Support alphanumeric and hyphenated words

[nuno-agostinho]

I am using the following words in my package:

• RNA-seq
• 1st
• 2nd
• EIF4G1

After inserting these words in inst/WORDLIST and running spelling::spell_check_package(), the function reports that the words seq, st, nd and EIF are misspelled.

Currently, my WORDLIST includes the words seq, st, nd and EIF to avoid triggering the spell checker, but I would prefer to include the full words. Thanks.

[jmbarbone]

I have the same issue, picked up with ordinal indicators. It looks like this is a problem with the hunspell parser:

hunspell::hunspell_parse(c("1st", "RNA-seq", "EIF4G1"))
#> [[1]]
#> [1] "st"
#> 
#> [[2]]
#> [1] "RNA" "seq"
#> 
#> [[3]]
#> [1] "EIF" "G"

^{Created on 2021-02-06 by the reprex package (v0.3.0)}

[jmbarbone]

Implementing a pre filter right before the parse here could work:

spelling/R/check-files.R

Lines 118 to 123 in a2b5f29

	spell_check_file_plain <- function(path, format, dict){
	lines <- readLines(path, warn = FALSE, encoding = 'UTF-8')
	words <- hunspell::hunspell_parse(lines, format = format, dict = dict)
	text <- vapply(words, paste, character(1), collapse = " ")
	spell_check_plain(text, dict = dict)
	}

It feels like more of a quick-fix because it parses with strsplit() then paste()s back together before being sent to the actual parsing function.

ignore_words <- c("1st", "RNA-seq", "EIF4G1")

lines <- c(
  "This is the 1st line.  It has first written in it.",
  "The second has RNA-seq inside. But does not use RNAseq -- without the '-'",
  "EIF4G1 but not EIF4G1fdsadf is used",
  "This line's words are fine!"
)

pre_filter_plain <- function(lines, ignore = character()) {
  word_list <- strsplit(lines, "([^-[:alnum:][:punct:]])")
  
  vapply(
    word_list,
    function(i) {
      paste(i[!i %in% ignore], collapse = " ")
    },
    character(1)
  )
}

pre_filter_plain(lines, ignore_words)
#> [1] "This is the line.  It has first written in it."                   
#> [2] "The second has inside. But does not use RNAseq -- without the '-'"
#> [3] "but not EIF4G1fdsadf is used"                                     
#> [4] "This line's words are fine!"

^{Created on 2021-02-06 by the reprex package (v0.3.0)}