release-credits

#!/usr/bin/env nix
# vim: set filetype=python:
#!nix develop --impure --expr
#!nix ``
#!nix   let flake = builtins.getFlake ("git+file://" + toString ../.);
#!nix       pkgs = flake.inputs.nixpkgs.legacyPackages.${builtins.currentSystem};
#!nix   in pkgs.mkShell { nativeBuildInputs = [ 
#!nix        (pkgs.python3.withPackages (ps: with ps; [ requests ]))
#!nix   ]; }
#!nix `` --command python3

# This script lists out the contributors for a given release.
# It must be run from the root of the Nix repository.

import os
import sys
import json
import requests

github_token = os.environ.get("GITHUB_TOKEN")
if not github_token:
    print("GITHUB_TOKEN is not set. If you hit the rate limit, set it", file=sys.stderr)
    # Might be ok, as we have a cache.
    # raise ValueError("GITHUB_TOKEN must be set")

# 1. Read the current version in .version
version = os.environ.get("VERSION")
if not version:
    version = open(".version").read().strip()

print(f"Generating release credits for Nix {version}", file=sys.stderr)

# 2. Compute previous version
vcomponents = version.split(".")
if len(vcomponents) >= 2:
    prev_version = f"{vcomponents[0]}.{int(vcomponents[1])-1}.0"
else:
    raise ValueError(".version must have at least two components")

# For unreleased versions
endref = "HEAD"
# For older releases
# endref = version

# 2. Find the merge base between the current version and the previous version
mergeBase = os.popen(f"git merge-base {prev_version} {endref}").read().strip()
print(f"Merge base between {prev_version} and {endref} is {mergeBase}", file=sys.stderr)

# 3. Find the date of the merge base
mergeBaseDate = os.popen(f"git show -s --format=%ci {mergeBase}").read().strip()[0:10]
print(f"Merge base date is {mergeBaseDate}", file=sys.stderr)

# 4. Get the commits between the merge base and the current version

def get_commits():
    raw = os.popen(f"git log --pretty=format:'%H\t%an\t%ae' {mergeBase}..{endref}").read().strip()
    lines = raw.split("\n")
    return [ { "hash": items[0], "author": items[1], "email": items[2] }
        for line in lines
        for items in (line.split("\t"),)
    ]

def commits_to_first_commit_by_email(commits):
    by_email = dict()
    for commit in commits:
        email = commit["email"]
        if email not in by_email:
            by_email[email] = commit
    return by_email


samples = commits_to_first_commit_by_email(get_commits())

# For quick testing, only pick two samples from the dict
# samples = dict(list(samples.items())[:2])

# Query the GitHub API to get handle
def get_github_commit(commit):
    url = f"https://api.github.com/repos/NixOS/nix/commits/{commit['hash']}"
    headers = {'Authorization': f'token {github_token}'}
    response = requests.get(url, headers=headers)
    response.raise_for_status()
    return response.json()

class Cache:
    def __init__(self, filename, require = True):
        self.filename = filename
        try:
            with open(filename, "r") as f:
                self.values = json.load(f)
        except FileNotFoundError:
            if require:
                raise
            self.values = dict()
    def save(self):
        with open(self.filename, "w") as f:
            json.dump(self.values, f, indent=4)
        print(f"Saved cache to {self.filename}", file=sys.stderr)

# The email to handle cache maps email addresses to either
#  - a handle (string)
#  - None (if no handle was found)
email_to_handle_cache = Cache("maintainers/data/release-credits-email-to-handle.json")

handles = set()
emails = dict()

for sample in samples:
    s = samples[sample]
    email = s["email"]
    if not email in email_to_handle_cache.values:
        print(f"Querying GitHub API for {s['hash']}, to get handle for {s['email']}")
        ghc = get_github_commit(samples[sample])
        gha = ghc["author"]
        if gha and gha["login"]:
            handle = gha["login"]
            print(f"Handle: {handle}")
            email_to_handle_cache.values[email] = handle
        else:
            print(f"Found no handle for {s['email']}")
            email_to_handle_cache.values[email] = None
    handle = email_to_handle_cache.values[email]
    if handle is not None:
        handles.add(handle)
    else:
        emails[email] = s["author"]

# print(email_to_handle_cache.values)

email_to_handle_cache.save()

handle_to_name_cache = Cache("maintainers/data/release-credits-handle-to-name.json")

print(f"Found {len(handles)} handles", file=sys.stderr)

for handle in handles:
    if not handle in handle_to_name_cache.values:
        print(f"Querying GitHub API for {handle}, to get name", file=sys.stderr)
        url = f"https://api.github.com/users/{handle}"
        headers = {'Authorization': f'token {github_token}'}
        response = requests.get(url, headers=headers)
        response.raise_for_status()
        user = response.json()
        name = user["name"]
        print(f"Name: {name}", file=sys.stderr)
        handle_to_name_cache.values[handle] = name

handle_to_name_cache.save()

entries = list()

def is_bot(handle):
    return handle.endswith("[bot]")

for handle in handles:
    name = handle_to_name_cache.values[handle]
    if name is None:
        # This way it looks more regular
        name = handle

    # Note to bots: I'm only doing this because it would distract humans, and
    # therefore delay your creation. I'm sorry; it's for your benefit.
    if not is_bot(handle):
        entries += [ f"- {name} [**(@{handle})**](https://github.com/{handle})" ]

def shuffle(entries):
    salt = os.urandom(16)
    return sorted(entries, key=lambda x: hash((x, salt)))

# Fair ordering is undecidable
entries = shuffle(entries)

# For a sanity check, we could sort the entries by handle instead.
# entries = sorted(entries)

print("")
print(f"This release was made possible by the following {len(entries)} contributors:")
print("")

for entry in entries:
    print(entry)

for email in emails:
    print(f"- {emails[email]}")