fluentd/lib/plugin/output.rbのextract_placeholders調査

_benchmark-result

Warming up --------------------------------------
    rewrite2 && full     9.576k i/100ms
     rewrite && full     9.081k i/100ms
        orig && full     8.088k i/100ms
Calculating -------------------------------------
    rewrite2 && full    100.151k (± 4.4%) i/s -    507.528k in   5.078509s
     rewrite && full     94.131k (± 5.0%) i/s -    472.212k in   5.030891s
        orig && full     83.549k (± 3.9%) i/s -    420.576k in   5.041819s

Comparison:
    rewrite2 && full:   100151.5 i/s
     rewrite && full:    94130.8 i/s - same-ish: difference falls within error
        orig && full:    83548.8 i/s - 1.20x  slower

Warming up --------------------------------------
  rewrite2 && simple    37.682k i/100ms
   rewrite && simple    35.668k i/100ms
      orig && simple    26.577k i/100ms
Calculating -------------------------------------
  rewrite2 && simple    423.057k (± 3.4%) i/s -      2.148M in   5.083244s
   rewrite && simple    405.711k (± 2.7%) i/s -      2.033M in   5.014899s
      orig && simple    296.869k (± 2.7%) i/s -      1.488M in   5.017274s

Comparison:
  rewrite2 && simple:   423056.7 i/s
   rewrite && simple:   405711.0 i/s - same-ish: difference falls within error
      orig && simple:   296868.6 i/s - 1.43x  slower

extract_placeholders.rb

require "benchmark/ips"

Metadata = Struct.new(:tag, :variables)
CHUNK_KEY_PLACEHOLDER_PATTERN = /\$\{[-_.@a-zA-Z0-9]+\}/
CHUNK_TAG_PLACEHOLDER_PATTERN = /\$\{(tag(?:\[\d+\])?)\}/

FULL_PATTERN = '/mypath/%Y/%m/%d/%H-%M/${tag}/${tag[1]}/${tag[2]}/${key1}/${key2}/tail'
SIMPLE_PATTERN = '/mypath'
SAMPLE_METADATA = Metadata.new('fluentd.test.output', {key1: 'value1', key2: 'value2'})

def rewrite2(str, metadata, chunk_keys, chunk_key_tag)
  rvalue = str

  # ${tag}, ${tag[0]}, ${tag[1]}, ...
  if chunk_key_tag
    if rvalue =~ CHUNK_TAG_PLACEHOLDER_PATTERN
      hash = {'${tag}' => metadata.tag}
      metadata.tag.split('.').each_with_index do |part, i|
        hash["${tag[#{i}]}"] = part
      end
      rvalue = rvalue.gsub(CHUNK_TAG_PLACEHOLDER_PATTERN, hash)
    end
  elsif rvalue =~ CHUNK_TAG_PLACEHOLDER_PATTERN
    log.warn "tag placeholder '#{$1}' not replaced. tag:#{metadata.tag}, template:#{str}"
  end

  # ${a_chunk_key}, ...
  if chunk_keys.empty?
    if rvalue =~ CHUNK_KEY_PLACEHOLDER_PATTERN
      log.warn "chunk key placeholder '#{$1}' not replaced. templace:#{str}"
    end
  elsif metadata.variables
    hash = {'${tag}' => '${tag}'}
    chunk_keys.each do |key|
      hash["${#{key}}"] = metadata.variables[key.to_sym]
    end
    rvalue = rvalue.gsub(CHUNK_KEY_PLACEHOLDER_PATTERN, hash)
  end
  rvalue
end

def rewrite1(str, metadata, chunk_keys, chunk_key_tag)
  rvalue = str

  # ${tag}, ${tag[0]}, ${tag[1]}, ...
  if chunk_key_tag
    if rvalue =~ CHUNK_TAG_PLACEHOLDER_PATTERN
      rvalue = rvalue.gsub('${tag}', metadata.tag)
      metadata.tag.split('.').each_with_index do |part, i|
        rvalue = rvalue.gsub("${tag[#{i}]}", part)
      end
      rvalue = rvalue.gsub(CHUNK_TAG_PLACEHOLDER_PATTERN, '')
    end
  elsif rvalue =~ CHUNK_TAG_PLACEHOLDER_PATTERN
    log.warn "tag placeholder '#{$1}' not replaced. tag:#{metadata.tag}, template:#{str}"
  end

  # ${a_chunk_key}, ...
  if chunk_keys.empty?
    if rvalue =~ CHUNK_KEY_PLACEHOLDER_PATTERN
      log.warn "chunk key placeholder '#{$1}' not replaced. templace:#{str}"
    end
  elsif metadata.variables
    chunk_keys.each do |key|
      rvalue = rvalue.gsub("${#{key}}", metadata.variables[key.to_sym])
    end
    rvalue = rvalue.gsub(CHUNK_KEY_PLACEHOLDER_PATTERN, {'${tag}' => '${tag}'})
  end
  rvalue
end

def original(str, metadata, chunk_keys, chunk_key_tag)
  rvalue = str.dup

  # ${tag}, ${tag[0]}, ${tag[1]}, ...
  if chunk_key_tag
    if str.include?('${tag}')
      rvalue = rvalue.gsub('${tag}', metadata.tag)
    end
    if str =~ CHUNK_TAG_PLACEHOLDER_PATTERN
      hash = {}
      metadata.tag.split('.').each_with_index do |part, i|
        hash["${tag[#{i}]}"] = part
      end
      rvalue = rvalue.gsub(CHUNK_TAG_PLACEHOLDER_PATTERN, hash)
    end
    if rvalue =~ CHUNK_TAG_PLACEHOLDER_PATTERN
      log.warn "tag placeholder '#{$1}' not replaced. tag:#{metadata.tag}, template:#{str}"
    end
  end

  # ${a_chunk_key}, ...
  if !chunk_keys.empty? && metadata.variables
    hash = {'${tag}' => '${tag}'} # not to erase this wrongly
    chunk_keys.each do |key|
      hash["${#{key}}"] = metadata.variables[key.to_sym]
    end
    rvalue = rvalue.gsub(CHUNK_KEY_PLACEHOLDER_PATTERN, hash)
  end
  if rvalue =~ CHUNK_KEY_PLACEHOLDER_PATTERN
    log.warn "chunk key placeholder '#{$1}' not replaced. templace:#{str}"
  end

  rvalue
end

metadata = SAMPLE_METADATA
chunk_keys = [:key1, :key2]
chunk_key_tag = true

Benchmark.ips do |x|
  x.report("rewrite2 && full") { rewrite2(FULL_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.report("rewrite && full") { rewrite1(FULL_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.report("orig && full") { original(FULL_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.compare!
end

Benchmark.ips do |x|
  x.report("rewrite2 && simple") { rewrite2(SIMPLE_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.report("rewrite && simple") { rewrite1(SIMPLE_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.report("orig && simple") { original(SIMPLE_PATTERN, metadata, chunk_keys, chunk_key_tag) }
  x.compare!
end