noctbot/bot/src/commands/aggregators/kiss_diary.ts

import request from "request-promise-native";
import { Response } from "request";
import { INewsItem, IAggregator } from "./aggregator";
import { getRepository } from "typeorm";
import { AggroNewsItem } from "@shared/db/entity/AggroNewsItem";
import cheerio from "cheerio";

const urlPattern = /diary\.php\?no=(\d+)/i;
const kissDiaryRoot = "http://www.kisskiss.tv/kiss";
const FEED_NAME = "kisskisstv-diary";

async function aggregate() {
    let repo = getRepository(AggroNewsItem);

    let lastPost = await repo.findOne({
        select: [ "newsId" ],
        where: { feedName: FEED_NAME },
        order: { newsId: "DESC" }
    });

    if(!lastPost)
        lastPost = repo.create({
            newsId: 0
        });
    
    try {
        let mainPageRes = await request(`${kissDiaryRoot}/diary.php`, {resolveWithFullResponse: true}) as Response;
        
        if(mainPageRes.statusCode != 200)
            return [];

        let rootNode = cheerio.load(mainPageRes.body);
        let diaryEntryNames = rootNode("table.blog_frame_top");
        
        if(diaryEntryNames.length == 0) {
            console.log("[KISS DIARY] Failed to find listing!");
        }

        let diaryTexts = rootNode("div.blog_frame_middle");
        let items = diaryEntryNames.map((i, e) => ({ table: e, content: diaryTexts.get(i) }));
        let result : INewsItem[] = [];
        let latestEntry = lastPost.newsId;

        for(let {table, content} of items.get() as {table: CheerioElement, content: CheerioElement}[]) {
            let a = cheerio(table).find("a");
            let link = a.attr("href");
            let matches = link ? urlPattern.exec(link) : false;
            if(!matches)
                continue;
            
            let id = +matches[1];
            if(id <= lastPost.newsId)
                continue;

            if(id > latestEntry)
                latestEntry = id;

            let diaryLink = `${kissDiaryRoot}/${link}`;
            let contentCh = cheerio(content);
            let title = a.text();
            
            let bottomFrame = contentCh.find("div.blog_data");
            bottomFrame.remove();

            result.push({
                newsId: id,
                feedId: FEED_NAME,
                link: diaryLink,
                title: title,
                author: "KISS BLOG",
                contents: contentCh.html() ?? "",
                embedColor: 0xf4c100,
                needsTranslation: true
            });
        }

        return result;
    } catch(err) {
        return [];
    }
}

export default {
    aggregate: aggregate
} as IAggregator;