feat(route): Penguin Random House (#10287)

* Initial commit of route Penguin Random House

* WIP: parse books into templates

* Updated radar info

* Added articles route

* Renamed the route and added docs

* Update lib/v2/penguin-random-house/articles.js according to suggestion

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* Changed category to Reading, added En docs

* Changed pubDate format

* Update lib/v2/penguin-random-house/router.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* Update lib/v2/penguin-random-house/articles.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* Update lib/v2/penguin-random-house/thereaddown.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* Parse time

* Fallback on failed regex matches

* Fallback date to undefined

* Update lib/v2/penguin-random-house/utils.js

Co-authored-by: Tony <TonyRL@users.noreply.github.com>

* fix: radar domain

docs/en/reading.md

@@ -57,6 +57,16 @@ Eg:<https://kakuyomu.jp/works/1177354054883783581>
 
 </RouteEn>
 
+## Penguin Random House
+
+### Book Lists
+
+<RouteEn author="StevenRCE0" example="/penguin-random-house/the-read-down" path="/penguin-random-house/the-read-down" />
+
+### Articles
+
+<RouteEn author="StevenRCE0" example="/penguin-random-house/articles" path="/penguin-random-house/articles" />
+
 ## syosetu
 
 ### chapter

docs/reading.md

@@ -69,6 +69,16 @@ pageClass: routes
 
 </Route>
 
+## Penguin Random House
+
+### Book Lists
+
+<Route author="StevenRCE0" example="/penguin-random-house/the-read-down" path="/penguin-random-house/the-read-down" />
+
+### Articles
+
+<Route author="StevenRCE0" example="/penguin-random-house/articles" path="/penguin-random-house/articles" />
+
 ## SoBooks
 
 ### 首页

lib/v2/diskanalyzer/whats-new.js

@@ -37,6 +37,7 @@ module.exports = async (ctx) => {
                 link: currentUrl,
                 description,
                 pubDate: parseDate(title.match(/\((.*)\)/)[1], ['D MMMM YYYY', 'D MMM YYYY']),
+                guid: title,
             };
         });
 

lib/v2/penguin-random-house/articles.js

@@ -0,0 +1,27 @@
+const utils = require('./utils');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const link = 'https://www.penguinrandomhouse.com/articles/';
+    const res = await got(link);
+    const $ = cheerio.load(res.data);
+
+    const itemArray = $('.archive-module-half-container,.archive-module-third-container')
+        .map(function () {
+            return {
+                url: $(this).find('a').attr('href'),
+                title: $(this).find('.archive-module-text').first().text(),
+            };
+        })
+        .get();
+
+    const out = await utils.parseList(itemArray, ctx, utils.parseArticle);
+
+    ctx.state.data = {
+        title: 'Penguin Random House Articles',
+        link,
+        description: 'In-depth interviews, author essays, fascinating essays. Go deeper into the books you love.',
+        item: out,
+    };
+};

lib/v2/penguin-random-house/maintainer.js

@@ -0,0 +1,4 @@
+module.exports = {
+    '/the-read-down': ['StevenRCE0'],
+    '/articles': ['StevenRCE0'],
+};

lib/v2/penguin-random-house/radar.js

@@ -0,0 +1,19 @@
+module.exports = {
+    'penguinrandomhouse.com': {
+        _name: 'Penguin Random House',
+        '.': [
+            {
+                title: 'Penguin Random House Book Lists',
+                docs: 'https://docs.rsshub.app/reading.html#penguin-random-house',
+                source: ['/the-read-down'],
+                target: '/penguin-random-house/the-read-down',
+            },
+            {
+                title: 'Penguin Random House Articles',
+                docs: 'https://docs.rsshub.app/reading.html#penguin-random-house',
+                source: ['/articles'],
+                target: '/penguin-random-house/articles',
+            },
+        ],
+    },
+};

lib/v2/penguin-random-house/router.js

@@ -0,0 +1,4 @@
+module.exports = function (router) {
+    router.get('/articles', require('./articles'));
+    router.get('/the-read-down', require('./thereaddown'));
+};

lib/v2/penguin-random-house/templates/articleHeader.art

@@ -0,0 +1,5 @@
+<p>
+<img src="{{ imageSrc }}" alt="{{ imageAlt }}"/>
+<br>
+{{ description }}
+</p>

lib/v2/penguin-random-house/templates/book.art

@@ -0,0 +1,4 @@
+<img src="{{ imageSrc }}" alt="{{ imageAlt }}"/>
+<h1>{{ title }}</h1>
+<h3>{{ author }}</h3>
+<p>{{ description }}</p>

lib/v2/penguin-random-house/thereaddown.js

@@ -0,0 +1,27 @@
+const utils = require('./utils');
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+
+module.exports = async (ctx) => {
+    const link = 'https://www.penguinrandomhouse.com/the-read-down/';
+    const res = await got(link);
+    const $ = cheerio.load(res.data);
+
+    const itemArray = $('.archive-module-half-container,.archive-module-third-container')
+        .map(function () {
+            return {
+                url: $(this).find('a').attr('href'),
+                title: $(this).find('.archive-module-text').first().text(),
+            };
+        })
+        .get();
+
+    const out = await utils.parseList(itemArray, ctx, utils.parseBooks);
+
+    ctx.state.data = {
+        title: 'Penguin Random House Book Lists',
+        link,
+        description: 'Never wonder what to read next! Check out these lists to find your next favorite book.',
+        item: out,
+    };
+};

lib/v2/penguin-random-house/utils.js

@@ -0,0 +1,119 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { parseDate } = require('@/utils/parse-date');
+
+const parseBookInList = (element) => {
+    const $ = cheerio.load(element);
+    const title = $('h2').first().text();
+    const author = $('h2.author').first().text();
+    const description = $('.desc').first().text();
+    let imageSrc = $('img.cover__backcover').attr('src');
+    let imageAlt = $('img.cover__backcover').attr('alt');
+    if (!imageSrc) {
+        imageSrc = $('img.img-responsive').attr('data-src');
+        imageAlt = $('img.img-responsive').attr('alt');
+    }
+
+    return art(path.join(__dirname, 'templates/book.art'), {
+        title,
+        author,
+        description,
+        imageSrc,
+        imageAlt,
+    });
+};
+
+const parsePubDate = (data) => {
+    const dateString = data('script')
+        .get()
+        .filter((element) => {
+            const fullString = element.children[0];
+            if (!fullString || !fullString.data) {
+                return false;
+            }
+            return fullString.data.includes('post_date');
+        })[0];
+    if (dateString.length === 0) {
+        return;
+    }
+
+    const dateMatch = dateString.children[0].data.match(/(?<="post_date":").*?(?=")/);
+    if (!dateMatch) {
+        return;
+    }
+
+    return parseDate(dateMatch[0]);
+};
+
+const parseBooks = (element) => {
+    const $ = cheerio.load(element);
+    const description = $('h2.read-down-text').first().html();
+    let mainBlock = '';
+
+    $('.awesome-list>li').map((i, element) => {
+        const appending = parseBookInList(element);
+        mainBlock += appending;
+        return appending;
+    });
+
+    return {
+        description,
+        content: mainBlock,
+        pubDate: parsePubDate($),
+    };
+};
+
+const parseArticle = (element) => {
+    const $ = cheerio.load(element);
+    const description = $('h2.hdr-smalltxt').first().html();
+    const imageSrc = $('div.img-block>img').first().attr('src');
+    const imageAlt = $('div.img-block>img').first().attr('alt');
+    let mainBlock = '';
+
+    const descriptionBlock = art(path.join(__dirname, 'templates/articleHeader.art'), {
+        description,
+        imageSrc,
+        imageAlt,
+    });
+
+    $('div.main-content>p,div.main-content>ul').map((i, element) => {
+        const appending = cheerio.load(element);
+        mainBlock += appending.html();
+        return appending;
+    });
+
+    return {
+        description: descriptionBlock,
+        content: mainBlock,
+        pubDate: parsePubDate($),
+    };
+};
+
+const parseList = (items, ctx, contentParser) =>
+    Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.url, async () => {
+                const itemRes = await got(item.url);
+
+                const itemPage = itemRes.data;
+                const itemParsed = contentParser(itemPage);
+
+                const result = {
+                    title: item.title,
+                    description: itemParsed.description + '<br>' + itemParsed.content,
+                    pubDate: itemParsed.pubDate,
+                    link: item.url,
+                };
+
+                return result;
+            })
+        )
+    );
+
+module.exports = {
+    parseList,
+    parseBooks,
+    parseArticle,
+};