website-api 1.1.9 → 1.1.10

package/dist/src/capabilities/proxy.d.ts

@@ -0,0 +1,27 @@
+/**
+ * Normalise a user-supplied proxy value into a URL string suitable for
+ * `HTTP_PROXY`/`HTTPS_PROXY` env vars. Mirrors chrome-cdp-manager's
+ * `parseProxy` logic so both the browser and HTTP transport see the same proxy.
+ *
+ * Accepted forms:
+ *   - `true` / `"default"` → `socks5://127.0.0.1:1080`
+ *   - `"1080"`             → `socks5://127.0.0.1:1080`
+ *   - `"host:port"`        → `socks5://host:port`
+ *   - full URL             → as given
+ *
+ * Returns `null` on invalid input so callers can skip silently.
+ */
+export declare function normalizeProxyUrl(value: string | boolean): string | null;
+/**
+ * Configure Node's global `fetch` to route through a proxy. Call once before
+ * any HTTP capability requests are made.
+ *
+ * Uses `http.setGlobalProxyFromEnv()` (Node ≥ 22) which teaches the built-in
+ * undici-backed `fetch` to honour `HTTP_PROXY` / `HTTPS_PROXY` env vars.
+ *
+ * This complements the browser-transport proxy (chrome-cdp-manager passes
+ * `--proxy-server` to Chrome) by also routing direct HTTP capability requests.
+ *
+ * @returns The resolved proxy URL if applied, or `null` if skipped.
+ */
+export declare function applyHttpProxy(proxyValue: string | boolean | undefined, debug?: boolean): string | null;

package/dist/src/capabilities/proxy.js

@@ -0,0 +1,71 @@
+import { setGlobalProxyFromEnv } from "node:http";
+// mark: Constants
+const DEFAULT_PROXY_SCHEME = "socks5";
+const DEFAULT_PROXY_HOST = "127.0.0.1";
+const DEFAULT_PROXY_PORT = 1080;
+const DEFAULT_PROXY = `${DEFAULT_PROXY_SCHEME}://${DEFAULT_PROXY_HOST}:${DEFAULT_PROXY_PORT}`;
+// mark: normalizeProxyUrl
+/**
+ * Normalise a user-supplied proxy value into a URL string suitable for
+ * `HTTP_PROXY`/`HTTPS_PROXY` env vars. Mirrors chrome-cdp-manager's
+ * `parseProxy` logic so both the browser and HTTP transport see the same proxy.
+ *
+ * Accepted forms:
+ *   - `true` / `"default"` → `socks5://127.0.0.1:1080`
+ *   - `"1080"`             → `socks5://127.0.0.1:1080`
+ *   - `"host:port"`        → `socks5://host:port`
+ *   - full URL             → as given
+ *
+ * Returns `null` on invalid input so callers can skip silently.
+ */
+export function normalizeProxyUrl(value) {
+    if (value === true)
+        return DEFAULT_PROXY;
+    let raw = String(value).trim();
+    if (raw === "" || raw.toLowerCase() === "default")
+        return DEFAULT_PROXY;
+    // Port-only shorthand: "1080"
+    if (/^\d+$/.test(raw)) {
+        return `${DEFAULT_PROXY_SCHEME}://${DEFAULT_PROXY_HOST}:${raw}`;
+    }
+    // host:port shorthand without scheme
+    if (!raw.includes("://")) {
+        raw = `${DEFAULT_PROXY_SCHEME}://${raw}`;
+    }
+    try {
+        const url = new URL(raw);
+        const scheme = url.protocol.replace(/:$/, "").toLowerCase();
+        const host = url.hostname || DEFAULT_PROXY_HOST;
+        const port = url.port ? Number(url.port) : scheme.startsWith("socks") ? DEFAULT_PROXY_PORT : 8080;
+        return `${scheme}://${host}:${port}`;
+    }
+    catch {
+        return null;
+    }
+}
+// mark: applyHttpProxy
+/**
+ * Configure Node's global `fetch` to route through a proxy. Call once before
+ * any HTTP capability requests are made.
+ *
+ * Uses `http.setGlobalProxyFromEnv()` (Node ≥ 22) which teaches the built-in
+ * undici-backed `fetch` to honour `HTTP_PROXY` / `HTTPS_PROXY` env vars.
+ *
+ * This complements the browser-transport proxy (chrome-cdp-manager passes
+ * `--proxy-server` to Chrome) by also routing direct HTTP capability requests.
+ *
+ * @returns The resolved proxy URL if applied, or `null` if skipped.
+ */
+export function applyHttpProxy(proxyValue, debug = false) {
+    if (!proxyValue)
+        return null;
+    const proxyUrl = normalizeProxyUrl(proxyValue);
+    if (!proxyUrl)
+        return null;
+    process.env.HTTP_PROXY = proxyUrl;
+    process.env.HTTPS_PROXY = proxyUrl;
+    setGlobalProxyFromEnv();
+    if (debug)
+        console.log(`[debug] HTTP proxy set to ${proxyUrl}`);
+    return proxyUrl;
+}

package/dist/src/core/context.js

@@ -1,9 +1,12 @@
+// mark: Imports
 import { resolve } from "node:path";
 import { connectChrome } from "../capabilities/browser.js";
 import { buildCookieString, resolveCookies, resolveCredentials, resolveUserAgent, } from "../capabilities/cookies.js";
 import { createSaver } from "../capabilities/download.js";
 import { applyFingerprint } from "../capabilities/fingerprint.js";
 import { createHttp } from "../capabilities/http.js";
+import { applyHttpProxy } from "../capabilities/proxy.js";
+// mark: Helpers
 /** Treats `1`/`true`/`yes`/`on` (case-insensitive) as truthy for env flags. */
 function isTruthyEnv(value) {
     if (!value)
@@ -23,6 +26,7 @@ function resolveHeadless(headed, headlessEnv) {
         return isTruthyEnv(headlessEnv);
     return true;
 }
+// mark: createContext
 /**
  * Builds the lazy capability context handed to `site.run(ctx)`. Nothing
  * expensive happens until a capability is touched: cookies are read on first
@@ -57,6 +61,8 @@ export function createContext(site, options = {}, providers = {}) {
         }
         return credentialsCache;
     };
+    // ── proxy for HTTP transport ──
+    applyHttpProxy(options.proxy ?? env.CDP_PROXY, debug);
     const http = createHttp({ fetchImpl: providers.fetchImpl, cookieString, userAgent, debug });
     // Host-provided cheerio: extensions can't resolve their own node_modules, so
     // they parse HTML through this instead of importing cheerio. Lazily imported

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "website-api",
-  "version": "1.1.9",
+  "version": "1.1.10",
   "description": "CLI and library to query website private APIs with your real logged-in Chrome session",
   "main": "./dist/src/website-api.js",
   "types": "./dist/src/website-api.d.ts",