Suspend the NextGraph suites, and smoke the ActivityPods stack

scripts/validate.ts

@@ -1,190 +1,101 @@
 #!/usr/bin/env bun
 /**
- * validate.ts — Validation matrix (broker-level tests only, no @ui/@e2e).
+ * validate.ts — the Festipod validation matrix.
  *
- * Default run (no flags) — key subset only, fast:
- *   (a) Polyfill unit tests        (@ng-eventually/client — bun test)
- *   (b) Polyfill e2e real-broker   (@ng-eventually/client — bun run e2e/run.ts)
- *   (c) Festipod @data KEY SUBSET  (cucumber --name regex covering terrain bugs)
- *   (d) Festipod @multibrowser     (cucumber --tags @multibrowser)
- *   (e) Festipod @smoke            (cucumber --tags @smoke — boot connecté rend)
- *   (f) Festipod @wip [informational only, non-blocking]
+ * SCOPE (2026-10-09, after the ActivityPods-first switch). Two layers are in
+ * the gate:
  *
- * With --full flag:
- *   (c) becomes full @data suite   (cucumber --tags @data)
+ *   (a) Festipod @ui  — the UI layer. The `@ui` tag does not exist in this repo;
+ *       the layer is exactly the complement of the polyfill/broker tags, and its
+ *       command is, verbatim:
+ *         node --import tsx/esm node_modules/@cucumber/cucumber/bin/cucumber.js \
+ *           --config cucumber.json \
+ *           --tags 'not @wip and not @humain and not @data and not @e2e and not @multibrowser'
+ *       `cucumber.json` additionally excludes `@suspendu` and `@e2e-activitypods`
+ *       (Cucumber ANDs the config tags with the CLI ones), so this stays the
+ *       stack-free UI layer exactly as verified at 125 scenarios / 0 failed.
  *
- * Each step runs even if the previous one failed (--bail mode is OFF).
- * Exit code is non-zero if any non-informational step has failures.
+ *   (b) Festipod @e2e-activitypods — the smoke over the LOCAL ActivityPods stack
+ *       (pod provider + Festipod backend + the app served for real). Distinct
+ *       from the suspended `@e2e`: no wallet, no broker, so `cucumber.json`
+ *       excludes it. Its config is `cucumber.smoke.json`; its reproducible setup
+ *       is the concept note `data-layer`, `cookbook_local-activitypods-platform.md`.
+ *       Requires the stack to be UP (the app server and Chromium are started
+ *       here); an unreachable stack fails the layer rather than skipping it.
  *
- * Profile rotation: both Playwright profiles are rotated before @data and
- * @multibrowser when their size exceeds BLOAT_THRESHOLD_MB (default 50 MB),
- * to avoid the sparql_query hang described in caveat_wallet-bloat-hang.
+ * SUSPENDED, NOT RUN HERE (documented skip, code preserved — nothing deleted):
+ *   - `@data`, `@e2e`, `@multibrowser`, `@shared-wallet`, `@smoke` scenarios are
+ *     polyfill/broker-bound: they sign the shared NextGraph wallet through the
+ *     harness (`src/shared/test-harness/harness-ng.tsx`). They are tagged
+ *     `@suspendu` in the feature files and excluded there, awaiting the NextGraph
+ *     workstream.
+ *   - the SDK's own tests (nextgraph/ng-eventually-js) are NOT Festipod's and are
+ *     no longer run from here (concept `bdd-testing`,
+ *     `rule_tests-validate-festipod-not-the-sdk`).
+ *
+ * A smoke layer that cannot reach its stack FAILS, loudly — it never silently
+ * skips. Exit code is non-zero if any layer fails.
  */
 
 import { spawnSync } from "child_process";
-import * as fs from "fs";
-import * as path from "path";
 
 // ─── Config ────────────────────────────────────────────────────────────────
 
-const FESTIPOD_DIR = "/home/sylvain/projects/festipod/festipod";
-const POLYFILL_DIR =
-  "/home/sylvain/projects/nextgraph/ng-eventually-js/packages/client";
+const REPO_DIR = process.cwd();
 
-const FESTIPOD_PROFILE = path.join(FESTIPOD_DIR, ".playwright-profile");
-const POLYFILL_PROFILE = path.join(
-  POLYFILL_DIR,
-  "e2e",
-  ".playwright-profile-lib",
-);
+/** The UI layer, verbatim — see the header. */
+const UI_TAGS = "not @wip and not @humain and not @data and not @e2e and not @multibrowser";
 
-/** Rotate profile when it exceeds this many MB (caveat_wallet-bloat-hang). */
-const BLOAT_THRESHOLD_MB = 50;
+/** The ActivityPods smoke, minus its documented suspends. */
+const APODS_TAGS = "@e2e-activitypods and not @suspendu";
 
-/**
- * Per-step timeouts:
- * - polyfill unit/e2e: short steps, keep 10 min
- * - @data key subset: generous — BeforeAll + 8 scenarios, ~15 min margin
- * - @data full: full suite, ~35 min margin
- * - @multibrowser: 7 scenarios, ~10 min margin
- * - @wip: informational, 10 min
- */
-const TIMEOUT_POLYFILL_UNIT_MS = 10 * 60 * 1000;  // 10 min
-const TIMEOUT_POLYFILL_E2E_MS  = 10 * 60 * 1000;  // 10 min
-const TIMEOUT_DATA_KEY_MS      = 15 * 60 * 1000;  // 15 min (key subset)
-const TIMEOUT_DATA_FULL_MS     = 35 * 60 * 1000;  // 35 min (--full)
-const TIMEOUT_MULTIBROWSER_MS  = 10 * 60 * 1000;  // 10 min
-const TIMEOUT_SMOKE_MS         = 10 * 60 * 1000;  // 10 min (1 @e2e boot scenario)
-const TIMEOUT_WIP_MS           = 10 * 60 * 1000;  // 10 min
-
-/**
- * Key-subset --name regex: matches exactly the 8 scenarios that cover the
- * known terrain bugs (inscription, désinscription, isolation, reconnexion,
- * compteur dérivé, créateur ne participe pas, auth vide, auth distinctes).
- *
- * Uses a single cucumber invocation so BeforeAll (broker login) runs once.
- *
- * French accent chars must be URL-safe in the regex — cucumber uses JS
- * RegExp, which handles unicode natively; we pass the literal string.
- */
-const DATA_KEY_NAME_REGEX = [
-  "S'inscrire à un événement",
-  "Se désinscrire d'un événement$",
-  "Une identité fraîche ne voit pas la participation d'une autre",
-  "Une page fraîche pour la même identité relit ses propres données",
-  "Le créateur ne participe pas automatiquement à son événement",
-  "L'inscription fait converger le compteur dérivé du propriétaire",
-  "Un portefeuille connecté est vide par défaut",
-  "Les données du portefeuille sont distinctes des données par défaut",
-].join("|");
+const TIMEOUT_UI_MS = 15 * 60 * 1000; // 15 min
+const TIMEOUT_APODS_MS = 10 * 60 * 1000; // 10 min
 
 // ─── Helpers ───────────────────────────────────────────────────────────────
 
-function dirSizeMB(dir: string): number {
-  if (!fs.existsSync(dir)) return 0;
-  try {
-    const result = spawnSync("du", ["-sm", dir], { encoding: "utf-8" });
-    const line = result.stdout.trim().split("\n")[0] ?? "";
-    return parseInt(line.split("\t")[0] ?? "0", 10);
-  } catch {
-    return 0;
-  }
-}
-
-/**
- * Remove any stale Chromium singleton files from `profilePath`. Chromium refuses
- * to launch (ProcessSingleton error) if SingletonLock, SingletonCookie, or
- * SingletonSocket are left over from a previous crashed run. Idempotent — safe to
- * call even when the profile does not exist yet.
- */
-function cleanSingletons(profilePath: string, label: string): void {
-  if (!fs.existsSync(profilePath)) return;
-  const singletons = ["SingletonLock", "SingletonCookie", "SingletonSocket"];
-  for (const name of singletons) {
-    const p = path.join(profilePath, name);
-    if (fs.existsSync(p)) {
-      try {
-        fs.rmSync(p, { force: true });
-        console.log(`[rotate] ${label}: removed stale ${name}.`);
-      } catch {
-        // Non-fatal: if we can't remove it, launch will fail with a clear error
-      }
-    }
-  }
-}
-
-function rotateProfile(profilePath: string, label: string): void {
-  const sizeMB = dirSizeMB(profilePath);
-  if (sizeMB > BLOAT_THRESHOLD_MB) {
-    console.log(
-      `[rotate] ${label}: ${sizeMB}MB > ${BLOAT_THRESHOLD_MB}MB — rotating profile...`,
-    );
-    try {
-      fs.rmSync(profilePath, { recursive: true, force: true });
-      console.log(`[rotate] ${label}: profile removed. Will be recreated.`);
-    } catch (e) {
-      console.warn(`[rotate] ${label}: failed to remove profile: ${e}`);
-    }
-  } else {
-    // Even if we keep the profile, remove any stale Chromium singleton files left
-    // by a previous crashed run — Chromium refuses to launch if they exist.
-    cleanSingletons(profilePath, label);
-    console.log(
-      `[rotate] ${label}: ${sizeMB}MB — below threshold, keeping profile.`,
-    );
-  }
-}
-
 interface StepResult {
   label: string;
   status: "passed" | "failed" | "error";
-  /** Lines to show in the summary (failed scenario names, FAIL lines, etc.) */
   failures: string[];
-  /** Raw exit code */
   exitCode: number;
   durationMs: number;
 }
 
-/**
- * Run a command and capture its output. Returns the result with parsed
- * pass/fail summary. Never throws — all errors are captured in StepResult.
- */
-function runStep(
-  label: string,
-  cmd: string,
-  args: string[],
-  cwd: string,
-  timeoutMs: number,
-  extraEnv: Record<string, string> = {},
-): StepResult {
+/** The Cucumber invocation every layer shares (same form as the UI gate command). */
+function cucumberArgs(config: string, tags: string): string[] {
+  return [
+    "--import",
+    "tsx/esm",
+    "node_modules/@cucumber/cucumber/bin/cucumber.js",
+    "--config",
+    config,
+    "--tags",
+    tags,
+  ];
+}
+
+function runStep(label: string, args: string[], timeoutMs: number): StepResult {
   const t0 = Date.now();
-  console.log(`\n${"═".repeat(60)}`);
+  console.log(`\n${"═".repeat(64)}`);
   console.log(`▶  ${label}`);
-  console.log(`   ${cmd} ${args.join(" ")}  (cwd: ${cwd})`);
-  console.log(`   timeout: ${Math.round(timeoutMs / 60000)}min`);
-  console.log(`${"═".repeat(60)}`);
+  console.log(`   node ${args.join(" ")}`);
+  console.log(`${"═".repeat(64)}`);
 
-  const env = { ...process.env, ...extraEnv };
-
-  const result = spawnSync(cmd, args, {
-    cwd,
-    env,
+  const result = spawnSync("node", args, {
+    cwd: REPO_DIR,
+    env: process.env,
     encoding: "utf-8",
     timeout: timeoutMs,
-    maxBuffer: 20 * 1024 * 1024, // 20MB
+    maxBuffer: 20 * 1024 * 1024,
   });
 
   const durationMs = Date.now() - t0;
-  const stdout = result.stdout ?? "";
-  const stderr = result.stderr ?? "";
-  const combined = stdout + "\n" + stderr;
-
-  // Print output in real-time equivalent (post-hoc since spawnSync)
-  if (stdout) process.stdout.write(stdout);
-  if (stderr) process.stderr.write(stderr);
+  const combined = (result.stdout ?? "") + "\n" + (result.stderr ?? "");
+  process.stdout.write(result.stdout ?? "");
+  process.stderr.write(result.stderr ?? "");
 
   if (result.error) {
-    console.error(`[${label}] process error:`, result.error.message);
     return {
       label,
       status: "error",
@@ -193,298 +104,66 @@ function runStep(
       durationMs,
     };
   }
-
-  const exitCode = result.status ?? 1;
-  const failures = extractFailures(combined, label);
-  const status = exitCode === 0 ? "passed" : "failed";
-
-  return { label, status, failures, exitCode, durationMs };
+  return {
+    label,
+    status: (result.status ?? 1) === 0 ? "passed" : "failed",
+    failures: extractFailures(combined),
+    exitCode: result.status ?? 1,
+    durationMs,
+  };
 }
 
-/**
- * Extract meaningful failure lines from combined stdout+stderr.
- * Heuristics per step type (cucumber scenario names, FAIL lines, etc.).
- */
-function extractFailures(output: string, label: string): string[] {
+/** The failure lines worth surfacing: failed scenario names and the summary. */
+function extractFailures(output: string): string[] {
   const lines = output.split("\n");
   const failures: string[] = [];
-
-  if (label.includes("polyfill:unit")) {
-    // bun test output: lines starting with "✗" or "FAIL" or "fail"
-    for (const line of lines) {
-      const l = line.trim();
-      if (/^(✗|✕|FAIL|fail)\s/.test(l) || l.includes("tests failed")) {
-        failures.push(l);
-      }
-    }
-    // Also capture summary line "N passed, M failed"
-    const summary = lines.find(
-      (l) => l.includes("passed") && l.includes("failed"),
-    );
-    if (summary) failures.push(summary.trim());
-  } else if (label.includes("polyfill:e2e")) {
-    // e2e/run.ts output: lines starting with "  [FAIL]"
-    for (const line of lines) {
-      const l = line.trim();
-      if (l.startsWith("[FAIL]")) failures.push(l);
-    }
-    // Summary: "N passed / M failed" style
-    const summary = lines.find(
-      (l) => l.includes("passed") || l.includes("failed"),
-    );
-    if (summary && !failures.includes(summary.trim()))
-      failures.push(summary.trim());
-  } else {
-    // Cucumber steps: look for "✗" scenario lines, "FAILED" scenario names,
-    // or lines beginning with "✖" / "×" / "Scenario:" after a failure tag
-    for (const line of lines) {
-      const l = line.trim();
-      if (
-        /^(✗|✕|×|✖)\s/.test(l) ||
-        l.startsWith("✘") ||
-        l.includes("# Scénario:") ||
-        l.includes("# Scenario:") ||
-        (l.startsWith("F") && l.length === 1) // progress-bar failure tick
-      ) {
-        if (l.length > 1) failures.push(l);
-      }
-    }
-    // Cucumber "N scenarios (M failed)" summary
-    const summary = lines.find((l) =>
-      /\d+ sc[eé]narios?.*(failed|undefined)/.test(l),
-    );
-    if (summary) failures.push(summary.trim());
-    // Individual scenario fail lines: "✗ Scenario name (features/...)"
-    for (const line of lines) {
-      const l = line.trim();
-      if (l.startsWith("✗") || l.startsWith("✕")) {
-        if (!failures.includes(l)) failures.push(l);
-      }
+  for (const line of lines) {
+    const l = line.trim();
+    if (/^(✗|✕|×|✖|✘)\s/.test(l)) failures.push(l);
+    if (/\d+ sc[eé]narios?.*(failed|undefined)/.test(l) && !failures.includes(l)) {
+      failures.push(l);
     }
   }
-
   return failures.filter(Boolean);
 }
 
 function fmtDuration(ms: number): string {
   if (ms < 60_000) return `${(ms / 1000).toFixed(1)}s`;
-  const m = Math.floor(ms / 60_000);
-  const s = ((ms % 60_000) / 1000).toFixed(0);
-  return `${m}m${s}s`;
+  return `${Math.floor(ms / 60_000)}m${((ms % 60_000) / 1000).toFixed(0)}s`;
 }
 
-function printMatrix(
-  steps: StepResult[],
-  wipResult: StepResult | null,
-  fullMode: boolean,
-): void {
+function printMatrix(steps: StepResult[]): void {
   console.log("\n");
   console.log("╔══════════════════════════════════════════════════════════╗");
-  console.log("║                  VALIDATION MATRIX                      ║");
-  if (fullMode) {
-    console.log("║              (mode: --full, @data complet)               ║");
-  } else {
-    console.log("║           (mode: défaut, sous-ensemble clé)              ║");
-  }
-  console.log("╚══════════════════════════════════════════════════════════╝");
-  console.log("");
-
+  console.log("║                  VALIDATION MATRIX                       ║");
+  console.log("╚══════════════════════════════════════════════════════════╝\n");
   const maxLabel = Math.max(...steps.map((s) => s.label.length));
-
   for (const step of steps) {
     const icon = step.status === "passed" ? "✅" : step.status === "failed" ? "❌" : "⚠️ ";
-    const pad = step.label.padEnd(maxLabel + 2);
-    console.log(`  ${icon}  ${pad}  [${fmtDuration(step.durationMs)}]`);
-    for (const f of step.failures) {
-      console.log(`       ↳ ${f}`);
-    }
-  }
-
-  if (wipResult) {
-    console.log("");
-    console.log("  ── @wip (informational, non-blocking) ──────────────────");
-    const icon =
-      wipResult.status === "passed"
-        ? "✅"
-        : wipResult.status === "failed"
-          ? "❌"
-          : "⚠️ ";
-    const pad = wipResult.label.padEnd(maxLabel + 2);
-    console.log(`  ${icon}  ${pad}  [${fmtDuration(wipResult.durationMs)}]`);
-    for (const f of wipResult.failures) {
-      console.log(`       ↳ ${f}`);
-    }
-  }
-
-  console.log("");
-  const allPassed = steps.every((s) => s.status === "passed");
-  const totalMs = steps.reduce((sum, s) => sum + s.durationMs, 0) +
-    (wipResult?.durationMs ?? 0);
-  if (allPassed) {
-    console.log("  🟢  ALL STEPS PASSED");
-  } else {
-    const failed = steps.filter((s) => s.status !== "passed");
-    console.log(`  🔴  ${failed.length} STEP(S) FAILED: ${failed.map((s) => s.label).join(", ")}`);
-  }
-  console.log(`  ⏱   Total: ${fmtDuration(totalMs)}`);
-  if (!fullMode) {
-    console.log("  ℹ️   Pour @data complet : bun run validate -- --full");
+    console.log(`  ${icon}  ${step.label.padEnd(maxLabel + 2)}  [${fmtDuration(step.durationMs)}]`);
+    for (const f of step.failures) console.log(`       ↳ ${f}`);
   }
   console.log("");
-}
-
-// ─── Cucumber command builder ───────────────────────────────────────────────
-
-function cucumberArgsByTags(tags: string): string[] {
-  return [
-    "--import",
-    "tsx/esm",
-    "node_modules/.bin/cucumber-js",
-    "--config",
-    "cucumber.json",
-    "--tags",
-    tags,
-  ];
-}
-
-function cucumberArgsByName(nameRegex: string): string[] {
-  return [
-    "--import",
-    "tsx/esm",
-    "node_modules/.bin/cucumber-js",
-    "--config",
-    "cucumber.json",
-    "--tags",
-    "@data",
-    "--name",
-    nameRegex,
-  ];
+  console.log("  ℹ️   @data/@e2e/@multibrowser/@shared-wallet/@smoke : suspendus (workstream NextGraph).");
+  const failed = steps.filter((s) => s.status !== "passed");
+  console.log(failed.length === 0
+    ? "  🟢  ALL LAYERS PASSED"
+    : `  🔴  ${failed.length} LAYER(S) FAILED: ${failed.map((s) => s.label).join(", ")}`);
+  console.log(`  ⏱   Total: ${fmtDuration(steps.reduce((sum, s) => sum + s.durationMs, 0))}`);
+  console.log("");
 }
 
 // ─── Main ──────────────────────────────────────────────────────────────────
 
-async function main(): Promise<void> {
-  const args = process.argv.slice(2);
-  const fullMode = args.includes("--full");
-
-  if (fullMode) {
-    console.log("🔍  Festipod — Full Validation Run (--full : @data complet)");
-  } else {
-    console.log("🔍  Festipod — Validation Run (sous-ensemble clé)");
-    console.log("    Pour @data complet : bun run validate -- --full");
-  }
-  console.log(`    Festipod: ${FESTIPOD_DIR}`);
-  console.log(`    Polyfill: ${POLYFILL_DIR}`);
-  console.log("");
-
-  // ── Profile rotation AVANT les étapes broker ──────────────────────────────
-  console.log("── Profile rotation check (avant @data et @multibrowser) ────");
-  rotateProfile(FESTIPOD_PROFILE, "festipod");
-  rotateProfile(POLYFILL_PROFILE, "polyfill-lib");
+function main(): void {
+  console.log(`🔍  Festipod — validation (repo: ${REPO_DIR})`);
 
   const steps: StepResult[] = [];
+  steps.push(runStep("festipod:@ui", cucumberArgs("cucumber.json", UI_TAGS), TIMEOUT_UI_MS));
+  steps.push(runStep("festipod:@e2e-activitypods", cucumberArgs("cucumber.smoke.json", APODS_TAGS), TIMEOUT_APODS_MS));
 
-  // ── (a) Polyfill unit tests ───────────────────────────────────────────────
-  steps.push(
-    runStep(
-      "polyfill:unit",
-      "bun",
-      ["test"],
-      POLYFILL_DIR,
-      TIMEOUT_POLYFILL_UNIT_MS,
-    ),
-  );
-
-  // ── (b) Polyfill e2e real broker ──────────────────────────────────────────
-  // Clean singleton files immediately before launching Chromium — guards against
-  // any file left by polyfill:unit (unlikely but defensive) or by a previous
-  // interrupted run that the initial rotateProfile call ran before.
-  cleanSingletons(POLYFILL_PROFILE, "polyfill-lib (pre-e2e)");
-  steps.push(
-    runStep(
-      "polyfill:e2e",
-      "bun",
-      ["run", "e2e/run.ts"],
-      POLYFILL_DIR,
-      TIMEOUT_POLYFILL_E2E_MS,
-    ),
-  );
-
-  // ── (c) Festipod @data ────────────────────────────────────────────────────
-  if (fullMode) {
-    // --full : lance tout @data
-    steps.push(
-      runStep(
-        "festipod:@data (complet)",
-        "node",
-        cucumberArgsByTags("@data"),
-        FESTIPOD_DIR,
-        TIMEOUT_DATA_FULL_MS,
-      ),
-    );
-  } else {
-    // défaut : sous-ensemble clé en UNE invocation (BeforeAll partagé)
-    steps.push(
-      runStep(
-        "festipod:@data (clé)",
-        "node",
-        cucumberArgsByName(DATA_KEY_NAME_REGEX),
-        FESTIPOD_DIR,
-        TIMEOUT_DATA_KEY_MS,
-      ),
-    );
-  }
-
-  // ── (d) Festipod @multibrowser ────────────────────────────────────────────
-  // Exclude @wip: a scenario tagged @wip @multibrowser (e.g. the reactive
-  // cross-session scenario, blocked by the shared-wallet structural limit) must
-  // not gate the baseline — it flows into the informational @wip pass below.
-  steps.push(
-    runStep(
-      "festipod:@multibrowser",
-      "node",
-      cucumberArgsByTags("@multibrowser and not @wip"),
-      FESTIPOD_DIR,
-      TIMEOUT_MULTIBROWSER_MS,
-    ),
-  );
-
-  // ── (e) Festipod @smoke — boot connecté rend / page blanche ───────────────
-  // Un seul scénario @e2e : boote le VRAI App, se connecte au broker, et vérifie
-  // que l'accueil connecté rend du contenu d'app réel SANS erreur runtime. Garde
-  // la CLASSE « crash de rendu une fois connecté » (page blanche). On ne lance
-  // QUE @smoke (pas tout @e2e) pour garder le run par défaut rapide.
-  // Nettoie les singletons Chromium juste avant, comme les autres étapes broker.
-  cleanSingletons(FESTIPOD_PROFILE, "festipod (pre-@smoke)");
-  steps.push(
-    runStep(
-      "festipod:@smoke",
-      "node",
-      cucumberArgsByTags("@smoke and not @wip"),
-      FESTIPOD_DIR,
-      TIMEOUT_SMOKE_MS,
-    ),
-  );
-
-  // ── (f) Festipod @wip [informational] ────────────────────────────────────
-  console.log("\n── @wip informational pass (non-blocking) ──────────────────");
-  const wipResult = runStep(
-    "festipod:@wip",
-    "node",
-    cucumberArgsByTags("@wip"),
-    FESTIPOD_DIR,
-    TIMEOUT_WIP_MS,
-  );
-
-  // ── Matrix ────────────────────────────────────────────────────────────────
-  printMatrix(steps, wipResult, fullMode);
-
-  // ── Exit code ─────────────────────────────────────────────────────────────
-  const anyFailed = steps.some((s) => s.status !== "passed");
-  process.exit(anyFailed ? 1 : 0);
+  printMatrix(steps);
+  process.exit(steps.some((s) => s.status !== "passed") ? 1 : 0);
 }
 
-main().catch((e) => {
-  console.error("validate.ts: unhandled error:", e);
-  process.exit(1);
-});
+main();