Process `ghc -v` output, summarising where time is spent.

example-output.txt

# ./ghc-performance-categorizer.py ghci-v-output-1.txt --sort-by time

TASK                            time ms   mem MB
------------------------------------------------
systool:cpp                           4        2
systool:merge-objects                11        3
Simplify                             38       68
ByteCodeGen                          70      164
initializing                         72      119
systool:as                          175       68
Chasing                             705     1453
Parser                              775     1964
systool:linker                      812      357
CoreTidy                           1149     1405
systool:cc                         1489       58
CorePrep                           1821     2203
Simplifier                         8040    11801
Renamer/typechecker               20184    26724
CodeGen                           43917    81220
Desugar                           58124   117565
------------------------------------------------
TOTAL                            137387   245175

ghc-performance-categorizer.py

#! /usr/bin/env python3

import argparse
import fileinput
import re

from collections import defaultdict
from operator import itemgetter


def main():
  parser = argparse.ArgumentParser(description='Process `ghc -v` output, summarising where time is spent.')
  parser.add_argument('files', metavar='FILE', nargs='*', help='files to read, if empty, stdin is used')
  parser.add_argument('--sort-by', dest='sort_by', choices=['time', 'memory'], help='Sort table by this column')

  args = parser.parse_args()

  # Example lines:
  #     !!! ByteCodeGen [Ghci1]: finished in 0.02 milliseconds, allocated 0.037 megabytes
  #     !!! Desugar [Test.MallocInfo]: finished in 1.46 milliseconds, allocated 3.290 megabytes
  #     !!! Simplify [expr]: finished in 0.72 milliseconds, allocated 2.717 megabytes
  #     !!! systool:as: finished in 0.37 milliseconds, allocated 0.219 megabytes
  r = re.compile(r'!!! (?P<task>\S+).* finished in (?P<time_milliseconds>\S+) milliseconds, allocated (?P<memory_megabytes>\S+) megabytes')
  times = defaultdict(float)
  memories = defaultdict(float)
  # If you would call fileinput.input() without files it would try to process all arguments.
  # We pass '-' as only file when argparse got no files which will cause fileinput to read from stdin
  # From: https://gist.github.com/martinth/ed991fb8cdcac3dfadf7
  for line in fileinput.input(files=args.files if len(args.files) > 0 else ('-', )):
    l = line.strip('\n')
    res = r.match(l)
    if res is not None:
      d = res.groupdict()

      task = d['task'].rstrip(':')
      time_milliseconds = float(d['time_milliseconds'])
      memory_megabytes = float(d['memory_megabytes'])

      times[task] += time_milliseconds
      memories[task] += memory_megabytes

  total_time = sum([t for _, t in times.items()])
  total_mem = sum([t for _, t in memories.items()])

  tasks = times.keys()
  if args.sort_by == 'time':
    tasks = sorted(times.keys(), key=lambda t: times[t])
  elif args.sort_by == 'memory':
    tasks = sorted(times.keys(), key=lambda t: memories[t])

  def format_line(task: str, time_milliseconds: float, memory_megabytes: float):
    time = f"{time_milliseconds:4.0f}"
    mem = f"{memory_megabytes:4.0f}"
    return f"{task.ljust(30)} {time.rjust(8)} {mem.rjust(8)}"

  print("TASK                            time ms   mem MB")
  print("------------------------------------------------")
  for task in tasks:
    print(format_line(task, times[task], memories[task]))
  print("------------------------------------------------")
  print(format_line("TOTAL", total_time, total_mem))


if __name__ == '__main__':
  main()