feat: Add chat embeds v1
Youtube and Website metadata embeds
This commit is contained in:
@@ -4,18 +4,28 @@ import { resolveRuntimePath } from '../runtime-paths';
|
||||
|
||||
export type ServerHttpProtocol = 'http' | 'https';
|
||||
|
||||
export interface LinkPreviewConfig {
|
||||
enabled: boolean;
|
||||
cacheTtlMinutes: number;
|
||||
maxCacheSizeMb: number;
|
||||
}
|
||||
|
||||
export interface ServerVariablesConfig {
|
||||
klipyApiKey: string;
|
||||
releaseManifestUrl: string;
|
||||
serverPort: number;
|
||||
serverProtocol: ServerHttpProtocol;
|
||||
serverHost: string;
|
||||
linkPreview: LinkPreviewConfig;
|
||||
}
|
||||
|
||||
const DATA_DIR = resolveRuntimePath('data');
|
||||
const VARIABLES_FILE = path.join(DATA_DIR, 'variables.json');
|
||||
const DEFAULT_SERVER_PORT = 3001;
|
||||
const DEFAULT_SERVER_PROTOCOL: ServerHttpProtocol = 'http';
|
||||
const DEFAULT_LINK_PREVIEW_CACHE_TTL_MINUTES = 7200;
|
||||
const DEFAULT_LINK_PREVIEW_MAX_CACHE_SIZE_MB = 50;
|
||||
const HARD_MAX_CACHE_SIZE_MB = 50;
|
||||
|
||||
function normalizeKlipyApiKey(value: unknown): string {
|
||||
return typeof value === 'string' ? value.trim() : '';
|
||||
@@ -66,6 +76,27 @@ function normalizeServerPort(value: unknown, fallback = DEFAULT_SERVER_PORT): nu
|
||||
: fallback;
|
||||
}
|
||||
|
||||
function normalizeLinkPreviewConfig(value: unknown): LinkPreviewConfig {
|
||||
const raw = (value && typeof value === 'object' && !Array.isArray(value))
|
||||
? value as Record<string, unknown>
|
||||
: {};
|
||||
const enabled = typeof raw.enabled === 'boolean'
|
||||
? raw.enabled
|
||||
: true;
|
||||
const cacheTtl = typeof raw.cacheTtlMinutes === 'number'
|
||||
&& Number.isFinite(raw.cacheTtlMinutes)
|
||||
&& raw.cacheTtlMinutes >= 0
|
||||
? raw.cacheTtlMinutes
|
||||
: DEFAULT_LINK_PREVIEW_CACHE_TTL_MINUTES;
|
||||
const maxSize = typeof raw.maxCacheSizeMb === 'number'
|
||||
&& Number.isFinite(raw.maxCacheSizeMb)
|
||||
&& raw.maxCacheSizeMb >= 0
|
||||
? Math.min(raw.maxCacheSizeMb, HARD_MAX_CACHE_SIZE_MB)
|
||||
: DEFAULT_LINK_PREVIEW_MAX_CACHE_SIZE_MB;
|
||||
|
||||
return { enabled, cacheTtlMinutes: cacheTtl, maxCacheSizeMb: maxSize };
|
||||
}
|
||||
|
||||
function hasEnvironmentOverride(value: string | undefined): value is string {
|
||||
return typeof value === 'string' && value.trim().length > 0;
|
||||
}
|
||||
@@ -111,7 +142,8 @@ export function ensureVariablesConfig(): ServerVariablesConfig {
|
||||
releaseManifestUrl: normalizeReleaseManifestUrl(remainingParsed.releaseManifestUrl),
|
||||
serverPort: normalizeServerPort(remainingParsed.serverPort),
|
||||
serverProtocol: normalizeServerProtocol(remainingParsed.serverProtocol),
|
||||
serverHost: normalizeServerHost(remainingParsed.serverHost ?? legacyServerIpAddress)
|
||||
serverHost: normalizeServerHost(remainingParsed.serverHost ?? legacyServerIpAddress),
|
||||
linkPreview: normalizeLinkPreviewConfig(remainingParsed.linkPreview)
|
||||
};
|
||||
const nextContents = JSON.stringify(normalized, null, 2) + '\n';
|
||||
|
||||
@@ -124,7 +156,8 @@ export function ensureVariablesConfig(): ServerVariablesConfig {
|
||||
releaseManifestUrl: normalized.releaseManifestUrl,
|
||||
serverPort: normalized.serverPort,
|
||||
serverProtocol: normalized.serverProtocol,
|
||||
serverHost: normalized.serverHost
|
||||
serverHost: normalized.serverHost,
|
||||
linkPreview: normalized.linkPreview
|
||||
};
|
||||
}
|
||||
|
||||
@@ -169,3 +202,7 @@ export function getServerHost(): string | undefined {
|
||||
export function isHttpsServerEnabled(): boolean {
|
||||
return getServerProtocol() === 'https';
|
||||
}
|
||||
|
||||
export function getLinkPreviewConfig(): LinkPreviewConfig {
|
||||
return getVariablesConfig().linkPreview;
|
||||
}
|
||||
|
||||
@@ -1,6 +1,7 @@
|
||||
import { Express } from 'express';
|
||||
import healthRouter from './health';
|
||||
import klipyRouter from './klipy';
|
||||
import linkMetadataRouter from './link-metadata';
|
||||
import proxyRouter from './proxy';
|
||||
import usersRouter from './users';
|
||||
import serversRouter from './servers';
|
||||
@@ -10,6 +11,7 @@ import { invitesApiRouter, invitePageRouter } from './invites';
|
||||
export function registerRoutes(app: Express): void {
|
||||
app.use('/api', healthRouter);
|
||||
app.use('/api', klipyRouter);
|
||||
app.use('/api', linkMetadataRouter);
|
||||
app.use('/api', proxyRouter);
|
||||
app.use('/api/users', usersRouter);
|
||||
app.use('/api/servers', serversRouter);
|
||||
|
||||
292
server/src/routes/link-metadata.ts
Normal file
292
server/src/routes/link-metadata.ts
Normal file
@@ -0,0 +1,292 @@
|
||||
import { Router } from 'express';
|
||||
import { getLinkPreviewConfig } from '../config/variables';
|
||||
import { resolveAndValidateHost, safeFetch } from './ssrf-guard';
|
||||
|
||||
const router = Router();
|
||||
const REQUEST_TIMEOUT_MS = 8000;
|
||||
const MAX_HTML_BYTES = 512 * 1024;
|
||||
const BYTES_PER_MB = 1024 * 1024;
|
||||
const MAX_FIELD_LENGTH = 512;
|
||||
|
||||
interface CachedMetadata {
|
||||
title?: string;
|
||||
description?: string;
|
||||
imageUrl?: string;
|
||||
siteName?: string;
|
||||
failed?: boolean;
|
||||
cachedAt: number;
|
||||
}
|
||||
|
||||
const metadataCache = new Map<string, CachedMetadata>();
|
||||
|
||||
let cacheByteEstimate = 0;
|
||||
|
||||
function estimateEntryBytes(key: string, entry: CachedMetadata): number {
|
||||
let bytes = key.length * 2;
|
||||
|
||||
if (entry.title)
|
||||
bytes += entry.title.length * 2;
|
||||
|
||||
if (entry.description)
|
||||
bytes += entry.description.length * 2;
|
||||
|
||||
if (entry.imageUrl)
|
||||
bytes += entry.imageUrl.length * 2;
|
||||
|
||||
if (entry.siteName)
|
||||
bytes += entry.siteName.length * 2;
|
||||
|
||||
return bytes + 64;
|
||||
}
|
||||
|
||||
function cacheSet(key: string, entry: CachedMetadata): void {
|
||||
const config = getLinkPreviewConfig();
|
||||
const maxBytes = config.maxCacheSizeMb * BYTES_PER_MB;
|
||||
|
||||
if (metadataCache.has(key)) {
|
||||
const existing = metadataCache.get(key) as CachedMetadata;
|
||||
|
||||
cacheByteEstimate -= estimateEntryBytes(key, existing);
|
||||
}
|
||||
|
||||
const entryBytes = estimateEntryBytes(key, entry);
|
||||
|
||||
while (cacheByteEstimate + entryBytes > maxBytes && metadataCache.size > 0) {
|
||||
const oldest = metadataCache.keys().next().value as string;
|
||||
const oldestEntry = metadataCache.get(oldest) as CachedMetadata;
|
||||
|
||||
cacheByteEstimate -= estimateEntryBytes(oldest, oldestEntry);
|
||||
metadataCache.delete(oldest);
|
||||
}
|
||||
|
||||
metadataCache.set(key, entry);
|
||||
cacheByteEstimate += entryBytes;
|
||||
}
|
||||
|
||||
function truncateField(value: string | undefined): string | undefined {
|
||||
if (!value)
|
||||
return value;
|
||||
|
||||
if (value.length <= MAX_FIELD_LENGTH)
|
||||
return value;
|
||||
|
||||
return value.slice(0, MAX_FIELD_LENGTH);
|
||||
}
|
||||
|
||||
function sanitizeImageUrl(rawUrl: string | undefined, baseUrl: string): string | undefined {
|
||||
if (!rawUrl)
|
||||
return undefined;
|
||||
|
||||
try {
|
||||
const resolved = new URL(rawUrl, baseUrl);
|
||||
|
||||
if (resolved.protocol !== 'http:' && resolved.protocol !== 'https:')
|
||||
return undefined;
|
||||
|
||||
return resolved.href;
|
||||
} catch {
|
||||
return undefined;
|
||||
}
|
||||
}
|
||||
|
||||
function getMetaContent(html: string, patterns: RegExp[]): string | undefined {
|
||||
for (const pattern of patterns) {
|
||||
const match = pattern.exec(html);
|
||||
|
||||
if (match?.[1])
|
||||
return decodeHtmlEntities(match[1].trim());
|
||||
}
|
||||
|
||||
return undefined;
|
||||
}
|
||||
|
||||
function decodeHtmlEntities(text: string): string {
|
||||
return text
|
||||
.replace(/&/g, '&')
|
||||
.replace(/</g, '<')
|
||||
.replace(/>/g, '>')
|
||||
.replace(/"/g, '"')
|
||||
.replace(/'/g, "'")
|
||||
.replace(/'/g, "'")
|
||||
.replace(///g, '/');
|
||||
}
|
||||
|
||||
function parseMetadata(html: string, url: string): CachedMetadata {
|
||||
const title = getMetaContent(html, [
|
||||
/<meta[^>]+property=["']og:title["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:title["']/i,
|
||||
/<meta[^>]+name=["']twitter:title["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+name=["']twitter:title["']/i,
|
||||
/<title[^>]*>([^<]+)<\/title>/i
|
||||
]);
|
||||
const description = getMetaContent(html, [
|
||||
/<meta[^>]+property=["']og:description["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:description["']/i,
|
||||
/<meta[^>]+name=["']twitter:description["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+name=["']twitter:description["']/i,
|
||||
/<meta[^>]+name=["']description["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+name=["']description["']/i
|
||||
]);
|
||||
const rawImageUrl = getMetaContent(html, [
|
||||
/<meta[^>]+property=["']og:image["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:image["']/i,
|
||||
/<meta[^>]+name=["']twitter:image["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+name=["']twitter:image["']/i
|
||||
]);
|
||||
const siteNamePatterns = [
|
||||
// eslint-disable-next-line @stylistic/js/array-element-newline
|
||||
/<meta[^>]+property=["']og:site_name["'][^>]+content=["']([^"']+)["']/i,
|
||||
/<meta[^>]+content=["']([^"']+)["'][^>]+property=["']og:site_name["']/i
|
||||
];
|
||||
const siteName = getMetaContent(html, siteNamePatterns);
|
||||
const imageUrl = sanitizeImageUrl(rawImageUrl, url);
|
||||
|
||||
return {
|
||||
title: truncateField(title),
|
||||
description: truncateField(description),
|
||||
imageUrl,
|
||||
siteName: truncateField(siteName),
|
||||
cachedAt: Date.now()
|
||||
};
|
||||
}
|
||||
|
||||
function evictExpired(): void {
|
||||
const config = getLinkPreviewConfig();
|
||||
|
||||
if (config.cacheTtlMinutes === 0) {
|
||||
cacheByteEstimate = 0;
|
||||
metadataCache.clear();
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
const ttlMs = config.cacheTtlMinutes * 60 * 1000;
|
||||
const now = Date.now();
|
||||
|
||||
for (const [key, entry] of metadataCache) {
|
||||
if (now - entry.cachedAt > ttlMs) {
|
||||
cacheByteEstimate -= estimateEntryBytes(key, entry);
|
||||
metadataCache.delete(key);
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
router.get('/link-metadata', async (req, res) => {
|
||||
try {
|
||||
const config = getLinkPreviewConfig();
|
||||
|
||||
if (!config.enabled) {
|
||||
return res.status(403).json({ error: 'Link previews are disabled' });
|
||||
}
|
||||
|
||||
const url = String(req.query.url || '');
|
||||
|
||||
if (!/^https?:\/\//i.test(url)) {
|
||||
return res.status(400).json({ error: 'Invalid URL' });
|
||||
}
|
||||
|
||||
const hostAllowed = await resolveAndValidateHost(url);
|
||||
|
||||
if (!hostAllowed) {
|
||||
return res.status(400).json({ error: 'URL resolves to a blocked address' });
|
||||
}
|
||||
|
||||
evictExpired();
|
||||
|
||||
const cached = metadataCache.get(url);
|
||||
|
||||
if (cached) {
|
||||
const { cachedAt, ...metadata } = cached;
|
||||
|
||||
console.log(`[Link Metadata] Cache hit for ${url} (cached at ${new Date(cachedAt).toISOString()})`);
|
||||
return res.json(metadata);
|
||||
}
|
||||
|
||||
console.log(`[Link Metadata] Cache miss for ${url}. Fetching...`);
|
||||
|
||||
const controller = new AbortController();
|
||||
const timeout = setTimeout(() => controller.abort(), REQUEST_TIMEOUT_MS);
|
||||
const response = await safeFetch(url, {
|
||||
signal: controller.signal,
|
||||
headers: {
|
||||
'Accept': 'text/html',
|
||||
'User-Agent': 'MetoYou-LinkPreview/1.0'
|
||||
}
|
||||
});
|
||||
|
||||
clearTimeout(timeout);
|
||||
|
||||
if (!response || !response.ok) {
|
||||
const failed: CachedMetadata = { failed: true, cachedAt: Date.now() };
|
||||
|
||||
cacheSet(url, failed);
|
||||
|
||||
return res.json({ failed: true });
|
||||
}
|
||||
|
||||
const contentType = response.headers.get('content-type') || '';
|
||||
|
||||
if (!contentType.includes('text/html')) {
|
||||
const failed: CachedMetadata = { failed: true, cachedAt: Date.now() };
|
||||
|
||||
cacheSet(url, failed);
|
||||
|
||||
return res.json({ failed: true });
|
||||
}
|
||||
|
||||
const reader = response.body?.getReader();
|
||||
|
||||
if (!reader) {
|
||||
const failed: CachedMetadata = { failed: true, cachedAt: Date.now() };
|
||||
|
||||
cacheSet(url, failed);
|
||||
|
||||
return res.json({ failed: true });
|
||||
}
|
||||
|
||||
const chunks: Uint8Array[] = [];
|
||||
|
||||
let totalBytes = 0;
|
||||
let done = false;
|
||||
|
||||
while (!done) {
|
||||
const result = await reader.read();
|
||||
|
||||
done = result.done;
|
||||
|
||||
if (result.value) {
|
||||
chunks.push(result.value);
|
||||
totalBytes += result.value.length;
|
||||
|
||||
if (totalBytes > MAX_HTML_BYTES) {
|
||||
reader.cancel();
|
||||
break;
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
const html = Buffer.concat(chunks).toString('utf-8');
|
||||
const metadata = parseMetadata(html, url);
|
||||
|
||||
cacheSet(url, metadata);
|
||||
|
||||
const { cachedAt, ...result } = metadata;
|
||||
|
||||
res.json(result);
|
||||
} catch (err) {
|
||||
const url = String(req.query.url || '');
|
||||
|
||||
if (url) {
|
||||
cacheSet(url, { failed: true, cachedAt: Date.now() });
|
||||
}
|
||||
|
||||
if ((err as { name?: string })?.name === 'AbortError') {
|
||||
return res.json({ failed: true });
|
||||
}
|
||||
|
||||
console.error('Link metadata error:', err);
|
||||
res.json({ failed: true });
|
||||
}
|
||||
});
|
||||
|
||||
export default router;
|
||||
@@ -1,4 +1,5 @@
|
||||
import { Router } from 'express';
|
||||
import { resolveAndValidateHost, safeFetch } from './ssrf-guard';
|
||||
|
||||
const router = Router();
|
||||
|
||||
@@ -10,14 +11,20 @@ router.get('/image-proxy', async (req, res) => {
|
||||
return res.status(400).json({ error: 'Invalid URL' });
|
||||
}
|
||||
|
||||
const hostAllowed = await resolveAndValidateHost(url);
|
||||
|
||||
if (!hostAllowed) {
|
||||
return res.status(400).json({ error: 'URL resolves to a blocked address' });
|
||||
}
|
||||
|
||||
const controller = new AbortController();
|
||||
const timeout = setTimeout(() => controller.abort(), 8000);
|
||||
const response = await fetch(url, { redirect: 'follow', signal: controller.signal });
|
||||
const response = await safeFetch(url, { signal: controller.signal });
|
||||
|
||||
clearTimeout(timeout);
|
||||
|
||||
if (!response.ok) {
|
||||
return res.status(response.status).end();
|
||||
if (!response || !response.ok) {
|
||||
return res.status(response?.status ?? 502).end();
|
||||
}
|
||||
|
||||
const contentType = response.headers.get('content-type') || '';
|
||||
|
||||
119
server/src/routes/ssrf-guard.ts
Normal file
119
server/src/routes/ssrf-guard.ts
Normal file
@@ -0,0 +1,119 @@
|
||||
import { lookup } from 'dns/promises';
|
||||
|
||||
const MAX_REDIRECTS = 5;
|
||||
|
||||
function isPrivateIp(ip: string): boolean {
|
||||
if (
|
||||
ip === '127.0.0.1' ||
|
||||
ip === '::1' ||
|
||||
ip === '0.0.0.0' ||
|
||||
ip === '::'
|
||||
)
|
||||
return true;
|
||||
|
||||
// 10.x.x.x
|
||||
if (ip.startsWith('10.'))
|
||||
return true;
|
||||
|
||||
// 172.16.0.0 - 172.31.255.255
|
||||
if (ip.startsWith('172.')) {
|
||||
const second = parseInt(ip.split('.')[1], 10);
|
||||
|
||||
if (second >= 16 && second <= 31)
|
||||
return true;
|
||||
}
|
||||
|
||||
// 192.168.x.x
|
||||
if (ip.startsWith('192.168.'))
|
||||
return true;
|
||||
|
||||
// 169.254.x.x (link-local, AWS metadata)
|
||||
if (ip.startsWith('169.254.'))
|
||||
return true;
|
||||
|
||||
// IPv6 private ranges (fc00::/7, fe80::/10)
|
||||
const lower = ip.toLowerCase();
|
||||
|
||||
if (lower.startsWith('fc') || lower.startsWith('fd') || lower.startsWith('fe80'))
|
||||
return true;
|
||||
|
||||
return false;
|
||||
}
|
||||
|
||||
export async function resolveAndValidateHost(url: string): Promise<boolean> {
|
||||
let hostname: string;
|
||||
|
||||
try {
|
||||
hostname = new URL(url).hostname;
|
||||
} catch {
|
||||
return false;
|
||||
}
|
||||
|
||||
// Block obvious private hostnames
|
||||
if (hostname === 'localhost' || hostname === 'metadata.google.internal')
|
||||
return false;
|
||||
|
||||
// If hostname is already an IP literal, check it directly
|
||||
if (/^[\d.]+$/.test(hostname) || hostname.startsWith('['))
|
||||
return !isPrivateIp(hostname.replace(/[[\]]/g, ''));
|
||||
|
||||
try {
|
||||
const { address } = await lookup(hostname);
|
||||
|
||||
return !isPrivateIp(address);
|
||||
} catch {
|
||||
return false;
|
||||
}
|
||||
}
|
||||
|
||||
export interface SafeFetchOptions {
|
||||
signal?: AbortSignal;
|
||||
headers?: Record<string, string>;
|
||||
}
|
||||
|
||||
/**
|
||||
* Fetches a URL while following redirects safely, validating each
|
||||
* hop against SSRF (private/reserved IPs, blocked hostnames).
|
||||
*
|
||||
* The caller must validate the initial URL with `resolveAndValidateHost`
|
||||
* before calling this function.
|
||||
*/
|
||||
export async function safeFetch(url: string, options: SafeFetchOptions = {}): Promise<Response | undefined> {
|
||||
let currentUrl = url;
|
||||
let response: Response | undefined;
|
||||
|
||||
for (let redirects = 0; redirects <= MAX_REDIRECTS; redirects++) {
|
||||
response = await fetch(currentUrl, {
|
||||
redirect: 'manual',
|
||||
signal: options.signal,
|
||||
headers: options.headers
|
||||
});
|
||||
|
||||
const location = response.headers.get('location');
|
||||
|
||||
if (response.status >= 300 && response.status < 400 && location) {
|
||||
let nextUrl: string;
|
||||
|
||||
try {
|
||||
nextUrl = new URL(location, currentUrl).href;
|
||||
} catch {
|
||||
break;
|
||||
}
|
||||
|
||||
if (!/^https?:\/\//i.test(nextUrl))
|
||||
break;
|
||||
|
||||
const redirectAllowed = await resolveAndValidateHost(nextUrl);
|
||||
|
||||
if (!redirectAllowed)
|
||||
break;
|
||||
|
||||
currentUrl = nextUrl;
|
||||
continue;
|
||||
}
|
||||
|
||||
break;
|
||||
}
|
||||
|
||||
return response;
|
||||
}
|
||||
Reference in New Issue
Block a user