Scraping Line Today's home page system

This took me some time, but they use a fancy system for pulling news data.

Endpoint on news.yuanhau.com aka this repo (Cached results)

/api/home/uuid_lt/feed?query=${query}

Fetches the uuid and returns back with the news

Main endpoint

For local Taiwan news they use this url: https://today.line.me/_next/data/v1/tw/v3/tab/domestic.json?tabs=domestic

From _next? I thought that is static? I mean it maybe is, it is just providing with the URLs that the client will be fetching to the server, which is a bit fun.

Here is a JSON snippet:

{
  "id": "the-news-id",
  "type": "HIGHLIGHT",
  "containerStyle": "Header",
  "name": "國內話題：topic",
  "source": "LISTING",
  "header": {
    "title": "the top title here",
    "hasCompositeTitle": false,
    "subTitle": "the news subtitle here"
  },
  "listings": [
    {
      "id": "1feef7d2-3acc-495d-becd-3ef4de6a92ce",
      "offset": 0,
      "length": 10
    }
  ]
},

We can ignore everything else, other than the strange UUID in the json. Well, this is the key we need to fetch their api 🤩

Fetch news URLs

Here is the fancy URL: https://today.line.me/api/v6/listings/{the-uuid-you-got-in-the-listings-json-file}/?country=tw&offset=0&length=24

This api can be used for fetching the news from them, however, there is an issue, the max length is only just 24 (yes, I tried it only can return 27* when requesting for 1000)

And viewing the JSON, oh would you look at that.

{
  "id": "news-id",
  "title": "news-title",
  "publisher": "news-publisher",
  "publisherId": "101366",
  "publishTimeUnix": 1747670221000,
  "contentType": "GENERAL",
  "thumbnail": {
    "type": "IMAGE",
    "hash": "0hvoq7de5NKUBMTjcMHM5WF3QYJTF_KDNJbixudj4ddXJnYm4ecX16Iz0edWwydjsTbH9vdm5IJ3EyKjtBeA"
  },
  "url": {
    "hash": "8nlkYeV"
  },
  "categoryId": 100262,
  "categoryName": "國內",
  "shortDescription": "The article's short description"
},

The url hash is just what we needed to use my scraper :D

You can query it by using: https://news.yuanhau.com/api/news/get/lt/8nlkYeV (Also videos are in the list, so avoid that) or just try this https://today.line.me/tw/v2/article/8nlkYeV

and that's it, I've bypassed Line's attempt to block people like me. :)

More to this debuckle

Apperently, there is something called a "hybrid listing" which is s simple recommendation system from here: https://today.line.me/webapi/recommendation/hybrid/listings/${id}?country=tw&maxVideoCount=0&offset=0&length=70&optOut=false, the ID still can be obtained via the same main endpoint, but just different things. Unlike the other api endpoint, it has a higher limit of 70, so you can do more things with it, and this endpoint is even easier to parse via json, just look at it.

{
  "id": "id",
  "items": [
    {
      "id": "news-id",
      "title": "title",
      "publisher": "publisher",
      "publisherId": "100005",
      "publishTimeUnix": 1747712924000,
      "contentType": "GENERAL",
      "thumbnail": {
        "type": "IMAGE",
        "hash": "image hash"
      },
      "url": {
        "hash": "67676767",
        "url": "https://today.line.me/tw/v2/article/67767676"
      },
      "categoryId": 100260,
      "categoryName": "cattype"
    }
  ]
}

This is 100% easier to work with, and with a another extra, I can easily search shitty news terms. Also there is as category type??? What?

Also the id can just work with the following pattern in regex: news_cat:[a-zA-Z0-9]{24}, there is also top_foryou:[a-zA-Z0-9]{24}

Hybrid listings?

news_cat
top_foryou

3.5 KiB Raw Blame History Unescape Escape