[reopen] add feed for net-updates of aozora.gr.jp (#910)

* add feed for net-updates of aozora.gr.jp

* .gitignore: filt vim swap file

* fix a line in .gitignore; try change char in ReadMe

* try fix ReadMe

.gitignore

@@ -12,3 +12,4 @@ config/config.js
 yarn-error.log
 tmp
 newrelic_agent.log
+*.swp

docs/README.md

@@ -1580,7 +1580,7 @@ IATA 国际航空运输协会机场代码, 参见[维基百科 国际航空运
 | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- | ---- |
 | 0    | 111  | 109  | 110  | 113  | 114  | 115  | 112  | 116  |
 
-## 网络小说
+## 小说·文学
 
 ### 笔趣阁
 
@@ -1619,6 +1619,14 @@ IATA 国际航空运输协会机场代码, 参见[维基百科 国际航空运
 
 </route>
 
+### 青空文庫
+
+<route name="青空文庫新着リスト" author="sgqy" example="/aozora/newbook/10" path="/aozora/newbook/:count?" :paramsDesc="['更新数量. 设置每次下载列表大小. 范围是 1 到 50.']">
+
+书籍网站每日一更. 信息更新时间为书籍最初出版时间, 排序可能不符合网络发表时间, 请认准未读消息.
+
+</route>
+
 ## 中国驻外使领馆
 
 ### 大使馆

router.js

@@ -656,6 +656,9 @@ router.get('/laosiji/feed', require('./routes/laosiji/feed'));
 // 99% Invisible
 router.get('/99percentinvisible/transcript', require('./routes/99percentinvisible/transcript'));
 
+// 青空文庫
+router.get('/aozora/newbook/:count?', require('./routes/aozora/newbook'));
+
 // Hermes UK
 router.get('/parcel/hermesuk/:tracking', require('./routes/parcel/hermesuk'));
 

routes/aozora/newbook.js

@@ -0,0 +1,103 @@
+// Warning: The author knows nothing about javascript!
+
+// params:
+// count? : count of new-books
+
+const axios = require('../../utils/axios'); // get web content
+const cheerio = require('cheerio'); // html parser
+const he = require('he');
+
+const base_url = 'https://www.aozora.gr.jp/index_pages/';
+module.exports = async (ctx) => {
+    // get the update list
+    const response = await axios({
+        method: 'get',
+        url: base_url + 'whatsnew1.html',
+    });
+    const data = response.data; // content is html format
+
+    const $ = cheerio.load(data);
+    const list = $('table.list tr');
+
+    // get how many new-books. amount in this page is 50
+    let count = parseInt(ctx.params.count);
+    if (Number.isNaN(count) || count < 1) {
+        count = 10; // default count of new-book list
+    } else if (count > 50) {
+        count = 50;
+    }
+
+    // parse book urls
+    const detail_urls = [];
+    for (let i = 1; i < count + 1; ++i) {
+        // i = 1: first tr is table title, ignore
+        const link =
+            base_url +
+            $(list[i])
+                .find('a')
+                .attr('href');
+        detail_urls.push(link);
+    }
+
+    // get book-cards
+    const responses = await Promise.all(detail_urls.map((url) => axios(url)));
+    const cards = responses.map(({ data }) => data);
+
+    // get real data to feed
+    const book_list = [];
+    for (let i = 0; i < count; ++i) {
+        const $ = cheerio.load(cards[i]);
+        const link = $('meta[property="og:url"]').attr('content');
+        const link_dir = link.replace(/\/[^/]*$/g, '/');
+        const title_info = $('table[summary="タイトルデータ"] > tbody > tr');
+        let author = '';
+        let title = '';
+        let title_sub = '';
+        for (let j = 0; j < title_info.length; ++j) {
+            const tmp = he.decode($(title_info[j]).html()); // should convert from escaped to unicode
+            if (tmp.includes('作品名：')) {
+                title = $(title_info[j])
+                    .find('td:nth-child(2)')
+                    .text();
+            }
+            if (tmp.includes('副題：')) {
+                title_sub = $(title_info[j])
+                    .find('td:nth-child(2)')
+                    .text();
+            }
+            if (tmp.includes('著者名：')) {
+                author = $(title_info[j])
+                    .find('td:nth-child(2)')
+                    .text();
+            }
+        }
+        if (title_sub !== '') {
+            title += '　——　' + title_sub;
+        }
+        const pub_date_raw = $('table[summary="底本データ"] > tbody > tr:nth-child(3) > td:nth-child(2)').text();
+        const pub_date_num = pub_date_raw.replace(/（.*）|日/g, '').replace(/[年月]/g, '-');
+        const pub_date = new Date(pub_date_num).toUTCString();
+        const full_text_relative_link = $('table.download > tbody > tr:nth-child(3) > td:nth-child(3) > a').attr('href');
+        const full_text_link = link_dir + full_text_relative_link;
+        const full_text_link_html = '<a href="' + full_text_link + '">いますぐXHTML版で読む</a><br>';
+        const summury = $('table[summary="作品データ"]')
+            .html()
+            .replace(/href="/g, 'href="' + link_dir);
+
+        const item = {
+            title: title,
+            author: author,
+            pubDate: pub_date,
+            link: link,
+            description: full_text_link_html + summury,
+        };
+        book_list.push(item);
+    }
+
+    // feed the data
+    ctx.state.data = {
+        title: '青空文庫新着リスト',
+        link: base_url + 'whatsnew1.html',
+        item: book_list,
+    };
+};