Archive the jssoup one for now

This commit is contained in:
yuanhau 2025-05-18 11:02:32 +08:00
parent 5f78e8c58a
commit 0bcb646cc0
4 changed files with 45 additions and 19 deletions

View file

@ -1,20 +1,7 @@
import lineToday from "~/server/scrape/line_today";
export default defineEventHandler(async (event) => {
const slug = getRouterParam(event, "slug");
return new Promise((resolve, reject) => {
const pythonProcess = spawn("python3", ["scraping/hot_articles.py"]);
let dataString = "";
pythonProcess.stdout.on("data", (data) => {
dataString += data.toString();
});
pythonProcess.stderr.on("data", (data) => {
console.error(`Error: ${data}`);
});
pythonProcess.on("close", (code) => {
resolve({ status: "completed", output: dataString });
});
});
const data = await lineToday(slug);
console.log(data);
return data;
});

View file

@ -1,4 +1,5 @@
import JSSoup from "jssoup";
//import cheerio from "cheerio";
async function lineToday(slug: string) {
const url = "https://today.line.me/tw/v2/article/" + slug;
@ -16,8 +17,20 @@ async function lineToday(slug: string) {
"Cache-Control": "max-age=0",
},
});
const data = fetchPageCode.text();
const soup = new JSSoup(data);
// 幹 又忘了 await
const data = await fetchPageCode.text();
// 加 await? no.
// AHHH I NEED TO CHANGE TO SOMETHING ELSE.
const soup = new JSSoup(data, false);
const titlesoup = soup.find("h1", "entityTitle");
const title = titlesoup.text.replaceAll("\n", "");
const article = soup.find("article", "news-content");
const paragraph = article.text;
return {
title: title,
paragraph: paragraph,
};
}
export default lineToday;