timeout option to restrict maximum time spent on every image

The original commit discourse/image_optim@8bf3c0e, see discussion in resolved PRs.

Resolves toy#21, resolves toy#148, resolves toy#149, resolves toy#162, resolves toy#184, resolves toy#189.

Co-authored-by: Blake Erickson <[email protected]>

CHANGELOG.markdown

@@ -2,6 +2,8 @@
 
 ## unreleased
 
+* Add `timeout` option to restrict maximum time spent on every image [#21](https://github.com/toy/image_optim/issues/21) [#148](https://github.com/toy/image_optim/pull/148) [#149](https://github.com/toy/image_optim/pull/149) [#162](https://github.com/toy/image_optim/pull/162) [#184](https://github.com/toy/image_optim/pull/184) [#189](https://github.com/toy/image_optim/pull/189) [@tgxworld](https://github.com/tgxworld) [@oblakeerickson](https://github.com/oblakeerickson) [@toy](https://github.com/toy)
+
 ## v0.29.0 (2021-04-28)
 
 * Require at least ruby 1.9.3 [@toy](https://github.com/toy)

README.markdown

@@ -291,6 +291,7 @@ optipng:
 * `:allow_lossy` — Allow lossy workers and optimizations *(defaults to `false`)*
 * `:cache_dir` — Configure cache directory
 * `:cache_worker_digests` - Also cache worker digests along with original file digest and worker options: updating workers invalidates cache
+* `:timeout` — Maximum time in seconds to spend on one image, note multithreading and cache *(defaults to unlimited)*
 
 Worker can be disabled by passing `false` instead of options hash or by setting option `:disable` to `true`.
 

lib/image_optim.rb

@@ -3,10 +3,12 @@
 require 'image_optim/bin_resolver'
 require 'image_optim/cache'
 require 'image_optim/config'
+require 'image_optim/errors'
 require 'image_optim/handler'
 require 'image_optim/image_meta'
 require 'image_optim/optimized_path'
 require 'image_optim/path'
+require 'image_optim/timer'
 require 'image_optim/worker'
 require 'in_threads'
 require 'shellwords'
@@ -46,6 +48,9 @@ class ImageOptim
   # Cache worker digests
   attr_reader :cache_worker_digests
 
+  # Timeout in seconds for each image
+  attr_reader :timeout
+
   # Initialize workers, specify options using worker underscored name:
   #
   # pass false to disable worker
@@ -78,6 +83,7 @@ def initialize(options = {})
       allow_lossy
       cache_dir
       cache_worker_digests
+      timeout
     ].each do |name|
       instance_variable_set(:"@#{name}", config.send(name))
       $stderr << "#{name}: #{send(name)}\n" if verbose
@@ -110,11 +116,17 @@ def optimize_image(original)
     return unless (workers = workers_for_image(original))
 
     optimized = @cache.fetch(original) do
+      timer = timeout && Timer.new(timeout)
+
       Handler.for(original) do |handler|
-        workers.each do |worker|
-          handler.process do |src, dst|
-            worker.optimize(src, dst)
+        begin
+          workers.each do |worker|
+            handler.process do |src, dst|
+              worker.optimize(src, dst, :timeout => timer)
+            end
           end
+        rescue Errors::TimeoutExceeded
+          handler.result
         end
       end
     end

lib/image_optim/cache.rb

@@ -21,6 +21,11 @@ def initialize(image_optim, workers_by_format)
           "#{bin.name}[#{bin.digest}]"
         end.sort!.uniq.join(', ')]
       end]
+      @global_options = begin
+        options = {}
+        options[:timeout] = image_optim.timeout if image_optim.timeout
+        options.empty? ? '' : options.inspect
+      end
     end
 
     def fetch(original)
@@ -68,6 +73,7 @@ def digest(path, format)
       digest = Digest::SHA1.file(path)
       digest.update options_by_format(format)
       digest.update bins_by_format(format) if @cache_worker_digests
+      digest.update @global_options
       s = digest.hexdigest
       "#{s[0..1]}/#{s[2..-1]}"
     end

lib/image_optim/cmd.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
+require 'image_optim/errors'
 require 'English'
 
 class ImageOptim
@@ -10,11 +11,30 @@ class << self
       # Return success status
       # Will raise SignalException if process was interrupted
       def run(*args)
-        success = system(*args)
+        if args.last.is_a?(Hash) && (timeout = args.last.delete(:timeout))
+          args.last[Gem.win_platform? ? :new_pgroup : :pgroup] = true
 
-        check_status!
+          pid = Process.spawn(*args)
+
+          waiter = Process.detach(pid)
+          if waiter.join(timeout)
+            status = waiter.value
+
+            check_status!(status)
+
+            status.success?
+          else
+            cleanup(pid, waiter)
 
-        success
+            fail Errors::TimeoutExceeded
+          end
+        else
+          success = system(*args)
+
+          check_status!
+
+          success
+        end
       end
 
       # Run using backtick
@@ -30,9 +50,7 @@ def capture(cmd)
 
     private
 
-      def check_status!
-        status = $CHILD_STATUS
-
+      def check_status!(status = $CHILD_STATUS)
         return unless status.signaled?
 
         # jruby incorrectly returns true for `signaled?` if process exits with
@@ -46,6 +64,27 @@ def check_status!
 
         fail SignalException, status.termsig
       end
+
+      def cleanup(pid, waiter)
+        if Gem.win_platform?
+          kill('KILL', pid)
+        else
+          kill('-TERM', pid)
+
+          # Allow 10 seconds for the process to exit
+          waiter.join(10)
+
+          kill('-KILL', pid)
+        end
+
+        waiter.join
+      end
+
+      def kill(signal, pid)
+        Process.kill(signal, pid)
+      rescue Errno::ESRCH, Errno::EPERM
+        # expected
+      end
     end
   end
 end

lib/image_optim/config.rb

@@ -117,6 +117,14 @@ def threads
       end
     end
 
+    # Timeout in seconds for each image:
+    # * not set by default and for `nil`
+    # * otherwise converted to float
+    def timeout
+      timeout = get!(:timeout)
+      timeout ? timeout.to_f : nil
+    end
+
     # Verbose mode, converted to boolean
     def verbose
       !!get!(:verbose)

lib/image_optim/elapsed_time.rb

@@ -0,0 +1,26 @@
+# frozen_string_literal: true
+
+class ImageOptim
+  # Use Process.clock_gettime if available to get time more fitting to calculate elapsed time
+  module ElapsedTime
+    CLOCK_NAME = %w[
+      CLOCK_UPTIME_RAW
+      CLOCK_UPTIME
+      CLOCK_MONOTONIC_RAW
+      CLOCK_MONOTONIC
+      CLOCK_REALTIME
+    ].find{ |name| Process.const_defined?(name) }
+
+    CLOCK_ID = CLOCK_NAME && Process.const_get(CLOCK_NAME)
+
+  module_function
+
+    def now
+      if CLOCK_ID
+        Process.clock_gettime(CLOCK_ID)
+      else
+        Time.now.to_f
+      end
+    end
+  end
+end

lib/image_optim/errors.rb

@@ -0,0 +1,9 @@
+# frozen_string_literal: true
+
+class ImageOptim
+  class Error < StandardError; end
+
+  module Errors
+    class TimeoutExceeded < Error; end
+  end
+end

lib/image_optim/runner/option_parser.rb

@@ -182,6 +182,10 @@ def wrap_regex(width)
     options[:allow_lossy] = allow_lossy
   end
 
+  op.on('--timeout N', Float, 'Maximum time in seconds to spend on one image') do |timeout|
+    options[:timeout] = timeout
+  end
+
   op.separator nil
 
   ImageOptim::Worker.klasses.each_with_index do |klass, i|

lib/image_optim/timer.rb

@@ -0,0 +1,25 @@
+# frozen_string_literal: true
+
+require 'image_optim/elapsed_time'
+
+class ImageOptim
+  # Hold start time and timeout
+  class Timer
+    include ElapsedTime
+
+    def initialize(seconds)
+      @start = now
+      @seconds = seconds
+    end
+
+    def elapsed
+      now - @start
+    end
+
+    def left
+      @seconds - elapsed
+    end
+
+    alias_method :to_f, :left
+  end
+end

lib/image_optim/worker.rb

@@ -3,6 +3,7 @@
 
 require 'image_optim/cmd'
 require 'image_optim/configuration_error'
+require 'image_optim/elapsed_time'
 require 'image_optim/path'
 require 'image_optim/worker/class_methods'
 require 'shellwords'
@@ -41,7 +42,7 @@ def options
 
     # Optimize image at src, output at dst, must be overriden in subclass
     # return true on success
-    def optimize(_src, _dst)
+    def optimize(_src, _dst, options = {})
       fail NotImplementedError, "implement method optimize in #{self.class}"
     end
 
@@ -122,33 +123,44 @@ def wrap_resolver_error_message(message)
     end
 
     # Run command setting priority and hiding output
-    def execute(bin, *arguments)
+    def execute(bin, arguments, options)
       resolve_bin!(bin)
 
       cmd_args = [bin, *arguments].map(&:to_s)
 
-      start = Time.now
-
-      success = run_command(cmd_args)
-
       if @image_optim.verbose
-        seconds = Time.now - start
-        $stderr << "#{success ? '✓' : '✗'} #{seconds}s #{cmd_args.shelljoin}\n"
+        run_command_verbose(cmd_args, options)
+      else
+        run_command(cmd_args, options)
       end
-
-      success
     end
 
     # Run command defining environment, setting nice level, removing output and
     # reraising signal exception
-    def run_command(cmd_args)
+    def run_command(cmd_args, options)
       args = [
         {'PATH' => @image_optim.env_path},
         *%W[nice -n #{@image_optim.nice}],
         *cmd_args,
-        {:out => Path::NULL, :err => Path::NULL},
+        options.merge(:out => Path::NULL, :err => Path::NULL),
       ]
       Cmd.run(*args)
     end
+
+    # Wrap run_command and output status, elapsed time and command
+    def run_command_verbose(cmd_args, options)
+      start = ElapsedTime.now
+
+      begin
+        success = run_command(cmd_args, options)
+        status = success ? '✓' : '✗'
+        success
+      rescue Errors::TimeoutExceeded
+        status = 'timeout'
+        raise
+      ensure
+        $stderr << format("%s %.1fs %s\n", status, ElapsedTime.now - start, cmd_args.shelljoin)
+      end
+    end
   end
 end

lib/image_optim/worker/advpng.rb

@@ -21,7 +21,7 @@ def run_order
         4
       end
 
-      def optimize(src, dst)
+      def optimize(src, dst, options = {})
         src.copy(dst)
         args = %W[
           --recompress
@@ -30,7 +30,7 @@ def optimize(src, dst)
           --
           #{dst}
         ]
-        execute(:advpng, *args) && optimized?(src, dst)
+        execute(:advpng, args, options) && optimized?(src, dst)
       end
     end
   end

lib/image_optim/worker/gifsicle.rb

@@ -37,7 +37,7 @@ def self.init(image_optim, options = {})
       CAREFUL_OPTION =
       option(:careful, false, 'Avoid bugs with some software'){ |v| !!v }
 
-      def optimize(src, dst)
+      def optimize(src, dst, options = {})
         args = %W[
           --output=#{dst}
           --no-comments
@@ -58,7 +58,7 @@ def optimize(src, dst)
         end
         args.unshift '--careful' if careful
         args.unshift "--optimize=#{level}" if level
-        execute(:gifsicle, *args) && optimized?(src, dst)
+        execute(:gifsicle, args, options) && optimized?(src, dst)
       end
     end
   end

lib/image_optim/worker/jhead.rb

@@ -25,7 +25,7 @@ def used_bins
         [:jhead, :jpegtran]
       end
 
-      def optimize(src, dst)
+      def optimize(src, dst, options = {})
         return false unless oriented?(src)
 
         src.copy(dst)
@@ -34,7 +34,7 @@ def optimize(src, dst)
           #{dst}
         ]
         resolve_bin!(:jpegtran)
-        execute(:jhead, *args) && dst.size?
+        execute(:jhead, args, options) && dst.size?
       end
 
     private