Summarize R Environment Variables

summarize_environment_yaml.R

library(ggplot2)
library(yaml)

# Function to summarize a ggplot object
# TODO: dendrograms with hclust do not work
summarize_ggplot <- function(object) {
  wrap <- function(x) paste(
    paste(strwrap(x, exdent = 2), collapse = "\n"),
    "\n", sep = ""
  )
  
  summary_list <- list()
  
  if (!is.null(object$data)) {
    summary_list$data <- list(
      columns = names(object$data),
      dimensions = list(rows = nrow(object$data), columns = ncol(object$data))
    )
  }
  
  if (length(object$layers) > 0) {
    summary_list$layers <- paste(sapply(object$layers, function(layer) class(layer$geom)[1]), collapse = ", ")
  }
  
  return(summary_list)
}

# Function to summarize environment and format as YAML
summarize_environment_yaml <- function(func_char_limit = 50) {
  # Get the names of all objects in the global environment
  obj_names <- ls(envir = .GlobalEnv)
  
  # Initialize a list to store summaries
  obj_summaries <- list()
  
  # Loop through each object and summarize
  for (obj_name in obj_names) {
    obj <- get(obj_name, envir = .GlobalEnv) # Retrieve the object
    obj_type <- class(obj)[1]               # Get the object's primary class
    obj_size <- format(object.size(obj), units = "auto") # Get the object's size
    
    # Generate the appropriate summary structure
    obj_summary <- tryCatch(
      {
        if (is.data.frame(obj) || is.matrix(obj)) {
          list(
            dimensions = list(
              rows = nrow(obj),
              columns = ncol(obj)
            )
          )
        } else if (is.character(obj)) {
          list(
            length = length(obj),
            class = "character",
            mode = mode(obj)
          )
        } else if (is.numeric(obj)) {
          summary_values <- as.list(summary(obj))
          names(summary_values) <- tolower(gsub("\\.", "", names(summary_values)))
          summary_values
        } else if (is.logical(obj)) {
          list(
            mode = "logical",
            "FALSE" = sum(!obj, na.rm = TRUE),
            "TRUE" = sum(obj, na.rm = TRUE)
          )
        } else if (is.factor(obj)) {
          summary_values <- summary(obj)
          factor_list <- as.list(as.character(summary_values))
          names(factor_list) <- names(summary_values)
          factor_list
        } else if (is.function(obj)) {
          full_code <- paste(capture.output(obj), collapse = " ")
          truncated_code <- substr(full_code, 1, func_char_limit)
          if (nchar(full_code) > func_char_limit) {
            truncated_code <- paste0(truncated_code, "...")
          }
          list(
            code = truncated_code
          )
        } else if (is.ggplot(obj)) {
          summarize_ggplot(obj)
        } else {
          list(summary = "Summary not available")
        }
      },
      error = function(e) list(summary = "Summary not available")
    )
    
    # Add to summaries list
    obj_summaries[[obj_name]] <- list(
      type = obj_type,
      summary = obj_summary,
      size = obj_size
    )
  }
  
  # Convert the list to YAML and return
  as.yaml(obj_summaries)
}

# Generate the YAML summary with a customizable function character limit
environment_yaml <- summarize_environment_yaml(func_char_limit = 100) # Change character limit here
cat(environment_yaml)