#!/usr/bin/env python2

import csv
import datetime
import locale
import os.path
import re
import sys

# For strptime.
locale.setlocale(locale.LC_ALL, "C")

# Dec 01 20:57:53.000 [notice] Bootstrapped 0%: Starting
bootstrapped_re = re.compile(r'^(\w+ \d+ \d\d:\d\d:\d\d\.\d\d\d) \[\w+\] Bootstrapped (\d+)%')

csvW = csv.DictWriter(sys.stdout, fieldnames=("timestamp", "site", "runid", "nickname", "percent"))
csvW.writeheader()

def process_log(f, site, runid, nickname):
    for line in f:
        m = bootstrapped_re.match(line)
        if m is not None:

            timestamp = datetime.datetime.strptime(m.group(1), "%b %d %H:%M:%S.%f")
            # tor logs don't contain the year, so grab it from the runid.
            timestamp = timestamp.replace(year=int(runid[:4]))
            percent = m.group(2)

            row = {
                "timestamp": timestamp.strftime("%Y-%m-%d %H:%M:%S.%f"),
                "site": site,
                "runid": runid,
                "nickname": nickname,
                "percent": percent,
            }
            csvW.writerow(row)

for filename in sys.stdin:
    filename = filename.strip()

    nickname, ext = os.path.splitext(os.path.basename(filename))

    if ext != ".log":
        continue
    if nickname == "main":
        continue

    parent = os.path.dirname(filename)
    runid = os.path.basename(parent)
    parent = os.path.dirname(parent)
    site = os.path.basename(parent)

    with open(filename) as f:
        process_log(f, site, runid, nickname)