tor-browser

The Tor Browser
git clone https://git.dasho.dev/tor-browser.git
Log | Files | Refs | README | LICENSE

benchmark.py (4219B)

      1 #!/usr/bin/env python
      2 
      3 import getpass
      4 import json
      5 import pickle
      6 import subprocess
      7 import sys
      8 import time
      9 import zlib
     10 
     11 
     12 CORPUS_FILE = "corpus.pkl"
     13 
     14 REPEAT = 10
     15 
     16 WB, ML = 12, 5  # defaults used as a reference
     17 
     18 
     19 def _corpus():
     20    OAUTH_TOKEN = getpass.getpass("OAuth Token? ")
     21    COMMIT_API = (
     22        f'curl -H "Authorization: token {OAUTH_TOKEN}" '
     23        f"https://api.github.com/repos/python-websockets/websockets/git/commits/:sha"
     24    )
     25 
     26    commits = []
     27 
     28    head = subprocess.check_output("git rev-parse HEAD", shell=True).decode().strip()
     29    todo = [head]
     30    seen = set()
     31 
     32    while todo:
     33        sha = todo.pop(0)
     34        commit = subprocess.check_output(COMMIT_API.replace(":sha", sha), shell=True)
     35        commits.append(commit)
     36        seen.add(sha)
     37        for parent in json.loads(commit)["parents"]:
     38            sha = parent["sha"]
     39            if sha not in seen and sha not in todo:
     40                todo.append(sha)
     41        time.sleep(1)  # rate throttling
     42 
     43    return commits
     44 
     45 
     46 def corpus():
     47    data = _corpus()
     48    with open(CORPUS_FILE, "wb") as handle:
     49        pickle.dump(data, handle)
     50 
     51 
     52 def _run(data):
     53    size = {}
     54    duration = {}
     55 
     56    for wbits in range(9, 16):
     57        size[wbits] = {}
     58        duration[wbits] = {}
     59 
     60        for memLevel in range(1, 10):
     61            encoder = zlib.compressobj(wbits=-wbits, memLevel=memLevel)
     62            encoded = []
     63 
     64            t0 = time.perf_counter()
     65 
     66            for _ in range(REPEAT):
     67                for item in data:
     68                    if isinstance(item, str):
     69                        item = item.encode("utf-8")
     70                    # Taken from PerMessageDeflate.encode
     71                    item = encoder.compress(item) + encoder.flush(zlib.Z_SYNC_FLUSH)
     72                    if item.endswith(b"\x00\x00\xff\xff"):
     73                        item = item[:-4]
     74                    encoded.append(item)
     75 
     76            t1 = time.perf_counter()
     77 
     78            size[wbits][memLevel] = sum(len(item) for item in encoded)
     79            duration[wbits][memLevel] = (t1 - t0) / REPEAT
     80 
     81    raw_size = sum(len(item) for item in data)
     82 
     83    print("=" * 79)
     84    print("Compression ratio")
     85    print("=" * 79)
     86    print("\t".join(["wb \\ ml"] + [str(memLevel) for memLevel in range(1, 10)]))
     87    for wbits in range(9, 16):
     88        print(
     89            "\t".join(
     90                [str(wbits)]
     91                + [
     92                    f"{100 * (1 - size[wbits][memLevel] / raw_size):.1f}%"
     93                    for memLevel in range(1, 10)
     94                ]
     95            )
     96        )
     97    print("=" * 79)
     98    print()
     99 
    100    print("=" * 79)
    101    print("CPU time")
    102    print("=" * 79)
    103    print("\t".join(["wb \\ ml"] + [str(memLevel) for memLevel in range(1, 10)]))
    104    for wbits in range(9, 16):
    105        print(
    106            "\t".join(
    107                [str(wbits)]
    108                + [
    109                    f"{1000 * duration[wbits][memLevel]:.1f}ms"
    110                    for memLevel in range(1, 10)
    111                ]
    112            )
    113        )
    114    print("=" * 79)
    115    print()
    116 
    117    print("=" * 79)
    118    print(f"Size vs. {WB} \\ {ML}")
    119    print("=" * 79)
    120    print("\t".join(["wb \\ ml"] + [str(memLevel) for memLevel in range(1, 10)]))
    121    for wbits in range(9, 16):
    122        print(
    123            "\t".join(
    124                [str(wbits)]
    125                + [
    126                    f"{100 * (size[wbits][memLevel] / size[WB][ML] - 1):.1f}%"
    127                    for memLevel in range(1, 10)
    128                ]
    129            )
    130        )
    131    print("=" * 79)
    132    print()
    133 
    134    print("=" * 79)
    135    print(f"Time vs. {WB} \\ {ML}")
    136    print("=" * 79)
    137    print("\t".join(["wb \\ ml"] + [str(memLevel) for memLevel in range(1, 10)]))
    138    for wbits in range(9, 16):
    139        print(
    140            "\t".join(
    141                [str(wbits)]
    142                + [
    143                    f"{100 * (duration[wbits][memLevel] / duration[WB][ML] - 1):.1f}%"
    144                    for memLevel in range(1, 10)
    145                ]
    146            )
    147        )
    148    print("=" * 79)
    149    print()
    150 
    151 
    152 def run():
    153    with open(CORPUS_FILE, "rb") as handle:
    154        data = pickle.load(handle)
    155    _run(data)
    156 
    157 
    158 try:
    159    run = globals()[sys.argv[1]]
    160 except (KeyError, IndexError):
    161    print(f"Usage: {sys.argv[0]} [corpus|run]")
    162 else:
    163    run()