feat(route): add nytimes book bestseller (DIYgod#7332)

KhaledAlkilane89 · May 25, 2021 · 94dd2b5 · 94dd2b5
1 parent 8bb8e1f
commit 94dd2b5
Show file tree

Hide file tree

Showing 4 changed files with 109 additions and 0 deletions.
diff --git a/docs/en/traditional-media.md b/docs/en/traditional-media.md
@@ -256,6 +256,25 @@ Provides a better reading experience (full text articles) over the official one.
 
 </RouteEn>
 
+### Best Seller Books
+
+
+<RouteEn author="melvinto" example="/nytimes/book/combined-print-and-e-book-nonfiction" path="/nytimes/book/:category?"/>
+
+| Category | 
+| -------- | 
+| combined-print-and-e-book-nonfiction |
+| hardcover-nonfiction| 
+| paperback-nonfiction| 
+| advice-how-to-and-miscellaneous| 
+| combined-print-and-e-book-fiction|
+| hardcover-fiction|
+| trade-fiction-paperback| 
+| childrens-middle-grade-hardcover| 
+| picture-books|
+| series-books|
+| young-adult-hardcover| 
+
 ## The Wall Street Journal (WSJ)
 
 ### News

diff --git a/docs/traditional-media.md b/docs/traditional-media.md
@@ -660,6 +660,24 @@ category 对应的关键词有
 
 <Route author="xyqfer" example="/nytimes/morning_post" path="/nytimes/morning_post"/>
 
+### 畅销书排行榜
+
+<Route author="melvinto" example="/nytimes/book/combined-print-and-e-book-nonfiction" path="/nytimes/book/:category?"/>
+
+| Category | 中文 |
+| -------- | -------- |
+| combined-print-and-e-book-nonfiction | 非虚构类 - 综合 |
+| hardcover-nonfiction| 非虚构类 - 精装本 |
+| paperback-nonfiction| 非虚构类 - 平装本 |
+| advice-how-to-and-miscellaneous| 工具类 |
+| combined-print-and-e-book-fiction| 虚构类 - 综合 |
+| hardcover-fiction| 虚构类 - 精装本 | 
+| trade-fiction-paperback| 虚构类 - 平装本 | 
+| childrens-middle-grade-hardcover| 儿童 - 中年级 |
+| picture-books| 儿童 - 绘本 | 
+| series-books| 儿童 - 系列图书 |
+| young-adult-hardcover| 青少年 |
+
 ## 澎湃新闻
 
 ### 首页头条

diff --git a/lib/router.js b/lib/router.js
@@ -337,6 +337,7 @@ router.get('/yande.re/post/popular_recent/:period', require('./routes/yande.re/p
 
 // 纽约时报
 router.get('/nytimes/morning_post', require('./routes/nytimes/morning_post'));
+router.get('/nytimes/book/:category?', require('./routes/nytimes/book.js'));
 router.get('/nytimes/:lang?', require('./routes/nytimes/index'));
 
 // 3dm

diff --git a/lib/routes/nytimes/book.js b/lib/routes/nytimes/book.js
@@ -0,0 +1,71 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+
+const categoryList = {
+    "combined-print-and-e-book-nonfiction" : "非虚构类 - 综合",
+    "hardcover-nonfiction": "非虚构类 - 精装本",
+    "paperback-nonfiction": "非虚构类 - 平装本",
+    "advice-how-to-and-miscellaneous": "工具类",
+    "combined-print-and-e-book-fiction": "虚构类 - 综合",
+    "hardcover-fiction": "虚构类 - 精装本",
+    "trade-fiction-paperback": "虚构类 - 平装本",
+    "childrens-middle-grade-hardcover": "儿童 - 中年级",
+    "picture-books": "儿童 - 绘本",
+    "series-books": "儿童 - 系列图书",
+    "young-adult-hardcover": "青少年"
+};
+
+module.exports = async (ctx) => {
+    const category = ctx.params.category || "combined-print-and-e-book-nonfiction";
+
+    const url = `https://www.nytimes.com/books/best-sellers/${category}`;
+
+    let items = [];
+    let dataTitle = '';
+    if (categoryList[category]) {
+        const response = await got({
+            method: 'get',
+            url,
+        });
+        const data = response.data;
+        const $ = cheerio.load(data);
+        dataTitle = $('h1').eq(0).text();
+
+        items = $('article[itemprop=itemListElement]')
+            .map((index, elem) => {
+                const $item = $(elem);
+                const firstInfo = $item.find('p').eq(0).text();
+                const $name = $item.find('h3[itemprop=name]');
+                const author = $item.find('p[itemprop=author]').text();
+                const publisher = $item.find('p[itemprop=publisher]').text();
+                const description = $item.find('p[itemprop=description]').text();
+                const imageLink = $item.find('img[itemprop=image]').attr('src');
+                const $link = $item.find('ul[aria-label="Links to Book Retailers"]');
+                const links = $link.find('a').toArray();
+
+                let primaryLink = links.length > 0 ? $(links[0]).attr('href') : "";
+
+                for (const link of links) {
+                    const l = $(link);
+                    if (l.text() === "Amazon") {
+                        primaryLink = l.attr('href');
+                        break;
+                    }
+                }
+
+                return {
+                    title: `${index + 1}: ${$name.text()}`,
+                    author,
+                    description: `<figure><img src="${imageLink}" alt="test"/><figcaption><span>${description}</span></figcaption></figure><br/>${firstInfo}<br/>Author: ${author}<br/>Publisher: ${publisher}`,
+                    link: primaryLink
+                };
+            })
+            .get();
+    }
+
+    ctx.state.data = {
+        title: `The New York Times Best Sellers - ${dataTitle}`,
+        link: url,
+        item: items,
+    };
+};