Files
TREK/server/src/middleware/globalMiddleware.ts
T
Maurice 141afe708d fix(journey): let the browser read a picked clip, so it previews and gets a poster
A clip picked in the entry editor is previewed through a <video> on an object
URL, and its poster frame is grabbed the same way before anything is uploaded.
The CSP never named media-src, so the browser fell back to default-src, which
does not admit blob: at all. The preview stayed empty and the poster capture
failed silently, so every clip landed without a poster and its thumbnail route
answered 404 (issue #2341, the part that survived 4.3.0). None of it showed in
development, where Vite serves the document and helmet never sees it.

media-src now carries 'self' and blob:. For the clips that still come up
without a poster, because the browser cannot decode them, the entry card and
the editor tile draw the play badge on a black tile instead of asking for a
thumbnail that is 404 by design, the way the gallery already did.

Closes #2341
2026-09-21 10:35:37 +02:00

338 lines
16 KiB
TypeScript

import express, { Request, Response, NextFunction } from 'express';
import compression from 'compression';
import cors from 'cors';
import helmet from 'helmet';
import cookieParser from 'cookie-parser';
import { readEnv, type AppEnv } from '../app-config';
import { logDebug, logWarn, logError } from '../nest/audit/audit-log.logger';
/**
* Field names redacted from request-log query/body dumps (case-insensitive —
* lookup lowercases the key first). `secretaccesskey` covers the S3 storage
* backend's secret field (storage-admin PUT bodies land in the same debug
* log line as any other request body).
*
* Exported (module-level, not per-request) so it can be unit-tested directly
* without threading LOG_LEVEL through the real logger pipeline.
*/
export const SENSITIVE_KEYS = new Set([
'password',
'new_password',
'current_password',
'token',
'jwt',
'authorization',
'cookie',
'client_secret',
'mfa_token',
'code',
'smtp_pass',
'secretaccesskey',
]);
/**
* Suffixes that make a field sensitive whatever its prefix. Matched on the end of
* the key, not anywhere in it: `accessKeyId` is an identifier and stays readable,
* while `refresh_token`, `code_verifier`, `mapbox_access_token` and the whole
* `*_api_key` family are secrets nobody needs in a debug log.
*/
const SENSITIVE_SUFFIXES = ['_token', '_key', '_secret', '_password', '_pass', '_verifier'];
function isSensitiveName(name: string): boolean {
const lower = name.toLowerCase();
return SENSITIVE_KEYS.has(lower) || SENSITIVE_SUFFIXES.some((suffix) => lower.endsWith(suffix));
}
/** Deep-redacts every key in `SENSITIVE_KEYS` (case-insensitive) from a request-log value. */
export function redact(value: unknown): unknown {
if (!value || typeof value !== 'object') return value;
if (Array.isArray(value)) return (value as unknown[]).map(redact);
const entries = value as Record<string, unknown>;
// PUT /api/settings names the setting instead of using it as the field:
// {key: 'carto_api_key', value: '<secret>'}. Neither field name is sensitive
// on its own, so without this the secret goes into the debug log in cleartext.
const namedSecret = typeof entries.key === 'string' && isSensitiveName(entries.key);
const out: Record<string, unknown> = {};
for (const [k, v] of Object.entries(entries)) {
out[k] = isSensitiveName(k) || (namedSecret && k === 'value') ? '[REDACTED]' : redact(v);
}
return out;
}
/**
* The global request pipeline shared by the legacy Express app and the NestJS
* instance. Both mount the *exact same* config so a request hitting a migrated
* Nest route is protected identically to one hitting the legacy fallback
* (helmet/CSP, CORS, HSTS, forced-HTTPS, the global MFA policy and request
* logging). Keeping it in one place is what makes the strangler dispatch
* behaviourally transparent — and is the prerequisite for retiring Express,
* since the Nest instance must carry the whole shell on its own.
*
* Body parsing is NOT here: the Nest instance does it, with the limit pinned in
* bootstrap.ts. The flag that used to switch a second parser on had exactly one
* caller, which passed false.
*/
/**
* Turns a configured routing base URL into the origin CSP needs.
*
* Only the origin: a CSP source with a path only matches that path prefix, and a router
* is asked at several of them (`/route/v1/...`, `/table/v1/...`). Anything that is not an
* absolute http(s) URL is dropped rather than passed through — a bad value in a settings
* row must not be able to widen the policy.
*/
export function routingCspOrigins(baseUrls: (string | null | undefined)[]): string[] {
const origins = new Set<string>();
for (const raw of baseUrls) {
if (!raw) continue;
try {
const url = new URL(raw.trim());
if (url.protocol !== 'http:' && url.protocol !== 'https:') continue;
origins.add(url.origin);
} catch {
// Not a URL; the client falls back to the public hosts, so the policy stays as it was.
}
}
return [...origins];
}
export function applyGlobalMiddleware(
app: express.Application,
opts: {
http?: AppEnv['http'];
/**
* Extra origins the browser may talk to, from the instance's own settings — today the
* self-hosted routing engine (#1797). Read once at apply time like everything else
* here, so changing it takes a restart; the alternative is a database read on every
* request for a value that changes about once in a deployment's life.
*/
extraConnectSrc?: string[];
} = {},
): void {
// The whole pipeline is configured at APPLY time (the per-request closures
// capture these values), so a snapshot is the correct semantic. bootstrap
// threads in the DI-loaded httpConfig; direct callers fall back to an
// apply-time readEnv() — same values, same freeze point.
const { http = readEnv().http, extraConnectSrc = [] } = opts;
const { nodeEnv, isProduction } = readEnv().app;
// Trust first proxy (nginx/Docker) for correct req.ip
if (isProduction || http.trustProxyRaw) {
app.set('trust proxy', http.trustProxy);
}
// Compress responses (gzip via Accept-Encoding). The Atlas admin-0 country
// GeoJSON is ~30 MB uncompressed, which stalls/aborts (~8s → net::ERR_FAILED)
// behind reverse proxies and Cloudflare Tunnel (#1254); gzip brings it to ~4 MB.
// SSE responses (the /mcp StreamableHTTP transport) must NOT be buffered, so
// they are excluded explicitly.
app.use(
compression({
filter: (req, res) => {
const type = res.getHeader('Content-Type');
if (typeof type === 'string' && type.includes('text/event-stream')) return false;
return compression.filter(req, res);
},
}),
);
const allowedOrigins = http.corsOrigins;
let corsOrigin: cors.CorsOptions['origin'];
if (allowedOrigins) {
corsOrigin = (origin: string | undefined, callback: (err: Error | null, allow?: boolean) => void) => {
if (!origin || allowedOrigins.includes(origin)) callback(null, true);
else callback(new Error('Not allowed by CORS'));
};
} else if (isProduction) {
corsOrigin = false;
} else {
corsOrigin = true;
}
const shouldForceHttps = http.forceHttps;
// HSTS is worth enabling any time we're serving production traffic,
// not only when FORCE_HTTPS is set. Self-hosters behind Traefik /
// Caddy / Cloudflare Tunnel typically leave FORCE_HTTPS unset (the
// proxy handles the redirect for them), and the previous "HSTS off by
// default" meant those instances never advertised HSTS at all.
//
// `includeSubDomains` stays OFF by default on purpose: an instance
// running on an apex domain would otherwise force HTTPS on every
// sibling subdomain the same operator may still be running over plain
// HTTP. Operators who want the stricter policy opt in with
// `HSTS_INCLUDE_SUBDOMAINS=true`.
// nodeEnv compared case-sensitively here on purpose (legacy parity — the
// trust-proxy/CORS checks above are the case-insensitive ones).
const hstsActive = shouldForceHttps || nodeEnv === 'production';
const hstsIncludeSubdomains = http.hstsIncludeSubdomains;
// RFC 8414 / RFC 9728 / RFC 7591: discovery docs and DCR are world-readable/writable.
// /mcp needs open CORS so external MCP clients (ChatGPT, Claude.ai, Inspector) can call it
// with Bearer tokens from any origin. /oauth/register and /oauth/authorize need it for
// browser-based DCR/authorization preflights — the global cors({ origin: false }) would
// answer OPTIONS without Access-Control-Allow-Origin before the SDK's own cors() runs.
// All /.well-known/* paths get open CORS so clients probing openid-configuration or the
// RFC 8414 path-suffixed AS metadata form don't get CORS-blocked (they get 404 JSON instead).
//
// `exposedHeaders` is load-bearing, not cosmetic. Without Access-Control-Expose-Headers the
// Fetch spec forbids a browser-context client (Claude Desktop connectors, Claude.ai, MCP
// Inspector) from *reading* Mcp-Session-Id off the initialize response — so it can never echo
// the header back, every request looks like a fresh initialize, and the server mints a new
// session per tool call until the per-user cap wedges the connection. Same reasoning for
// WWW-Authenticate, which carries the RFC 9728 resource-metadata challenge that drives
// OAuth discovery.
app.use(
(req: Request, _res: Response, next: NextFunction) => {
if (
req.path.startsWith('/.well-known/') ||
req.path === '/oauth/register' ||
req.path === '/oauth/authorize' ||
req.path === '/oauth/userinfo' ||
req.path === '/mcp'
) {
cors({
origin: '*',
credentials: false,
exposedHeaders: ['Mcp-Session-Id', 'MCP-Protocol-Version', 'WWW-Authenticate'],
})(req, _res, next);
} else {
next();
}
},
);
app.use(cors({ origin: corsOrigin, credentials: true }));
app.use(helmet({
contentSecurityPolicy: {
directives: {
defaultSrc: ["'self'"],
// 'unsafe-eval' is load-bearing, not leftover: heic-to's libheif build
// initialises embind through new Function(), and that is what converts
// an iPhone .heic the moment somebody picks one. 'wasm-unsafe-eval'
// alone was tried first and was not enough (93b51a0b). The package
// ships a CSP-safe entry point at heic-to/csp; dropping this directive
// means switching client/src/utils/convertHeic.ts over to it and
// verifying a real .heic upload in a browser, not just deleting the
// string here.
scriptSrc: ["'self'", "'wasm-unsafe-eval'", "'unsafe-eval'"],
styleSrc: ["'self'", "'unsafe-inline'", "https://fonts.googleapis.com", "https://unpkg.com"],
imgSrc: ["'self'", "data:", "blob:", "https:"],
connectSrc: [
"'self'", "ws:", "wss:",
"https://nominatim.openstreetmap.org", "https://overpass-api.de",
"https://places.googleapis.com", "https://api.openweathermap.org",
"https://en.wikipedia.org", "https://commons.wikimedia.org",
// Both forms here too: CARTO documents the apex host on its key page,
// so that is the template users paste in, and the {s} sharded form is
// what TREK ships (#2054).
"https://basemaps.cartocdn.com", "https://*.basemaps.cartocdn.com",
// Both forms: a CSP wildcard host never matches the apex, and OSM
// serves everything from the bare tile.openstreetmap.org since it
// retired the a/b/c/d shards (#1733). The sharded hosts stay listed
// for tile templates users saved before that.
"https://tile.openstreetmap.org", "https://*.tile.openstreetmap.org",
// The other two raster presets TREK ships. `mode: 'no-cors'` relaxes
// CORS, not CSP, so without these the tile prefetch is refused in the
// document and never reaches the Service Worker that would cache it
// (#2180). routing.openstreetmap.de below is a different host.
"https://tile.openstreetmap.de", "https://tiles.stadiamaps.com",
// The imagery host, for the same reason and one more. Leaflet fetches a tile
// as an <img>, which img-src's blanket `https:` waves through, so the satellite
// view worked on Leaflet with this host missing. A GL map reads the raster
// through fetch to hand it to WebGL, and the prefetch does too, so both were
// refused here while nothing in the app could see it: the switch flipped, the
// layer went on, and no tile ever arrived (#2307).
"https://server.arcgisonline.com",
// Amap's raster tiles, for an install whose users are in China, for the
// same reason as the hosts above (#2180). Road (webrd01..04) and
// satellite (webst01..04) are numbered shards of one domain, so the
// wildcard is the whole list.
"https://*.is.autonavi.com",
"https://unpkg.com", "https://open-meteo.com", "https://api.open-meteo.com",
"https://geocoding-api.open-meteo.com", "https://api.frankfurter.dev",
"https://router.project-osrm.org/route/v1/", "https://routing.openstreetmap.de/",
// The second routing engine, shipped as a default the same way the OSRM hosts
// above are. It is asked only when a leg should avoid tolls, motorways or a
// ferry — which the OSRM hosts answer with HTTP 400, because their car profile
// carries no excludable classes. Origin only, no path: unlike OSRM this one is
// a POST to /route and would grow more endpoints if isochrones ever land.
"https://valhalla1.openstreetmap.de",
"https://api.mapbox.com", "https://*.tiles.mapbox.com", "https://events.mapbox.com",
"https://tiles.openfreemap.org",
// A self-hosted routing engine, when the instance has one configured. Without
// this the browser blocks it silently: no error the app can catch, just legs
// that never route (#1797).
...extraConnectSrc,
],
workerSrc: ["'self'", "blob:"],
childSrc: ["'self'", "blob:"],
// blob: because a picked clip is previewed and its poster frame grabbed
// through a <video> on an object URL, before any byte reaches the server.
// Unset, this fell back to default-src, which refuses blob: outright: the
// editor showed nothing and every clip landed without a poster, so its
// thumbnail answered 404 (#2341). Invisible in dev, where Vite serves the
// document without this header.
mediaSrc: ["'self'", "blob:"],
fontSrc: ["'self'", "https://fonts.gstatic.com", "data:"],
// 'self' so same-origin file previews can embed PDFs via <object>/<embed>
// (Firefox/Chrome enforce object-src; 'none' broke inline PDF previews there).
objectSrc: ["'self'"],
// 'self' so the app can embed same-origin, sandboxed plugin frames
// (/plugin-frame/*). Those frames are sandboxed WITHOUT allow-same-origin,
// so they run at an opaque origin and get their own locked-down CSP.
frameSrc: ["'self'"],
frameAncestors: ["'self'"],
// Restrict <form> submission targets (form-action has no default-src
// fallback, so it must be set explicitly).
formAction: ["'self'"],
upgradeInsecureRequests: shouldForceHttps ? [] : null
}
},
crossOriginEmbedderPolicy: false,
hsts: hstsActive ? { maxAge: 31536000, includeSubDomains: hstsIncludeSubdomains } : false,
referrerPolicy: { policy: 'strict-origin-when-cross-origin' },
}));
// The instance's own hostname, when the operator configured one. The redirect
// below is a 301, so echoing a client-supplied Host header would let a stranger
// park a foreign origin in a visitor's cache. Falls back to the request Host for
// instances that never set APP_URL, which is what it always did.
const configuredHost = (() => {
const url = readEnv().app.appUrl;
try {
return url ? new URL(url).host : null;
} catch {
return null;
}
})();
if (shouldForceHttps) {
app.use((req: Request, res: Response, next: NextFunction) => {
if (req.path === '/api/health') return next();
if (req.secure || req.headers['x-forwarded-proto'] === 'https') return next();
res.redirect(301, 'https://' + (configuredHost ?? req.headers.host) + req.url);
});
}
app.use(cookieParser());
// Request logging with sensitive field redaction (SENSITIVE_KEYS/redact above)
app.use((req: Request, res: Response, next: NextFunction) => {
if (req.path === '/api/health') return next();
const startedAt = Date.now();
res.on('finish', () => {
const ms = Date.now() - startedAt;
if (res.statusCode >= 500) {
logError(`${req.method} ${req.path} ${res.statusCode} ${ms}ms ip=${req.ip}`);
} else if (res.statusCode === 401 || res.statusCode === 403) {
logDebug(`${req.method} ${req.path} ${res.statusCode} ${ms}ms ip=${req.ip}`);
} else if (res.statusCode >= 400) {
logWarn(`${req.method} ${req.path} ${res.statusCode} ${ms}ms ip=${req.ip}`);
}
const q = Object.keys(req.query).length ? ` query=${JSON.stringify(redact(req.query))}` : '';
const b = req.body && Object.keys(req.body).length ? ` body=${JSON.stringify(redact(req.body))}` : '';
logDebug(`${req.method} ${req.path} ${res.statusCode} ${ms}ms ip=${req.ip}${q}${b}`);
});
next();
});
}