#!/usr/bin/python3
import os,sys,re,functools

def compute_failure(s):
  it = s['current']['interval']
  if it not in s['failure']: s['failure'][it] = []
  s['failure'][it].append(round(s['current']['max_pkt'] * s['current']['interval'] / 1000 / 60))

def compute_circuit_distri(s):
  s['current']['lats'] = l = sorted(s['current']['lats'])
  s['current']['distri'] = d = {}
  perc = lambda p: l[round(p * (len(l) - 1))]
  for x in [0, 0.25, 0.5, 0.75, 0.99, 0.999, 1]:
    d[str(x)] = perc(x)

def compute_interval_distri(s):
  for inter, val in s['per_interval']:
    s['per_interval_res'][inter] = {}
    s['per_interval_res'][inter]['agg'] = sorted([].extend([ elem['lats'] for elem in val]))

def extract_measlat(log, s):
  s['current']['max_pkt'] = 0
  s['current']['lats'] = []
  with open(log) as f:
    for l in f:
      x = re.search(r'Packet (\d+) latency (\d+)µs with', l)
      if x:
        lat = x.groups()[0]
        s['current']['max_pkt'] = max(s['current']['max_pkt'], int(lat))
        s['current']['lats'].append(lat)

def extract_info(inf, s):
  with open(inf) as f:
    full = ''.join(f.readlines())
    x = re.search(r'orig-server (\d+) (\d+) \d+', full)
    s['current']['npkt'] = int(x.groups()[0])
    s['current']['interval'] = int(x.groups()[1])

def extract_folder(p, s):
  extract_info(p + '/info.txt', s)
  extract_measlat(p + '/log/client-measlat-stdout.log', s)

  compute_failure(s)
  compute_circuit_distri(s) if s['current']['interval'] == 40

def categorize(folder, s):
  s[folder] = s['current']

  i = s['current']['interval']
  if i not in s['per_interval']: s['per_interval'][i] = []
  s['per_interval'][i].append(current)

def extract(p, s):
  item_count = functools.reduce(lambda acc, prev: acc + 1, os.listdir(p), 0)

  counter = 0
  print("extracting...")
  for folder in os.listdir(p):
    s['current'] = {}
    extract_folder(p + '/' + folder, s)
    categorize(folder, s)
    counter += 1
    progress = round(counter / item_count * 100)
    print(f"{progress}%", end="\r")
  print("done")

def compute_global(s):
  print("computing global...")
  compute_interval_distri(s)

def analyze_failure(s):
  with open('jan_failure.csv', 'w') as f:
    f.write(f"rate,duration,ecdf\n")
    for k, v in s['failure'].items():
      v = sorted(v)
      total = len(v)
      rate = round(1000 / k)
      score = 0
      f.write(f"{rate},0,0\n")
      for idx,e in enumerate(v,start=1):
        if e >= 90: 
          f.write(f"{rate},90,{score}\n")
          break
        score = idx/total
        f.write(f"{rate},{e},{score}\n")

def analyze(s):
  print("analyzing...")
  analyze_failure(s)

  state = {'failure': {}, 'per_interval': {}, 'per_interval_res': {}}
extract(sys.argv[1], state)
compute_global(state)
analyze(state)

#for key, value in state.items():
#  print(value)