#!/usr/bin/env python3
import sys, re, os, glob

# Parse mimalloc-bench logs and print CSV: file,allocator,procs,test,throughput
# Assumes lines like: "Throughput =   1234567 operations per second, ..."

def infer_allocator_from_file(path):
    fn = os.path.basename(path)
    if fn.startswith('hakmem_'): return 'hakmem'
    if fn.startswith('mimalloc_'): return 'mimalloc'
    if fn.startswith('jemalloc_'): return 'jemalloc'
    if fn.startswith('system_'): return 'system'
    # fallback: try substring
    for k in ('hakmem','mimalloc','jemalloc','system'):
        if k in fn: return k
    return 'unknown'

def infer_procs_from_file(path):
    m = re.search(r'procs=([0-9,]+)', os.path.basename(path))
    return m.group(1) if m else ''

def parse_file(path):
    alloc = infer_allocator_from_file(path)
    procs = infer_procs_from_file(path)
    test = ''
    with open(path,'r',errors='ignore') as f:
        for line in f:
            tl = line.strip()
            m = re.search(r'^Test\s*:\s*(\S+)', tl)
            if m: test = m.group(1)
            m2 = re.search(r'Throughput\s*=\s*([0-9]+)\s+operations per second', tl)
            if m2:
                thr = int(m2.group(1))
                yield (path, alloc, procs, test, thr)

def main():
    if len(sys.argv) != 2:
        print(f"usage: {sys.argv[0]} <log_dir>")
        sys.exit(1)
    logdir = sys.argv[1]
    files = sorted(glob.glob(os.path.join(logdir,'*.log')))
    print('file,allocator,procs,test,throughput_ops_per_sec')
    for fp in files:
        for rec in parse_file(fp):
            print(','.join(str(x) for x in rec))

if __name__ == '__main__':
    main()