

- Packages ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------   
 ! package     * version     date       lib source
 P cli           2.5.0.9000  2021-05-23 [?] local
 R digest        0.6.27      <NA>       [?] <NA>
 R evaluate      0.14        <NA>       [?] <NA>
 R fastmap       1.1.0.9000  <NA>       [?] <NA>
 R htmltools     0.5.1.9005  <NA>       [?] <NA>
 R jsonlite      1.7.2       <NA>       [?] <NA>
 R knitr         1.33        <NA>       [?] <NA>
 R renv          0.13.2-29   <NA>       [?] <NA>
 R rlang         0.4.11.9000 <NA>       [?] <NA>
 R rmarkdown     2.8         <NA>       [?] <NA>
   sessioninfo   1.1.1.9000  2021-06-03 [2] local
   withr         2.4.2.9000  2021-05-28 [2] local
 R xfun          0.22        <NA>       [?] <NA>
 R yaml          2.2.1       <NA>       [?] <NA>

[kevinushey]

Thanks! Should be fixed up now.

(the issue was indeed that calls to glue::glue() with non-default open and close delimiters was tripping up renv)