perf-subsample.hs

#!/usr/bin/env stack
-- stack --resolver lts-13.29 script --package regex-tdfa,conduit,bytestring,random-fu

-- Given a perf 'script', as emitted by 'perf script -i perf.data', subsample
-- at 0.1 probability to make the script small enough to load into e.g.
-- speedscope.

{-# LANGUAGE OverloadedStrings, ScopedTypeVariables #-}

import Conduit
import Data.ByteString (ByteString)
import Data.Random
import Data.Random.Distribution
import Data.Random.Distribution.Bernoulli

main = runConduit $ stdinC .| linesUnboundedAsciiC .| splitSamples .| unlinesAsciiC .| stdoutC

splitSamples :: ConduitT ByteString ByteString IO ()
splitSamples = do
  done <- nullC
  if not done
    then do
    take :: Bool <- lift $ sample $ bernoulli (0.1 :: Float)
    if take
    then do
      takeWhileC (/= "")
      takeC 1
      splitSamples
    else do
      dropWhileC (/= "")
      dropC 1
      splitSamples
  else pure ()