git-sync.js

/**
 * git-sync.js
 * Syncs the ./data directory with a private GitHub repository.
 *
 * On startup:  clones the repo into ./data (or pulls if already cloned)
 * After writes: debounced git add + commit + push (batches rapid writes)
 *
 * Environment variables (set as HF Space Secrets):
 *   GITHUB_TOKEN    – GitHub Personal Access Token (with repo scope)
 *   GITHUB_USERNAME – GitHub username
 *   GITHUB_REPO    – e.g. "username/usagi-data"
 *   GIT_USER_NAME  – (optional) commit author name, default "UsagiBot"
 *   GIT_USER_EMAIL – (optional) commit author email, default "bot@usagi.local"
 */

const { execSync, exec } = require("child_process");
const path = require("path");

const GITHUB_TOKEN = process.env.GITHUB_TOKEN || "";
const GITHUB_USERNAME = process.env.GITHUB_USERNAME || process.env.GIT_USER_NAME || "";
const GITHUB_REPO = process.env.GITHUB_REPO || "";
const GIT_USER_NAME = process.env.GIT_USER_NAME || "UsagiBot";
const GIT_USER_EMAIL = process.env.GIT_USER_EMAIL || "bot@usagi.local";

// Debounce: wait 15 seconds after last write before pushing
const PUSH_DEBOUNCE_MS = parseInt(process.env.PUSH_DEBOUNCE_MS) || 15000;

const DATA_DIR = path.join(__dirname, "data");

let _pushTimer = null;
let _pushing = false;
let _pendingPush = false;

function isConfigured() {
  return !!(GITHUB_TOKEN && GITHUB_USERNAME && GITHUB_REPO);
}

function getRepoUrl() {
  return `https://${GITHUB_USERNAME}:${GITHUB_TOKEN}@github.com/${GITHUB_REPO}.git`;
}

function git(cmd, cwd = DATA_DIR) {
  return execSync(`git ${cmd}`, {
    cwd,
    encoding: "utf-8",
    stdio: ["pipe", "pipe", "pipe"],
    timeout: 60000,
  }).trim();
}

function gitAsync(cmd, cwd = DATA_DIR) {
  return new Promise((resolve, reject) => {
    exec(`git ${cmd}`, { cwd, encoding: "utf-8", timeout: 60000 }, (err, stdout, stderr) => {
      if (err) reject(err);
      else resolve((stdout || "").trim());
    });
  });
}

/**
 * Clone the data repo on startup, or pull latest if already cloned.
 * If the repo is empty / doesn't exist yet, initialize it.
 */
function initSync() {
  if (!isConfigured()) {
    console.log("⚠️  Git sync disabled (GITHUB_TOKEN or GITHUB_REPO not set)");
    return false;
  }

  const fs = require("fs");

  try {
    // Check if data dir is already a git repo
    const isGitRepo =
      fs.existsSync(path.join(DATA_DIR, ".git")) ||
      fs.existsSync(path.join(DATA_DIR, ".git", "HEAD"));

    if (isGitRepo) {
      console.log("📥 Pulling latest data from GitHub...");
      try {
        git("pull --rebase --autostash origin main");
        console.log("✅ Data pulled successfully");
      } catch (e) {
        // If pull fails (e.g. diverged), force reset to remote
        console.log("⚠️  Pull failed, resetting to remote...");
        try {
          git("fetch origin");
          git("reset --hard origin/main");
        } catch (e2) {
          console.log("⚠️  Reset also failed, continuing with local data");
        }
      }
    } else {
      // Clone the repo
      console.log(`📥 Cloning data repo: ${GITHUB_REPO}...`);

      // Remove data dir if it exists (has local files but no git)
      const hadLocalData = fs.existsSync(DATA_DIR);
      const localFiles = {};

      if (hadLocalData) {
        // Backup local files before clone
        const files = fs.readdirSync(DATA_DIR);
        for (const f of files) {
          const fp = path.join(DATA_DIR, f);
          if (fs.statSync(fp).isFile()) {
            localFiles[f] = fs.readFileSync(fp, "utf-8");
          }
        }
        fs.rmSync(DATA_DIR, { recursive: true, force: true });
      }

      try {
        // Clone into data dir
        execSync(`git clone ${getRepoUrl()} data`, {
          cwd: __dirname,
          encoding: "utf-8",
          stdio: ["pipe", "pipe", "pipe"],
          timeout: 60000,
        });
        console.log("✅ Data repo cloned successfully");
      } catch (cloneErr) {
        // If repo is empty or doesn't exist, init fresh
        console.log("⚠️  Clone failed, initializing fresh repo...");
        fs.mkdirSync(DATA_DIR, { recursive: true });
        git("init", DATA_DIR);
        git(`remote add origin ${getRepoUrl()}`, DATA_DIR);
        git("checkout -b main", DATA_DIR);
      }

      // Restore local files that weren't in the cloned repo
      if (hadLocalData) {
        for (const [filename, content] of Object.entries(localFiles)) {
          const fp = path.join(DATA_DIR, filename);
          if (!fs.existsSync(fp)) {
            fs.writeFileSync(fp, content, "utf-8");
            console.log(`  📄 Restored local file: ${filename}`);
          }
        }
      }
    }

    // Configure git user
    try {
      git(`config user.name "${GIT_USER_NAME}"`);
      git(`config user.email "${GIT_USER_EMAIL}"`);
    } catch (e) {}

    return true;
  } catch (e) {
    console.error("❌ Git sync init failed:", e.message);
    return false;
  }
}

/**
 * Schedule a debounced push. Call this after every database write.
 * Multiple rapid writes will be batched into a single commit+push.
 */
function scheduleSync() {
  if (!isConfigured()) return;

  if (_pushTimer) clearTimeout(_pushTimer);

  _pushTimer = setTimeout(() => {
    _pushTimer = null;
    doSync();
  }, PUSH_DEBOUNCE_MS);
}

/**
 * Actually commit and push all changes.
 */
async function doSync() {
  if (_pushing) {
    _pendingPush = true;
    return;
  }

  _pushing = true;

  try {
    // Check for changes
    const status = await gitAsync("status --porcelain");
    if (!status) {
      _pushing = false;
      return; // nothing to commit
    }

    await gitAsync("add -A");

    const timestamp = new Date().toISOString().replace("T", " ").slice(0, 19);
    const changedFiles = status
      .split("\n")
      .map((l) => l.trim().split(/\s+/).pop())
      .join(", ");

    await gitAsync(`commit -m "sync: ${changedFiles} [${timestamp}]"`);

    // Push with retry
    let retries = 3;
    while (retries > 0) {
      try {
        await gitAsync("push origin main");
        console.log(`📤 Data synced to GitHub (${changedFiles})`);
        break;
      } catch (pushErr) {
        retries--;
        if (retries > 0) {
          // Pull and retry
          try {
            await gitAsync("pull --rebase origin main");
          } catch (e) {}
        } else {
          console.error("❌ Push failed after retries:", pushErr.message);
        }
      }
    }
  } catch (e) {
    // Don't crash on sync errors
    if (!e.message?.includes("nothing to commit")) {
      console.error("❌ Sync error:", e.message);
    }
  } finally {
    _pushing = false;

    // If another write happened while we were pushing, push again
    if (_pendingPush) {
      _pendingPush = false;
      setTimeout(doSync, 3000);
    }
  }
}

/**
 * Force an immediate sync (e.g. on shutdown).
 */
async function forceSync() {
  if (!isConfigured()) return;
  if (_pushTimer) {
    clearTimeout(_pushTimer);
    _pushTimer = null;
  }
  await doSync();
}

module.exports = { initSync, scheduleSync, forceSync, isConfigured };