pancakecounter/table.py

#!/usr/bin/env python3

import sys
import re
from datetime import datetime
from collections import defaultdict
from functools import partial

if sys.stdin.isatty():
	print("Hint: stdin is a terminal, you may want to do `./make_table.py < verifications.log` instead.", file=sys.stderr)

users = {}

for line_nr, line in enumerate(sys.stdin, start=1):
	line = line.rstrip()

	m = re.fullmatch(r"([^ ]+) ([^ ]+) at ([^ ]+) verified by ([^ ]+) at ([^ ]+)", line)
	if m:
		awardee = m.group(1)
		post_id = m.group(2)
		post_time = datetime.fromisoformat(m.group(3))
		verifier = m.group(4)
		verification_time = datetime.fromisoformat(m.group(5))

		if awardee not in users:
			users[awardee] = {}
		if post_id not in users[awardee]:
			users[awardee][post_id] = defaultdict(dict)
		users[awardee][post_id][verifier] = verification_time
		continue

	m = re.fullmatch(r"([^ ]+) ([^ ]+) verification removed by ([^ ]+)", line)
	if m:
		awardee = m.group(1)
		post_id = m.group(2)
		verifier = m.group(3)

		try:
			del users[awardee][post_id][verifier]
		except KeyError:
			print(f"Line {line_nr}: trying to remove non-existing verification by {verifier}, looks like the file is corrupt!", file=sys.stderr)

		# If no more verifiers left, clean up the dicts, this makes it easier to count
		if not users[awardee][post_id]:
			del users[awardee][post_id]
		if not users[awardee]:
			del users[awardee]

		continue

	print(f"Couldn't parse line {line_nr}, looks like the file is corrupt: {line}", file=sys.stderr)

for username, user_posts in users.items():
	print(f"{username}: {len(user_posts)}")