Activity

1. laxamanaj

   self-assigned this

   on Feb 19, 2025
2. laxamanaj

   added

   documentationImprovements or additions to documentationImprovements or additions to documentation

   on Feb 19, 2025
3. laxamanaj

   linked a pull request that will close this issuefirst commit for listing R packages function #20on Mar 29, 2025
4. pstorozenko commented on May 28, 2025

   

   pstorozenko

   on May 28, 2025

   More actions

   Hi, we discussed at Appsilon how to extract all packages and function from the script and those are our findings.

   Insights

   1. NCmisc::list.functions.in.file() -> lists all functions from the script. It needs library() calls to be executed first. For example:

   # main.R
   library(dplyr)
   internal_function <- function() {
     return("This is an internal function")
   }
   internal_function()
   df <- iris %>%
     filter(Petal.Length > 1.5) %>%
     select(Species, Petal.Length)
   
   print(head(df))

   Gives

   

   Without loading the library it gives

   

   To find all library/require calls you can regular expressions, but for more reliability you can use XML Xpath queries or treesitter.

   This is not {box} compatible though.
   2. For box you can take inspiration from box.linters::box_usage_linter() and box.linters::box_unused_att_pkg_fun_linter() functions implementations in box.linters package.

   A PoC for NCmisc version

   In details, I attach the PoC. Note that since this PoC works on regular expressions, it fails on lines like # library(...), i.e., it recognises them as imports.

   Details

   get_functions_df <- function(file) {
     # Extract packages from file (both library and require)
     code <- readLines(file)
     
     # Match library() and require() with various quote styles
     lib_pattern <- '(library|require)\\s*\\(\\s*["\']?([^"\'\\)\\s]+)["\']?\\s*\\)'
     matches <- regmatches(code, gregexpr(lib_pattern, code))
     
     # Extract package names
     pkgs <- unique(unlist(lapply(matches, function(m) {
       gsub(lib_pattern, '\\2', m)
     })))
     
     # Load packages
     invisible(lapply(pkgs, function(p) {
       suppressPackageStartupMessages(require(p, character.only = TRUE, quietly = TRUE))
     }))
     
     # Get functions from NCmisc
     func_list <- NCmisc::list.functions.in.file(file)
     
     # Process into clean data frame
     df_list <- list()
     
     for (name in names(func_list)) {
       funcs <- func_list[[name]]
       if (length(funcs) == 0) next
       
       # Handle multiple packages
       if (grepl("^c\\(", name)) {
         # Extract individual package names
         pkgs <- gsub('"', '', unlist(strsplit(gsub("c\\(|\\)", "", name), ", ")))
         pkg_str <- paste(pkgs, collapse = ", ")
       } else {
         pkg_str <- gsub('"', '', name)
       }
       
       # Clean up package names
       pkg_str <- gsub("character\\(0\\)", "user_defined", pkg_str)
       
       df_list[[length(df_list) + 1]] <- data.frame(
         func_name = funcs,
         package = pkg_str,
         stringsAsFactors = FALSE
       )
     }
     
     # Combine and sort
     result <- do.call(rbind, df_list)
     result[order(result$package, result$func_name), ]
   }
   
   # Example usage:
   df <- get_functions_df("main.R")
   print(df)

   For the main.R from above we get, note that filter is in both dplyr and stats

   ➜ Rscript script.R
             func_name                      package
   3           library                 package:base
   4             print                 package:base
   5            return                 package:base
   6            select                package:dplyr
   1            filter package:dplyr, package:stats
   7              head                package:utils
   2 internal_function                 user_defined
   

   I hope it helps anyone working on this in the future.
5. laxamanaj commented on Jun 6, 2025

   

   laxamanaj

   on Jun 6, 2025

   CollaboratorAuthor

   More actions

   Hi, we discussed at Appsilon how to extract all packages and function from the script and those are our findings.

   Insights

   1. NCmisc::list.functions.in.file() -> lists all functions from the script. It needs library() calls to be executed first. For example:

   main.R

   library(dplyr)
   internal_function <- function() {
   return("This is an internal function")
   }
   internal_function()
   df <- iris %>%
   filter(Petal.Length > 1.5) %>%
   select(Species, Petal.Length)

   print(head(df))
   Gives

   

   Without loading the library it gives

   

   To find all library/require calls you can regular expressions, but for more reliability you can use XML Xpath queries or treesitter.

   This is not {box} compatible though. 2. For box you can take inspiration from box.linters::box_usage_linter() and box.linters::box_unused_att_pkg_fun_linter() functions implementations in box.linters package.

   A PoC for NCmisc version

   In details, I attach the PoC. Note that since this PoC works on regular expressions, it fails on lines like # library(...), i.e., it recognises them as imports.

   get_functions_df <- function(file) {

   Extract packages from file (both library and require)

   code <- readLines(file)

   Match library() and require() with various quote styles

   lib_pattern <- '(library|require)\s*\(\s*["']?([^"\'\\)\\s]+)["']?\s*\)'
   matches <- regmatches(code, gregexpr(lib_pattern, code))

   Extract package names

   pkgs <- unique(unlist(lapply(matches, function(m) {
   gsub(lib_pattern, '\2', m)
   })))

   Load packages

   invisible(lapply(pkgs, function(p) {
   suppressPackageStartupMessages(require(p, character.only = TRUE, quietly = TRUE))
   }))

   Get functions from NCmisc

   func_list <- NCmisc::list.functions.in.file(file)

   Process into clean data frame

   df_list <- list()

   for (name in names(func_list)) {
   funcs <- func_list[[name]]
   if (length(funcs) == 0) next

   # Handle multiple packages
   if (grepl("^c\\(", name)) {
     # Extract individual package names
     pkgs <- gsub('"', '', unlist(strsplit(gsub("c\\(|\\)", "", name), ", ")))
     pkg_str <- paste(pkgs, collapse = ", ")
   } else {
     pkg_str <- gsub('"', '', name)
   }
   
   # Clean up package names
   pkg_str <- gsub("character\\(0\\)", "user_defined", pkg_str)
   
   df_list[[length(df_list) + 1]] <- data.frame(
     func_name = funcs,
     package = pkg_str,
     stringsAsFactors = FALSE
   )
   

   }

   Combine and sort

   result <- do.call(rbind, df_list)
   result[order(result$package, result$func_name), ]
   }

   Example usage:

   df <- get_functions_df("main.R")
   print(df)
   For the main.R from above we get, note that filter is in both dplyr and stats

   ➜ Rscript script.R
             func_name                      package
   3           library                 package:base
   4             print                 package:base
   5            return                 package:base
   6            select                package:dplyr
   1            filter package:dplyr, package:stats
   7              head                package:utils
   2 internal_function                 user_defined
   

   I hope it helps anyone working on this in the future.

   Hi, @pstorozenko .

   I appreciate you looking into this. I like how you are able to extract both the functions and the packages that they are being called from.

   This could potentially be helpful information to merge with what I did here in #20 .

   We can take this back to our R Consortium colleagues to get their feedback to see if they still want this.