Add 'clean' rake task (#33)

* clean rake task

* Extract explaining method and improve style

* Pass count, use explaining kwargs

* Update output_path_test.rb

* Fix mtime of files

* Use different file names per test

* Enforce mtime

* Ensure files are removed after test

Co-authored-by: David Heinemeier Hansson <david@basecamp.com>

lib/propshaft/output_path.rb

@@ -0,0 +1,61 @@
+require "propshaft/asset"
+
+class Propshaft::OutputPath
+  attr_reader :path, :manifest
+
+  def initialize(path, manifest)
+    @path, @manifest = path, manifest
+  end
+
+  def clean(count, age)
+    asset_versions = files.group_by { |_, attrs| attrs[:logical_path] }
+    asset_versions.each do |logical_path, versions|
+      current = manifest[logical_path]
+
+      versions
+        .reject { |path, _| current && path == current }
+        .sort_by { |_, attrs| attrs[:mtime] }
+        .reverse
+        .each_with_index
+        .drop_while { |(_, attrs), index| fresh_version_within_limit(attrs[:mtime], count, expires_at: age, limit: index) }
+        .each { |(path, _), _| remove(path) }
+    end
+  end
+
+  def files
+    Hash.new.tap do |files|
+      all_files_from_tree(path).each do |file|
+        digested_path = file.relative_path_from(path)
+        logical_path, digest = extract_path_and_digest(digested_path)
+
+        files[digested_path.to_s] = {
+          logical_path: logical_path.to_s,
+          digest: digest,
+          mtime: File.mtime(file)
+        }
+      end
+    end
+  end
+
+  private
+    def fresh_version_within_limit(mtime, count, expires_at:, limit:)
+      modified_at = [ 0, Time.now - mtime ].max
+      modified_at < expires_at || limit < count
+    end
+
+    def remove(path)
+      FileUtils.rm(@path.join(path))
+      Propshaft.logger.info "Removed #{path}"
+    end
+
+    def all_files_from_tree(path)
+      path.children.flat_map { |child| child.directory? ? all_files_from_tree(child) : child }
+    end
+
+    def extract_path_and_digest(digested_path)
+      digest = digested_path.to_s[/-([0-9a-f]{7,128})\.(?!digested)[^.]+\z/, 1]
+      path = digest ? digested_path.sub("-#{digest}", "") : digested_path
+
+      [path, digest]
+    end
+end

lib/propshaft/processor.rb

@@ -1,3 +1,5 @@
+require "propshaft/output_path"
+
 class Propshaft::Processor
   MANIFEST_FILENAME = ".manifest.json"
 
@@ -18,6 +20,10 @@ def clobber
     FileUtils.rm_r(output_path) if File.exist?(output_path)
   end
 
+  def clean
+    OutputPath.new(output_path, load_path.manifest).clean(2, 1.hour)
+  end
+
   private
     def ensure_output_path_exists
       FileUtils.mkdir_p output_path

lib/propshaft/railtie.rb

@@ -57,6 +57,11 @@ class Railtie < ::Rails::Railtie
 
         desc "Remove config.assets.output_path"
         task clobber: :environment do
+          Rails.application.assets.processor.clobber
+        end
+
+        desc "Removes old files in config.assets.output_path"
+        task clean: :environment do
           Rails.application.assets.processor.clean
         end
 

test/propshaft/output_path_test.rb

@@ -0,0 +1,74 @@
+require "test_helper"
+require "minitest/mock"
+require "propshaft/asset"
+require "propshaft/load_path"
+require "propshaft/output_path"
+
+class Propshaft::OutputPathTest < ActiveSupport::TestCase
+  setup do
+    @manifest    = {
+      ".manifest.json": ".manifest.json",
+      "one.txt": "one-f2e1ec14d6856e1958083094170ca6119c529a73.txt"
+    }.stringify_keys
+    @output_path = Propshaft::OutputPath.new(Pathname.new("#{__dir__}/../fixtures/output"), @manifest)
+  end
+
+  test "files" do
+    files = @output_path.files
+
+    file = files["one-f2e1ec14d6856e1958083094170ca6119c529a73.txt"]
+    assert_equal "one.txt", file[:logical_path]
+    assert_equal "f2e1ec14d6856e1958083094170ca6119c529a73", file[:digest]
+    assert file[:mtime].is_a?(Time)
+  end
+
+  test "clean always keeps most current versions" do
+    @output_path.clean(0, 0)
+    assert @output_path.path.join(@manifest["one.txt"])
+    assert @output_path.path.join(@manifest[".manifest.json"])
+  end
+
+  test "clean keeps versions of assets that no longer exist" do
+    removed = output_asset("no-longer-in-manifest.txt", "current")
+    @output_path.clean(1, 0)
+    assert File.exists?(removed)
+  ensure
+    FileUtils.rm(removed) if File.exists?(removed)
+  end
+
+  test "clean keeps the correct number of versions" do
+    old     = output_asset("by_count.txt", "old", created_at: Time.now - 300)
+    current = output_asset("by_count.txt", "current", created_at: Time.now - 180)
+
+    @output_path.clean(1, 0)
+
+    assert File.exists?(current)
+    assert_not File.exists?(old)
+  ensure
+    FileUtils.rm(old) if File.exists?(old)
+    FileUtils.rm(current) if File.exists?(current)
+  end
+
+  test "clean keeps all versions under a certain age" do
+    old     = output_asset("by_age.txt", "old")
+    current = output_asset("by_age.txt", "current")
+
+    @output_path.clean(0, 3600)
+
+    assert File.exists?(current)
+    assert File.exists?(old)
+  ensure
+    FileUtils.rm(old) if File.exists?(old)
+    FileUtils.rm(current) if File.exists?(current)
+  end
+
+  private
+    def output_asset(filename, content, created_at: Time.now)
+      asset = Propshaft::Asset.new(nil, logical_path: filename)
+      asset.stub :content, content do
+        output_path = @output_path.path.join(asset.digested_path)
+        `touch -mt #{created_at.strftime('%y%m%d%H%M')} #{output_path}`
+        output_path
+      end
+    end
+end