From b9cd043c92b2b3381bd2df92b0e1e1c7a0334718 Mon Sep 17 00:00:00 2001
From: Barret Schloerke <barret@posit.co>
Date: Wed, 14 Feb 2024 12:26:19 -0500
Subject: [PATCH] Add `prune` param to `setup-r-dependencies` to prune extra R
 packages only when `cache="always"`

---
 setup-r-dependencies/README.md   |  4 ++-
 setup-r-dependencies/action.yaml | 43 +++++++++++++++++++++++++++++---
 2 files changed, 43 insertions(+), 4 deletions(-)

diff --git a/setup-r-dependencies/README.md b/setup-r-dependencies/README.md
index 7d4a377ea..b09ca3b41 100644
--- a/setup-r-dependencies/README.md
+++ b/setup-r-dependencies/README.md
@@ -14,7 +14,7 @@ This action install dependencies for the current R environment based on the DESC
 Inputs available
 
 - `cache` - default `true`. Whether packages should be cached across runs or
-  not.
+  not. If `"always"` is provided, the package cache will be saved even if the Action fails.
 - `cache-version` - default `1`. If you need to invalidate the existing
   cache pass any other number and a new cache will be used. Ignored if
   `cache: false`. Note that you can also [delete caches
@@ -40,6 +40,8 @@ Inputs available
   you need quoting. Defaults to `FALSE`.
 - `working-directory` - default `'.'`. If the DESCRIPTION file is not in the
   root directory of your repository.
+- `prune` - default `true`. Used when `cache="always"`. Whether to prune the R
+  library of packages not found within the lock file used for installation.
 
 Basic:
 ```yaml
diff --git a/setup-r-dependencies/action.yaml b/setup-r-dependencies/action.yaml
index 0db13733c..3ad0b934e 100644
--- a/setup-r-dependencies/action.yaml
+++ b/setup-r-dependencies/action.yaml
@@ -3,7 +3,7 @@ description: 'Action to setup installation tools and install R dependencies'
 author: 'Jim Hester'
 inputs:
   cache:
-    description: 'A boolean value indicating whether packages should be cached (on success) from one to the other. If `"always"` is provided, the package cache will be saved even if the Action fails.'
+    description: 'Whether packages should be cached across runs or not. If `"always"` is provided, the package cache will be saved even if the Action fails.'
     required: true
     default: true
   cache-version:
@@ -29,6 +29,9 @@ inputs:
   upgrade:
     description: 'Whether to install the latest available versions of the dependencies. Must be an R expression. See the README for details if you need quoting.'
     default: 'FALSE'
+  prune:
+    description: 'Whether to prune the R library of packages not found within the lock file used for installation.'
+    default: true
 runs:
   using: "composite"
   steps:
@@ -145,13 +148,47 @@ runs:
           restore-keys: ${{ steps.cache-args.outputs.restore-keys }}
 
       - name: Install dependencies
+        id: install-pkgs
         run: |
           # Install/Update packages
           cat("::group::Install/update packages\n")
           Sys.setenv("PKGCACHE_HTTP_VERSION" = "2")
           library(pak, lib.loc = Sys.getenv("R_LIB_FOR_PAK"))
-          pak::lockfile_install(".github/pkg.lock")
-          ## Clean up lock file
+          pkgs <- pak::lockfile_install(".github/pkg.lock")
+          # Store the installed packages
+          packages <- paste0(pkgs$package, collapse = "|")
+          cat("packages=", packages, "\n", file = Sys.getenv("GITHUB_OUTPUT"), sep = "", append = TRUE)
+          cat("::endgroup::\n")
+        shell: Rscript {0}
+        working-directory: ${{ inputs.working-directory }}
+
+      - name: Prune unused R packages
+        if: inputs.cache == 'always' && inputs.prune == 'true'
+        run: |
+          # Prune packages
+          cat("::group::Prune unnecessary packages\n")
+          installed_df <- as.data.frame(installed.packages(), stringsAsFactors = FALSE)
+          installed_packages <- installed_df$Package
+          core_pkgs <- installed_packages[!is.na(installed_df$Priority)]
+          pak_pkgs <- c(
+            "pak",
+            strsplit("${{ steps.install-pkgs.outputs.packages }}", "|", fixed = TRUE)[[1]]
+          )
+          to_remove_pkgs <- setdiff(installed_packages, c(core_pkgs, pak_pkgs))
+          if (length(to_remove_pkgs) > 0) {
+            cat("Removing unnecessary packages: ", paste0(to_remove_pkgs, collapse = ", "), "\n")
+            remove.packages(to_remove_pkgs)
+          } else {
+            cat("No unnecessary packages to remove\n")
+          }
+          cat("::endgroup::\n")
+        shell: Rscript {0}
+        working-directory: ${{ inputs.working-directory }}
+
+      - name: Clean up lock file
+        run: |
+          # Clean up lock file
+          cat("::group::Clean up lock file\n")
           unlink(".github/pkg.lock")
           cat("::endgroup::\n")
         shell: Rscript {0}