@anthusai/papyrus 2.0.0-next.5 → 2.0.0-next.50

package/lib/reader-moderation.ts

@@ -0,0 +1,220 @@
+/**
+ * The data side of the Administration "Moderation" panel for the reader-steering plugin: rows from the staff-readable
+ * models, shaped into a queue, a hidden list and the recent action log. Pure functions, no network: the panel loads the
+ * rows and calls the `moderateReader` mutation; everything it shows is decided here so it can be tested.
+ */
+
+export type ModerationTarget = "submission" | "comment" | "reader";
+export type ModerationActionName = "approve" | "hide" | "unhide" | "ban" | "unban" | "limit" | "promote" | "undo";
+
+export type SubmissionRow = {
+  id: string;
+  canonicalUrl: string;
+  domain: string;
+  title: string;
+  kind?: string | null;
+  note?: string | null;
+  readerId: string;
+  state: string;
+  moderation?: string | null;
+  reportCount?: number | null;
+  signals?: unknown;
+  itemId?: string | null;
+  createdAt: string;
+};
+
+export type CommentRow = {
+  id: string;
+  subjectKey: string;
+  readerId: string;
+  body: string;
+  parentId?: string | null;
+  state: string;
+  moderation?: string | null;
+  reportCount?: number | null;
+  signals?: unknown;
+  createdAt: string;
+};
+
+export type ReportRow = { id: string; targetType: string; target: string; reporterReaderId: string; reason: string; state: string; createdAt: string };
+
+export type ActionRow = {
+  id: string;
+  by: string;
+  actorSub: string;
+  action: string;
+  targetType: string;
+  target: string;
+  reason?: string | null;
+  fromState: string;
+  toState: string;
+  undoOf?: string | null;
+  undoneBy?: string | null;
+  at: string;
+};
+
+export type Signal = { source: string; policy: string; detector: string; severity: string; excerpt: string };
+
+export type ModerationRow = {
+  kind: "submission" | "comment";
+  id: string;
+  readerId: string;
+  handle: string | null;
+  createdAt: string;
+  state: string;
+  title: string;
+  url: string | null;
+  text: string;
+  subject: string | null;
+  reports: ReportRow[];
+  signals: Signal[];
+  canPromote: boolean;
+};
+
+export type ActionEntry = ActionRow & { label: string; undoable: boolean; byLabel: string };
+
+const SEVERITY_ORDER: Record<string, number> = { error: 0, warn: 1, info: 2 };
+
+export function parseSignals(value: unknown): Signal[] {
+  let parsed: unknown = value;
+  if (typeof value === "string") {
+    try {
+      parsed = JSON.parse(value);
+    } catch {
+      return [];
+    }
+  }
+  if (!Array.isArray(parsed)) return [];
+  return parsed
+    .filter((entry): entry is Record<string, unknown> => typeof entry === "object" && entry !== null)
+    .map((entry) => ({
+      source: String(entry.source ?? ""),
+      policy: String(entry.policy ?? ""),
+      detector: String(entry.detector ?? ""),
+      severity: String(entry.severity ?? "info"),
+      excerpt: String(entry.excerpt ?? ""),
+    }))
+    .sort((left, right) => (SEVERITY_ORDER[left.severity] ?? 3) - (SEVERITY_ORDER[right.severity] ?? 3));
+}
+
+/** An address a moderator may follow: http or https only, never `javascript:` or `data:`. */
+export function safeHttpUrl(value: string | null | undefined): string | null {
+  if (!value) return null;
+  try {
+    const url = new URL(value);
+    return url.protocol === "http:" || url.protocol === "https:" ? url.href : null;
+  } catch {
+    return null;
+  }
+}
+
+function newestFirst<T extends { createdAt: string; id: string }>(rows: T[]): T[] {
+  return [...rows].sort((left, right) => (left.createdAt === right.createdAt ? right.id.localeCompare(left.id) : right.createdAt.localeCompare(left.createdAt)));
+}
+
+function reportsByTarget(reports: ReportRow[]): Map<string, ReportRow[]> {
+  const grouped = new Map<string, ReportRow[]>();
+  for (const report of reports) {
+    if (report.state !== "open") continue;
+    grouped.set(report.target, [...(grouped.get(report.target) ?? []), report]);
+  }
+  return grouped;
+}
+
+export function submissionRow(row: SubmissionRow, handles: Record<string, string>, reports: Map<string, ReportRow[]>): ModerationRow {
+  return {
+    kind: "submission",
+    id: row.id,
+    readerId: row.readerId,
+    handle: handles[row.readerId] ?? null,
+    createdAt: row.createdAt,
+    state: row.state,
+    title: row.title,
+    url: safeHttpUrl(row.canonicalUrl),
+    text: row.note ?? "",
+    subject: row.itemId ?? row.id,
+    reports: reports.get(row.id) ?? [],
+    signals: parseSignals(row.signals),
+    canPromote: row.state === "pending",
+  };
+}
+
+export function commentRow(row: CommentRow, handles: Record<string, string>, reports: Map<string, ReportRow[]>): ModerationRow {
+  return {
+    kind: "comment",
+    id: row.id,
+    readerId: row.readerId,
+    handle: handles[row.readerId] ?? null,
+    createdAt: row.createdAt,
+    state: row.state,
+    title: "",
+    url: null,
+    text: row.body,
+    subject: row.subjectKey,
+    reports: reports.get(row.id) ?? [],
+    signals: parseSignals(row.signals),
+    canPromote: false,
+  };
+}
+
+/** Everything waiting for a decision: new items and reported ones, newest first. Deleted comments are the author's own and are not in it. */
+export function buildQueue(input: { submissions: SubmissionRow[]; comments: CommentRow[]; reports: ReportRow[]; handles: Record<string, string> }): ModerationRow[] {
+  const reports = reportsByTarget(input.reports);
+  const rows = [
+    ...input.submissions.filter((row) => (row.moderation ?? "new") === "new").map((row) => submissionRow(row, input.handles, reports)),
+    ...input.comments.filter((row) => (row.moderation ?? "new") === "new" && row.state !== "deleted").map((row) => commentRow(row, input.handles, reports)),
+  ];
+  return newestFirst(rows);
+}
+
+/** Items a moderator or a pre-moderation hold keeps out of the public files, newest first, so they can be brought back. */
+export function buildHidden(input: { submissions: SubmissionRow[]; comments: CommentRow[]; reports: ReportRow[]; handles: Record<string, string> }): ModerationRow[] {
+  const reports = reportsByTarget(input.reports);
+  return newestFirst([
+    ...input.submissions.filter((row) => row.state === "hidden").map((row) => submissionRow(row, input.handles, reports)),
+    ...input.comments.filter((row) => row.state === "hidden").map((row) => commentRow(row, input.handles, reports)),
+  ]);
+}
+
+const ACTION_LABELS: Record<string, string> = {
+  approve: "Approved",
+  hide: "Hidden",
+  unhide: "Shown again",
+  ban: "Banned",
+  unban: "Unbanned",
+  limit: "Limited",
+  promote: "Promoted to a draft",
+  undo: "Undid an action",
+};
+
+/** The recent log, newest first. An action can be undone while nothing has undone it, and a promotion never. */
+export function buildActionLog(actions: ActionRow[], limit = 50): ActionEntry[] {
+  return [...actions]
+    .sort((left, right) => (left.at === right.at ? right.id.localeCompare(left.id) : right.at.localeCompare(left.at)))
+    .slice(0, limit)
+    .map((entry) => ({
+      ...entry,
+      label: ACTION_LABELS[entry.action] ?? entry.action,
+      byLabel: entry.by === "ai" ? "AI moderator" : "A moderator",
+      undoable: !entry.undoneBy && entry.action !== "promote",
+    }));
+}
+
+export type Receipt = { status?: string | null; id?: string | null; error?: string | null; message?: string | null };
+
+export function receiptMessage(receipt: Receipt | null | undefined): string {
+  if (!receipt) return "The action did not answer. Reload and check.";
+  if (receipt.status === "ok") return "Done.";
+  return receipt.message || "The action was refused.";
+}
+
+export function moderationModeText(mode: unknown): string {
+  if (mode === "off") return "Moderation is off for this site: nothing waits in a queue. You can still hide content and ban readers.";
+  if (mode === "human") return "New submissions and comments are public at once and wait here for a decision, unless this site holds them until approved.";
+  if (mode === "ai") return "An AI moderator reads each new submission and comment and allows or hides it, with a reason in the log. Its decisions are public. You can undo any of them, and what you decide it never judges again.";
+  return "Moderation is on.";
+}
+
+export function shortReader(readerId: string, handle: string | null): string {
+  return handle ? `${handle} (${readerId.slice(0, 8)})` : `reader ${readerId.slice(0, 8)}`;
+}

package/lib/reader-retirement-route.ts

@@ -0,0 +1,89 @@
+import { NextResponse, type NextRequest } from "next/server";
+import { GetParameterCommand, SSMClient } from "@aws-sdk/client-ssm";
+import { SiteOriginNotConfiguredError, publicSiteOrigin, siteOriginNotConfiguredResponse } from "./site-origin";
+import { resolveProspectusSignInRuntime, type ProspectusSignInRuntime } from "./prospectus-session";
+import { ProspectusAccountsError } from "./prospectus-site-identity";
+import { secretsMatch } from "./revalidate-secret";
+
+export const READER_POLL_SECRET_PARAMETER_ENV = "PAPYRUS_READER_POLL_SECRET_PARAMETER";
+export const READER_POLL_SECRET_HEADER = "x-papyrus-reader-poll-secret";
+export const READER_POLL_SECRET_CACHE_MILLISECONDS = 5 * 60 * 1000;
+const CURSOR_PATTERN = /^[A-Za-z0-9._~=+/-]{1,512}$/;
+const NO_STORE_HEADERS = { "cache-control": "no-store" };
+
+export interface ReaderRetirementRouteOptions {
+  environment?: Record<string, string | undefined>;
+  readSecret?: (parameterName: string) => Promise<string | null>;
+  resolveRuntime?: (siteOrigin: string) => Promise<Pick<ProspectusSignInRuntime, "accounts"> | null>;
+  now?: number;
+}
+
+type CachedSecret = { parameterName: string; value: string; expiresAt: number };
+let cachedSecret: CachedSecret | null = null;
+let defaultSsmClient: SSMClient | null = null;
+
+export function resetReaderPollSecretCache(): void {
+  cachedSecret = null;
+}
+
+async function readSecretFromSsm(parameterName: string): Promise<string | null> {
+  defaultSsmClient ??= new SSMClient({});
+  const response = await defaultSsmClient.send(new GetParameterCommand({ Name: parameterName, WithDecryption: true }));
+  return response.Parameter?.Value?.trim() || null;
+}
+
+async function loadReaderPollSecret(options: ReaderRetirementRouteOptions, environment: Record<string, string | undefined>): Promise<string | null> {
+  const parameterName = environment[READER_POLL_SECRET_PARAMETER_ENV]?.trim();
+  if (!parameterName) return null;
+  const now = options.now ?? Date.now();
+  if (cachedSecret && cachedSecret.parameterName === parameterName && cachedSecret.expiresAt > now) return cachedSecret.value;
+  try {
+    const value = (await (options.readSecret ?? readSecretFromSsm)(parameterName))?.trim();
+    if (!value) return null;
+    cachedSecret = { parameterName, value, expiresAt: now + READER_POLL_SECRET_CACHE_MILLISECONDS };
+    return value;
+  } catch (error) {
+    console.error(`Reader poll secret could not be read (${error instanceof Error ? error.name : "error"}).`);
+    return null;
+  }
+}
+
+function failure(status: number, error: string, message: string): NextResponse {
+  return NextResponse.json({ error, message }, { status, headers: NO_STORE_HEADERS });
+}
+
+/**
+ * The scheduled snapshot run's way to Prospectus's retired-subject route. Prospectus admits one caller per site, this
+ * app's compute role, so the run cannot call it directly: it calls this route with a shared secret from SSM, and the route
+ * calls Prospectus as the compute role and passes the answer through. Anyone without the secret gets 401, and a site
+ * without the secret parameter gets 503, so this is never open.
+ */
+export async function handleReaderRetiredSubjects(request: NextRequest, options: ReaderRetirementRouteOptions = {}): Promise<NextResponse> {
+  const environment = options.environment ?? process.env;
+  const expected = await loadReaderPollSecret(options, environment);
+  if (!expected) return failure(503, "reader_poll_unavailable", "The retirement poll is not configured for this site.");
+  if (!secretsMatch(request.headers.get(READER_POLL_SECRET_HEADER), expected)) return failure(401, "unauthorized", "Missing or wrong poll secret.");
+
+  const since = request.nextUrl.searchParams.get("since") ?? undefined;
+  const unknownParameters = [...request.nextUrl.searchParams.keys()].filter((key) => key !== "since");
+  if (unknownParameters.length > 0 || (since !== undefined && !CURSOR_PATTERN.test(since))) {
+    return failure(400, "invalid_since", "Send only a since cursor.");
+  }
+  let siteOrigin: string;
+  try {
+    siteOrigin = publicSiteOrigin(request, environment);
+  } catch (error) {
+    if (error instanceof SiteOriginNotConfiguredError) return siteOriginNotConfiguredResponse(error);
+    throw error;
+  }
+  const runtime = await (options.resolveRuntime ?? ((origin) => resolveProspectusSignInRuntime(origin, { environment })))(siteOrigin);
+  if (!runtime) return failure(503, "prospectus_sign_in_unavailable", "Prospectus is not configured for this site.");
+  try {
+    const page = await runtime.accounts.retiredSiteSubjects(since);
+    return NextResponse.json({ retired: page.retired, cursor: page.cursor, more: page.more }, { headers: NO_STORE_HEADERS });
+  } catch (error) {
+    if (error instanceof ProspectusAccountsError) return failure(error.status === 400 ? 400 : 502, error.code, error.message);
+    console.error(`Prospectus retired subjects failed (${error instanceof Error ? error.name : "error"}).`);
+    return failure(502, "prospectus_unavailable", "Prospectus could not be reached.");
+  }
+}

package/lib/reader-subject.md

@@ -0,0 +1,29 @@
+# Reader subject
+
+`readerSubject(session)` in `reader-subject.ts` is the only place a reader's identity for the reader API is
+decided. It returns the per-site pairwise `siteSubject` (`ss1_` plus 43 base64url characters) that Prospectus
+puts on `person.siteSubject`, or `null`.
+
+## Production
+
+Only the `siteSubject` the sealed session cookie carries counts. While Prospectus does not return one, the
+accessor returns `null`, every reader route answers `503 reader_identity_unavailable`, and the reader features
+stay off. There is no other flag. The global Prospectus person id is never used as a reader key.
+
+## Development and test only
+
+When `SITE_ENV` is `development` or `test`, `NODE_ENV` is not `production` and `PAPYRUS_READER_DEV_SECRET`
+holds at least 16 characters, a session without a `siteSubject` gets the interim value
+`ssdev_` plus `base64url(HMAC-SHA-256(devSecret, personId))`.
+
+That path never runs in production. No remap from `ssdev_` values to `ss1_` values is ever needed: production
+reader writes stay closed until `siteSubject` is live, so no production row is ever keyed on an interim value.
+The `reader-actions` function also refuses `ssdev_` subjects unless the site was synthesized with a
+development or test `SITE_ENV`.
+
+## Staff
+
+`readerIsStaff(session)` reads the `cognito:groups` claim from the access token inside the sealed session
+cookie (`editor` or `admin`). The cookie is written by the site's own callback route, so the token is trusted
+the way the rest of the session is. The reader API passes the result to the function as `isStaff`; clients never
+send it.

package/lib/reader-subject.ts

@@ -0,0 +1,53 @@
+import { createHmac } from "node:crypto";
+import type { ProspectusSignedInSession } from "./prospectus-session";
+
+export const SITE_SUBJECT_PATTERN = /^ss1_[A-Za-z0-9_-]{43}$/;
+export const INTERIM_SUBJECT_PATTERN = /^ssdev_[A-Za-z0-9_-]{43}$/;
+export const READER_DEV_SECRET_ENV = "PAPYRUS_READER_DEV_SECRET";
+export const READER_STAFF_GROUPS = ["editor", "admin"] as const;
+
+const MINIMUM_DEV_SECRET_LENGTH = 16;
+const INTERIM_SUBJECT_PREFIX = "ssdev_";
+const INTERIM_ENVIRONMENTS = ["development", "test"];
+
+type EnvironmentVariables = Record<string, string | undefined>;
+
+export function interimSubjectsAllowed(environment: EnvironmentVariables = process.env): boolean {
+  if (environment.NODE_ENV === "production") return false;
+  const siteEnvironment = (environment.SITE_ENV ?? "").trim();
+  if (!INTERIM_ENVIRONMENTS.includes(siteEnvironment)) return false;
+  return (environment[READER_DEV_SECRET_ENV] ?? "").length >= MINIMUM_DEV_SECRET_LENGTH;
+}
+
+export function interimSubject(personId: string, devSecret: string): string {
+  return INTERIM_SUBJECT_PREFIX + createHmac("sha256", devSecret).update(personId).digest("base64url");
+}
+
+/**
+ * The reader's per-site pairwise id, or null. Production reads only the siteSubject Prospectus returned;
+ * the interim `ssdev_` value exists in development and test only (lib/reader-subject.md).
+ */
+export function readerSubject(
+  session: Pick<ProspectusSignedInSession, "personId" | "siteSubject"> | null | undefined,
+  environment: EnvironmentVariables = process.env,
+): string | null {
+  if (!session) return null;
+  if (typeof session.siteSubject === "string" && SITE_SUBJECT_PATTERN.test(session.siteSubject)) return session.siteSubject;
+  if (!interimSubjectsAllowed(environment) || !session.personId) return null;
+  return interimSubject(session.personId, environment[READER_DEV_SECRET_ENV] as string);
+}
+
+/** Staff membership from the Cognito groups claim of the access token that the sealed session cookie carries. */
+export function readerIsStaff(session: Pick<ProspectusSignedInSession, "accessToken"> | null | undefined): boolean {
+  const token = session?.accessToken;
+  if (typeof token !== "string") return false;
+  const payload = token.split(".")[1];
+  if (!payload) return false;
+  try {
+    const claims = JSON.parse(Buffer.from(payload, "base64url").toString("utf8")) as { "cognito:groups"?: unknown };
+    const groups = claims["cognito:groups"];
+    return Array.isArray(groups) && groups.some((group) => (READER_STAFF_GROUPS as readonly unknown[]).includes(group));
+  } catch {
+    return false;
+  }
+}

package/lib/relevance-status.ts

@@ -52,10 +52,10 @@ export const DEMO_CYCLOTRON_STATUS: CyclotronStatus = {
   },
   calibration: { saysSure: null, isRight: null, gapPoints: null, curve: [] },
   reviewRate: {
-    state: "onboarding", rate: 1, reason: "Onboarding: every decision is reviewed until the cyclotron earns a lower rate.",
+    state: "onboarding", rate: 1, reason: "Getting started: every suggestion is reviewed until suggestions earn a lower review rate.",
     auditFloor: 0.05, confidenceThreshold: 0.7, override: null,
     nextCheckAfterDecisions: 99,
-    nextStep: "Steps down to 50% after 2 more windows of 100 decisions with accuracy on confident decisions of at least 85% (at least 20 reviewed) and a calibration gap under 5 points. Next check in 99 decisions.",
+    nextStep: "Steps down to 50% after 2 more windows of 100 decisions with accuracy on confident decisions of at least 85% (at least 20 reviewed) and a gap under 5 points between how sure suggestions say they are and how often they are right. Next check in 99 decisions.",
     expectedReviewsPerWeek: 1,
   },
   lastChange: null,

package/lib/site-brand-tenant.ts

@@ -9,14 +9,19 @@ export function resolveBrandConfigPath(relativePath: string): string {
   return path.isAbsolute(relativePath) ? relativePath : path.join(process.cwd(), relativePath);
 }
 
-export function getBrandSteeringConfigPath(brand: SiteBrand = resolveActiveSiteBrand()): string {
-  return resolveBrandConfigPath(brand.steeringConfigPath);
+/** Optional repo files a brand may still name; undefined means the site keeps that configuration as backend data and nothing is looked up. */
+function optionalBrandConfigPath(relativePath: string | undefined): string | undefined {
+  return relativePath ? resolveBrandConfigPath(relativePath) : undefined;
 }
 
-export function getBrandNewsroomSectionsConfigPath(brand: SiteBrand = resolveActiveSiteBrand()): string {
-  return resolveBrandConfigPath(brand.newsroomSectionsConfigPath);
+export function getBrandSteeringConfigPath(brand: SiteBrand = resolveActiveSiteBrand()): string | undefined {
+  return optionalBrandConfigPath(brand.steeringConfigPath);
 }
 
-export function getBrandAnalysisProfilesPath(brand: SiteBrand = resolveActiveSiteBrand()): string {
-  return resolveBrandConfigPath(brand.analysisProfilesPath);
+export function getBrandNewsroomSectionsConfigPath(brand: SiteBrand = resolveActiveSiteBrand()): string | undefined {
+  return optionalBrandConfigPath(brand.newsroomSectionsConfigPath);
+}
+
+export function getBrandAnalysisProfilesPath(brand: SiteBrand = resolveActiveSiteBrand()): string | undefined {
+  return optionalBrandConfigPath(brand.analysisProfilesPath);
 }

package/lib/site-brand.ts

@@ -142,9 +142,12 @@ export type SiteBrand = {
   readerBasePath?: string;
   opsChrome: OpsChrome;
   corpusKey: string;
-  steeringConfigPath: string;
-  newsroomSectionsConfigPath: string;
-  analysisProfilesPath: string;
+  /** Optional repo file for topic steering. Editorial and steering configuration is backend data; a data-driven site sets none and nothing is looked up. */
+  steeringConfigPath?: string;
+  /** Optional repo file for the local-preview demo sections only. Sections live in the backend. */
+  newsroomSectionsConfigPath?: string;
+  /** Optional repo file listing analysis profiles. */
+  analysisProfilesPath?: string;
   publicationName: string;
   components?: SiteBrandComponents;
   /**

package/lib/site-origin.ts

@@ -0,0 +1,67 @@
+import { NextResponse, type NextRequest } from "next/server";
+import { getSiteEnv } from "./site-env";
+
+type EnvironmentVariables = Record<string, string | undefined>;
+
+export const SITE_ORIGIN_ENV = "PAPYRUS_SITE_ORIGIN";
+
+/**
+ * The site's public origin cannot be read from the request on Amplify SSR (WEB_COMPUTE):
+ * there `request.nextUrl.origin` is the compute adapter's internal https://localhost:3000.
+ * It also must never be derived from Host or X-Forwarded-* headers, which a client can set.
+ * The one trusted source is PAPYRUS_SITE_ORIGIN, set by the app shell from the site's host.
+ */
+export class SiteOriginNotConfiguredError extends Error {
+  readonly status = 503;
+  readonly code = "site_origin_not_configured";
+}
+
+export function siteOriginNotConfiguredResponse(error: SiteOriginNotConfiguredError): NextResponse {
+  return NextResponse.json({ error: error.code, message: error.message }, { status: error.status, headers: { "cache-control": "no-store" } });
+}
+
+function isLoopbackHostname(hostname: string): boolean {
+  return hostname === "localhost" || hostname === "127.0.0.1" || hostname === "[::1]";
+}
+
+function configuredOrigin(value: string): string {
+  const invalid = () => new SiteOriginNotConfiguredError(`${SITE_ORIGIN_ENV} must be an https origin such as https://example.com with no path.`);
+  let parsed: URL;
+  try {
+    parsed = new URL(value);
+  } catch {
+    throw invalid();
+  }
+  if (parsed.protocol !== "https:" || parsed.origin !== value) throw invalid();
+  return parsed.origin;
+}
+
+function siteEnvOrNull(environment: EnvironmentVariables): string | null {
+  try {
+    return getSiteEnv(environment);
+  } catch {
+    return null;
+  }
+}
+
+/**
+ * The site's public origin, from trusted configuration.
+ * - PAPYRUS_SITE_ORIGIN set: it must be an https origin with no path; it is used as is.
+ * - Unset: `request.nextUrl.origin` only when it is loopback, SITE_ENV resolves to development
+ *   and NODE_ENV is not production (a local `next dev` server). Anything else, production and staging included, throws
+ *   SiteOriginNotConfiguredError so routes answer 503 instead of redirecting to localhost.
+ */
+export function publicSiteOrigin(request: Pick<NextRequest, "nextUrl">, environment: EnvironmentVariables = process.env): string {
+  const configured = (environment[SITE_ORIGIN_ENV] ?? "").trim();
+  if (configured) return configuredOrigin(configured);
+  const { hostname, origin } = request.nextUrl;
+  // SITE_ENV defaults to development when a runtime never received it, so NODE_ENV=production also closes the fallback.
+  const developing = siteEnvOrNull(environment) === "development" && environment.NODE_ENV !== "production";
+  if (isLoopbackHostname(hostname) && developing) return origin;
+  throw new SiteOriginNotConfiguredError(`${SITE_ORIGIN_ENV} is not set; this site cannot tell its public origin.`);
+}
+
+/** An absolute URL on the site's public origin, for redirects. Fails closed like publicSiteOrigin. */
+export function publicSiteUrl(path: string, request: Pick<NextRequest, "nextUrl">, environment: EnvironmentVariables = process.env): URL {
+  return new URL(path, publicSiteOrigin(request, environment));
+}

package/middleware.ts

@@ -5,6 +5,7 @@ import { newsroomHref } from "./lib/newsroom-base-path";
 import { getReaderBasePath } from "./lib/reader-base-path";
 import { isStagingGatedPath, isStaticOrApiPath } from "./lib/staging-gated-path";
 import { getSiteEnv, isIndexable } from "./lib/site-env";
+import { SiteOriginNotConfiguredError, publicSiteUrl, siteOriginNotConfiguredResponse } from "./lib/site-origin";
 import { getStagingAccess } from "./lib/staging-gate";
 import {
   isPreviewGatedPath,
@@ -38,11 +39,11 @@ function routeRequest(request: NextRequest): NextResponse {
   }
 
   if (pathname === "/newsroom" || pathname === "/newsroom/") {
-    return NextResponse.redirect(new URL(`/${request.nextUrl.search}`, request.url));
+    return NextResponse.redirect(publicSiteUrl(`/${request.nextUrl.search}`, request));
   }
   if (pathname.startsWith("/newsroom/")) {
     const target = pathname.slice("/newsroom".length) || "/";
-    return NextResponse.redirect(new URL(`${target}${request.nextUrl.search}`, request.url));
+    return NextResponse.redirect(publicSiteUrl(`${target}${request.nextUrl.search}`, request));
   }
 
   if (pathname === "/" || pathname === "") {
@@ -59,6 +60,16 @@ function routeRequest(request: NextRequest): NextResponse {
 }
 
 export async function middleware(request: NextRequest): Promise<NextResponse> {
+  try {
+    return await handleRequest(request);
+  } catch (error) {
+    // Redirects name the site's public origin (request.url is localhost on Amplify); without it, fail closed.
+    if (error instanceof SiteOriginNotConfiguredError) return withRobotsHeader(siteOriginNotConfiguredResponse(error));
+    throw error;
+  }
+}
+
+async function handleRequest(request: NextRequest): Promise<NextResponse> {
   let sessionResponse: NextResponse | null = null;
   const { pathname } = request.nextUrl;
   const gated = isStaticPreviewEnabled()
@@ -68,7 +79,7 @@ export async function middleware(request: NextRequest): Promise<NextResponse> {
     sessionResponse = NextResponse.next();
     const access = await getStagingAccess(request, sessionResponse);
     if (access === "anonymous") {
-      return withRobotsHeader(NextResponse.redirect(new URL(newsroomHref(), request.url)));
+      return withRobotsHeader(NextResponse.redirect(publicSiteUrl(newsroomHref(), request)));
     }
     if (access === "forbidden") {
       return withRobotsHeader(new NextResponse("Staging is limited to editors and admins.", { status: 403 }));