@@ -1,6 +1,7 @@ === 0.0.5 / 2009-11-08 * API CHANGE: Reworked how string encoding is specified. Default encoding for both reading and writing is now ASCII-8BIT unless specified in the API or within the file as a parser directive. +* Handle encoding errors more gracefully by replacing characters rather than raising exceptions. * Added a parser directive to set logfile encoding. * Versionomy is no longer a hard dependency-- it is now used only if available. History.rdoc @@ -267,6 +267,8 @@ module Sawmill # Specify an encoding for file data. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. # If not specified, reads raw bytes (e.g. defaults to 'ASCII-8BIT'). + # Note that the encoding may also be modified by the file itself, + # if an appropriate parser directive is encountered. # <tt>:internal_encoding</tt>:: # Specify an encoding to transcode to. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. @@ -298,6 +300,8 @@ module Sawmill # Specify an encoding for file data. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. # If not specified, reads raw bytes (e.g. defaults to 'ASCII-8BIT'). + # Note that the encoding may also be modified by the file itself, + # if an appropriate parser directive is encountered. # <tt>:internal_encoding</tt>:: # Specify an encoding to transcode to. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. @@ -332,6 +336,8 @@ module Sawmill # Specify an encoding for file data. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. # If not specified, reads raw bytes (e.g. defaults to 'ASCII-8BIT'). + # Note that the encoding may also be modified by the file itself, + # if an appropriate parser directive is encountered. # <tt>:internal_encoding</tt>:: # Specify an encoding to transcode to. (Ruby 1.9 only.) # You may specify an encoding name or an encoding object. @@ -353,10 +359,7 @@ module Sawmill end if encoding_ mode_ << ":#{encoding_.name}" - elsif internal_encoding_ - mode_ << ":#{::Encoding.default_external.name}" end - mode_ << ":#{internal_encoding_.name}" if internal_encoding_ else encoding_ = nil internal_encoding_ = nil @@ -376,7 +379,7 @@ module Sawmill else io_array_ << ::File.open(path_, mode_) encoding_array_ << nil - internal_encoding_array_ << nil + internal_encoding_array_ << internal_encoding_ end end end lib/sawmill/interface.rb @@ -47,6 +47,7 @@ module Sawmill DIRECTIVE_REGEXP = /^#\s+sawmill_format:\s+(\w+)=(.*)$/ ATTRIBUTE_REGEXP = /^([[:graph:]]+)\s([=+\/-])\s/ SUPPORTS_ENCODING = defined?(::Encoding) + ENCODING_OPTS = {:invalid => :replace, :undef => :replace} # :startdoc: @@ -197,7 +198,7 @@ module Sawmill str_ = @io.gets if str_ && SUPPORTS_ENCODING str_.force_encoding(@encoding) if @encoding - str_.encode!(@internal_encoding) if @internal_encoding + str_.encode!(@internal_encoding, ENCODING_OPTS) if @internal_encoding end str_ end lib/sawmill/parser.rb @@ -59,6 +59,11 @@ module Sawmill class Rotater + # :stopdoc: + SUPPORTS_ENCODING = defined?(::Encoding) + ENCODING_OPTS = {:invalid => :replace, :undef => :replace} + # :startdoc: + # Create a rotater using the given rotation strategy. # See Sawmill::Rotater::DateBasedLogFile and @@ -70,8 +75,26 @@ module Sawmill # # <tt>:omit_directives</tt>:: # If true, omit standard logfile directives. Default is false. + # <tt>:concurrent_writes</tt>:: + # Set this to true if you expect multiple processes to attempt to + # write to the same log file simultaneously. This option causes the + # rotater to surround writes with an acquisition of the cooperative + # filesystem lock (if available) for the logfile, in an attempt to + # prevent lines from interleaving in one another. Default is false. + # <tt>:encoding</tt>:: + # Specify an encoding for file data. (Ruby 1.9 only). + # You may pass either an encoding object or an encoding name. + # If not specified, writes raw bytes (e.g. defaults to ASCII-8BIT). def initialize(io_manager_, opts_={}) + @omit_directives = opts_.delete(:omit_directives) + @concurrent_writes = opts_.delete(:concurrent_writes) + if SUPPORTS_ENCODING + @encoding = opts_.delete(:encoding) + if @encoding && !@encoding.respond_to?(:name) + @encoding = ::Encoding.find(@encoding) + end + end if io_manager_.kind_of?(::Class) @io_manager = io_manager_.new(opts_) else @@ -98,6 +121,23 @@ module Sawmill end + def _write_to_stream(io_, str_) + if SUPPORTS_ENCODING && @encoding + str_ = str_.encode(@encoding, ENCODING_OPTS) + end + if @concurrent_writes + begin + io_.flock(::File::LOCK_EX) + io_.write(str_) + ensure + io_.flock(::File::LOCK_UN) + end + else + io_.write(str_) + end + end + + def _obtain_handle # :nodoc: handle_ = @io_manager.preferred_handle if @handles.include?(handle_) @@ -105,11 +145,11 @@ module Sawmill else io_ = @io_manager.open_handle(handle_) unless @omit_directives - io_.write("# sawmill_format: version=1\n") + _write_to_stream(io_, "# sawmill_format: version=1\n") if defined?(::Encoding) encoding_ = io_.encoding if encoding_ - io_.write("# sawmill_format: encoding=#{encoding_.name}\n") + _write_to_stream(io_, "# sawmill_format: encoding=#{encoding_.name}\n") end end end @@ -154,7 +194,7 @@ module Sawmill handle_ = _check_rotate_handle(handle_) end info_ = @handles[handle_] - info_[1].write(str_) + _write_to_stream(info_[1], str_) handle_ end end lib/sawmill/rotater.rb @@ -79,9 +79,6 @@ module Sawmill # <tt>:local_datestamps</tt>:: # If true, use the local timezone to create datestamps. # The default is to use UTC. - # <tt>:encoding</tt>:: - # Specify an encoding name for file data. (Ruby 1.9 only). - # If not specified, writes raw bytes (e.g. defaults to ASCII-8BIT). def initialize(options_) @turnover_frequency = options_[:turnover_frequency] || :none @@ -97,10 +94,6 @@ module Sawmill when :hourly then "%Y-%m-%d-%H" else nil end - @mode = 'a' - if defined?(::Encoding) && (encoding_ = options_[:encoding]) - @mode << ":#{encoding_}" - end end @@ -125,7 +118,7 @@ module Sawmill else path_ = @prefix+@suffix end - file_ = ::File.open(path_, @mode) + file_ = ::File.open(path_, 'a') file_.sync = true file_ end lib/sawmill/rotater/date_based_log_file.rb @@ -76,9 +76,6 @@ module Sawmill # The maximum number of old logfiles (files with indexes) to # keep. Files beyond this history size will be automatically # deleted. Default is 1. This value must be at least 1. - # <tt>:encoding</tt>:: - # Specify an encoding name for file data. (Ruby 1.9 only). - # If not specified, writes raw bytes (e.g. defaults to ASCII-8BIT). def initialize(options_) @max_logfile_size = options_[:max_file_size] || options_[:max_logfile_size] @@ -100,10 +97,6 @@ module Sawmill @preferred_handle = 0 @open_handles = {} @last_shift = ::Time.now - @mode = 'a' - if defined?(::Encoding) && (encoding_ = options_[:encoding]) - @mode << ":#{encoding_}" - end end @@ -122,7 +115,7 @@ module Sawmill else path_ = "#{@normal_path}.#{@preferred_handle-handle_-1}" end - file_ = ::File.open(path_, @mode) + file_ = ::File.open(path_, 'a') file_.sync = true @open_handles[handle_] = true file_ lib/sawmill/rotater/shifting_log_file.rb 906581908884d55f504890a2b8cdb47fc786abc5 Daniel Azuma dazuma@gmail.com http://github.com/dazuma/sawmill/commit/89b213963dc831bf72d4065d1cd9ec5a72172edf 89b213963dc831bf72d4065d1cd9ec5a72172edf 2009-11-08T17:18:48-08:00 2009-11-08T17:18:48-08:00 Some more encoding work 66022200f875dacf4d4f825c28fabb0666a346c5 Daniel Azuma dazuma@gmail.com