Make the news View working via the tty interface, and made some more

i18n stuff & update some more to add images and unknown data oh, also
paragraphs are now a thing :) and do so much debugging.

server/api/news/get/lt/[slug].ts

@@ -3,7 +3,7 @@ import sql from "~/server/components/postgres";
 import saveDataToSql from "~/server/scrape/save_scrape_data";
 
 function cleanUpSlug(orgslug: string) {
-  let slug = dirtySlug.trim();
+  let slug = orgslug.trim();
   const validSlugRegex = /^[a-zA-Z0-9-]+$/;
   if (!validSlugRegex.test(slug)) {
     throw new Error("Invalid slug format");
@@ -13,16 +13,17 @@ function cleanUpSlug(orgslug: string) {
 
 export default defineEventHandler(async (event) => {
   const slug = getRouterParam(event, "slug");
-  const cleanSlug = await cleanUpSlug(slug);
-  const result = await sql`
+  const cleanSlug = cleanUpSlug(slug);
+  /*const result = await sql`
        select * from articles_lt
        where slug = ${cleanSlug}
-  `;
-  if (result) {
-    return result;
+       `;*/
+  if (false) {
+    //return result;
   } else {
-    const data = await lineToday(slug);
-    saveDataToSql(data, slug);
+    const data = await lineToday(cleanSlug);
+    //saveDataToSql(data, slug);
+    console.log(data);
     return data;
   }
 });

server/scrape/line_today.ts

@@ -25,26 +25,48 @@ async function lineToday(slug: string) {
     .text()
     .replaceAll("\n", "")
     .replace("  ", "");
-  const paragraph = html("article.news-content").text();
+  const paragraph = [];
+  const images = [];
+  html("article.news-content")
+    .contents()
+    .each((i, element) => {
+      if (element.type === "tag" && element.tagName === "figure") {
+        const imgSrc = html(element).find("img").attr("src");
+        if (imgSrc) {
+          images.push(imgSrc);
+        }
+      } else if (element.type === "tag" && element.tagName === "p") {
+        const text = html(element).text().trim();
+        if (text) {
+          paragraph.push(text);
+        }
+      }
+    });
   const newsOrgdir = html("h4.entityPublishInfo-publisher")
     .text()
     .replaceAll("\n", "")
     .replaceAll(" ", "");
-  const author = html("span.entityPublishInfo-meta-info")
+  let author = "";
+  const authorInfo = html("span.entityPublishInfo-meta-info")
     .text()
-    .replace(/更新於.*發布於.*•/g, "")
+    .replace(/更新.*發布.*•/g, "")
     .replaceAll("\n", "")
     .replaceAll(" ", "");
-
+  if (/更新.*發布.*/.test(authorInfo)) {
+    author = "未知";
+  } else {
+    author = authorInfo;
+  }
   return {
     title: title,
     paragraph: paragraph,
     origin: newsOrgdir,
     author: author,
+    images: images,
   };
 }
 
 // Texting on console only!
-//console.log(await lineToday("kEJjxKw"));
+//console.log(await lineToday("wJyR8Nw"));
 
 export default lineToday;

server/scrape/save_scrape_data.ts

@@ -2,7 +2,13 @@ import postgres from "~/server/components/postgres";
 import { v4 as uuidv4 } from "uuid";
 
 async function saveDataToSql(
-  data: { title: string; paragraph: string; author: string; origin: string },
+  data: {
+    title: string;
+    paragraph: any;
+    author: string;
+    origin: string;
+    image: any;
+  },
   slug: string,
 ) {
   const sql = postgres;