module SemanticLogger

Return a logger for the supplied class or class_name

# File lib/semantic_logger/semantic_logger.rb, line 9
def self.[](klass)
  Logger.new(klass)
end

Add a new logging appender as a new destination for all log messages emitted from Semantic Logger

Appenders will be written to in the order that they are added

If a block is supplied then it will be used to customize the format of the messages sent to that appender. See SemanticLogger::Logger.new for more information on custom formatters

Parameters

file_name: [String]
  File name to write log messages to.

Or,
io: [IO]
  An IO Stream to log to.
  For example $stdout, $stderr, etc.

Or,
appender: [Symbol|SemanticLogger::Subscriber]
  A symbol identifying the appender to create.
  For example:
    :bugsnag, :elasticsearch, :graylog, :http, :mongodb, :new_relic, :splunk_http, :syslog, :wrapper
       Or,
  An instance of an appender derived from SemanticLogger::Subscriber
  For example:
    SemanticLogger::Appender::Http.new(url: 'http://localhost:8088/path')

Or,
logger: [Logger|Log4r]
  An instance of a Logger or a Log4r logger.

level: [:trace | :debug | :info | :warn | :error | :fatal]
  Override the log level for this appender.
  Default: SemanticLogger.default_level

formatter: [Symbol|Object|Proc]
  Any of the following symbol values: :default, :color, :json
    Or,
  An instance of a class that implements #call
    Or,
  A Proc to be used to format the output from this appender
  Default: :default

filter: [Regexp|Proc]
  RegExp: Only include log messages where the class name matches the supplied.
  regular expression. All other messages will be ignored.
  Proc: Only include log messages where the supplied Proc returns true
        The Proc must return true or false.

Examples:

# Send all logging output to Standard Out (Screen)
SemanticLogger.add_appender(io: $stdout)

# Send all logging output to a file
SemanticLogger.add_appender(file_name: 'logfile.log')

# Send all logging output to a file and only :info and above to standard output
SemanticLogger.add_appender(file_name: 'logfile.log')
SemanticLogger.add_appender(io: $stdout, level: :info)

Log to log4r, Logger, etc.:

# Send Semantic logging output to an existing logger
require 'logger'
require 'semantic_logger'

# Built-in Ruby logger
log = Logger.new($stdout)
log.level = Logger::DEBUG

SemanticLogger.default_level = :debug
SemanticLogger.add_appender(logger: log)

logger = SemanticLogger['Example']
logger.info "Hello World"
logger.debug("Login time", user: 'Joe', duration: 100, ip_address: '127.0.0.1')

# File lib/semantic_logger/semantic_logger.rb, line 166
def self.add_appender(**args, &block)
  appender = Logger.processor.appenders.add(**args, &block)
  # Start appender thread if it is not already running
  Logger.processor.start
  appender
end

Add signal handlers for Semantic Logger

Two signal handlers will be registered by default:

1. Changing the log_level:

The log level can be changed without restarting the process by sending the
log_level_signal, which by default is 'USR2'

When the log_level_signal is raised on this process, the global default log level
rotates through the following log levels in the following order, starting
from the current global default level:
  :warn, :info, :debug, :trace

If the current level is :trace it wraps around back to :warn

1. Logging a Ruby thread dump

When the signal is raised on this process, Semantic Logger will write the list
of threads to the log file, along with their back-traces when available

For JRuby users this thread dump differs form the standard QUIT triggered
Java thread dump which includes system threads and Java stack traces.

It is recommended to name any threads you create in the application, by
calling the following from within the thread itself:
   Thread.current.name = 'My Worker'

Also adds JRuby Garbage collection logging so that any garbage collections that exceed the time threshold will be logged. Default: 100 ms Currently only supported when running JRuby

Note:

To only register one of the signal handlers, set the other to nil
Set gc_log_microseconds to nil to not enable JRuby Garbage collections

# File lib/semantic_logger/semantic_logger.rb, line 279
def self.add_signal_handler(log_level_signal = "USR2", thread_dump_signal = "TTIN", gc_log_microseconds = 100_000)
  if log_level_signal
    Signal.trap(log_level_signal) do
      index     = default_level == :trace ? LEVELS.find_index(:error) : LEVELS.find_index(default_level)
      new_level = LEVELS[index - 1]
      self["SemanticLogger"].warn "Changed global default log level to #{new_level.inspect}"
      self.default_level = new_level
    end
  end

  if thread_dump_signal
    Signal.trap(thread_dump_signal) do
      logger = SemanticLogger["Thread Dump"]
      Thread.list.each do |thread|
        # MRI re-uses the main thread for signals, JRuby uses `SIGTTIN handler` thread.
        next if defined?(JRuby) && (thread == Thread.current)

        logger.backtrace(thread: thread)
      end
    end
  end

  if gc_log_microseconds && defined?(JRuby)
    listener = SemanticLogger::JRuby::GarbageCollectionLogger.new(gc_log_microseconds)
    Java::JavaLangManagement::ManagementFactory.getGarbageCollectorMXBeans.each do |gcbean|
      gcbean.add_notification_listener(listener, nil, nil)
    end
  end

  true
end

Returns [SemanticLogger::Subscriber] a copy of the list of active appenders for debugging etc. Use SemanticLogger.add_appender and SemanticLogger.remove_appender to manipulate the active appenders list

# File lib/semantic_logger/semantic_logger.rb, line 191
def self.appenders
  Logger.processor.appenders.to_a
end

Returns [String] name of this application for logging purposes Note: Not all appenders use `application`

# File lib/semantic_logger/semantic_logger.rb, line 65
def self.application
  @application
end

Override the default application

# File lib/semantic_logger/semantic_logger.rb, line 70
def self.application=(application)
  @application = application
end

Returns the current backtrace level

# File lib/semantic_logger/semantic_logger.rb, line 42
def self.backtrace_level
  @backtrace_level
end

Sets the level at which backtraces should be captured for every log message.

By enabling backtrace capture the filename and line number of where message was logged can be written to the log file. Additionally, the backtrace can be forwarded to error management services such as Bugsnag.

Warning:

Capturing backtraces is very expensive and should not be done all
the time. It is recommended to run it at :error level in production.

# File lib/semantic_logger/semantic_logger.rb, line 35
def self.backtrace_level=(level)
  @backtrace_level = level
  # For performance reasons pre-calculate the level index
  @backtrace_level_index = level.nil? ? 65_535 : Levels.index(level)
end

Returns the current backtrace level index For internal use only

# File lib/semantic_logger/semantic_logger.rb, line 48
def self.backtrace_level_index
  @backtrace_level_index
end

Clear out all previously registered appenders

# File lib/semantic_logger/semantic_logger.rb, line 183
def self.clear_appenders!
  Logger.processor.close
end

Close all appenders and flush any outstanding messages.

# File lib/semantic_logger/semantic_logger.rb, line 202
def self.close
  Logger.processor.close
end

Returns the global default log level

# File lib/semantic_logger/semantic_logger.rb, line 21
def self.default_level
  @default_level
end

Sets the global default log level

# File lib/semantic_logger/semantic_logger.rb, line 14
def self.default_level=(level)
  @default_level = level
  # For performance reasons pre-calculate the level index
  @default_level_index = Levels.index(level)
end

Returns [String] name of this environment for logging purposes Note: Not all appenders use `environment`

# File lib/semantic_logger/semantic_logger.rb, line 76
def self.environment
  @environment
end

Override the default environment

# File lib/semantic_logger/semantic_logger.rb, line 81
def self.environment=(environment)
  @environment = environment
end

If the tag being supplied is definitely a string then this fast tag api can be used for short lived tags

# File lib/semantic_logger/semantic_logger.rb, line 313
def self.fast_tag(tag)
  return yield if tag.nil? || tag == ""

  t = Thread.current[:semantic_logger_tags] ||= []
  begin
    t << tag
    yield
  ensure
    t.pop
  end
end

Flush all queued log entries disk, database, etc.

All queued log messages are written and then each appender is flushed in turn.

# File lib/semantic_logger/semantic_logger.rb, line 197
def self.flush
  Logger.processor.flush
end

Returns [String] name of this host for logging purposes Note: Not all appenders use `host`

# File lib/semantic_logger/semantic_logger.rb, line 54
def self.host
  @host ||= Socket.gethostname.force_encoding("UTF-8")
end

Override the default host name

# File lib/semantic_logger/semantic_logger.rb, line 59
def self.host=(host)
  @host = host
end

Returns the check_interval which is the number of messages between checks to determine if the appender thread is falling behind.

# File lib/semantic_logger/semantic_logger.rb, line 480
def self.lag_check_interval
  Logger.processor.lag_check_interval
end

Set the check_interval which is the number of messages between checks to determine if the appender thread is falling behind.

# File lib/semantic_logger/semantic_logger.rb, line 486
def self.lag_check_interval=(lag_check_interval)
  Logger.processor.lag_check_interval = lag_check_interval
end

Returns the amount of time in seconds to determine if the appender thread is falling behind.

# File lib/semantic_logger/semantic_logger.rb, line 492
def self.lag_threshold_s
  Logger.processor.lag_threshold_s
end

:nodoc

# File lib/semantic_logger/semantic_logger.rb, line 389
def self.named_tagged(hash)
  return yield if hash.nil? || hash.empty?
  raise(ArgumentError, "#named_tagged only accepts named parameters (Hash)") unless hash.is_a?(Hash)

  begin
    push_named_tags(hash)
    yield
  ensure
    pop_named_tags
  end
end

Returns [Hash] a copy of the named tags currently active for this thread.

# File lib/semantic_logger/semantic_logger.rb, line 402
def self.named_tags
  if (list = Thread.current[:semantic_logger_named_tags]) && !list.empty?
    if list.size > 1
      list.reduce({}) { |sum, h| sum.merge(h) }
    else
      list.first.clone
    end
  else
    {}
  end
end

Supply a callback to be called whenever a log entry is created. Useful for capturing appender specific context information.

Parameters
  object: [Object | Proc]
    [Proc] the block to call.
    [Object] any object on which to call #call.

Example:

SemanticLogger.on_log do |log|
  log.set_context(:honeybadger, Honeybadger.get_context)
end

Example:

module CaptureContext
  def call(log)
    log.set_context(:honeybadger, Honeybadger.get_context)
  end
end
SemanticLogger.on_log(CaptureContext)

Note:

• This callback is called within the thread of the application making the logging call.

• If these callbacks are slow they will slow down the application.

# File lib/semantic_logger/semantic_logger.rb, line 240
def self.on_log(object = nil, &block)
  Logger.subscribe(object, &block)
end

Remove specified number of tags from the current tag list

# File lib/semantic_logger/semantic_logger.rb, line 383
def self.pop_tags(quantity = 1)
  t = Thread.current[:semantic_logger_tags]
  t&.pop(quantity)
end

Add tags to the current scope

Note:

• This method does not flatten the array or remove any empty elements, or duplicates since the performance penalty is excessive.

• To get the flattening behavior use the slower api:

  `logger.push_tags`
  

# File lib/semantic_logger/semantic_logger.rb, line 377
def self.push_tags(*tags)
  (Thread.current[:semantic_logger_tags] ||= []).concat(tags)
  tags
end

Returns [Integer] the number of log entries waiting to be written to the appenders.

When this number grows it is because the logging appender thread is not able to write to the appenders fast enough. Either reduce the amount of logging, increase the log level, reduce the number of appenders, or look into speeding up the appenders themselves

# File lib/semantic_logger/semantic_logger.rb, line 474
def self.queue_size
  Logger.processor.queue.size
end

Remove an existing appender Currently only supports appender instances

# File lib/semantic_logger/semantic_logger.rb, line 175
def self.remove_appender(appender)
  return unless appender

  Logger.processor.appenders.delete(appender)
  appender.close
end

After forking an active process call SemanticLogger.reopen to re-open any open file handles etc to resources.

Note:

Not all appender's implement reopen.
Check the code for each appender you are using before relying on this behavior.

# File lib/semantic_logger/semantic_logger.rb, line 212
def self.reopen
  Logger.processor.reopen
end

Silence noisy log levels by changing the default_level within the block

This setting is thread-safe and only applies to the current thread

Any threads spawned within the block will not be affected by this setting

silence can be used to both raise and lower the log level within the supplied block.

Example:

# Perform trace level logging within the block when the default is higher
SemanticLogger.default_level = :info

logger.debug 'this will _not_ be logged'

SemanticLogger.silence(:trace) do
  logger.debug "this will be logged"
end

Parameters

new_level
  The new log level to apply within the block
  Default: :error

Example:

# Silence all logging for this thread below :error level
SemanticLogger.silence do
  logger.info "this will _not_ be logged"
  logger.warn "this neither"
  logger.error "but errors will be logged"
end

Note:

#silence does not affect any loggers which have had their log level set
explicitly. I.e. That do not rely on the global default level

# File lib/semantic_logger/semantic_logger.rb, line 460
def self.silence(new_level = :error)
  current_index                            = Thread.current[:semantic_logger_silence]
  Thread.current[:semantic_logger_silence] = Levels.index(new_level)
  yield
ensure
  Thread.current[:semantic_logger_silence] = current_index
end

Run Semantic Logger in Synchronous mode.

I.e. Instead of logging messages in a separate thread for better performance, log them using the current thread.

# File lib/semantic_logger/semantic_logger.rb, line 504
def self.sync!
  Logger.sync!
end

Running in synchronous mode?

# File lib/semantic_logger/semantic_logger.rb, line 509
def self.sync?
  Logger.sync?
end

Add the tags or named tags to the list of tags to log for this thread whilst the supplied block is active.

Returns result of block.

Tagged example:

SemanticLogger.tagged(12345, 'jack') do
  logger.debug('Hello World')
end

Named Tags (Hash) example:

SemanticLogger.tagged(tracking_number: 12345) do
  logger.debug('Hello World')
end

Notes:

• Tags should be a list without any empty values, or contain any array.

  • `logger.tagged` is a slower api that will flatten the example below: `logger.tagged([['first', nil], nil, ['more'], 'other'])`

  to the equivalent of:

  `logger.tagged('first', 'more', 'other')`
  

# File lib/semantic_logger/semantic_logger.rb, line 345
def self.tagged(*tags, &block)
  return yield if tags.empty?

  # Allow named tags to be passed into the logger
  if tags.size == 1
    tag = tags[0]
    return tag.is_a?(Hash) ? named_tagged(tag, &block) : fast_tag(tag, &block)
  end

  begin
    push_tags(*tags)
    yield
  ensure
    pop_tags(tags.size)
  end
end

Returns a copy of the [Array] of [String] tags currently active for this thread Returns nil if no tags are set

# File lib/semantic_logger/semantic_logger.rb, line 364
def self.tags
  # Since tags are stored on a per thread basis this list is thread-safe
  t = Thread.current[:semantic_logger_tags]
  t.nil? ? [] : t.clone
end