@davidpunya/web-scraper/library/system/nsfw.js

The library web scraper for Restfull API's

const axios = require("axios");
const cheerio = require("cheerio");
const fs = require("fs");
const qs = require("qs");
const path = require("path");
const util = require("util");
const fetch = (...args) => import('node-fetch').then(({default: fetch}) => fetch(...args));
const Tools = new (require("./tools"));

module.exports = class NSFW {
    nhentaiGet = async (code) => {
        try {
            const baseUrl = "https://nhentai.net";
            const galleryUrl = `${baseUrl}/g/${code}`;
            const { headers } = await axios.get(baseUrl);
            const cookie = headers["set-cookie"].join("; ");
            const response = await axios.get(galleryUrl, {
                headers: {
                    Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
                    Cookie: cookie,
                    Referer: baseUrl,
                    "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
                }
            });
            if (response.status === 404) {
                console.log(JSON.stringify({
                    status: 404,
                    creator: "David XD",
                    msg: "Cannot Find " + code + " :/"
                }));
                return console.log(JSON.stringify({
                    status: 404,
                    creator: "David XD",
                    msg: "Cannot Find " + code + " :/"
                }));
            };
            const $ = cheerio.load(response.data);
            const thumburl = $("#content #cover img").attr("data-src");
            const mediaID = thumburl.split("/")[4];
            const titleEN = {
                title: $("#info h1.title").text(),
                pretty: $("#info h1.title .pretty").text()
            };
            const titleJP = {
                title: $("#info h2.title").text(),
                pretty: $("#info h2.title .pretty").text()
            };
            const galleryId = $("#info #gallery_id").text().replace("#", "");
            const thumb = {
                img: thumburl,
                w: $("#cover img").attr("width"),
                h: $("#cover img").attr("height")
            };
            const numPages = $("#tags .tag-container").eq(7).find(".tags .name").text();
            const uploadTime = $("#tags tag-container").eq(8).find(".tags .time").text();
            const result = {
                status: 200,
                creator: "David XD",
                data: {
                    title: {
                        en: titleEN,
                        jp: titleJP
                    },
                    id: galleryId,
                    media_id: mediaID,
                    count: numPages,
                    upload_at: uploadTime,
                    images: {
                        thumb: thumb,
                        pages: []
                    },
                    tags: []
                }
            };
            $("#tags .tag-container").each((i, elem) => {
                const type = $(elem).text().trim().split("\t")[0].split("\n")[0].replace(":", "");
                if (type !== "Pages") {
                    $(elem).find(".tags a").each((_, tagElem) => {
                        result.data.tags.push({
                            type: type,
                            link: $(tagElem).attr("href"),
                            name: $(tagElem).find(".name").text(),
                            count: $(tagElem).find(".count").text()
                        });
                    });
                }
            });
            for (let i = 1; i <= numPages; i++) {
                const res = await axios.get(`${galleryUrl}/${i}`, {
                    headers: {
                        Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
                        Cookie: cookie,
                        Referer: galleryUrl,
                        "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
                    }
                });
                const $$ = cheerio.load(res.data);
                const img = $$("#image-container img");
                result.data.images.pages.push({
                    img: img.attr("src"),
                    w: img.attr("width"),
                    h: img.attr("height")
                });
            };
            //console.log(result);
            return result;
        } catch (error) {
            console.error(error);
            return JSON.stringify({
                status: 500,
                creator: "David XD",
                msg: "Something Error :/"
            });
        };
    };

    nhentaiPDF = async(code) => {
        try {
            const puki = await this.nhentaiGet(code);
            const data = puki.data;
            if (data === 0) {
                console.log({
                    status: 404,
                    crator: "David XD",
                    msg: "Cannot Find Code " + code + " :/"
                });
                return JSON.stringify({
                    status: 404,
                    crator: "David XD",
                    msg: "Cannot Find Code " + code + " :/"
                });
            };
            let title = data.title;
            let hal = data.images.pages.map(res => ({
                img: res.img,
                w: res.w,
                h: res.h
            }));
            let tag = data.tags.map(e => ({
                type: e.type,
                url: "https://nhentai.net" + e.link,
                name: e.name,
                caount: e.count
            }));
            let doc = [];
            data.images.pages.map(page => {
                doc.push(page.img);
            });
            let pdfBuffer = await (await Tools.toPDFBuffer(doc)).data;
            let result = {
                status: 200,
                cretor: "David XD",
                result: {
                    title: {
                        en: {
                            title: title.en.title,
                            pretty: title.en.pretty
                        },
                        jp: {
                            title: title.jp.title,
                            pretty: title.jp.pretty
                        }
                    },
                    id: data.id,
                    media_id: data.media_id,
                    pages_count: data.count,
                    images: {
                        thumb: {
                            img: data.images.thumb.img,
                            w: data.images.thumb.w,
                            h: data.images.thumb.h
                        },
                        pages: hal
                    },
                    tags: tag,
                    result_pdf: pdfBuffer,
                }
            };
            //console.log(result);
            return result;
        } catch (e) {
            console.log(e);
            return JSON.stringify({
                status: 500,
                creator: "David XD",
                msg: util.format(e)
            });
        };
    };

    nhentaiLatest = async() => {
        const baseUrl = "https://nhentai.net";
        const { headers } = await axios.get(baseUrl);
        const cookie = headers["set-cookie"].join("; ");
        let result = {
            status: 200,
            creator: "David XD",
            result: {
                popular_now: [],
                new_upload: []
            }
        }
        await axios({
            url: baseUrl,
            method: "GET",
            headers: {
                Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
                Cookie: cookie,
                Referer: baseUrl,
                "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
            }
        }).then((res) => {
            const $ = cheerio.load(res.data);
            $("#content > .container.index-container.index-popular > .gallery").each(function() {
                result.result.popular_now.push({
                    title: $(this).find(".caption").text().trim(),
                    url: baseUrl + $(this).find("a").attr("href"),
                    img: {
                        url: $(this).find("a > img").attr("data-src"),
                        w: $(this).find("a > img").attr("width"),
                        h: $(this).find("a > img").attr("height")
                    }
                });
            });
            $("#content > .container.index-container > .gallery").each(function() {
                result.result.new_upload.push({
                    title: $(this).find(".caption").text().trim(),
                    url: baseUrl + $(this).find("a").attr("href"),
                    img: {
                        url: $(this).find("a > img").attr("data-src"),
                        w: $(this).find("a > img").attr("width"),
                        h: $(this).find("a > img").attr("height")
                    }
                });
            });
        }).catch((err) => {
            console.error(err);
            return JSON.stringify({
                status: 500,
                creator: "David XD",
                msg: util.format(err)
            });
        });
        //console.log(result);
        return result;
    };

    nhentaiSearch = async(query, page) => {
        const baseUrl = "https://nhentai.net";
        const { headers } = await axios.get(baseUrl);
        const cookie = headers["set-cookie"].join("; ");
        let result = {
            status: 200,
            creator: "David XD",
            result: {
                count: "",
                data: []
            }
        };
        await axios({
            url: "https://nhentai.net/search/?q=" + query + "&page="+ page,
            method: "GET",
            headers: {
                Accept: "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
                "Cookie": cookie,
                Referer: baseUrl,
                "User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36"
            }
        }).then((res) => {
            const $ = cheerio.load(res.data);
            result.result.count = $("#content > h1").text().trim();
            $("#content > .container.index-container > .gallery").each(function() {
                result.result.data.push({
                    title: $(this).find(".caption").text().trim(),
                    url: baseUrl + $(this).find("a").attr("href"),
                    img: {
                        url: $(this).find("a > img").attr("data-src"),
                        w: $(this).find("a > img").attr("width"),
                        h: $(this).find("a > img").attr("height")
                    }
                });
            });
        }).catch((err) => {
            console.error(err);
            return JSON.stringify({
                status: 500,
                creator: "David XD",
                msg: util.format(err)
            });
        });
        //console.log(result);
        return result;
    };

    doujindesuLatest = async(page = 1) => {
        const baseUrl = "https://doujindesu.tv";
        const { headers } = await axios.get(baseUrl);
        const cookie = headers["set-cookie"].join("; ");
        const result = {
            status: 200,
            crator: "David XD",
            result: {}
        };
        await axios({
            url: "https://doujindesu.tv/doujin/page/" + page,
            method: "GET",
            headers: {
                "Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,image/apng,*/*;q=0.8,application/signed-exchange;v=b3;q=0.7",
                "Cookie": `PHPSESSID=giogppgp8ejplo0b01lehj0hp3; __cf_bm=K3n8NkZT7thzs9HNy1P2oupLZTWGrMPEOojFXLoRQt0-1719401109-1.0.1.1-vgJUrwX1EyQIyvidhy2ZRh5CHuDp4DLAZwmsEUyld5zUFQmkaWfzy9VDVxKBfong4HCYXRCjYBz48Kcb2ehdtw; cf_clearance=i.xy1Nmf1d0OMk.Y69wag5ZQE5eS01utMSaJpRt1ZtM-1719401110-1.0.1.1-EOrpjRxb2pQkFPnbwdt9Crkkn7bE0iTfJg0FT.vLFm._OsEIkebNFcSzwzvYCVR3TIMatAeYJ72PRt7FlDNk_Q; __PPU_CAIFRQ=ACZLEAAAAAAAAAAB; __PPU_CAIFRT=ACZLEAAAAABmfEkQ`,
                "Sec-Ch-Ua": `"Chromium";v="124", "Google Chrome";v="124", "Not-A.Brand";v="99"`,
                "User-Agent": "okhttp/4.9.3"
            }
        }).then((res) => {
            const $ = cheerio.load(res.data);
            const mg = $("#archives").eq(0).find(".entries > article");
            result.result = {
                latest_manga: $(mg).map(function() {
                    let baru = $(this).find(".metadata > .artists > .dtch > div").text().trim();
                    if (baru === "NEW") {
                        var isNew = true
                    }
                    return {
                        title: $(this).find("a").attr("title"),
                        type: $(this).find("a > .thumbnail > span").text().trim(),
                        chapter: $(this).find(".metadata > .artists > a > span").text().trim(),
                        new: isNew ? true : baru,
                        url: baseUrl + $(this).find("a").attr("href")
                    };
                }).get(),
            }
        }).catch((err) => {
            console.error(err);
            return {
                status: 500,
                creator: "David XD",
                msg: util.format(err)
            };
        });
        //console.log(result);
        return result;
    };
};