@bongos/core 1.21.175 → 1.21.177

package/scripts/migrate_cost_ledger.sh

@@ -1,178 +0,0 @@
-#!/usr/bin/env bash
-# scripts/migrate_cost_ledger.sh
-#
-# One-shot (idempotent) migration: fold any per-worktree
-# art/iterations/cost_ledger.jsonl files into the shared location at
-# ~/.config/otb/iterations/cost_ledger.jsonl, then rebuild cost_summary.json
-# from the merged ledger.
-#
-# Per GDS #140 (2026-05-10 Lars decision): the ledger lives next to
-# pms-session.json + GOOGLE_AI_STUDIO_KEY in $HOME so it survives worktree churn
-# the way the API key does.
-#
-# Safe to re-run: appends only ledger rows not already present in the shared
-# file (matched by (ts, model, kind, tile, attempt, cost_usd, notes) tuple).
-# Per-worktree files are renamed to *.migrated-<timestamp> instead of deleted,
-# so nothing is destroyed.
-#
-# Usage:
-#   scripts/migrate_cost_ledger.sh                  # search standard locations
-#   scripts/migrate_cost_ledger.sh <ledger> [...]   # explicit files to fold in
-#   OTB_LEDGER_DIR=/tmp/foo scripts/migrate_cost_ledger.sh   # override target
-
-set -euo pipefail
-
-SHARED_DIR="${OTB_LEDGER_DIR:-$HOME/.config/otb/iterations}"
-SHARED_LEDGER="$SHARED_DIR/cost_ledger.jsonl"
-SHARED_SUMMARY="$SHARED_DIR/cost_summary.json"
-
-mkdir -p "$SHARED_DIR"
-touch "$SHARED_LEDGER"
-
-# Build candidate list: explicit args win, otherwise sweep common locations.
-declare -a CANDIDATES=()
-if [ "$#" -gt 0 ]; then
-    CANDIDATES=("$@")
-else
-    # Repo root (this script lives in <root>/scripts/).
-    SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
-    REPO_ROOT="$(cd "$SCRIPT_DIR/.." && pwd)"
-
-    # 1. The current repo's per-worktree ledger.
-    [ -f "$REPO_ROOT/art/iterations/cost_ledger.jsonl" ] && \
-        CANDIDATES+=("$REPO_ROOT/art/iterations/cost_ledger.jsonl")
-
-    # 2. Any sibling worktrees of this repo.
-    if [ -d "$REPO_ROOT/.claude/worktrees" ]; then
-        while IFS= read -r p; do
-            CANDIDATES+=("$p")
-        done < <(find "$REPO_ROOT/.claude/worktrees" -mindepth 2 -maxdepth 4 \
-                      -type f -name cost_ledger.jsonl 2>/dev/null)
-    fi
-fi
-
-# De-dup the candidate list AND drop the shared-ledger path itself if it snuck in.
-# (Stays portable to macOS default bash 3.2 — no associative arrays.)
-SHARED_REAL="$(cd "$SHARED_DIR" && pwd)/cost_ledger.jsonl"
-SEEN_LIST=""
-FILTERED=()
-for c in "${CANDIDATES[@]+"${CANDIDATES[@]}"}"; do
-    [ -f "$c" ] || continue
-    REAL="$(cd "$(dirname "$c")" && pwd)/$(basename "$c")"
-    [ "$REAL" = "$SHARED_REAL" ] && continue
-    case "$SEEN_LIST" in
-        *"|$REAL|"*) continue ;;
-    esac
-    SEEN_LIST="$SEEN_LIST|$REAL|"
-    FILTERED+=("$REAL")
-done
-
-if [ "${#FILTERED[@]}" -eq 0 ]; then
-    echo "No per-worktree cost_ledger.jsonl files to migrate."
-    echo "Shared ledger: $SHARED_LEDGER"
-    # Still rebuild the summary so a fresh shared dir gets a summary file.
-    python3 "$(dirname "$0")/../modules/art-pipeline/pipeline/cost_ledger.py" status >/dev/null || true
-    exit 0
-fi
-
-echo "Migrating to: $SHARED_LEDGER"
-for f in "${FILTERED[@]}"; do echo "  source: $f"; done
-
-# Python does the dedup + merge + summary rebuild. Keeps the bash side simple
-# and gives us proper JSON-aware comparison.
-python3 - "$SHARED_LEDGER" "$SHARED_SUMMARY" "${FILTERED[@]}" <<'PYEOF'
-import json, os, sys, time
-from pathlib import Path
-
-shared_ledger = Path(sys.argv[1])
-shared_summary = Path(sys.argv[2])
-sources = [Path(p) for p in sys.argv[3:]]
-
-def row_key(r: dict) -> tuple:
-    # Identity: (ts, model, kind, tile, attempt, cost_usd, notes).
-    # No `id` column exists today, so dedup by content tuple.
-    return (
-        r.get("ts", ""),
-        r.get("model", ""),
-        r.get("kind", ""),
-        r.get("tile", ""),
-        r.get("attempt", 0),
-        round(float(r.get("cost_usd", 0.0)), 6),
-        r.get("notes", ""),
-    )
-
-existing_keys: set = set()
-if shared_ledger.exists():
-    with shared_ledger.open() as f:
-        for line in f:
-            line = line.strip()
-            if not line:
-                continue
-            try:
-                existing_keys.add(row_key(json.loads(line)))
-            except Exception:
-                pass
-
-added = 0
-skipped_dup = 0
-skipped_bad = 0
-ts_suffix = time.strftime("%Y%m%dT%H%M%S")
-
-with shared_ledger.open("a") as out:
-    for src in sources:
-        if not src.exists():
-            continue
-        with src.open() as f:
-            for line in f:
-                line_stripped = line.strip()
-                if not line_stripped:
-                    continue
-                try:
-                    row = json.loads(line_stripped)
-                except Exception:
-                    skipped_bad += 1
-                    continue
-                k = row_key(row)
-                if k in existing_keys:
-                    skipped_dup += 1
-                    continue
-                existing_keys.add(k)
-                out.write(line if line.endswith("\n") else line_stripped + "\n")
-                added += 1
-        # Rename source so a re-run doesn't re-process it AND nothing is destroyed.
-        try:
-            src.rename(src.with_suffix(src.suffix + f".migrated-{ts_suffix}"))
-            print(f"  archived: {src} → {src.name}.migrated-{ts_suffix}")
-        except OSError as e:
-            print(f"  warn: could not rename {src}: {e}", file=sys.stderr)
-
-# Rebuild summary from the merged ledger.
-total = 0.0
-n = 0
-with shared_ledger.open() as f:
-    for line in f:
-        line = line.strip()
-        if not line:
-            continue
-        try:
-            total += float(json.loads(line).get("cost_usd", 0.0))
-            n += 1
-        except Exception:
-            pass
-
-HARD_STOP = 80.00
-shared_summary.parent.mkdir(parents=True, exist_ok=True)
-shared_summary.write_text(json.dumps({
-    "total_usd": round(total, 4),
-    "calls": n,
-    "hard_stop_usd": HARD_STOP,
-    "remaining_usd": round(HARD_STOP - total, 4),
-    "updated_at": time.strftime("%Y-%m-%dT%H:%M:%S"),
-}, indent=2))
-
-print(f"\nMerged: {added} new rows, {skipped_dup} duplicates skipped, {skipped_bad} malformed lines skipped.")
-print(f"Shared ledger now: {n} rows, ${total:.4f} spent of ${HARD_STOP:.2f} cap.")
-PYEOF
-
-echo
-echo "Done. Verify with: python3 modules/art-pipeline/pipeline/cost_ledger.py status"

package/scripts/status-probe.mjs

@@ -1,213 +0,0 @@
-#!/usr/bin/env node
-// External uptime prober for example.com (incident 2026-06-03, task #565).
-//
-// WHY THIS EXISTS, AND WHY IT IS NOT src/bongos/uptime-poller.js:
-//   The in-process poller (src/bongos/uptime-poller.js) runs INSIDE the Node
-//   process on the droplet and probes 127.0.0.1. When the droplet is down,
-//   that poller is down too, so it records nothing — the outage becomes a data
-//   gap that renders as "100% up". A monitor cannot witness its own death.
-//   This script is the opposite: it runs on a GitHub-hosted runner (off our
-//   infrastructure) and probes the PUBLIC URL from the outside, exactly as a
-//   visitor would. It is the source of truth for real availability.
-//
-// WHAT IT DOES:
-//   - Probes a target URL up to N times (default 3) with a gap between tries.
-//     UP if any attempt returns HTTP 200; DOWN only if every attempt fails.
-//     The in-run retries suppress single transient blips, so a run that
-//     reports DOWN is a confirmed outage worth alerting on.
-//   - Appends the result to <dir>/history.jsonl and rewrites <dir>/status.json
-//     (a small snapshot + rolling uptime summary). These files live on the
-//     off-box `status-data` branch (see .github/workflows/status-probe.yml),
-//     so The Watch can read truthful uptime even while the droplet is dark.
-//   - Prunes history to the trailing 90 days (matches The Watch's window).
-//   - Writes `status=up|down` (+ latency, http_status) to $GITHUB_OUTPUT when
-//     run inside Actions, so the workflow can decide whether to alert.
-//
-// EXIT CODE: 0 on success regardless of up/down (the workflow owns alerting via
-//   the step output). Non-zero only on its own internal error.
-//
-// USAGE:
-//   node scripts/status-probe.mjs [--dir .] [--target https://example.com/healthz]
-//                                 [--attempts 3] [--timeout-ms 15000] [--gap-ms 10000]
-
-import { readFile, writeFile, mkdir } from 'node:fs/promises';
-import { existsSync } from 'node:fs';
-import path from 'node:path';
-import { pathToFileURL } from 'node:url';
-import { createRequire } from 'node:module';
-
-// Target + User-Agent come from the INSTANCE's own branding pack (task 1002757).
-// Both were founder literals, so another instance's uptime CI probed the founder
-// host and reported ITS health as their own. No literal fallback for the target:
-// a probe with nothing to probe must say so, not silently watch someone else.
-const requireCjs = createRequire(import.meta.url);
-const brandingSafe = () => { try { return requireCjs('../src/branding'); } catch { return null; } };
-const defaultTarget = () => {
-  const b = brandingSafe();
-  try {
-    const origin = b && b.branding().domains?.publicOrigin;
-    return origin ? `${String(origin).replace(/\/+$/, '')}/healthz` : null;
-  } catch { return null; }
-};
-const defaultUserAgent = () => {
-  const b = brandingSafe();
-  try { return `${b.userAgent().split('/')[0]}-status-probe/1.0 (+github-actions)`; }
-  catch { return 'cloud-bongos-status-probe/1.0 (+github-actions)'; }
-};
-
-const DEFAULTS = {
-  dir: '.',
-  target: defaultTarget(),
-  attempts: 3,
-  'timeout-ms': 15000,
-  'gap-ms': 10000,
-};
-const PRUNE_DAYS = 90;
-
-export function parseArgs(argv) {
-  const out = { ...DEFAULTS };
-  for (let i = 0; i < argv.length; i++) {
-    const a = argv[i];
-    if (a.startsWith('--')) {
-      const key = a.slice(2);
-      if (key in DEFAULTS) out[key] = argv[++i];
-    }
-  }
-  out.attempts = Number(out.attempts);
-  out['timeout-ms'] = Number(out['timeout-ms']);
-  out['gap-ms'] = Number(out['gap-ms']);
-  return out;
-}
-
-const sleep = (ms) => new Promise((r) => setTimeout(r, ms));
-
-// One HTTP attempt. Resolves { ok, httpStatus, latencyMs }. Never throws.
-async function attempt(target, timeoutMs) {
-  const ac = new AbortController();
-  const timer = setTimeout(() => ac.abort(), timeoutMs);
-  const start = Date.now();
-  try {
-    const res = await fetch(target, {
-      signal: ac.signal,
-      redirect: 'manual', // a 200 must come from the origin, not a redirect chain
-      headers: { 'User-Agent': defaultUserAgent() },
-    });
-    return { ok: res.status === 200, httpStatus: res.status, latencyMs: Date.now() - start };
-  } catch (_) {
-    return { ok: false, httpStatus: null, latencyMs: null };
-  } finally {
-    clearTimeout(timer);
-  }
-}
-
-// Probe with confirmation retries. UP if any attempt is a 200.
-async function probe({ target, attempts, timeoutMs, gapMs }) {
-  let last = { ok: false, httpStatus: null, latencyMs: null };
-  for (let i = 0; i < attempts; i++) {
-    const r = await attempt(target, timeoutMs);
-    if (r.ok) return { status: 'up', httpStatus: r.httpStatus, latencyMs: r.latencyMs, tries: i + 1 };
-    last = r;
-    if (i < attempts - 1) await sleep(gapMs);
-  }
-  return { status: 'down', httpStatus: last.httpStatus, latencyMs: last.latencyMs, tries: attempts };
-}
-
-export async function readHistory(file) {
-  if (!existsSync(file)) return [];
-  const text = await readFile(file, 'utf8');
-  const rows = [];
-  for (const line of text.split('\n')) {
-    const t = line.trim();
-    if (!t) continue;
-    try { rows.push(JSON.parse(t)); } catch (_) { /* skip corrupt line */ }
-  }
-  return rows;
-}
-
-export function uptimePct(rows, sinceMs) {
-  const inWindow = rows.filter((r) => Date.parse(r.t) >= sinceMs);
-  if (inWindow.length === 0) return null;
-  const up = inWindow.filter((r) => r.status === 'up').length;
-  return Math.round((up / inWindow.length) * 10000) / 100; // 2 d.p.
-}
-
-async function emitGithubOutput(kv) {
-  const f = process.env.GITHUB_OUTPUT;
-  if (!f) return;
-  const lines = Object.entries(kv).map(([k, v]) => `${k}=${v}`).join('\n') + '\n';
-  await writeFile(f, lines, { flag: 'a' });
-}
-
-async function main() {
-  const args = parseArgs(process.argv.slice(2));
-  const dir = path.resolve(args.dir);
-  await mkdir(dir, { recursive: true });
-  const historyFile = path.join(dir, 'history.jsonl');
-  const statusFile = path.join(dir, 'status.json');
-
-  // No target means the branding pack names no public origin and none was passed
-  // (task 1002757 — there is deliberately no founder-host fallback to probe).
-  if (!args.target) {
-    console.error('status-probe: no target. Pass --target <url>, or give this instance a domains.publicOrigin in config/branding.json.');
-    process.exitCode = 2;
-    return;
-  }
-
-  const result = await probe({
-    target: args.target,
-    attempts: args.attempts,
-    timeoutMs: args['timeout-ms'],
-    gapMs: args['gap-ms'],
-  });
-
-  const now = new Date();
-  const record = {
-    t: now.toISOString(),
-    status: result.status,
-    http_status: result.httpStatus,
-    latency_ms: result.latencyMs,
-    tries: result.tries,
-  };
-
-  // Append + prune to the trailing 90 days.
-  const cutoff = now.getTime() - PRUNE_DAYS * 24 * 60 * 60 * 1000;
-  const history = (await readHistory(historyFile)).filter((r) => Date.parse(r.t) >= cutoff);
-  history.push(record);
-  await writeFile(historyFile, history.map((r) => JSON.stringify(r)).join('\n') + '\n');
-
-  const dayMs = 24 * 60 * 60 * 1000;
-  const snapshot = {
-    generated_at: now.toISOString(),
-    target: args.target,
-    current: record,
-    summary: {
-      uptime_24h: uptimePct(history, now.getTime() - dayMs),
-      uptime_7d: uptimePct(history, now.getTime() - 7 * dayMs),
-      uptime_90d: uptimePct(history, now.getTime() - 90 * dayMs),
-      checks_total: history.length,
-    },
-  };
-  await writeFile(statusFile, JSON.stringify(snapshot, null, 2) + '\n');
-
-  await emitGithubOutput({
-    status: result.status,
-    http_status: result.httpStatus ?? '',
-    latency_ms: result.latencyMs ?? '',
-  });
-
-  const icon = result.status === 'up' ? '🟢' : '🔴';
-  console.log(
-    `${icon} ${result.status.toUpperCase()} — ${args.target} ` +
-    `(http=${result.httpStatus ?? 'n/a'}, latency=${result.latencyMs ?? 'n/a'}ms, tries=${result.tries})`
-  );
-  console.log(`   history: ${history.length} checks · 24h ${snapshot.summary.uptime_24h ?? 'n/a'}% · 90d ${snapshot.summary.uptime_90d ?? 'n/a'}%`);
-}
-
-// Run as a CLI only when invoked directly (not when imported by tests).
-const isMain = import.meta.url === pathToFileURL(process.argv[1] || '').href;
-if (isMain) {
-  main().catch((err) => {
-    console.error('[status-probe] internal error:', err.message);
-    process.exit(2);
-  });
-}

package/src/bongos/uptime-poller.js

@@ -1,93 +0,0 @@
-'use strict';
-
-// Uptime poller — runs inside the Node process, polls once per minute.
-// Writes results to uptime_checks. Prunes rows older than 90 days inline.
-//
-// Services checked:
-//   game_server — GET http://127.0.0.1:3000/healthz
-//   api         — GET http://127.0.0.1:3000/api/gds/healthz
-//   database    — direct Postgres query (SELECT 1)
-
-const http = require('node:http');
-const { pool } = require('./pool');
-// task 1003208: structured logging — was console.*. A core file logs through
-// the kernel logger directly; only MODULES go via api.logger (the doorway).
-const log = require('./logger').logger.child({ src: 'uptime-poller' });
-
-const INTERVAL_MS = 60_000;
-const TIMEOUT_MS  = 8_000;
-const PRUNE_DAYS  = 90;
-
-const SERVICES = [
-  {
-    name: 'game_server',
-    check: () => httpGet('http://127.0.0.1:3000/healthz'),
-  },
-  {
-    name: 'api',
-    check: () => httpGet('http://127.0.0.1:3000/api/gds/healthz'),
-  },
-  {
-    name: 'database',
-    check: () => dbPing(),
-  },
-];
-
-function httpGet(url) {
-  return new Promise((resolve) => {
-    const start = Date.now();
-    const req = http.get(url, { timeout: TIMEOUT_MS }, (res) => {
-      res.resume(); // drain
-      const latency = Date.now() - start;
-      const status = res.statusCode === 200 ? 'up'
-        : res.statusCode < 500 ? 'degraded'
-        : 'down';
-      resolve({ status, latency });
-    });
-    req.on('error', () => resolve({ status: 'down', latency: null }));
-    req.on('timeout', () => { req.destroy(); resolve({ status: 'down', latency: null }); });
-  });
-}
-
-async function dbPing() {
-  const start = Date.now();
-  try {
-    await pool.query('SELECT 1');
-    return { status: 'up', latency: Date.now() - start };
-  } catch (_) {
-    return { status: 'down', latency: null };
-  }
-}
-
-async function runChecks() {
-  for (const svc of SERVICES) {
-    try {
-      const { status, latency } = await svc.check();
-      await pool.query(
-        `INSERT INTO uptime_checks (service, status, latency_ms) VALUES ($1, $2, $3)`,
-        [svc.name, status, latency ?? null]
-      );
-    } catch (err) {
-      log.error(`[uptime] check failed for ${svc.name}:`, err.message);
-    }
-  }
-
-  // Prune old rows (fire-and-forget)
-  pool.query(
-    `DELETE FROM uptime_checks WHERE checked_at < now() - interval '${PRUNE_DAYS} days'`
-  ).catch((err) => log.error('[uptime] prune error:', err.message));
-}
-
-function start() {
-  // Stagger first check by 10s so the server finishes binding before we hit it
-  setTimeout(() => {
-    runChecks().catch((err) => log.error('[uptime] initial check error:', err.message));
-    setInterval(() => {
-      runChecks().catch((err) => log.error('[uptime] check error:', err.message));
-    }, INTERVAL_MS);
-  }, 10_000);
-
-  log.info('[uptime] poller started — checking every 60s');
-}
-
-module.exports = { start };