package_github.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Dead Software Rescue -- GitHub Mirror Packager
Creates provenance READMEs and tar.gz archives for each GitHub bare mirror.
"""

import json, tarfile, sys, io
from pathlib import Path
from datetime import datetime, timezone

sys.stdout = io.TextIOWrapper(sys.stdout.buffer, encoding='utf-8')

MIRRORS  = Path("C:/Users/Tyrone/Documents/dead-software-rescue/mirrors")
PACKAGES = Path("C:/Users/Tyrone/Documents/dead-software-rescue/packages")
CAPTURE_DATE = "2026-04-24"

# All 17 GitHub bare mirrors with metadata
GITHUB_PROJECTS = [
    # Tier 1 — Confirmed Dead
    {
        "dir": "Cryptocat.git",
        "name": "Cryptocat",
        "tier": "1",
        "tier_label": "Confirmed Dead — officially discontinued",
        "type": "Encrypted group chat (browser extension + desktop)",
        "death": "Discontinued February 2019",
        "reason": "Critical crypto vulnerability in 2013 destroyed trust. Never recovered despite a full rewrite.",
        "source": "https://github.com/cryptocat/cryptocat",
    },
    {
        "dir": "TorChat.git",
        "name": "TorChat",
        "tier": "1",
        "tier_label": "Confirmed Dead — officially discontinued",
        "type": "Tor-based P2P instant messaging",
        "death": "All development stopped after 2014",
        "reason": "Single developer vanished. Relied on deprecated Tor hidden service protocol.",
        "source": "https://github.com/prof7bit/TorChat",
    },
    {
        "dir": "Surespot.git",
        "name": "Surespot",
        "tier": "1",
        "tier_label": "Confirmed Dead — officially discontinued",
        "type": "E2E encrypted messenger for Android/iOS",
        "death": "Shut down July 31, 2022",
        "reason": "Solo developer, unsustainable maintenance burden.",
        "source": "https://github.com/DevlinBlankert/surespot-encrypted-messenger",
    },
    {
        "dir": "Ricochet Original.git",
        "name": "Ricochet (Original)",
        "tier": "1",
        "tier_label": "Confirmed Dead — officially discontinued",
        "type": "Tor-based anonymous IM",
        "death": "Development ended November 2016",
        "reason": "Relies on deprecated Tor v2 onion services. Ricochet Refresh is the community fork still alive.",
        "source": "https://github.com/ricochet-im/ricochet",
    },
    # Tier 2 — Effectively Dead
    {
        "dir": "PyBitmessage.git",
        "name": "PyBitmessage",
        "tier": "2",
        "tier_label": "Effectively Dead — zombie state",
        "type": "Decentralised serverless encrypted messaging protocol (Python)",
        "death": "Sporadic commits; effectively stalled",
        "reason": "Python 2 dependency (EOL 2020). No completed port to Python 3. Tiny dev team.",
        "source": "https://github.com/Bitmessage/PyBitmessage",
    },
    {
        "dir": "qTox.git",
        "name": "qTox",
        "tier": "2",
        "tier_label": "Effectively Dead — zombie state",
        "type": "P2P encrypted messenger, voice and video (Qt/C++)",
        "death": "Development of almost all Tox clients abandoned",
        "reason": "Developer drama, mass exodus. Protocol complexity. No sustainable funding.",
        "source": "https://github.com/qTox/qTox",
    },
    {
        "dir": "QKSMS.git",
        "name": "QKSMS",
        "tier": "2",
        "tier_label": "Effectively Dead — zombie state",
        "type": "Open-source SMS app for Android",
        "death": "Last commit June 2023",
        "reason": "Solo developer moved on. Android SMS API changes made maintenance painful.",
        "source": "https://github.com/moezbhatti/qksms",
    },
    {
        "dir": "Kontalk.git",
        "name": "Kontalk",
        "tier": "2",
        "tier_label": "Effectively Dead — zombie state",
        "type": "XMPP-based encrypted messenger (Android)",
        "death": "Minimal development, tiny user base",
        "reason": "Cannot compete with Signal. No funding. Volunteer burnout.",
        "source": "https://github.com/kontalk/androidclient",
    },
    {
        "dir": "LetsChat.git",
        "name": "Let's Chat",
        "tier": "2",
        "tier_label": "Effectively Dead — zombie state",
        "type": "Self-hosted team chat (Node.js + MongoDB)",
        "death": "No meaningful updates in years",
        "reason": "Superseded by Rocket.Chat, Mattermost, and Zulip.",
        "source": "https://github.com/sdelements/lets-chat",
    },
    # Tier 3 — Abandoned IRC/XMPP clients
    {
        "dir": "Subway.git",
        "name": "Subway (IRC)",
        "tier": "3",
        "tier_label": "Abandoned IRC/XMPP client",
        "type": "Web-based IRC client",
        "death": "Not currently functional and not under active development",
        "reason": "Developer recommends The Lounge instead.",
        "source": "https://github.com/thedjpetersen/subway",
    },
    {
        "dir": "CloudBot.git",
        "name": "CloudBot (IRC)",
        "tier": "3",
        "tier_label": "Abandoned IRC/XMPP client",
        "type": "Python IRC bot framework",
        "death": "Unmaintained for over a decade",
        "reason": "No active maintainers.",
        "source": "https://github.com/CloudBotIRC/CloudBot",
    },
    {
        "dir": "Blackened.git",
        "name": "Blackened (IRC)",
        "tier": "3",
        "tier_label": "Abandoned IRC/XMPP client",
        "type": "IRC client",
        "death": "Abandoned by developer",
        "reason": "Developer graduated, got a job, walked away.",
        "source": "https://github.com/tjensen/blackened",
    },
    {
        "dir": "irc-client-Haskell.git",
        "name": "irc-client (Haskell)",
        "tier": "3",
        "tier_label": "Abandoned IRC/XMPP client",
        "type": "Haskell IRC client library",
        "death": "Explicitly labelled abandonware by maintainer",
        "reason": "Maintainer self-declared it abandonware.",
        "source": "https://github.com/barrucadu/irc-client",
    },
    {
        "dir": "most-IRC.git",
        "name": "most (IRC)",
        "tier": "3",
        "tier_label": "Abandoned IRC/XMPP client",
        "type": "Python IRC client",
        "death": "Repo name literally includes 'abandoned'",
        "reason": "Moved to iivvoo-abandoned GitHub org by author.",
        "source": "https://github.com/iivvoo-abandoned/most",
    },
    # Tier 5 — Zombie P2P / Decentralised
    {
        "dir": "Friends-Chat.git",
        "name": "Friends Chat",
        "tier": "5",
        "tier_label": "Zombie P2P / Decentralised",
        "type": "P2P offline-friendly open-source chat (like Slack but P2P)",
        "death": "No recent development",
        "reason": "Niche use case, no sustained community.",
        "source": "https://github.com/moose-team/friends",
    },
    {
        "dir": "Cabal.git",
        "name": "Cabal",
        "tier": "5",
        "tier_label": "Zombie P2P / Decentralised",
        "type": "P2P community offline-first chat platform",
        "death": "Development sporadic",
        "reason": "Extremely niche, no funding path.",
        "source": "https://github.com/cabal-club/cabal",
    },
    {
        "dir": "DeChat.git",
        "name": "DeChat",
        "tier": "5",
        "tier_label": "Zombie P2P / Decentralised",
        "type": "Decentralised browser-extension chat",
        "death": "Proof of concept, never gained users",
        "reason": "No real development beyond initial demo.",
        "source": "https://github.com/agilgur5/DeChat",
    },
]


def dir_size_kb(path: Path) -> float:
    return round(sum(f.stat().st_size for f in path.rglob("*") if f.is_file()) / 1024, 1)


def count_objects(git_dir: Path) -> str:
    """Count loose objects in a bare git repo as a rough commit proxy."""
    try:
        objects = git_dir / "objects"
        count = sum(1 for f in objects.rglob("*") if f.is_file() and f.parent.name not in ("info", "pack"))
        packs = list((objects / "pack").glob("*.pack")) if (objects / "pack").exists() else []
        parts = []
        if count:
            parts.append(f"{count} loose objects")
        if packs:
            parts.append(f"{len(packs)} pack file(s)")
        return ", ".join(parts) if parts else "bare repo"
    except Exception:
        return "bare repo"


def make_readme(proj: dict, git_dir: Path) -> str:
    size_kb = dir_size_kb(git_dir)
    obj_info = count_objects(git_dir)
    lines = [
        "=" * 72,
        "DEAD SOFTWARE RESCUE — Preservation Package",
        "=" * 72,
        "",
        f"Project:       {proj['name']}",
        f"Tier:          {proj['tier']} — {proj['tier_label']}",
        f"Type:          {proj['type']}",
        f"Death/Status:  {proj['death']}",
        f"Death reason:  {proj['reason']}",
        f"Source:        {proj['source']}",
        "",
        "--- Preservation ---",
        f"Captured:      {CAPTURE_DATE} by Dead Software Rescue project",
        f"Packaged:      {datetime.now(timezone.utc).strftime('%Y-%m-%d')}",
        f"Method:        git clone --mirror (bare repository, full history preserved)",
        f"Archive size:  {size_kb} KB",
        f"Repo contents: {obj_info}",
        "",
        "--- Contents of this archive ---",
        f"  {git_dir.name}/   (bare git mirror — all branches, tags, full history)",
        "",
        "To use this mirror:",
        "  git clone <this_dir>/<project>.git  my-local-copy",
        "  (or: git clone --mirror to re-mirror elsewhere)",
        "",
        f"Original source: {proj['source']}",
        "",
        "This package was created for archival and research purposes.",
        "All software belongs to its respective authors and is subject to its original license.",
        "=" * 72,
    ]
    return "\n".join(lines) + "\n"


def pack_project(proj: dict) -> dict | None:
    git_dir = MIRRORS / proj["dir"]
    if not git_dir.exists():
        print(f"  SKIP — {git_dir} not found")
        return None

    out_name = f"{proj['dir'].replace('.git','').replace(' ', '_')}_rescue_{CAPTURE_DATE}.tar.gz"
    out_path = PACKAGES / out_name

    if out_path.exists():
        print(f"  EXISTS — {out_name}, skipping")
        return {"package": out_name, "skipped": True}

    readme_text = make_readme(proj, git_dir)
    readme_tmp = PACKAGES / f"_README_{proj['name']}.txt"
    readme_tmp.write_text(readme_text, encoding="utf-8")

    print(f"  Packing {out_name} ...")
    archive_root = proj["dir"].replace(".git", "").replace(" ", "_")
    with tarfile.open(out_path, "w:gz") as tar:
        tar.add(readme_tmp, arcname=f"{archive_root}/README.txt")
        tar.add(git_dir, arcname=f"{archive_root}/{proj['dir']}")

    readme_tmp.unlink()
    size_mb = round(out_path.stat().st_size / (1024 * 1024), 2)
    print(f"  -> {out_name}  ({size_mb} MB)")
    return {"package": out_name, "size_mb": size_mb}


def main():
    PACKAGES.mkdir(parents=True, exist_ok=True)

    print("\n" + "=" * 60)
    print("  Dead Software Rescue — GitHub Mirror Packager")
    print(f"  Projects: {len(GITHUB_PROJECTS)}")
    print(f"  Output:   {PACKAGES}")
    print("=" * 60)

    results = []
    for proj in GITHUB_PROJECTS:
        print(f"\n[{proj['name']}]")
        r = pack_project(proj)
        if r:
            results.append({**r, "name": proj["name"], "tier": proj["tier"]})

    packed   = [r for r in results if not r.get("skipped")]
    skipped  = [r for r in results if r.get("skipped")]
    total_mb = sum(r.get("size_mb", 0) for r in packed)

    print(f"\n{'='*60}")
    print(f"  Done.")
    print(f"  Newly packed: {len(packed)}")
    print(f"  Skipped (already existed): {len(skipped)}")
    print(f"  Total new size: {total_mb:.1f} MB")
    print(f"{'='*60}")

    # Append to existing package_log.json
    log_path = PACKAGES / "package_log.json"
    if log_path.exists():
        with open(log_path, encoding="utf-8") as f:
            log = json.load(f)
    else:
        log = {"packages": []}

    for r in packed:
        log["packages"].append({
            "name": r["name"],
            "tier": r["tier"],
            "source": "github",
            "package": r["package"],
            "size_mb": r.get("size_mb"),
            "packaged_at": datetime.now(timezone.utc).isoformat(),
        })

    with open(log_path, "w", encoding="utf-8") as f:
        json.dump(log, f, indent=2, ensure_ascii=False)
    print(f"  Log updated -> {log_path}")


if __name__ == "__main__":
    main()