core / oven / papers-git-poller.mjs

// papers-git-poller.mjs — polls git for papers/ changes, auto-triggers PDF builds
//
// Runs inside the oven server. Every POLL_INTERVAL_MS (default 60s), fetches
// origin/main and checks if any papers/ paths changed since the last
// successful build. If so, pulls and triggers startPapersBuild().
//
// Shares the git clone at GIT_REPO_DIR with native-git-poller.mjs.
// Uses a separate hash file (.last-papers-built-hash) to track state.

import { execFile } from "child_process";
import { promises as fs } from "fs";
import path from "path";

const POLL_INTERVAL_MS = parseInt(process.env.PAPERS_POLL_INTERVAL_MS || "60000", 10);
const GIT_REPO_DIR = process.env.NATIVE_GIT_DIR || "/opt/oven/native-git";
const BRANCH = process.env.NATIVE_GIT_BRANCH || "main";
const HASH_FILE = path.join(GIT_REPO_DIR, ".last-papers-built-hash");

// Paths that should trigger a papers rebuild (prefixes)
const TRIGGER_PREFIXES = [
  "papers/",
  "system/public/type/webfonts/", // font changes affect paper output
];

// File extensions that are LaTeX source files (trigger a build).
// Excludes .pdf and other build outputs to prevent infinite loops
// when the oven pushes built PDFs back to the repo.
const SOURCE_EXTENSIONS = [
  ".tex", ".bib", ".sty", ".cls", ".mjs", ".js", ".json",
  ".png", ".jpg", ".jpeg", ".svg", ".eps",  // figures
];

// Paths generated by the oven build — always ignore these
const IGNORE_PATTERNS = [
  "system/public/papers.aesthetic.computer/", // deployed PDFs + index.html
  "papers/metadata.json",
  "papers/BUILDLOG.md",
];

function isSourceChange(filePath) {
  // Synthetic force-rebuild paths always trigger
  if (filePath === "papers/force-rebuild") return true;
  // Ignore oven-generated output paths
  if (IGNORE_PATTERNS.some((pat) => filePath.startsWith(pat) || filePath === pat)) {
    return false;
  }
  // Must match a trigger prefix
  if (!TRIGGER_PREFIXES.some((prefix) => filePath.startsWith(prefix))) {
    return false;
  }
  // Font changes always trigger
  if (filePath.startsWith("system/public/type/webfonts/")) return true;
  // For papers/, only trigger on source file extensions (not .pdf, .log, .aux, etc.)
  const ext = filePath.slice(filePath.lastIndexOf(".")).toLowerCase();
  return SOURCE_EXTENSIONS.includes(ext);
}

let polling = false;
let timer = null;
let startBuildFn = null;
let logFn = (level, icon, msg) => console.log(`[papers-git-poller] ${msg}`);

function git(args, cwd = GIT_REPO_DIR) {
  return new Promise((resolve, reject) => {
    execFile("git", args, { cwd, timeout: 30_000 }, (err, stdout, stderr) => {
      if (err) {
        err.stderr = stderr;
        return reject(err);
      }
      resolve(stdout.trim());
    });
  });
}

async function readLastBuiltHash() {
  try {
    return (await fs.readFile(HASH_FILE, "utf8")).trim();
  } catch {
    return null;
  }
}

async function writeLastBuiltHash(hash) {
  await fs.writeFile(HASH_FILE, hash + "\n", "utf8");
}

async function poll() {
  if (polling) return;
  polling = true;

  try {
    // Fetch latest from origin
    await git(["fetch", "origin", BRANCH, "--quiet"]);

    const remoteHead = await git(["rev-parse", `origin/${BRANCH}`]);
    const lastBuilt = await readLastBuiltHash();

    if (remoteHead === lastBuilt) {
      polling = false;
      return;
    }

    // Check which files changed
    let changedPaths = "";
    if (lastBuilt) {
      try {
        const diffOutput = await git([
          "diff",
          "--name-only",
          lastBuilt,
          remoteHead,
        ]);
        changedPaths = diffOutput;
      } catch {
        // lastBuilt hash might not exist (force push, etc) — treat as full build
        changedPaths = "papers/force-rebuild";
      }
    } else {
      // First run — treat as full build
      changedPaths = "papers/force-rebuild";
    }

    // Filter to papers-relevant source paths (excludes .pdf and build outputs)
    const papersPaths = changedPaths
      .split("\n")
      .filter((p) => isSourceChange(p));

    if (papersPaths.length === 0) {
      // Changes exist but not in papers/ — update hash, skip build
      logFn("info", "⏭️", `New commits (${remoteHead.slice(0, 8)}) but no papers/ changes — skipping build`);
      await writeLastBuiltHash(remoteHead);
      polling = false;
      return;
    }

    // Pull the changes so cli.mjs works on up-to-date code
    await git(["checkout", BRANCH, "--quiet"]);
    await git(["merge", `origin/${BRANCH}`, "--ff-only", "--quiet"]);

    logFn(
      "info",
      "📄",
      `Papers changes detected (${remoteHead.slice(0, 8)}): ${papersPaths.length} file(s) — triggering PDF build`
    );

    // Trigger build
    const job = await startBuildFn({
      ref: remoteHead,
      changed_paths: papersPaths.join(","),
    });

    logFn(
      "info",
      "🚀",
      `Papers build ${job.id} started`
    );

    // Update hash after successfully starting the build
    await writeLastBuiltHash(remoteHead);
  } catch (err) {
    if (err?.code === "PAPERS_BUILD_BUSY") {
      logFn("info", "⏳", "Papers build already running — will retry next poll");
    } else {
      logFn(
        "error",
        "❌",
        `Papers git poll error: ${err.message}${err.stderr ? " | " + err.stderr.trim() : ""}`
      );
    }
  } finally {
    polling = false;
  }
}

// ── Public API ──────────────────────────────────────────────────────────────

export function startPoller({ startPapersBuild, addServerLog }) {
  startBuildFn = startPapersBuild;
  if (addServerLog) logFn = addServerLog;

  // Check that GIT_REPO_DIR exists before starting
  fs.access(GIT_REPO_DIR)
    .then(() => {
      logFn(
        "info",
        "📄",
        `Papers git poller started (every ${POLL_INTERVAL_MS / 1000}s, repo: ${GIT_REPO_DIR})`
      );
      // Stagger start vs native poller (native uses 5s delay, we use 15s)
      setTimeout(poll, 15000);
      timer = setInterval(poll, POLL_INTERVAL_MS);
    })
    .catch(() => {
      logFn(
        "error",
        "⚠️",
        `Papers git poller disabled — repo dir not found: ${GIT_REPO_DIR}. Run: git clone --branch main https://github.com/whistlegraph/aesthetic-computer.git ${GIT_REPO_DIR}`
      );
    });
}

export function stopPoller() {
  if (timer) {
    clearInterval(timer);
    timer = null;
    logFn("info", "🛑", "Papers git poller stopped");
  }
}

export function getPollerStatus() {
  return {
    running: timer !== null,
    intervalMs: POLL_INTERVAL_MS,
    repoDir: GIT_REPO_DIR,
    branch: BRANCH,
  };
}