4 files changed, 84 insertions, 12 deletions

diff --git a/lib/chef/cookbook/file_system_file_vendor.rb b/lib/chef/cookbook/file_system_file_vendor.rb
index 8896e3ed30..e351ec4702 100644
--- a/lib/chef/cookbook/file_system_file_vendor.rb
+++ b/lib/chef/cookbook/file_system_file_vendor.rb
@@ -31,6 +31,9 @@ class Chef
     # non-sensical.
     class FileSystemFileVendor < FileVendor
 
+      attr_reader :cookbook_name
+      attr_reader :repo_paths
+
       def initialize(manifest, *repo_paths)
         @cookbook_name = manifest[:cookbook_name]
         @repo_paths = repo_paths.flatten
diff --git a/lib/chef/cookbook/file_vendor.rb b/lib/chef/cookbook/file_vendor.rb
index 406f23ca25..b82b52f90c 100644
--- a/lib/chef/cookbook/file_vendor.rb
+++ b/lib/chef/cookbook/file_vendor.rb
@@ -24,15 +24,39 @@ class Chef
     # This class handles fetching of cookbook files based on specificity.
     class FileVendor
 
-      def self.on_create(&block)
-        @instance_creator = block
+      @vendor_class = nil
+      @initialization_options = nil
+
+      # Configures FileVendor to use the RemoteFileVendor implementation. After
+      # calling this, subsequent calls to create_from_manifest will return a
+      # RemoteFileVendor object initialized with the given http_client
+      def self.fetch_from_remote(http_client)
+        @vendor_class = RemoteFileVendor
+        @initialization_options = http_client
+      end
+
+      def self.fetch_from_disk(cookbook_paths)
+        @vendor_class = FileSystemFileVendor
+        @initialization_options = cookbook_paths
+      end
+
+      # Returns the implementation class that is currently configured, or `nil`
+      # if one has not been configured yet.
+      def self.vendor_class
+        @vendor_class
+      end
+
+      def self.initialization_options
+        @initialization_options
       end
 
       # Factory method that creates the appropriate kind of
       # Cookbook::FileVendor to serve the contents of the manifest
       def self.create_from_manifest(manifest)
-        raise "Must call Chef::Cookbook::FileVendor.on_create before calling create_from_manifest factory" unless defined?(@instance_creator)
-        @instance_creator.call(manifest)
+        if @vendor_class.nil?
+          raise "Must configure FileVendor to use a specific implementation before creating an instance"
+        end
+        @vendor_class.new(manifest, @initialization_options)
       end
 
       # Gets the on-disk location for the given cookbook file.
diff --git a/lib/chef/cookbook/remote_file_vendor.rb b/lib/chef/cookbook/remote_file_vendor.rb
index 49de62cf65..2ddce31001 100644
--- a/lib/chef/cookbook/remote_file_vendor.rb
+++ b/lib/chef/cookbook/remote_file_vendor.rb
@@ -25,6 +25,9 @@ class Chef
     # if not available, loading them from the remote server.
     class RemoteFileVendor < FileVendor
 
+      attr_reader :rest
+      attr_reader :cookbook_name
+
       def initialize(manifest, rest)
         @manifest = manifest
         @cookbook_name = @manifest[:cookbook_name]
diff --git a/lib/chef/cookbook/synchronizer.rb b/lib/chef/cookbook/synchronizer.rb
index 7df3f554bb..1b96d0510b 100644
--- a/lib/chef/cookbook/synchronizer.rb
+++ b/lib/chef/cookbook/synchronizer.rb
@@ -10,6 +10,8 @@ class Chef
   # cache.
   class CookbookCacheCleaner
 
+    attr_accessor :skip_removal
+
     # Setup a notification to clear the valid_cache_entries when a Chef client
     # run starts
     Chef::Client.when_run_starts do |run_status|
@@ -40,15 +42,17 @@ class Chef
     end
 
     def cleanup_file_cache
-      unless Chef::Config[:solo]
+      unless Chef::Config[:solo] || skip_removal
         # Delete each file in the cache that we didn't encounter in the
         # manifest.
-        cache.find(File.join(%w{cookbooks ** *})).each do |cache_filename|
+        cache.find(File.join(%w{cookbooks ** {*,.*}})).each do |cache_filename|
           unless @valid_cache_entries[cache_filename]
             Chef::Log.info("Removing #{cache_filename} from the cache; it is no longer needed by chef-client.")
             cache.delete(cache_filename)
           end
         end
+      else
+        Chef::Log.info("Skipping removal of unused files from the cache")
       end
     end
 
@@ -59,6 +63,8 @@ class Chef
   class CookbookSynchronizer
     CookbookFile = Struct.new(:cookbook, :segment, :manifest_record)
 
+    attr_accessor :remove_obsoleted_files
+
     def initialize(cookbooks_by_name, events)
       @eager_segments = Chef::CookbookVersion::COOKBOOK_SEGMENTS.dup
       unless Chef::Config[:no_lazy_load]
@@ -70,6 +76,7 @@ class Chef
       @cookbooks_by_name, @events = cookbooks_by_name, events
 
       @cookbook_full_file_paths = {}
+      @remove_obsoleted_files = true
     end
 
     def cache
@@ -92,6 +99,10 @@ class Chef
       @cookbooks_by_name.key?(cookbook_name)
     end
 
+    def cookbook_segment(cookbook_name, segment)
+      @cookbooks_by_name[cookbook_name].manifest[segment]
+    end
+
     def files
       @files ||= cookbooks.inject([]) do |memo, cookbook|
         @eager_segments.each do |segment|
@@ -169,12 +180,10 @@ class Chef
       @cookbook_full_file_paths[file.cookbook][file.segment] << full_path
     end
 
-    # Iterates over cached cookbooks' files, removing files belonging to
-    # cookbooks that don't appear in +cookbook_hash+
-    def clear_obsoleted_cookbooks
-      @events.cookbook_clean_start
-      # Remove all cookbooks no longer relevant to this node
-      cache.find(File.join(%w{cookbooks ** *})).each do |cache_file|
+    # remove cookbooks that are not referenced in the expanded run_list at all
+    # (if we have an override run_list we may not want to do this)
+    def remove_old_cookbooks
+      cache.find(File.join(%w{cookbooks ** {*,.*}})).each do |cache_file|
         cache_file =~ /^cookbooks\/([^\/]+)\//
         unless have_cookbook?($1)
           Chef::Log.info("Removing #{cache_file} from the cache; its cookbook is no longer needed on this client.")
@@ -182,6 +191,39 @@ class Chef
           @events.removed_cookbook_file(cache_file)
         end
       end
+    end
+
+    # remove deleted files in cookbooks that are being used on the node
+    def remove_deleted_files
+      cache.find(File.join(%w{cookbooks ** {*,.*}})).each do |cache_file|
+        md = cache_file.match(/^cookbooks\/([^\/]+)\/([^\/]+)\/(.*)/)
+        next unless md
+        ( cookbook_name, segment, file ) = md[1..3]
+        if have_cookbook?(cookbook_name)
+          manifest_segment = cookbook_segment(cookbook_name, segment)
+          if manifest_segment.select { |manifest_record| manifest_record["path"] == "#{segment}/#{file}" }.empty?
+            Chef::Log.info("Removing #{cache_file} from the cache; its is no longer in the cookbook manifest.")
+            cache.delete(cache_file)
+            @events.removed_cookbook_file(cache_file)
+          end
+        end
+      end
+    end
+
+    # Iterates over cached cookbooks' files, removing files belonging to
+    # cookbooks that don't appear in +cookbook_hash+
+    def clear_obsoleted_cookbooks
+      @events.cookbook_clean_start
+
+      if remove_obsoleted_files
+        remove_old_cookbooks
+      else
+        Chef::Log.info("Skipping removal of obsoleted cookbooks from the cache")
+        CookbookCacheCleaner.instance.skip_removal = true
+      end
+
+      remove_deleted_files
+
       @events.cookbook_clean_complete
     end