download.py

analyze.py

import os
import re

# ZA̡͊͠͝LGΌ ISͮ̂҉̯͈͕̹̘̱ TO͇̹̺ͅƝ̴ȳ̳ TH̘Ë͖́̉ ͠P̯͍̭O̚​N̐Y̡ H̸̡̪̯ͨ͊̽̅̾̎Ȩ̬̩̾͛ͪ̈́̀́͘ ̶̧̨̱̹̭̯ͧ̾ͬC̷̙̲̝͖ͭ̏ͥͮ͟Oͮ͏̮̪̝͍M̲̖͊̒ͪͩͬ̚̚͜Ȇ̴̟̟͙̞ͩ͌͝S̨̥̫͎̭ͯ̿̔̀ͅ
pattern = re.compile(r'<span class="CommitLog-author" title="(\w*@\w*.\w*)')

all_committers = set()

for f in os.listdir("log"):
    if "external" in f or "third_party" in f or "deps" in f:
        continue
    log = open(f"log/{f}", "r").read()
    commits = pattern.findall(log)
    repo_committers = set(commits)
    all_committers = set.union(all_committers, repo_committers)
    print(f"{f}: {repo_committers}")

print(len(all_committers))

download.py

import json
import requests

# cleaned up https://chromium.googlesource.com/?format=JSON
repos = json.load(open("repolist.json", "r"))

for name, details in repos.items():
    filename = name.replace("/", "_")
    url = details["clone_url"] + "/+log"
    log = requests.get(url).text
    with open(f"log/{filename}.html", "w") as f:
        f.write(log)
    print(name)