This script decodes Google News generated encoded, internal URLs for RSS items

decodeGoogleNewsUrl.ts

/**
 * This magically uses batchexecute protocol. It's not documented, but it works.
 * 
 * Licensed under: MIT License
 *
 * Copyright (c) 2024 Ruslan Gainutdinov
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included
 * in all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
const fetchDecodedBatchExecute = (id: string) => {
  const s =
    '[[["Fbv4je","[\\"garturlreq\\",[[\\"en-US\\",\\"US\\",[\\"FINANCE_TOP_INDICES\\",\\"WEB_TEST_1_0_0\\"],null,null,1,1,\\"US:en\\",null,180,null,null,null,null,null,0,null,null,[1608992183,723341000]],\\"en-US\\",\\"US\\",1,[2,3,4,8],1,0,\\"655000234\\",0,0,null,0],\\"' +
    id +
    '\\"]",null,"generic"]]]';

  return fetch("https://news.google.com/_/DotsSplashUi/data/batchexecute?" + "rpcids=Fbv4je", {
    headers: {
      "Content-Type": "application/x-www-form-urlencoded;charset=utf-8",
      Referrer: "https://news.google.com/"
    },
    body: "f.req=" + encodeURIComponent(s),
    method: "POST"
  })
    .then(e => e.text())
    .then(s => {
      const header = '[\\"garturlres\\",\\"';
      const footer = '\\",';
      if (!s.includes(header)) {
        throw new Error("header not found: " + s);
      }
      const start = s.substring(s.indexOf(header) + header.length);
      if (!start.includes(footer)) {
        throw new Error("footer not found");
      }
      const url = start.substring(0, start.indexOf(footer));
      return url;
    });
};

/**
 * Google News started generate encoded, internal URLs for RSS items
 * https://news.google.com/rss/search?q=New%20York%20when%3A30d&hl=en-US&gl=US&ceid=US:en
 *
 * This script decodes URLs into original one, for example URL
 * https://news.google.com/__i/rss/rd/articles/CBMiSGh0dHBzOi8vdGVjaGNydW5jaC5jb20vMjAyMi8xMC8yNy9uZXcteW9yay1wb3N0LWhhY2tlZC1vZmZlbnNpdmUtdHdlZXRzL9IBAA?oc=5
 *
 * contains this
 * https://techcrunch.com/2022/10/27/new-york-post-hacked-offensive-tweets/
 *
 * In path after articles/ goes Base64 encoded binary data
 *
 * Format is the following:
 * <prefix> <len bytes> <URL bytes> <len bytes> <amp URL bytes> [<suffix>]
 *
 * <prefix> - 0x08, 0x13, 0x22
 * <suffix> - 0xd2, 0x01, 0x00 (sometimes missing??)
 * <len bytes> - formatted as 0x40 or 0x81 0x01 sometimes
 *
 *
 * https://news.google.com/rss/articles/CBMiqwFBVV95cUxNMTRqdUZpNl9hQldXbGo2YVVLOGFQdkFLYldlMUxUVlNEaElsYjRRODVUMkF3R1RYdWxvT1NoVzdUYS0xSHg3eVdpTjdVODQ5cVJJLWt4dk9vZFBScVp2ZmpzQXZZRy1ncDM5c2tRbXBVVHVrQnpmMGVrQXNkQVItV3h4dVQ1V1BTbjhnM3k2ZUdPdnhVOFk1NmllNTZkdGJTbW9NX0k5U3E2Tkk?oc=5
 * https://news.google.com/rss/articles/CBMidkFVX3lxTFB1QmFsSi1Zc3dLQkpNLThKTXExWXBGWlE0eERJQ2hLRENIOFJzRTlsRnM1NS1Hc2FlbjdIMlZ3eWNQa0JqeVYzZGs1Y0hKaUtTUko2dmJabUtVMWZob0lNSFNCa3NLQ05ROGh4cVZfVTYyUDVxc2c?oc=5
 * https://news.google.com/rss/articles/CBMiqwFBVV95cUxNMTRqdUZpNl9hQldXbGo2YVVLOGFQdkFLYldlMUxUVlNEaElsYjRRODVUMkF3R1RYdWxvT1NoVzdUYS0xSHg3eVdpTjdVODQ5cVJJLWt4dk9vZFBScVp2ZmpzQXZZRy1ncDM5c2tRbXBVVHVrQnpmMGVrQXNkQVItV3h4dVQ1V1BTbjhnM3k2ZUdPdnhVOFk1NmllNTZkdGJTbW9NX0k5U3E2Tkk?oc=5
 *
 * FIXME: What will happen if URL more than 255 bytes??
 *
 * Licensed under: MIT License
 *
 * Copyright (c) 2022 Ruslan Gainutdinov
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included
 * in all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
 * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
export const decodeGoogleNewsUrl = async (sourceUrl: string) => {
  const url = new URL(sourceUrl);
  const path = url.pathname.split("/");
  if (
    url.hostname === "news.google.com" &&
    path.length > 1 &&
    path[path.length - 2] === "articles"
  ) {
    const base64 = path[path.length - 1];
    let str = atob(base64);

    const prefix = Buffer.from([0x08, 0x13, 0x22]).toString("binary");
    if (str.startsWith(prefix)) {
      str = str.substring(prefix.length);
    }

    const suffix = Buffer.from([0xd2, 0x01, 0x00]).toString("binary");
    if (str.endsWith(suffix)) {
      str = str.substring(0, str.length - suffix.length);
    }

    // One or two bytes to skip
    const bytes = Uint8Array.from(str, c => c.charCodeAt(0));
    const len = bytes.at(0)!;
    if (len >= 0x80) {
      str = str.substring(2, len + 2);
    } else {
      str = str.substring(1, len + 1);
    }

    if (str.startsWith("AU_yqL")) {
      // New style encoding, introduced in July 2024. Not yet known how to decode offline.
      const url = await fetchDecodedBatchExecute(base64);
      return url;
    }

    return str;
  } else {
    return sourceUrl;
  }
};

A lot slower, but this is what i used for a while (uses selenium to access the url and get the redirection, ignore the usage of ValueError, it's just laziness lol)

playwright version

from playwright.sync_api import TimeoutError as PlaywrightTimeoutError
from playwright.sync_api import sync_playwright

def get_correct_url(url: str) -> str:
    """
    Convert Google News URL to its original source URL using Playwright.

    Args:
        url (str): The input URL to process

    Returns:
        str: The resolved URL pointing to the original news source

    Raises:
        ValueError: If URL resolution fails or hits Google's rate limit
        Exception: If browser automation fails
        TimeoutError: If URL resolution timed out after 60 seconds
    """
    if not url.startswith("https://news.google.com"):
        return url

    with sync_playwright() as p:
        # Launch browser with specific configurations
        browser = p.chromium.launch(
            headless=True,
            args=[
                "--disable-gpu",
                "--no-sandbox",
                "--disable-dev-shm-usage",
            ],
        )

        try:
            context = browser.new_context(
                user_agent="Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/91.0.4472.124 Safari/537.36"
            )
            page = context.new_page()

            # Navigate to URL and wait for redirect
            page.goto(url, wait_until="networkidle")
            try:
                page.wait_for_url(
                    lambda url: not url.startswith("https://news.google.com"),
                    timeout=60000,
                )
            except PlaywrightTimeoutError:
                raise ValueError("URL resolution timed out after 60 seconds")

            final_url = page.url
            if "google.com/sorry" in final_url:
                raise ValueError("Rate limited by Google (HTTP 429)")
            if "news.google" in final_url:
                raise ValueError("Failed to resolve original news URL")
            return final_url

        except Exception as e:
            raise Exception(f"Browser automation failed: {str(e)}")
        finally:
            browser.close()

google_news_url = "https://news.google.com/rss/articles/CBMipgFBVV95cUxPWV9fTEI4cjh1RndwanpzNVliMUh6czg2X1RjeEN0YUctUmlZb0FyeV9oT3RWM1JrMGRodGtqTk1zV3pkNEpmdGNxc2lfd0c4LVpGVENvUDFMOEJqc0FCVVExSlRrQmI3TWZ2NUc4dy1EVXF4YnBLaGZ4cTFMQXFFM2JpanhDR3hoRmthUjVjdm1najZsaFh4a3lBbDladDZtVS1FMHFn?oc=5"

print(get_correct_url(google_news_url))

sadly this method as of 17/02/2025 no longer works.. specifically decode_urls() part.

eg using php, php test-google-decode.php
Array
(
[signature] => AV_R3eBjl9LmyZGQgthZngaDg8y_
[timestamp] => 1739777504
[gn_art_id] => CBMipgFBVV95cUxPWV9fTEI4cjh1RndwanpzNVliMUh6czg2X1RjeEN0YUctUmlZb0FyeV9oT3RWM1JrMGRodGtqTk1zV3pkNEpmdGNxc2lfd0c4LVpGVENvUDFMOEJqc0FCVVExSlRrQmI3TWZ2NUc4dy1EVXF4YnBLaGZ4cTFMQXFFM2JpanhDR3hoRmthUjVjdm1najZsaFh4a3lBbDladDZtVS1FMHFn
)
Array
(
[signature] => AV_R3eAbCwyvr0UU5JgU_ZZW4E9h
[timestamp] => 1739777504
[gn_art_id] => CBMi3AFBVV95cUxOX01TWDZZN2J5LWlmU3hudGZaRDh6a1dxUHMtalBEY1c0TlJSNlpieWxaUkxUU19MVTN3Y1BqaUZael83d1ctNXhaQUtPM0IyMFc4R3VydEtoMmFYMWpMU1Rtc3BjYmY4d3gxZHlMZG5NX0s1RmR2ZXI5YllvdzNSd2xkOFNCUTZTaEp3b0IxZEJZdVFLUDBNMC1wNGgwMGhjRG9HRFpRZU5BMFVIYjZCOWdWcHI1YzdoVHFWYnZSOEFwQ0NubGx3Rzd0SHN6OENKMXZUcHUxazA5WTIw
)
)]}'

[["wrb.fr","Fbv4je",null,null,null,[3],""],["wrb.fr","Fbv4je",null,null,null,[3],""],["di",16],["af.httprm",16,"5177547088804027398",9]]Array
(
[0] =>
[1] =>
)

array elements are now blank via decode_urls($articles_params,$user_agent); - I added a user agent string to php code version, also handling ssl certificates.

@HopiumCurrency dunno, but my JS implementation is still working.

It's still working in my JS implementation too, and will probably give me a lot of headaches if it stops working lol

I'm having a problem with order of the urls.

Basically, i get the URL from a previous array of Article objects. then when decoding, i need i to be at the same order of previous Article array. But when returning from /batchexecute it just loses the order. Anyone has a solution for this? For now i may need to not "batch run it" but run one by one...

fyi my current implementation, tips are welcome

I didn't find any key or id-like that i could use to re-map the ordering, sadly

did you find any tips? thank you

I'm having a problem with order of the urls.
Basically, i get the URL from a previous array of Article objects. then when decoding, i need i to be at the same order of previous Article array. But when returning from /batchexecute it just loses the order. Anyone has a solution for this? For now i may need to not "batch run it" but run one by one...
fyi my current implementation, tips are welcome

I didn't find any key or id-like that i could use to re-map the ordering, sadly

did you find any tips? thank you

I literally gave up and used proxies, running one by one rotating proxies and user agents. It was the only way i found to guarantee order

Hey guys,

Has the process been impacted by google's recent update ? https://www.seroundtable.com/google-news-automatically-generated-publications-39149.html

seems like RSS works still not sure about url decoding https://news.google.com/rss?query=google

Yes, it's still working