check for similarly named function arguments in an R package

check_similar_fn_args.R

Library(dplyr)

# Which package to check
trias_namespace <- asNamespace("trias")

# Character vector of all functions and their arguments, print method of base ls with str() on every object
trias_fns_args <- capture.output(utils::lsf.str(trias_namespace))

# Extract the functions
trias_functions <- stringr::str_extract(trias_fns_args, ".+(?= :)") %>%
  .[!is.na(.)]
# Extract their arguments
trias_arguments <-
  stringr::str_extract_all(
    stringr::str_c(trias_fns_args, collapse = " "),
    stringr::regex("(?<=function \\().*?\\)", dotall = TRUE, multiline = TRUE)
  ) %>%
  purrr::map(~ stringr::str_remove(.x, "\\)$")) %>%
  purrr::map(~ stringr::str_split(.x, stringr::fixed(","))) %>%
  purrr::map(~ purrr::map(.x, stringr::str_squish)) %>%
  purrr::flatten()


# Create a tibble with one row per argument, also column with the arguments
# witout defaults (not foolproof)
trias_fns_args_tbbl <- 
  tibble(trias_functions, trias_arguments) %>% 
  tidyr::unnest(trias_arguments) %>% 
  mutate(arg_only = stringr::str_remove(trias_arguments, " .+$"))

# Cluster based on string disntance
hc <- 
  stringdist::stringdistmatrix(trias_fns_args_tbbl$arg_only,
                             trias_fns_args_tbbl$arg_only) %>% 
  as.dist() %>% 
  hclust(method = "average")

# Reorder our tibble so the most similar arguments are together
trias_fns_args_tbbl[hc$order, ]

See also https://gist.github.com/PietrH/c27e9ad582af97505e365eab96d46aa6