use WuxiaWorld api & cheerio instead of puppeteer

This commit is contained in:
Rajarshee Chatterjee
2021-05-02 18:13:30 +05:30
parent 72d7adfb15
commit 4f6d025f73
+95 -93
View File
@@ -1,39 +1,37 @@
const express = require("express"); const express = require("express");
const cheerio = require("cheerio"); const cheerio = require("cheerio");
const request = require("request"); const request = require("request");
const chromium = require("chrome-aws-lambda");
const puppeteer = require("puppeteer");
const { scraper } = require("../../helper");
const baseUrl = "https://www.wuxiaworld.com/"; const baseUrl = "https://www.wuxiaworld.com/";
const router = express.Router(); const router = express.Router();
router.get("/novels/", async (req, res) => { router.get("/novels/", async (req, res) => {
let url = `https://www.wuxiaworld.com/novels`; let url = `https://www.wuxiaworld.com/api/novels`;
const body = await scraper(url);
let novels = []; request(url, (err, response, body) => {
if (err) throw err;
const $ = cheerio.load(body); let novels = [];
$(".novel-item").each(function (result) { let data = JSON.parse(body);
let novelName = $(this).find("div.novel-title").text(); data = data.items;
let novelCover = $(this).find("img").attr("src");
let novelUrl = $(this).find("a").attr("href");
novelUrl = novelUrl.replace("/novel/", "").concat("/");
novel = { data.map((novel) => {
extensionId: 7, let novelName = novel.name;
novelName, let novelCover = novel.coverUrl;
novelCover, let novelUrl = novel.slug;
novelUrl, novelUrl = novelUrl + "/";
};
novels.push(novel); novels.push({
extensionId: 7,
novelUrl,
novelName,
novelCover,
});
});
res.json(novels);
}); });
res.json(novels);
}); });
// Novel // Novel
@@ -41,67 +39,69 @@ router.get("/novels/", async (req, res) => {
router.get("/novel/:novelUrl", async (req, res) => { router.get("/novel/:novelUrl", async (req, res) => {
let novelUrl = req.params.novelUrl; let novelUrl = req.params.novelUrl;
let url = `${baseUrl}novel/${novelUrl}/`; let url = `${baseUrl}novel/${novelUrl}/`;
const body = await scraper(url);
const $ = cheerio.load(body); request(url, (err, response, body) => {
if (err) throw err;
const $ = cheerio.load(body);
let novel = {}; let novel = {};
novel.extensionId = 7; novel.extensionId = 7;
novel.sourceName = "WuxiaWorld"; novel.sourceName = "WuxiaWorld";
novel.sourceUrl = url; novel.sourceUrl = url;
novel.novelUrl = `${novelUrl}/`; novel.novelUrl = `${novelUrl}/`;
novel.novelName = $("h2").text(); novel.novelName = $("h2").text();
novel.novelCover = $("img.img-thumbnail").attr("src"); novel.novelCover = $("img.img-thumbnail").attr("src");
novel.novelSummary = $(".fr-view > p").text(); novel.novelSummary = $(".fr-view > p").text();
novel["Author(s)"] = $("div > dt") novel["Author(s)"] = $("div > dt")
.filter(function () { .filter(function () {
return $(this).text().trim() === "Author:"; return $(this).text().trim() === "Author:";
}) })
.next() .next()
.text(); .text();
let genres = []; let genres = [];
$(".genres") $(".genres")
.find("div") .find("div")
.each(function (res) { .each(function (res) {
genres.push($(this).find("a").text()); genres.push($(this).find("a").text());
});
novel["Genre(s)"] = genres.join(",");
novel["Artist(s)"] = null;
novel.Status = null;
let novelChapters = [];
$(".chapter-item").each(function (result) {
let chapterName = $(this).text();
chapterName = chapterName.replace(/[\t\n]/g, "");
let releaseDate = null;
let chapterUrl = $(this).find("a").attr("href");
chapterUrl = chapterUrl.replace(`/novel/${novelUrl}/`, "");
novelChapters.push({
chapterName,
releaseDate,
chapterUrl,
});
}); });
novel["Genre(s)"] = genres.join(","); novel.novelChapters = novelChapters;
novel["Artist(s)"] = null; res.json(novel);
novel.Status = null;
let novelChapters = [];
$(".chapter-item").each(function (result) {
let chapterName = $(this).text();
chapterName = chapterName.replace(/[\t\n]/g, "");
let releaseDate = null;
let chapterUrl = $(this).find("a").attr("href");
chapterUrl = chapterUrl.replace(`/novel/${novelUrl}/`, "");
novelChapters.push({
chapterName,
releaseDate,
chapterUrl,
});
}); });
novel.novelChapters = novelChapters;
res.json(novel);
}); });
router.get("/novel/:novelUrl/:chapterUrl", async (req, res) => { router.get("/novel/:novelUrl/:chapterUrl", async (req, res) => {
@@ -109,41 +109,43 @@ router.get("/novel/:novelUrl/:chapterUrl", async (req, res) => {
const novelUrl = `${req.params.novelUrl}/`; const novelUrl = `${req.params.novelUrl}/`;
const chapterUrl = `${req.params.chapterUrl}/`; const chapterUrl = `${req.params.chapterUrl}/`;
const body = await scraper(url);
const $ = cheerio.load(body); request(url, (err, response, body) => {
if (err) throw err;
const $ = cheerio.load(body);
let chapterName = $("#sidebar-toggler-container").next().text(); let chapterName = $("#sidebar-toggler-container").next().text();
chapterName = chapterName.replace(/[\t\n]/g, ""); chapterName = chapterName.replace(/[\t\n]/g, "");
let chapterText = $("#chapter-content").text(); let chapterText = $("#chapter-content").text();
let nextChapter = null; let nextChapter = null;
if ($("li.next > a").attr("href") !== "#") { if ($("li.next > a").attr("href") !== "#") {
nextChapter = $("li.next > a") nextChapter = $("li.next > a")
.attr("href") .attr("href")
.replace(`/novel/${novelUrl}`, ""); .replace(`/novel/${novelUrl}`, "");
} }
let prevChapter = null; let prevChapter = null;
if ($("li.prev > a").attr("href") !== "#") { if ($("li.prev > a").attr("href") !== "#") {
prevChapter = $("li.prev > a") prevChapter = $("li.prev > a")
.attr("href") .attr("href")
.replace(`/novel/${novelUrl}`, ""); .replace(`/novel/${novelUrl}`, "");
} }
chapter = { chapter = {
extensionId: 7, extensionId: 7,
novelUrl, novelUrl,
chapterUrl, chapterUrl,
chapterName, chapterName,
chapterText, chapterText,
nextChapter, nextChapter,
prevChapter, prevChapter,
}; };
res.json(chapter); res.json(chapter);
});
}); });
router.get("/search/", (req, res) => { router.get("/search/", (req, res) => {