#!/usr/bin/env python3
import glob, sys, os, re
import numpy as np
from datetime import datetime, timedelta

if len(sys.argv) == 1:
    print("GPT job statistics expects job directories as arguments")
    sys.exit(1)

skip_level = 0
start_datetime = None
end_datetime = None
i=1
while i < len(sys.argv):
    if sys.argv[i] == "-i":
        skip_level=int(sys.argv[i+1])
        del sys.argv[i:i+2]
    elif sys.argv[i] == "-s":
        if sys.argv[i+1][0] == "-":
            start_datetime = datetime.now() - timedelta(days=int(sys.argv[i+1][1:]))
            print("Consider only events after", start_datetime)
        del sys.argv[i:i+2]
    elif sys.argv[i] == "-e":
        if sys.argv[i+1][0] == "-":
            end_datetime = datetime.now() - timedelta(days=int(sys.argv[i+1][1:]))
            print("Consider only events before", end_datetime)
        del sys.argv[i:i+2]
    else:
        i+=1

stats = {
    "ndir" : 0,
    "jdir" : 0,
    "fdir" : 0,
    "tags" : {}
}
def analyze_directory(d):
    global stats

    stats["ndir"] += 1
    
    if os.path.exists(f"{d}/.started"):
        stats["jdir"] += 1

        start = open(f"{d}/.started").read().split("\n")
        start_time = datetime.strptime(start[0], '%a %b %d %H:%M:%S %Y')
        if start_datetime is not None:
            if start_time < start_datetime:
                return
        
        if os.path.exists(f"{d}/.completed"):
            end = open(f"{d}/.completed").read().split("\n")
            end_time = datetime.strptime(end[0], '%a %b %d %H:%M:%S %Y')
            if end_datetime is not None:
                if end_time > end_datetime:
                    return
        
            dt = (end_time - start_time).total_seconds()

            stats["fdir"] += 1

            # remove all numbers from d and replace them by single #
            tag = re.sub(r'\d+', '#', d)
            tag = "/".join(tag.split("/")[skip_level:])
            if tag not in stats["tags"]:
                stats["tags"][tag] = [dt]
            else:
                stats["tags"][tag].append(dt)

    else:
        for sd in os.scandir(d):
            if sd.is_dir():
                analyze_directory(sd.path)

for d in sys.argv[1:]:
    analyze_directory(d)
        
print(f"Analyzed {stats['fdir']} out of {stats['jdir']} job directories out of {stats['ndir']} scanned directories:")
print("-" * 120)

for tag in stats["tags"]:
    dt = stats["tags"][tag]
    dt_min = f"{min(dt):g}"
    dt_max = f"{max(dt):g}"
    dt_mean = f"{np.mean(dt):g}"
    dt_median = f"{np.median(dt):g}"
    print(f"{tag:50s} : {dt_min:>10s} - {dt_max:>10s} s  :  mean={dt_mean:>10s}    median={dt_median:>10s}")