Skip to contents

Keep selected objects in an environment and remove everything else

Source: R/KeepEnv.R
KeepEnv.Rd

Keeps only the objects specified in Keep within Env and removes all other objects in that environment. Optionally returns a summary of what was removed.

Usage

KeepEnv(
  Keep,
  Env = parent.frame(),
  DryRun = FALSE,
  Invert = FALSE,
  Quiet = FALSE
)

Arguments

Keep

Character vector of object names to keep.

Env

Environment to clean. Defaults to the calling environment.

DryRun

If TRUE, does not remove anything and only reports what would be removed.

Invert

If TRUE, removes only Keep and keeps everything else.

Quiet

If TRUE, suppresses messages.

Value

Invisible list with kept and removed vectors.

Details

A long analysis session accumulates intermediates - raw imports, temporary merges, loop variables, half-finished plots - and only a few of those objects are worth carrying forward. Saving the whole workspace preserves all of it, producing .RData files that are large, slow to load, and full of objects whose provenance nobody remembers.

This is the inverse of writing out a long rm() call: name the handful of objects that matter and everything else goes. Because the list is allow-list rather than deny-list, objects created after the code was written are cleaned up too, instead of surviving because nobody remembered to add them to the rm().

Run it with DryRun = TRUE first. It reports exactly what would be removed without touching anything, which is worth doing whenever the environment holds something expensive to recompute.

Examples

# An analysis environment: a few results among many intermediates
env_Analysis <- new.env()
local({
  df_Raw <- data.frame(id = 1:5, value = rnorm(5))
  df_Clean <- df_Raw[!is.na(df_Raw$value), ]
  tmp_merge <- df_Clean
  i <- 3
  scratch_vector <- 1:100
  model_Final <- lm(value ~ id, data = df_Clean)
  df_Results <- data.frame(term = "id", estimate = coef(model_Final)[2])
}, envir = env_Analysis)

ls(env_Analysis)
#> [1] "df_Clean"       "df_Raw"         "df_Results"     "i"             
#> [5] "model_Final"    "scratch_vector" "tmp_merge"     

# Check what would go, before anything is removed
preview <- KeepEnv(
  Keep = c("df_Results", "model_Final"),
  Env = env_Analysis,
  DryRun = TRUE
)
#> Dry run: would remove 5 object(s).
preview$removed
#> [1] "df_Clean"       "df_Raw"         "i"              "scratch_vector"
#> [5] "tmp_merge"     

# Then do it for real, and save a workspace holding only what matters
KeepEnv(c("df_Results", "model_Final"), Env = env_Analysis)
#> Removing 5 object(s).
ls(env_Analysis)
#> [1] "df_Results"  "model_Final"

save(list = ls(env_Analysis), envir = env_Analysis,
     file = file.path(tempdir(), "analysis_results.RData"))

# `Invert = TRUE`: drop the objects named, keep the rest
env_Other <- new.env()
assign("df_Huge", data.frame(x = 1:10), envir = env_Other)
assign("df_Small", data.frame(x = 1:2), envir = env_Other)
KeepEnv("df_Huge", Env = env_Other, Invert = TRUE)
#> Removing 1 object(s).
ls(env_Other)
#> [1] "df_Small"