pygments.rb 1.2.1 → 2.0.0

data/lib/pygments/popen.rb

@@ -1,9 +1,7 @@
-# coding: utf-8
 # frozen_string_literal: true
 
+require 'json'
 require 'open3'
-require 'multi_json'
-require 'timeout'
 require 'logger'
 require 'time'
 
@@ -15,38 +13,36 @@ end
 # Python process.
 module Pygments
   class Popen
-
-    def popen4(cmd)
-      stdin, stdout, stderr, wait_thr = Open3.popen3(cmd)
-      [wait_thr[:pid], stdin, stdout, stderr]
+    def popen4(argv)
+      stdin, stdout, stderr, wait_thr = Open3.popen3(*argv, { close_others: true })
+      while (pid = wait_thr.pid).nil? && wait_thr.alive?
+        # For unknown reasons, wait_thr.pid is not immediately available on JRuby
+      end
+      [pid, stdin, stdout, stderr]
     end
 
     # Get things started by opening a pipe to mentos (the freshmaker), a
     # Python process that talks to the Pygments library. We'll talk back and
     # forth across this pipe.
-    def start(pygments_path = File.expand_path('../../../vendor/pygments-main/', __FILE__))
+    def start(pygments_path = File.expand_path('../../vendor/pygments-main', __dir__))
       begin
-        @log = Logger.new(ENV['MENTOS_LOG'] ||= File::NULL)
+        @log = Logger.new(ENV['MENTOS_LOG'] || File::NULL)
         @log.level = Logger::INFO
-        @log.datetime_format = "%Y-%m-%d %H:%M "
-      rescue
-        @log = Logger.new(File::NULL)
+        @log.datetime_format = '%Y-%m-%d %H:%M '
       end
 
       ENV['PYGMENTS_PATH'] = pygments_path
 
       # Make sure we kill off the child when we're done
-      at_exit { stop "Exiting" }
+      at_exit { stop 'Exiting' }
 
       # A pipe to the mentos python process. #popen4 gives us
       # the pid and three IO objects to write and read.
-      script = "#{python_binary} #{File.expand_path('../mentos.py', __FILE__)}"
-      @pid, @in, @out, @err = popen4(script)
-      @log.info "Starting pid #{@pid} with fd #{@out.to_i} and python #{python_binary}."
-    end
-
-    def windows?
-      RUBY_PLATFORM =~ /mswin|mingw/
+      argv = [*python_binary, File.expand_path('mentos.py', __dir__)]
+      @pid, @in, @out, @err = popen4(argv)
+      @in.binmode
+      @out.binmode
+      @log.info "Starting pid #{@pid} with python #{python_binary}."
     end
 
     def python_binary
@@ -58,14 +54,13 @@ module Pygments
     end
 
     # Detect a suitable Python binary to use.
-    # Or return $PYGMENTS_RB_PYTHON if it's exists.
     def find_python_binary
-      if ENV['PYGMENTS_RB_PYTHON']
-        return which(ENV['PYGMENTS_RB_PYTHON'])
-      elsif windows? && which('py')
-        return 'py -2'
+      if Gem.win_platform?
+        return %w[py python3 python].first { |py| !which(py).nil? }
       end
-      return which('python2') || which('python')
+
+      # On non-Windows platforms, we simply rely on shebang
+      []
     end
 
     # Cross platform which command
@@ -73,12 +68,12 @@ module Pygments
     def which(command)
       exts = ENV['PATHEXT'] ? ENV['PATHEXT'].split(';') : ['']
       ENV['PATH'].split(File::PATH_SEPARATOR).each do |dir|
-        exts.each { |ext|
+        exts.each do |ext|
           path = File.join(dir, "#{command}#{ext}")
           return path if File.executable?(path) && !File.directory?(path)
-        }
+        end
       end
-      return nil
+      nil
     end
 
     # Stop the child process by issuing a kill -9.
@@ -94,14 +89,14 @@ module Pygments
     # the signal isn't sent); but we have permissions, and
     # we're not doing anything invalid here.
     def stop(reason)
-      if @pid
+      unless @pid.nil?
+        @log.info "Killing pid: #{@pid}. Reason: #{reason}"
         begin
           Process.kill('KILL', @pid)
           Process.waitpid(@pid)
         rescue Errno::ESRCH, Errno::ECHILD
         end
       end
-      @log.info "Killing pid: #{@pid.to_s}. Reason: #{reason}"
       @pid = nil
     end
 
@@ -116,30 +111,29 @@ module Pygments
     # Returns true if the child is alive.
     def alive?
       return true if defined?(@pid) && @pid && Process.kill(0, @pid)
+
       false
     rescue Errno::ENOENT, Errno::ESRCH
       false
     rescue Errno::EPERM
-      raise MentosError, "EPERM checking if child process is alive."
+      raise MentosError, 'EPERM checking if child process is alive.'
     end
 
     # Public: Get an array of available Pygments formatters
     #
     # Returns an array of formatters.
     def formatters
-      mentos(:get_all_formatters).inject(Hash.new) do | hash, (name, desc, aliases) |
+      mentos(:get_all_formatters).each_with_object({}) do |(name, desc, aliases), hash|
         # Remove the long-winded and repetitive 'Formatter' suffix
         name.sub!(/Formatter$/, '')
         hash[name] = {
-          :name => name,
-          :description => desc,
-          :aliases => aliases
+          name: name,
+          description: desc,
+          aliases: aliases
         }
-        hash
       end
     end
 
-
     # Public: Get all lexers from a serialized array. This avoids needing to spawn
     # mentos when it's not really needed (e.g., one-off jobs, loading the Rails env, etc).
     #
@@ -147,33 +141,25 @@ module Pygments
     #
     # Returns an array of lexers.
     def lexers
-      begin
-        lexer_file = File.expand_path('../../../lexers', __FILE__)
-        raw = File.open(lexer_file, "rb").read
-        Marshal.load(raw)
-      rescue Errno::ENOENT
-        raise MentosError, "Error loading lexer file. Was it created and vendored?"
-      end
+      lexer_file = File.expand_path('../../lexers', __dir__)
+      raw = File.open(lexer_file, 'rb').read
+      Marshal.load(raw)
+    rescue Errno::ENOENT
+      raise MentosError, 'Error loading lexer file. Was it created and vendored?'
     end
 
     # Public: Get back all available lexers from mentos itself
     #
     # Returns an array of lexers.
     def lexers!
-      mentos(:get_all_lexers).inject(Hash.new) do |hash, lxr|
+      mentos(:get_all_lexers).each_with_object({}) do |lxr, hash|
         name = lxr[0]
         hash[name] = {
-          :name => name,
-          :aliases => lxr[1],
-          :filenames => lxr[2],
-          :mimetypes => lxr[3]
+          name: name,
+          aliases: lxr[1],
+          filenames: lxr[2],
+          mimetypes: lxr[3]
         }
-        hash["dasm16"] = {:name=>"dasm16", :aliases=>["DASM16"], :filenames=>["*.dasm16", "*.dasm"], :mimetypes=>['text/x-dasm16']}
-        hash["Puppet"] = {:name=>"Puppet", :aliases=>["puppet"], :filenames=>["*.pp"], :mimetypes=>[]}
-        hash["Augeas"] = {:name=>"Augeas", :aliases=>["augeas"], :filenames=>["*.aug"], :mimetypes=>[]}
-        hash["TOML"]   = {:name=>"TOML",   :aliases=>["toml"],   :filenames=>["*.toml"], :mimetypes=>[]}
-        hash["Slash"]  = {:name=>"Slash",  :aliases=>["slash"],  :filenames=>["*.sl"], :mimetypes=>[]}
-        hash
       end
     end
 
@@ -188,7 +174,7 @@ module Pygments
     end
 
     # Public: Return css for highlighted code
-    def css(klass='', opts={})
+    def css(klass = '', opts = {})
       if klass.is_a?(Hash)
         opts = klass
         klass = ''
@@ -199,17 +185,13 @@ module Pygments
     # Public: Return the name of a lexer.
     def lexer_name_for(*args)
       # Pop off the last arg if it's a hash, which becomes our opts
-      if args.last.is_a?(Hash)
-        opts = args.pop
-      else
-        opts = {}
-      end
+      opts = if args.last.is_a?(Hash)
+               args.pop
+             else
+               {}
+             end
 
-      if args.last.is_a?(String)
-        code = args.pop
-      else
-        code = nil
-      end
+      code = (args.pop if args.last.is_a?(String))
 
       mentos(:lexer_name_for, args, opts, code)
     end
@@ -221,7 +203,7 @@ module Pygments
     #
     # Returns the highlighted string
     # or nil when the request to the Python process timed out.
-    def highlight(code, opts={})
+    def highlight(code, opts = {})
       # If the caller didn't give us any code, we have nothing to do,
       # so return right away.
       return code if code.nil? || code.empty?
@@ -234,191 +216,142 @@ module Pygments
 
       # Get back the string from mentos and force encoding if we can
       str = mentos(:highlight, nil, opts, code)
-      str.force_encoding(opts[:options][:outencoding]) if str.respond_to?(:force_encoding)
+      if str.respond_to?(:force_encoding)
+        str.force_encoding(opts[:options][:outencoding])
+      end
       str
     end
 
     private
 
+    def with_watchdog(timeout_time, error_message)
+      state_mutex = Mutex.new
+      state = :alive
+      wd_cleanup = ConditionVariable.new
+
+      watchdog = timeout_time > 0 ? Thread.new do
+        state_mutex.synchronize do
+          wd_cleanup.wait(state_mutex, timeout_time) if state != :finished
+          if state != :finished
+            @log.error error_message
+            stop error_message
+            state = :timeout
+          end
+        end
+      end : nil
+      begin
+        yield
+      ensure
+        if watchdog
+          state_mutex.synchronize do
+            state = :finished if state == :alive
+            # wake up watchdog thread
+            wd_cleanup.signal
+          end
+          watchdog.join
+        end
+
+        raise MentosError, error_message if state == :timeout
+      end
+    end
+
     # Our 'rpc'-ish request to mentos. Requires a method name, and then optional
     # args, kwargs, code.
-    def mentos(method, args=[], kwargs={}, original_code=nil)
+    def mentos(method, args = [], kwargs = {}, code = nil)
       # Open the pipe if necessary
       start unless alive?
 
-      begin
-        # Timeout requests that take too long.
-        # Invalid MENTOS_TIMEOUT results in just using default.
-        timeout_time = Integer(ENV["MENTOS_TIMEOUT"]) rescue 8
-
-        Timeout::timeout(timeout_time) do
-          # For sanity checking on both sides of the pipe when highlighting, we prepend and
-          # append an id.  mentos checks that these are 8 character ids and that they match.
-          # It then returns the id's back to Rubyland.
-          id = (0...8).map{65.+(rand(25)).chr}.join
-          code = add_ids(original_code, id) if original_code
-
-          # Add metadata to the header and generate it.
-          if code
-            bytesize = code.bytesize
-          else
-            bytesize = 0
-          end
+      # Timeout requests that take too long.
+      # Invalid MENTOS_TIMEOUT results in just using default.
+      timeout_time = kwargs.delete(:timeout)
+      if timeout_time.nil?
+        timeout_time = begin
+                         Integer(ENV['MENTOS_TIMEOUT'])
+                       rescue TypeError
+                         0
+                       end
+      end
 
-          kwargs.freeze
-          kwargs = kwargs.merge("fd" => @out.to_i, "id" => id, "bytes" => bytesize)
-          out_header = MultiJson.dump(:method => method, :args => args, :kwargs => kwargs)
+      # Add metadata to the header and generate it.
+      bytesize = if code
+                   code.bytesize
+                 else
+                   0
+                 end
 
+      kwargs.freeze
+      kwargs = kwargs.merge('bytes' => bytesize)
+      out_header = JSON.generate(method: method, args: args, kwargs: kwargs)
+
+      begin
+        res = with_watchdog(timeout_time, "Timeout on a mentos #{method} call") do
           # Get the size of the header itself and write that.
-          bits = get_fixed_bits_from_header(out_header)
-          @in.write(bits)
+          @in.write([out_header.bytesize].pack('N'))
+          @log.info "Size out: #{out_header.bytesize}"
 
           # mentos is now waiting for the header, and, potentially, code.
-          write_data(out_header, code)
+          @in.write(out_header)
+          @log.info "Out header: #{out_header}"
+          @in.write(code) unless code.nil?
 
-          check_for_error
+          @in.flush
 
           # mentos will now return data to us. First it sends the header.
-          header = get_header
 
-          # Now handle the header, any read any more data required.
-          res = handle_header_and_return(header, id)
-
-          # Finally, return what we got.
-          return_result(res, method)
-        end
-      rescue Timeout::Error
-        # If we timeout, we need to clear out the pipe and start over.
-        @log.error "Timeout on a mentos #{method} call"
-        stop "Timeout on mentos #{method} call."
-      end
-
-    rescue Errno::EPIPE, EOFError
-    stop "EPIPE"
-    raise MentosError, "EPIPE"
-    end
-
-    def check_for_error
-      return if @err.closed?
+          header_len_bytes = @out.read(4)
+          if header_len_bytes.nil?
+            raise Errno::EPIPE, %(Failed to read response from Python process on a mentos #{method} call)
+          end
 
-      timeout_time = 0.25 # set a very little timeout so that we do not hang the parser
+          header_len = header_len_bytes.unpack('N')[0]
+          @log.info "Size in: #{header_len}"
+          header = @out.read(header_len)
 
-      Timeout::timeout(timeout_time) do
-        error_msg = @err.read
+          # Now handle the header, any read any more data required.
+          handle_header_and_return(header)
+        end
 
-        unless error_msg.empty?
-          @log.error "Error running python script: #{error_msg}"
-          stop "Error running python script: #{error_msg}"
-          raise MentosError, error_msg
+        # Finally, return what we got.
+        return_result(res, method)
+      rescue Errno::EPIPE => e
+        begin
+          error_msg = @err.read
+          @log.error "Error running Python script: #{error_msg}"
+          stop "Error running Python script: #{error_msg}"
+          raise MentosError, %(#{e}: #{error_msg})
+        rescue Errno::EPIPE
+          @log.error e.to_s
+          stop e.to_s
+          raise e
         end
+      rescue StandardError => e
+        @log.error e.to_s
+        stop e.to_s
+        raise e
       end
-    rescue Timeout::Error
-      # during the specified time no error were found
-      @err.close
     end
 
-
     # Based on the header we receive, determine if we need
     # to read more bytes, and read those bytes if necessary.
     #
-    # Then, do a sanity check with the ids.
-    #
-    # Returns a result — either highlighted text or metadata.
-    def handle_header_and_return(header, id)
+    # Returns a result - either highlighted text or metadata.
+    def handle_header_and_return(header)
       if header
+        @log.info "In header: #{header}"
         header = header_to_json(header)
         bytes = header[:bytes]
 
         # Read more bytes (the actual response body)
         res = @out.read(bytes.to_i)
 
-        if header[:method] == "highlight"
+        if header[:method] == 'highlight'
           # Make sure we have a result back; else consider this an error.
-          if res.nil?
-            @log.warn "No highlight result back from mentos."
-            stop "No highlight result back from mentos."
-            raise MentosError, "No highlight result back from mentos."
-          end
-
-          # Remove the newline from Python
-          res = res[0..-2]
-          @log.info "Highlight in process."
-
-          # Get the id's
-          start_id = res[0..7]
-          end_id = res[-8..-1]
-
-          # Sanity check.
-          if not (start_id == id and end_id == id)
-            @log.error "ID's did not match. Aborting."
-            stop "ID's did not match. Aborting."
-            raise MentosError, "ID's did not match. Aborting."
-          else
-            # We're good. Remove the padding
-            res = res[10..-11]
-            @log.info "Highlighting complete."
-            res
-          end
+          raise MentosError, 'No highlight result back from mentos.' if res.nil?
         end
-        res
-      else
-        @log.error "No header data back."
-        stop "No header data back."
-        raise MentosError, "No header received back."
-      end
-    end
 
-    # With the code, prepend the id (with two spaces to avoid escaping weirdness if
-    # the following text starts with a slash (like terminal code), and append the
-    # id, with two padding also. This means we are sending over the 8 characters +
-    # code + 8 characters.
-    def add_ids(code, id)
-      (id + "  #{code}  #{id}").freeze
-    end
-
-    # Write data to mentos, the Python process.
-    #
-    # Returns nothing.
-    def write_data(out_header, code=nil)
-      @in.write(out_header)
-      @log.info "Out header: #{out_header}"
-      @in.write(code) if code
-    end
-
-    # Sanity check for size (32-arity of 0's and 1's)
-    def size_check(size)
-      size_regex = /[0-1]{32}/
-      if size_regex.match(size)
-        true
+        res
       else
-        false
-      end
-    end
-
-    # Read the header via the pipe.
-    #
-    # Returns a header.
-    def get_header
-      begin
-        size = @out.read(33)
-        size = size[0..-2]
-
-        # Sanity check the size
-        if not size_check(size)
-          @log.error "Size returned from mentos.py invalid."
-          stop "Size returned from mentos.py invalid."
-          raise MentosError, "Size returned from mentos.py invalid."
-        end
-
-        # Read the amount of bytes we should be expecting. We first
-        # convert the string of bits into an integer.
-        header_bytes = size.to_s.to_i(2) + 1
-        @log.info "Size in: #{size.to_s} (#{header_bytes.to_s})"
-        @out.read(header_bytes)
-      rescue
-        @log.error "Failed to get header."
-        stop "Failed to get header."
-        raise MentosError, "Failed to get header."
+        raise MentosError, 'No header received back.'
       end
     end
 
@@ -426,7 +359,7 @@ module Pygments
     # want them, text otherwise.
     def return_result(res, method)
       unless method == :lexer_name_for || method == :highlight || method == :css
-        res = MultiJson.load(res, :symbolize_keys => true)
+        res = JSON.parse(res, symbolize_names: true)
       end
       res = res.rstrip if res.class == String
       res
@@ -434,27 +367,13 @@ module Pygments
 
     # Convert a text header into JSON for easy access.
     def header_to_json(header)
-      @log.info "[In header: #{header} "
-      header = MultiJson.load(header, :symbolize_keys => true)
+      header = JSON.parse(header, symbolize_names: true)
 
       if header[:error]
-        # Raise this as a Ruby exception of the MentosError class.
-        # Stop so we don't leave the pipe in an inconsistent state.
-        @log.error "Failed to convert header to JSON."
-        stop header[:error]
         raise MentosError, header[:error]
       else
         header
       end
     end
-
-    def get_fixed_bits_from_header(out_header)
-      size = out_header.bytesize
-
-      # Fixed 32 bits to represent the int. We return a string
-      # representation: e.g, "00000000000000000000000000011110"
-      Array.new(32) { |i| size[i] }.reverse!.join
-    end
   end
 end
-

data/lib/pygments/version.rb

@@ -1,5 +1,5 @@
-# coding: utf-8
 # frozen_string_literal: true
+
 module Pygments
-  VERSION = '1.2.1'
+  VERSION = '2.0.0'
 end

data/lib/pygments.rb

@@ -1,10 +1,9 @@
-# coding: utf-8
 # frozen_string_literal: true
+
 require File.join(File.dirname(__FILE__), 'pygments/popen')
 require 'forwardable'
 
 module Pygments
-
   autoload :Lexer, 'pygments/lexer'
 
   class << self
@@ -16,14 +15,14 @@ module Pygments
     end
 
     def_delegators :engine,
-      :formatters,
-      :lexers,
-      :lexers!,
-      :filters,
-      :styles,
-      :css,
-      :lexer_name_for,
-      :highlight,
-      :start
+                   :formatters,
+                   :lexers,
+                   :lexers!,
+                   :filters,
+                   :styles,
+                   :css,
+                   :lexer_name_for,
+                   :highlight,
+                   :start
   end
 end

data/pygments.rb.gemspec

@@ -1,25 +1,27 @@
-require File.expand_path('../lib/pygments/version', __FILE__)
+# frozen_string_literal: true
+
+require File.expand_path('lib/pygments/version', __dir__)
 
 Gem::Specification.new do |s|
   s.name = 'pygments.rb'
   s.version = Pygments::VERSION
 
   s.summary = 'pygments wrapper for ruby'
-  s.description = 'pygments.rb exposes the pygments syntax highlighter to Ruby'
+  s.description = 'pygments.rb is a Ruby wrapper for Pygments syntax highlighter'
 
-  s.homepage = 'https://github.com/tmm1/pygments.rb'
-  s.has_rdoc = false
+  s.homepage = 'https://github.com/pygments/pygments.rb'
+  s.required_ruby_version = '>= 2.3.0'
 
-  s.authors = ['Aman Gupta', 'Ted Nyman']
-  s.email = ['aman@tmm1.net']
+  s.authors = ['Aman Gupta', 'Ted Nyman', 'Marat Radchenko']
+  s.email = ['marat@slonopotamus.org']
   s.license = 'MIT'
 
-  s.add_dependency 'multi_json', '>= 1.0.0'
-  s.add_development_dependency 'rake-compiler', '~> 0.7.6'
-  s.add_development_dependency 'test-unit', '~> 3.0.0'
+  s.add_development_dependency 'rake', '~> 13.0.0'
+  s.add_development_dependency 'rubocop', '~> 0.81.0'
+  s.add_development_dependency 'test-unit', '~> 3.3.0'
 
   # s.extensions = ['ext/extconf.rb']
   s.require_paths = ['lib']
 
-  s.files = `git ls-files`.split("\n").select { |f| !File.symlink?(f) } + ['lexers']
+  s.files = `git ls-files`.split("\n").reject { |f| File.symlink?(f) } + ['lexers']
 end