Merge pull request #836 from DIYgod/master

[pull] master from diygod:master

docs/en/journal.md

@@ -331,21 +331,18 @@ Return results from 2020
 
 <RouteEn author="Fatpandac" example="/oup/journals/adaptation" path="/oup/journals/:name" :paramsDesc="['short name for a journal, can be found in URL']" anticrawler="1"/>
 
-## Proceedings of The National Academy of Sciences (PNAS)
+## Proceedings of The National Academy of Sciences
 
-### Latest Articles - Articles by Topic
-
-### Proceedings of The National Academy of Sciences (PNAS) - Latest Articles
+### Journal
 
-<RouteEn author="emdoe" example="/pnas/Applied Mathematics" path="/pnas/:topic" :paramsDesc="['topic name', 'obtain it from pnas.org (new research in ...)']" />
+<RouteEn author="emdoe HenryQW y9c" example="/pnas/latest" path="/pnas/:topicPath*" :paramsDesc="['Topic path, support **Featured Topics**, **Articles By Topic** and [**Collected Papers**](https://www.pnas.org/about/collected-papers), `latest` by default']" radar="1" anticrawler="1" puppeteer="1" supportScihub="1">
 
-- Using router (`/pnas/` + Topic of Interest) to query latest research paper for a certain topic from PNAS journal.
-    If the `:topic` parameter is blank, or equal to 'latest', then all the latest papers will return.
+::: tip Tips
+Some topics require adding `topic/` to `topicPath` like [`/pnas/topic/app-math`](https://rsshub.app/pnas/topic/app-math) and some don't like [`/pnas/biophysics-and-computational-biology`](https://rsshub.app/pnas/biophysics-and-computational-biology)
+:::
 
 </RouteEn>
 
-<RouteEn author="emdoe HenryQW" example="/pnas/Applied Mathematics" path="/pnas/:topic" :paramsDesc="['topic name', 'obtain it from pnas.org (new research in ...). `Latest` will include all.']" supportScihub="1"/>
-
 ## PubMed
 
 ### Trending articles

docs/journal.md

@@ -302,14 +302,15 @@ You can get all short name of a journal from <https://www.nature.com/siteindex>
 
 <Route author="Fatpandac" example="/oup/journals/adaptation" path="/oup/journals/:name" :paramsDesc="['期刊名称缩写，可以在网址中得到']" anticrawler="1"/>
 
-## PNAS
+## Proceedings of The National Academy of Sciences
 
-### 最新文章（可筛选领域）
+### 期刊
 
-<Route author="emdoe y9c" example="/pnas/Applied Mathematics" path="/pnas/:topic" :paramsDesc="['领域名称','可从 pnas.org 获得']" supportScihub="1"/>
+<Route author="emdoe HenryQW y9c" example="/pnas/latest" path="/pnas/:topicPath*" :paramsDesc="['领域路径，支持 **Featured Topics**、**Articles By Topic** 及 [**Collected Papers**](https://www.pnas.org/about/collected-papers), 预设为 `latest`']" radar="1" anticrawler="1" puppeteer="1" supportScihub="1">
 
--   通过 `/pnas/` + “领域名称” 来获取对应 “领域” 的最新文章（Latest Research）。
-    若参数置空（`/pnas`）或为 latest（`/pnas/latest`），则默认获取全部文章。
+::: tip Tips
+有些领域需要在 `topicPath` 中添加 `topic/`，如 [`/pnas/topic/app-math`](https://rsshub.app/pnas/topic/app-math)，有些则不需要，如 [`/pnas/biophysics-and-computational-biology`](https://rsshub.app/pnas/biophysics-and-computational-biology)
+:::
 
 </Route>
 

docs/other.md

@@ -721,7 +721,7 @@ type 为 all 时，category 参数不支持 cost 和 free
 
 ### 今日油价查询
 
-<Route author="TonyRL" example="/qiyoujiage/shanghai" path="/qiyoujiage/:path*" :paramsDesc="['路径']">
+<Route author="TonyRL" example="/qiyoujiage/shanghai" path="/qiyoujiage/:path+" :paramsDesc="['路径']">
 
 ::: tip 提示
 

lib/router.js

@@ -2165,7 +2165,7 @@ router.get('/elife/:tid', lazyloadRouteHandler('./routes/elife/index'));
 router.get('/ieee/author/:aid/:sortType/:count?', lazyloadRouteHandler('./routes/ieee/author'));
 
 // PNAS [Sci Journal]
-router.get('/pnas/:topic?', lazyloadRouteHandler('./routes/pnas/index'));
+// router.get('/pnas/:topic?', lazyloadRouteHandler('./routes/pnas/index'));
 
 // cell [Sci Journal]
 router.get('/cell/cell/:category', lazyloadRouteHandler('./routes/cell/cell/index'));

lib/v2/pnas/index.js

@@ -0,0 +1,101 @@
+const cheerio = require('cheerio');
+const got = require('@/utils/got');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+const { setCookies } = require('@/utils/puppeteer-utils');
+const { CookieJar } = require('tough-cookie');
+const logger = require('@/utils/logger');
+
+module.exports = async (ctx) => {
+    const baseUrl = 'https://www.pnas.org';
+    const { topicPath } = ctx.params;
+    const link = `${baseUrl}/${topicPath ? topicPath : 'latest'}`;
+
+    let cookieJar = await ctx.cache.get('pnas:cookieJar');
+    const cacheMiss = !cookieJar;
+    cookieJar = cacheMiss ? new CookieJar() : CookieJar.fromJSON(cookieJar);
+    const { data: res } = await got(link, {
+        cookieJar,
+    });
+    if (cacheMiss) {
+        await ctx.cache.set('pnas:cookieJar', cookieJar.toJSON());
+    }
+
+    const $ = cheerio.load(res);
+    const list = $('.card--row-reversed .card-content')
+        .toArray()
+        .map((item) => {
+            item = $(item);
+            const a = item.find('.article-title a');
+            return {
+                title: a.text(),
+                link: new URL(a.attr('href'), baseUrl).href,
+                pubDate: parseDate(item.find('.card__meta__date').text()),
+            };
+        });
+
+    const browser = await require('@/utils/puppeteer')();
+
+    const out = await Promise.all(
+        list.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const page = await browser.newPage();
+                await setCookies(page, await cookieJar.getCookieString(item.link), '.pnas.org');
+                await page.setRequestInterception(true);
+                page.on('request', (request) => {
+                    request.resourceType() === 'document' ? request.continue() : request.abort();
+                });
+                logger.debug(`Requesting ${item.link}`);
+                await page.goto(item.link, {
+                    waitUntil: 'domcontentloaded',
+                    referer: link,
+                });
+                await page.waitForSelector('.core-container');
+
+                const res = await page.evaluate(() => document.documentElement.innerHTML);
+                await page.close();
+
+                const $ = cheerio.load(res);
+                const PNASdataLayer = JSON.parse(
+                    $('script')
+                        .text()
+                        .match(/PNASdataLayer =(.*?);/)[1]
+                );
+
+                $('.signup-alert-ad, .citations-truncation button').remove();
+
+                const { keywords, topic } = PNASdataLayer.page.attributes;
+
+                item.category = [...keywords, topic];
+                item.author = PNASdataLayer.page.pageInfo.author;
+                item.doi = PNASdataLayer.page.pageInfo.DOI;
+                item.description = art(path.join(__dirname, 'templates', 'article.art'), {
+                    access: PNASdataLayer.user.access === 'yes',
+                    //
+                    abstracts: $('#abstracts .core-container').html(),
+                    //
+                    articleBody: $('[property=articleBody]').html(),
+                    //
+                    dataAvailability: $('#data-availability').html(),
+                    acknowledgments: $('#acknowledgments').html(),
+                    supplementaryMaterials: $('#supplementary-materials').html(),
+                    bibliography: $('#bibliography').html(),
+                });
+
+                return item;
+            })
+        )
+    );
+
+    browser.close();
+
+    ctx.state.data = {
+        title: `${$('.banner-widget__content h1').text()} - PNAS`,
+        description: $('.banner-widget__content p').text(),
+        image: 'https://www.pnas.org/favicon.ico',
+        language: 'en-US',
+        link,
+        item: out,
+    };
+};

lib/v2/pnas/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:topicPath*': ['emdoe', 'y9c'],
+};

lib/v2/pnas/radar.js

@@ -0,0 +1,13 @@
+module.exports = {
+    'pnas.org': {
+        _name: 'Proceedings of the National Academy of Sciences',
+        '.': [
+            {
+                title: '期刊',
+                docs: 'https://docs.rsshub.app/journal.html#proceedings-of-the-national-academy-of-sciences',
+                source: ['/*topicPath'],
+                target: '/pnas/:topicPath',
+            },
+        ],
+    },
+};

lib/v2/pnas/router.js

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

lib/v2/pnas/templates/article.art

@@ -0,0 +1,8 @@
+{{ if abstracts }}{{@ abstracts }}{{ /if }}
+
+{{ if access }}{{@ articleBody }}{{ /if }}
+
+{{ if dataAvailability }}{{@ dataAvailability }}{{ /if }}
+{{ if acknowledgments }}{{@ acknowledgments }}{{ /if }}
+{{ if supplementaryMaterials }}{{@ supplementaryMaterials }}{{ /if }}
+{{ if bibliography }}{{@ bibliography }}{{ /if }}