rsshub 1.0.0-master.fccb918 → 1.0.0-master.fd1d453

package/lib/v2/hafu/utils.js

@@ -0,0 +1,205 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+const typeMap = {
+    ggtz: { url: 'https://www.hafu.edu.cn/index/ggtz.htm', root: 'https://www.hafu.edu.cn/', title: '河南财院 - 公告通知', parseFn: ggtzParse },
+    jwc: { url: 'https://jwc.hafu.edu.cn/tzgg.htm', root: 'https://jwc.hafu.edu.cn/', title: '河南财院 教务处 - 公告通知', parseFn: jwcParse },
+    zsjyc: { url: 'https://zsjyc.hafu.edu.cn/tztg.htm', root: 'https://zsjyc.hafu.edu.cn/', title: '河南财院 招生就业处 - 公告通知', parseFn: zsjycParse },
+};
+// Number of get articles
+let limit = 10;
+
+module.exports = async (ctx, type) => {
+    const link = typeMap[type].url;
+    const title = typeMap[type].title;
+
+    const response = await got(link);
+    const $ = cheerio.load(response.data);
+
+    limit = ctx.query.limit || limit;
+    const resultList = await typeMap[type].parseFn(ctx, $);
+
+    return {
+        title,
+        link,
+        resultList,
+    };
+};
+
+async function tryGetFullText(href, link, type) {
+    let articleData = '';
+    let description = '';
+    // for some unexpected href link
+    try {
+        const articleRes = await got(link);
+        articleData = cheerio.load(articleRes.data);
+        // fullText
+        let articleBody = articleData('div[class=v_news_content]').html();
+        // attachments
+        if (articleData('[id^=nattach]').length !== 0) {
+            articleBody = tryGetAttachments(articleData, articleBody, type);
+        }
+
+        description = art(path.join(__dirname, 'templates/hafu.art'), articleBody)();
+    } catch {
+        description = href;
+    }
+
+    return { articleData, description };
+}
+
+function tryGetAttachments(articleData, articleBody, type) {
+    if (type === 'ggtz') {
+        articleData(`[id^=nattach]`)
+            .prev()
+            .map((_, item) => {
+                const href = articleData(item).attr('href').slice(1);
+                const link = typeMap.ggtz.root + href;
+                const title = articleData(item).text();
+                articleBody += '<br/>';
+                articleBody += `<a href=${link}>${title}</a>`;
+                return null;
+            });
+    } else {
+        articleData('[id^=nattach]')
+            .parent()
+            .prev()
+            .map((_, item) => {
+                const href = articleData(item).find('a').attr('href').slice(1);
+                const link = typeMap[type].root + href;
+                const title = articleData(item).find('a').find('span').text();
+                articleBody += '<br/>';
+                articleBody += `<a href=${link}> ${title} </a>`;
+                return null;
+            });
+    }
+
+    return articleBody;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.article(author), 3.detailed(pubDate)
+async function ggtzParse(ctx, $) {
+    const data = $('a[class=c269582]').parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                // .slice(3) for cut out str '../' in original link
+                const href = $(item).find('a[class=c269582]').attr('href').slice(3);
+                const link = typeMap.ggtz.root + href;
+                const title = $(item).find('a[class=c269582]').attr('title');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'ggtz');
+                    let author = '';
+                    let pubDate = '';
+                    if (articleData instanceof Function) {
+                        const header = articleData('h1').next().text();
+                        const index = header.indexOf('日期');
+
+                        author = header.substring(0, index - 2) || '';
+
+                        const date = header.substring(index + 3, index + 19);
+                        pubDate = parseDate(date, 'YYYY-MM-DD HH:mm');
+                    } else {
+                        const date = $(item).find('a[class=c269582_date]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD');
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author,
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.article(author)
+async function jwcParse(ctx, $) {
+    const data = $('a[class=c259713]').parent().parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                const href = $(item).find('a[class=c259713]').attr('href');
+                const link = typeMap.jwc.root + href;
+                const title = $(item).find('a[class=c259713]').attr('title');
+
+                const date = $(item).find('span[class=timestyle259713]').text();
+                const pubDate = parseDate(date, 'YYYY/MM/DD');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'jwc');
+
+                    let author = '';
+                    if (articleData instanceof Function) {
+                        author = articleData('span[class=authorstyle259690]').text();
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author: '供稿单位：' + author,
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}
+// A. got from hostPage     1.article(link), 2.article(title), 3.(pubDate)
+// B. got from articlePage  1.description(fullText), 2.detailed(pubDate)
+async function zsjycParse(ctx, $) {
+    const data = $('a[class=c127701]').parent().parent().slice(0, limit);
+    const resultItems = await Promise.all(
+        data
+            .map(async (_, item) => {
+                const href = $(item).find('a[class=c127701]').attr('href');
+                const link = typeMap.zsjyc.root + href;
+
+                const title = $(item).find('a[class=c127701]').attr('title');
+
+                const result = await ctx.cache.tryGet(link, async () => {
+                    const { articleData, description } = await tryGetFullText(href, link, 'zsjyc');
+
+                    let pubDate = '';
+                    if (articleData instanceof Function) {
+                        const date = articleData('span[class=timestyle127702]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD HH:mm');
+                    } else {
+                        const date = $(item).find('a[class=c269582_date]').text();
+                        pubDate = parseDate(date, 'YYYY-MM-DD');
+                    }
+
+                    return {
+                        title,
+                        description,
+                        pubDate: timezone(pubDate, +8),
+                        link,
+                        author: '供稿单位：招生就业处',
+                    };
+                });
+
+                return result;
+            })
+            .get()
+    );
+
+    return resultItems;
+}

package/lib/v2/hakkatv/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/news/:type?': ['TonyRL'],
+};

package/lib/v2/hakkatv/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hakkatv.org.tw': {
+        _name: '客家電視台',
+        '.': [
+            {
+                title: '新聞首頁 - 客家電視台',
+                docs: 'https://docs.rsshub.app/traditional-media.html#ke-jia-dian-shi-tai',
+                source: '/news',
+                target: '/hakkatv/news',
+            },
+        ],
+    },
+};

package/lib/v2/hakkatv/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/news/:type?', require('./type'));
+};

package/lib/v2/hakkatv/type.js

@@ -0,0 +1,67 @@
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const timezone = require('@/utils/timezone');
+
+const typeMap = ['hakka', 'political', 'medical', 'local', 'international'];
+const baseUrl = 'https://www.hakkatv.org.tw';
+const apiUrl = 'https://api.hakkatv.org.tw';
+
+module.exports = async (ctx) => {
+    const { type } = ctx.params;
+
+    let allData;
+    if (!type) {
+        allData = await Promise.all(
+            typeMap.map(async (t) => {
+                const { data } = await got(`${apiUrl}/api/news/index`, {
+                    searchParams: {
+                        per: 4,
+                        'sort[created_at]': 'desc',
+                        type: t,
+                        keywords: '',
+                    },
+                });
+                return data.data;
+            })
+        );
+    } else {
+        allData = (
+            await got(`${apiUrl}/api/news/index`, {
+                searchParams: {
+                    per: 4,
+                    'sort[created_at]': 'desc',
+                    type,
+                    keywords: '',
+                },
+            })
+        ).data.data;
+    }
+
+    const list = [].concat(...allData).map((item) => ({
+        title: item.title,
+        pubDate: timezone(parseDate(item.created_at), +8),
+        author: item.author,
+        link: `${baseUrl}/news-detail/${item.id}`,
+        id: item.id,
+    }));
+
+    const items = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const { data } = await got(`${apiUrl}/api/news/read/${item.id}`);
+                item.category = data.tag.map((t) => t.tag);
+                item.description = data.content.replace(/\n/g, '<br>');
+                delete item.id;
+                return item;
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '新聞首頁 - 客家電視台',
+        description: '客家電視是屬於全民、以至於全世界客家族群的頻道，亦是為傳播客家文化而存在，定位為「全體客家族群之媒體」。',
+        link: `${baseUrl}/news`,
+        language: 'zh-TW',
+        item: items,
+    };
+};

package/lib/v2/hashnode/blog.js

@@ -0,0 +1,62 @@
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+
+const baseApiUrl = 'https://api.hashnode.com';
+
+module.exports = async (ctx) => {
+    const username = ctx.params.username;
+    if (!username) {
+        return;
+    }
+
+    const query = `
+    {
+        user(username: "${username}") {
+            publication {
+                posts{
+                    slug
+                    title
+                    brief
+                    coverImage
+                    dateAdded
+                }
+            }
+        }
+    }
+    `;
+
+    const userUrl = `https://${username}.hashnode.dev`;
+    const response = await got({
+        method: 'POST',
+        url: baseApiUrl,
+        headers: {
+            Referer: userUrl,
+            'Content-type': 'application/json',
+        },
+        body: JSON.stringify({ query }),
+    });
+
+    const publication = response.data.data.user.publication;
+    if (!publication) {
+        return;
+    }
+
+    const list = publication.posts;
+    ctx.state.data = {
+        title: `Hashnode by ${username}`,
+        link: userUrl,
+        item: list
+            .map((item) => ({
+                title: item.title,
+                description: art(path.join(__dirname, 'templates/description.art'), {
+                    image: item.coverImage,
+                    brief: item.brief,
+                }),
+                pubDate: parseDate(item.dateAdded),
+                link: `${userUrl}/${item.slug}`,
+            }))
+            .filter((item) => item !== ''),
+    };
+};

package/lib/v2/hashnode/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/blog/:username': ['hnrainll'],
+};

package/lib/v2/hashnode/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'hashnode.dev': {
+        _name: 'hashnode',
+        '.': [
+            {
+                title: 'Hashnode Blog',
+                docs: 'https://docs.rsshub.app/blog.html#hashnode',
+                source: '/',
+                target: '/hashnode/blog/:username',
+            },
+        ],
+    },
+};

package/lib/v2/hashnode/router.js

@@ -0,0 +1,3 @@
+module.exports = (router) => {
+    router.get('/blog/:username', require('./blog'));
+};

package/lib/v2/hashnode/templates/description.art

@@ -0,0 +1,2 @@
+<img src="{{ image }}" />
+{{@ brief }}

package/lib/v2/hdu/cs/notice.js

@@ -0,0 +1,54 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const link = 'https://computer.hdu.edu.cn';
+const host = 'https://computer.hdu.edu.cn/6738/list.htm';
+
+const getSingleRecord = async () => {
+    const res = await got(host);
+
+    const $ = cheerio.load(res.data);
+    const list = $('.posts-list').find('li');
+
+    return (
+        list &&
+        list
+            .map((index, item) => {
+                item = $(item);
+                const dateTxt = item.find('.date').text();
+                const date = dateTxt.slice(1, dateTxt.length - 1);
+                return {
+                    title: item.find('a').text(),
+                    pubDate: parseDate(date),
+                    link: link + item.find('a').attr('href'),
+                };
+            })
+            .get()
+    );
+};
+
+module.exports = async (ctx) => {
+    const items = await getSingleRecord();
+    const out = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+                return {
+                    title: item.title,
+                    link: item.link,
+                    description: $('.wp_articlecontent').html(),
+                    pubDate: item.pubDate,
+                };
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '杭电计算机-通知公告',
+        description: '杭州电子科技大学计算机学院-通知公告',
+        link: host,
+        item: out,
+    };
+};

package/lib/v2/hdu/cs/pg.js

@@ -0,0 +1,54 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+
+const link = 'https://computer.hdu.edu.cn';
+const host = 'https://computer.hdu.edu.cn/6769/list.htm';
+
+const getSingleRecord = async () => {
+    const res = await got(host);
+
+    const $ = cheerio.load(res.data);
+    const list = $('.posts-list').find('li');
+
+    return (
+        list &&
+        list
+            .map((index, item) => {
+                item = $(item);
+                const dateTxt = item.find('.date').text();
+                const date = dateTxt.slice(1, dateTxt.length - 1);
+                return {
+                    title: item.find('a').text(),
+                    pubDate: parseDate(date),
+                    link: link + item.find('a').attr('href'),
+                };
+            })
+            .get()
+    );
+};
+
+module.exports = async (ctx) => {
+    const items = await getSingleRecord();
+    const out = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const response = await got(item.link);
+                const $ = cheerio.load(response.data);
+                return {
+                    title: item.title,
+                    link: item.link,
+                    description: $('.wp_articlecontent').html(),
+                    pubDate: item.pubDate,
+                };
+            })
+        )
+    );
+
+    ctx.state.data = {
+        title: '杭电计算机-研究生通知',
+        description: '杭州电子科技大学计算机学院-研究生教学通知',
+        link: host,
+        item: out,
+    };
+};

package/lib/v2/hdu/maintainer.js

@@ -1,3 +1,4 @@
 module.exports = {
     '/cs': ['legr4ndk'],
+    '/cs/pg': ['legr4ndk'],
 };

package/lib/v2/hdu/radar.js

@@ -8,6 +8,12 @@ module.exports = {
                 source: '/6738/list.htm',
                 target: '/hdu/cs',
             },
+            {
+                title: '杭电计算机-研究生通知',
+                docs: 'https://docs.rsshub.app/university.html#hang-zhou-dian-zi-ke-ji-da-xue',
+                source: '/6769/list.htm',
+                target: '/hdu/cs/pg',
+            },
         ],
     },
 };

package/lib/v2/hdu/router.js

@@ -1,4 +1,5 @@
 module.exports = function (router) {
     // 杭州电子科技大学
-    router.get('/cs', require('./cs'));
+    router.get('/cs', require('./cs/notice'));
+    router.get('/cs/pg', require('./cs/pg'));
 };

package/lib/v2/heu/gx/list.js

@@ -47,7 +47,7 @@ module.exports = async (ctx) => {
         })
         .get();
 
-    await Promise.all(
+    const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 if (item.link.indexOf('info') !== -1) {
@@ -68,6 +68,6 @@ module.exports = async (ctx) => {
     ctx.state.data = {
         title: '工学-' + bigTitle,
         link: toUrl,
-        item: list,
+        item: items,
     };
 };

package/lib/v2/heu/job/list.js

@@ -44,7 +44,7 @@ module.exports = async (ctx) => {
         })
         .get();
 
-    await Promise.all(
+    const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 if (item.link.indexOf('HrbeuJY') !== -1) {
@@ -65,6 +65,6 @@ module.exports = async (ctx) => {
     ctx.state.data = {
         title: '就业服务平台-' + idMap[id].name,
         link: idMap[id].url,
-        item: list,
+        item: items,
     };
 };

package/lib/v2/heu/uae/list.js

@@ -35,7 +35,7 @@ module.exports = async (ctx) => {
         })
         .get();
 
-    await Promise.all(
+    const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 if (item.link.indexOf('page.htm') !== -1) {
@@ -57,6 +57,6 @@ module.exports = async (ctx) => {
     ctx.state.data = {
         title: '水声学院-' + bigTitle,
         link: rootUrl.concat('/', id, '/list.htm'),
-        item: list,
+        item: items,
     };
 };

package/lib/v2/heu/yjsy/list.js

@@ -39,7 +39,7 @@ module.exports = async (ctx) => {
         })
         .get();
 
-    await Promise.all(
+    const items = await Promise.all(
         list.map((item) =>
             ctx.cache.tryGet(item.link, async () => {
                 if (item.link.indexOf('page.htm') !== -1) {
@@ -60,6 +60,6 @@ module.exports = async (ctx) => {
     ctx.state.data = {
         title: '研究生院-' + bigTitle,
         link: rootUrl.concat('/', id, '/list.htm'),
-        item: list,
+        item: items,
     };
 };

package/lib/v2/hjedd/article.js

@@ -0,0 +1,27 @@
+const got = require('@/utils/got');
+const { rootUrl, ProcessItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const node = ctx.params.node ?? '258';
+    const type = ctx.params.type ?? '1';
+
+    const currentUrl = rootUrl;
+    const nodeUrl = `${rootUrl}/api/topic/node?nodeId=${node}`;
+    const apiUrl = `${rootUrl}/api/topic/node/topics?page=1&nodeId=${node}&type=${type}&limit=${ctx.query.limit}`;
+
+    const response = await got({
+        method: 'get',
+        url: nodeUrl,
+    });
+
+    const data = response.data.data;
+
+    const items = await ProcessItems(apiUrl, ctx.query.limit, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: `海角社区 - ${data.name}`,
+        link: `${currentUrl}/article?nodeId=${node}`,
+        item: items,
+        description: data.description.replace(/<\/?p>/, ''),
+    };
+};

package/lib/v2/hjedd/hot.js

@@ -0,0 +1,14 @@
+const { rootUrl, ProcessItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const currentUrl = rootUrl;
+    const apiUrl = `${rootUrl}/api/topic/hot/topics?page=1`;
+
+    const items = await ProcessItems(apiUrl, ctx.query.limit, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: '海角社区 - 热门',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/hjedd/latest.js

@@ -0,0 +1,14 @@
+const { rootUrl, ProcessItems } = require('./utils');
+
+module.exports = async (ctx) => {
+    const currentUrl = rootUrl;
+    const apiUrl = `${rootUrl}/api/topic/node/topics?page=1&nodeId=0&type=1`;
+
+    const items = await ProcessItems(apiUrl, ctx.query.limit, ctx.cache.tryGet);
+
+    ctx.state.data = {
+        title: '海角社区 - 最新',
+        link: currentUrl,
+        item: items,
+    };
+};

package/lib/v2/hjedd/maintainer.js

@@ -0,0 +1,10 @@
+module.exports = {
+    '/event': ['nczitzk'],
+    '/hot': ['nczitzk'],
+    '/latest': ['nczitzk'],
+    '/news': ['nczitzk'],
+    '/notice': ['nczitzk'],
+    '/original': ['nczitzk'],
+    '/top': ['nczitzk'],
+    '/:node?/:type?': ['nczitzk'],
+};