added digitimes, chinatimes, xfastest

app.js

@@ -7,8 +7,7 @@ const port = 3001;
 app.use(cors());
 
 app.get("/test", async (req, res) => {
-  var data = await scrapeContent(decodeURI(req.query.url));
-  res.json({ data });
+  res.json(await scrapeContent(decodeURI(req.query.url)));
 });
 
 app.listen(port, () => {

example.http

@@ -2,7 +2,7 @@ GET http://localhost:3000/test?url=https%3A%2F%2Fmoney.udn.com%2Fmoney%2Fstory%2
 
 
 ###
-GET http://localhost:3000/test?url=https%3A%2F%2Fudn.com%2Fnews%2Fstory%2F7240%2F8145957 HTTP/1.1
+GET http://localhost:3001/test?url=https%3A%2F%2Fwww.xfastest.com%2Fthread-290981-1-1.html HTTP/1.1
 
 
 ###

lib/scrapeContent.js

@@ -131,7 +131,9 @@ async function scrapeContent(link) {
           !text.startsWith("googletag") &&
           !text.startsWith("▲") &&
           !text.startsWith("訂閱手機王，快速掌握") &&
-          !text.startsWith("現在，你也可以同步追蹤")
+          !text.startsWith("現在，你也可以同步追蹤") &&
+          !text.includes("訂閱追蹤") &&
+          !text.includes("手機王網友")
       );
       content = textArray;
       break;
@@ -193,7 +195,10 @@ async function scrapeContent(link) {
       const udnresult = udnmatch ? udnmatch[1] : "";
 
       date_source_author =
-        $("time.article-content__time").text().replace(/\//g, "-").slice(0, 10) +
+        $("time.article-content__time")
+          .text()
+          .replace(/\//g, "-")
+          .slice(0, 10) +
         " / 經濟日報 / " +
         $("a[href^='/news/reporter/']").text();
       content = [];
@@ -203,31 +208,118 @@ async function scrapeContent(link) {
         }
       });
       break;
-///////
-
+    ///////
 
+    ////digitimes
+    case link.includes("digitimes"):
+      title = $("h1.news-title").text();
+      date_source_author =
+        $("time").text() + " / 電子時報 / " + $("font").first().text();
+      content = [];
+      break;
+    /////
 
+    /////chinatimes
     case link.includes("chinatimes"):
+      title = $("h1.article-title").text();
+      date_source_author =
+        $("span.date").first().text() +
+        " / " +
+        $("div.source").text() +
+        " / " +
+        $("div.author").text();
+
+      content = [];
+      $("div.article-body p").each((index, element) => {
+        if (!$(element).text().trim() == "") {
+          content.push($(element).text().trim());
+        }
+      });
       break;
+    //////
+
     case link.includes("ctimes"):
       break;
+
+    ///////kocpc
     case link.includes("kocpc"):
+      title = $("h1.jeg_post_title").text();
+      date_source_author =
+        $('a[href="' + link + '"]')
+          .text()
+          .replace(/(\d{4}) 年 (\d{2}) 月 (\d{2}) 日/, "$1-$2-$3") +
+        " / 電腦王阿達 / " +
+        $("div.jeg_meta_author a").text();
+      //Zi_ad_ar_iR
+      content = [];
+      $("div.Zi_ad_ar_iR p").each((index, element) => {
+        if (
+          !$(element).text().trim() == "" &&
+          !$(element).text().startsWith("<img")
+        ) {
+          content.push($(element).text().trim());
+        }
+      });
+      //kocpc not done
       break;
+    //////
+
     case link.includes("3c.ltn"):
+      title = ""
+      date_source_author = ""
+      content = []
       break;
+
     case link.includes("ec.ltn"):
       break;
-    case link.includes("ltn"):
-      break;
+
+
+    //////XFastest
     case link.includes("xfastest"):
+      const formattedDate = $('span[title^="202"]')
+        .attr("title")
+        .replace(
+          /(\d{4})-(\d{1,2})-(\d{1,2}) .*/,
+          (match, year, month, day) => {
+            // Pad month and day with leading zeros if necessary
+            return `${year}-${month.padStart(2, "0")}-${day.padStart(2, "0")}`;
+          }
+        );
+      title = $("span#thread_subject").text();
+      date_source_author =
+        formattedDate +
+        " / XFastest / " +
+        $("a[href^='space-uid']").last().text();
+      content = [];
+      $('td[id^="post"]').each((index, element) => {
+        // Get innerHTML of the selected <td> element
+        const innerHTML = $(element).html();
+
+        // Split the innerHTML by <br></br> tag
+        content = innerHTML
+          .split(/<\/?[^>]+>/)
+          .map((item) => item.trim()) // Trim all elements
+          .filter((item) => item !== "")
+          .filter((item) => !item.includes("&nbsp;"))
+          .filter((item) => item !== "上傳")
+          .filter((item) => item !== "消息來源")
+          .filter((item) => !item.includes("jpg"))
+          .filter((item) => !item.startsWith("("))
+          .filter((item) => item !== "下載附件")
+          .filter((item) => item !== "保存到相冊");
+      });
+
       break;
+    //////
+
     case link.includes("cnyes"):
       break;
+
     case link.includes("moneydj"):
-        break;
-    case link.includes("investor"):
       break;
 
+    case link.includes("investor"):
+      break;
   }
 
   return { title, date_source_author, link, content };

package.json

@@ -12,6 +12,7 @@
   "license": "ISC",
   "packageManager": "[email protected]+sha512.140036830124618d624a2187b50d04289d5a087f326c9edfc0ccd733d76c4f52c3a313d4fc148794a2a9d81553016004e6742e8cf850670268a7387fc220c903",
   "dependencies": {
+    "cheerio": "1.0.0-rc.12",
     "cors": "^2.8.5",
     "docx": "^8.5.0",
     "express": "^4.19.2",