Save a warmed R package cache for faster installs in failing Actions

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

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 from one to the other'
+    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:
@@ -111,29 +114,99 @@ runs:
         shell: Rscript {0}
         working-directory: ${{ inputs.working-directory }}
 
-      - name: Restore R package cache
+      - name: Restore R package cache keys
+        id: cache-args
+        if: inputs.cache != 'false'
+        shell: bash
+        run: |
+          # Make cache args
+          RESTORE_KEY="${{ format('{0}-{1}-{2}', steps.install.outputs.os-version, steps.install.outputs.r-version, inputs.cache-version) }}"
+          PRIMARY_KEY="$RESTORE_KEY-${{ hashFiles(format('{0}/.github/pkg.lock', inputs.working-directory )) }}"
+          echo "key=$PRIMARY_KEY" >> $GITHUB_OUTPUT
+          echo "restore-keys=$RESTORE_KEY" >> $GITHUB_OUTPUT
+
+          # Make cache path
+          echo "path<<EOF" >> $GITHUB_OUTPUT
+          echo '${{ env.R_LIBS_USER }}/*' >> $GITHUB_OUTPUT
+          echo '!${{ env.R_LIBS_USER }}/pak' >> $GITHUB_OUTPUT
+          echo "EOF" >> $GITHUB_OUTPUT
+
+      - name: Restore R package cache (and save cache on successful Action)
         if: inputs.cache == 'true'
         uses: actions/cache@v4
         with:
-          path: |
-            ${{ env.R_LIBS_USER }}/*
-            !${{ env.R_LIBS_USER }}/pak
-          key: ${{ format('{0}-{1}-{2}-{3}', steps.install.outputs.os-version, steps.install.outputs.r-version, inputs.cache-version, hashFiles(format('{0}/.github/pkg.lock', inputs.working-directory ))) }}
-          restore-keys: ${{ steps.install.outputs.os-version }}-${{ steps.install.outputs.r-version }}-${{inputs.cache-version }}-
+          path: ${{ steps.cache-args.outputs.path }}
+          key: ${{ steps.cache-args.outputs.key }}
+          restore-keys: ${{ steps.cache-args.outputs.restore-keys }}
+
+      - name: Restore R package cache
+        if: inputs.cache == 'always'
+        id: r-pkg-cache
+        uses: actions/cache/restore@v3
+        with:
+          path: ${{ steps.cache-args.outputs.path }}
+          key: ${{ steps.cache-args.outputs.key }}
+          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 <- c(
+            "translations", # Core package on windows
+            installed_packages[!is.na(installed_df$Priority)]
+          )
+          pak_pkgs <- c(
+            "pak", # Never remove pak
+            strsplit("${{ 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}
         working-directory: ${{ inputs.working-directory }}
 
+      # Step used to help speed up package installation on troublesome Actions
+      - name: Save R package cache
+        # If `actions/cache/restore@v3` restored from `cache-primary-key`, then `cache-hit == 'true'`; If so, do not save cache
+        if: inputs.cache == 'always' && steps.r-pkg-cache.outputs.cache-hit != 'true'
+        uses: actions/cache/save@v3
+        with:
+          path: ${{ steps.cache-args.outputs.path }}
+          key: ${{ steps.r-pkg-cache.outputs.cache-primary-key }}
+
       - name: Session info
         run: |
           # Session info