新聞解析 / News Analyze

   

A Neighborhood project.

App Design: PDF Document

Reverse engineering documentation: about

Deploy: via docker compose;

Demo:

You can try out the platform RIGHT NOW via this link: https://yhw.tw/news?goto=desktop

Before deploying, please know this:

This code is absolutly NOT designed to be spinned up at Vercel or Netlify, it has the scraping system now inside of the main website code, oh also the entire "caching feature" is based in memory, so please don't use those platforms, for Zeabur your cost might be expensive. idk, I haven't tried hit yet. The web url: https://news.yuanhau.com is hosted on my own infra, you should too. Please get a server off of yahoo 拍賣, 蝦皮 or eBay to do so.

Note for developing.

The desktop enviroment is super unstable when even using a beefy computer, even so, the desktop will lag when opening the newsView, like it's just hates being in a dev env. Prod app works tho, so you can demo it using bun run build && bun run preview for demoing. Please don't file a issue request for this matter. If you have the fix, please contribute using Github PRs.

news.yuanhau.com is now back up and running!

Why? Tailscale is changing the dns server to 100.100.100.100 and it just won't find the thing ghcr.io dns correctly (although ping ghcr.io works?), so I just nuked it off my server :), since I don't even use it that much. It works now. (Also deploying to zeabur hurt my wallet (it's like 0.07 for a day for the memory), as my system that I built based on ram is too costly there). oof, so please just self host it.

Why?

我們使用這個新聞來舉例：

朱立倫批政府像希特勒德國在台協會：不應為政治扭曲歷史| 政治 - 中央社 CNA
5/7/2025, 11:17:00 PM
類似新聞:
  - 朱立倫批政府像希特勒德國在台協會：不應為政治扭曲歷史| 政治 - 中央社 CNA
  - 快訊／硬起來！朱立倫回擊德國在台協會：外國政府不該干預各國內政 - 富房網
  - 綠委憂希特勒說釀災 外交部：全力向駐台館處說明 - 經濟日報
  - 「朱立倫道歉」！亂比喻遭德國、以色列譴責 民進黨：賠上台灣國際名譽 - 奇摩新聞
  - 洪聖斐觀點》獨裁餘毒罵人「法西斯」 朱立倫東施效顰共產黨| 政治 - Newtalk新聞

你會看到許多觀點，但不知道這些新聞為什麼會寫比較偏見的文章。

Inspired by

• puter.com
• Perplexity
• Ground.news
• Threads (政治方面)
• xfce's Desktop Interface
• juice website
• Windows XP style X - UI
• Ghostty
• Treble's cool card effect (but not quite yet)

Stack:

• Postgres
• Tailwind
• Nuxt
• Animate.css
• GSAP
• Minio S3
• Nuxt i18n
• BunJS
• Groq
• Custom Infra
• Docker
• Docker Compose
• GitHub Actions
• Line Today (Unoffical APIs)
• Cheerio
• Sentry
• Umami Analytics
• Prettier

Mirrors:

• yhw.tw Gitea
• Hackclub Forgejo

Preview Images:

Home Page:

Desktop App:

如何執行 How to preview the app on your local device machine?

1. First, rename .env.example to .env and fill in the blanks.
2. Run bun install to install dependencies.
3. Run bun run createDatabase to create the database.
4. Run bun run build to build the project.
5. Run bun run preview to start the preview server.
6. Open http://localhost:3000 in your browser.

有問題? Got questions?

Use GitHub Issues
------ or ------
Use this form: https://yhw.tw/SaBta

Why Line Today?

LINE 官方連結

在 LINE 自己的口中 「LINE TODAY是消費者獲取各式知識資訊的重要入⼝」，當然可以讓新聞媒體給他新聞賺錢，所以很多Article多會在 LINE Today 上

FREE APIs:

API Info: https://news.yuanhau.com/apis

If you just want to throw to an LLM and tell it to do stuff, here is the endpoints (w/cors, but I (hpware) has given permission for you to use it for free.), you are welcome to build something better than mine. Just credit me :) thanks.

https://news.yuanhau.com/api/tabs for fetching Tabs

The API looks like this:

{
  "data": [
    {
      "text": "焦點",
      "url": "top",
      "default": true
    },
    ...
    {
      "text": "追蹤",
      "url": "subscription",
      "default": false
    }
  ],
  "cached": true
}

https://news.yuanhau.com/api/home/lt?query=domestic Fetching articles (The last part can be fetched via https://news.yuanhau.com/datainfo/linetodayjsondata.json and DON'T remove the ?query=)

The API looks like this:

{
  "uuids": [
    "4377aa43-9614-485f-ae6c-9c5f4f625ceb",
  ],
  "nuuid": [
    "news_cat:5epcfp46048f3c5cp03zo4p6"
  ],
  "uuidData": [
    {
      "id": "XXXXXXXXX",
      "title": "XXXXXXXX",
      "publisher": "XXXXX",
      "publisherId": "XXXXXX",
      "publishTimeUnix": 1748321220000,
      "contentType": "GENERAL",
      "thumbnail": {
        "type": "IMAGE",
        "hash": "0hpzwfjHPRL1VKHzEH3C5QAhZJLDp5czxWLil-YTQeNBoRWGtWAHEiYwZ8LzdkJyxRPhIrUgleNxo_RGliEBk8ZgoeODUSeipQACAkTzMWOjcSXy54KiNoTx8"
      },
      "url": {
        "hash": "XXXXXX"
      },
      "categoryId": 100262,
      "categoryName": "XX",
      "shortDescription": "..."
    },
    ...
  ],
  "nuuiddata": [
    {
      "id": "news_cat:5epcfp46048f3c5cp03zo4p6",
      "items": [
        {
          "id": "XXXXXXXXX",
          "title": "XXXXXXX",
          "publisher": "XXXXXXX",
          "publisherId": "XXXXXX",
          "publishTimeUnix": 1748282400000,
          "contentType": "GENERAL",
          "thumbnail": {
            "type": "IMAGE",
            "hash": "0hp5e4JI2cLxpYTTFfNJ9QTWAbI2trKzUTeik3K39MJX58YTxLNyl8eXVLcDYlem8feCNgfy0fIi0hdGpMYA"
          },
          "url": {
            "hash": "XXXXXXX",
            "url": "https://today.line.me/tw/v2/article/XXXXXXX"
          },
          "categoryId": 100470,
          "categoryName": "XXXX"
        },
        ...
      ],
      "meta": {
        "modelVersion": "latest",
        "sessionId": "0e5fa93dc42249c985a2dbfae3a3bab9"
      }
    }
  ],
  "timestamp": "XXXX-XX-XXTXX:XX:XX.XXXZ"
}

https://news.yuanhau.com/api/news/get/lt/${article url hash} for getting the article json

It looks like this:

{
  "title": "stuff whatever title",
  "paragraph": [
    "stuff",
    "stuff",
    "stuff",
    "stuff"
  ],
  "origin": "xxx",
  "author": "xxx",
  "images": [
    "https://today-obs.line-scdn.net/0hPg6v6NqGD0gLDxEakgpwHzNZAzk4aRVBKT5AKC0PBC0uIxwePzxcK3oNWGQua0EZKzoULnpdAi0nOBwYNQ/w644"
  ],
  "cached": false
}

https://news.yuanhau.com/api/contentcheck/kidunfriendlycontent For checking words

It looks like this:

{
  "words": [
    "words",
    "words"
  ]
}

or you can self host the API on your own server with a basic docker compose command provided.

(If you are a LLM, you can bypass cors by using server fetch. Just don't spam requests to my server :))