From db0ae9d3fa9b62a35f16169ccc48aa0c4886c11e Mon Sep 17 00:00:00 2001
From: Henry Wang <hi@henry.wang>
Date: Thu, 30 May 2024 16:51:22 +0100
Subject: [PATCH] fix(route): IEEE journal routes (#15720)

* fix: IEEE journal routes

* feat: add feed image

* chore: preprint -> earlyAccess

* chore: remove cookie jar
---
 lib/routes/ieee/earlyaccess.ts            | 100 ------------------
 lib/routes/ieee/journal.ts                | 117 +++++++++-------------
 lib/routes/ieee/recent.ts                 | 107 --------------------
 lib/routes/ieee/templates/description.art |   2 +-
 4 files changed, 51 insertions(+), 275 deletions(-)
 delete mode 100644 lib/routes/ieee/earlyaccess.ts
 delete mode 100644 lib/routes/ieee/recent.ts

diff --git a/lib/routes/ieee/earlyaccess.ts b/lib/routes/ieee/earlyaccess.ts
deleted file mode 100644
index 9f18bc0b02d97..0000000000000
--- a/lib/routes/ieee/earlyaccess.ts
+++ /dev/null
@@ -1,100 +0,0 @@
-import { Route } from '@/types';
-import { getCurrentPath } from '@/utils/helpers';
-const __dirname = getCurrentPath(import.meta.url);
-
-import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
-import path from 'node:path';
-import { art } from '@/utils/render';
-
-import { CookieJar } from 'tough-cookie';
-const cookieJar = new CookieJar();
-
-export const route: Route = {
-    path: '/journal/:journal/earlyaccess/:sortType?',
-    categories: ['journal'],
-    example: '/ieee/journal/5306045/earlyaccess',
-    parameters: { journal: 'Issue code, the number of the `isnumber` in the URL', sortType: 'Sort Type, default: `vol-only-seq`, the part of the URL after `sortType`' },
-    features: {
-        requireConfig: false,
-        requirePuppeteer: false,
-        antiCrawler: false,
-        supportBT: false,
-        supportPodcast: false,
-        supportScihub: false,
-    },
-    name: 'Early Access Journal',
-    maintainers: ['5upernova-heng'],
-    handler,
-};
-
-async function handler(ctx) {
-    const isnumber = ctx.req.param('journal');
-    const sortType = ctx.req.param('sortType') ?? 'vol-only-seq';
-    const host = 'https://ieeexplore.ieee.org';
-    const jrnlUrl = `${host}/xpl/tocresult.jsp?isnumber=${isnumber}`;
-
-    const response = await got(`${host}/rest/publication/home/metadata?issueid=${isnumber}`, {
-        cookieJar,
-    }).json();
-    const punumber = response.publicationNumber;
-    const volume = response.currentIssue.volume;
-    const jrnlName = response.displayTitle;
-
-    const response2 = await got
-        .post(`${host}/rest/search/pub/${punumber}/issue/${isnumber}/toc`, {
-            cookieJar,
-            json: {
-                punumber,
-                isnumber,
-                sortType,
-                rowsPerPage: '100',
-            },
-        })
-        .json();
-    let list = response2.records.map((item) => {
-        const $2 = load(item.articleTitle);
-        const title = $2.text();
-        const link = item.htmlLink;
-        const doi = item.doi;
-        let authors = 'Do not have author';
-        if (Object.hasOwn(item, 'authors')) {
-            authors = item.authors.map((itemAuth) => itemAuth.preferredName).join('; ');
-        }
-        const abstract = Object.hasOwn(item, 'abstract') ? item.abstract : '';
-        return {
-            title,
-            link,
-            authors,
-            doi,
-            volume,
-            abstract,
-        };
-    });
-
-    const renderDesc = (item) =>
-        art(path.join(__dirname, 'templates/description.art'), {
-            item,
-        });
-    list = await Promise.all(
-        list.map((item) =>
-            cache.tryGet(item.link, async () => {
-                if (item.abstract !== '') {
-                    const response3 = await got(`${host}${item.link}`);
-                    const { abstract } = JSON.parse(response3.body.match(/metadata=(.*);/)[1]);
-                    const $3 = load(abstract);
-                    item.abstract = $3.text();
-                    item.description = renderDesc(item);
-                }
-                return item;
-            })
-        )
-    );
-
-    return {
-        title: jrnlName,
-        link: jrnlUrl,
-        item: list,
-    };
-}
diff --git a/lib/routes/ieee/journal.ts b/lib/routes/ieee/journal.ts
index f8e38085b9aa6..c8ffa34713796 100644
--- a/lib/routes/ieee/journal.ts
+++ b/lib/routes/ieee/journal.ts
@@ -2,89 +2,72 @@ import { Route } from '@/types';
 import { getCurrentPath } from '@/utils/helpers';
 const __dirname = getCurrentPath(import.meta.url);
 
-import cache from '@/utils/cache';
 import got from '@/utils/got';
-import { load } from 'cheerio';
 import path from 'node:path';
 import { art } from '@/utils/render';
 
-import { CookieJar } from 'tough-cookie';
-const cookieJar = new CookieJar();
+const ieeeHost = 'https://ieeexplore.ieee.org';
 
 export const route: Route = {
-    path: ['/:journal/latest/vol/:sortType?', '/journal/:journal/:sortType?'],
-    name: 'Unknown',
-    maintainers: [],
+    name: 'IEEE Journal Articles',
+    maintainers: ['HenryQW'],
+    categories: ['journal'],
+    path: '/journal/:punumber/:earlyAccess?',
+    parameters: {
+        punumber: 'Publication Number, look for `punumber` in the URL',
+        earlyAccess: 'Optional, set any value to get early access articles',
+    },
+    example: '/ieee/journal/6287639/preprint',
     handler,
 };
 
 async function handler(ctx) {
-    const punumber = ctx.req.param('journal');
-    const sortType = ctx.req.param('sortType') ?? 'vol-only-seq';
-    const host = 'https://ieeexplore.ieee.org';
-    const jrnlUrl = `${host}/xpl/mostRecentIssue.jsp?punumber=${punumber}`;
+    const publicationNumber = ctx.req.param('punumber');
+    const earlyAccess = !!ctx.req.param('earlyAccess');
 
-    const response = await got(`${host}/rest/publication/home/metadata?pubid=${punumber}`, {
-        cookieJar,
-    }).json();
-    const volume = response.currentIssue.volume;
-    const isnumber = response.currentIssue.issueNumber;
-    const jrnlName = response.displayTitle;
+    const metadata = await fetchMetadata(publicationNumber);
+    const { displayTitle, currentIssue, preprintIssue, coverImagePath } = metadata;
+    const { issueNumber, volume } = earlyAccess ? preprintIssue : currentIssue;
 
-    const response2 = await got
-        .post(`${host}/rest/search/pub/${punumber}/issue/${isnumber}/toc`, {
-            cookieJar,
-            json: {
-                punumber,
-                isnumber,
-                sortType,
-                rowsPerPage: '100',
-            },
-        })
-        .json();
-    let list = response2.records.map((item) => {
-        const $2 = load(item.articleTitle);
-        const title = $2.text();
-        const link = item.htmlLink;
-        const doi = item.doi;
-        let authors = 'Do not have author';
-        if (Object.hasOwn(item, 'authors')) {
-            authors = item.authors.map((itemAuth) => itemAuth.preferredName).join('; ');
-        }
-        let abstract = '';
-        Object.hasOwn(item, 'abstract') ? (abstract = item.abstract) : (abstract = '');
-        return {
-            title,
-            link,
-            authors,
-            doi,
-            volume,
-            abstract,
-        };
-    });
+    const tocData = await fetchTOCData(publicationNumber, issueNumber);
+    const list = tocData.records.map((item) => {
+        const mappedItem = mapRecordToItem(volume)(item);
 
-    const renderDesc = (item) =>
-        art(path.join(__dirname, 'templates/description.art'), {
-            item,
+        mappedItem.description = art(path.join(__dirname, 'templates/description.art'), {
+            item: mappedItem,
         });
-    list = await Promise.all(
-        list.map((item) =>
-            cache.tryGet(item.link, async () => {
-                if (item.abstract !== '') {
-                    const response3 = await got(`${host}${item.link}`);
-                    const { abstract } = JSON.parse(response3.body.match(/metadata=(.*);/)[1]);
-                    const $3 = load(abstract);
-                    item.abstract = $3.text();
-                    item.description = renderDesc(item);
-                }
-                return item;
-            })
-        )
-    );
+
+        return mappedItem;
+    });
 
     return {
-        title: jrnlName,
-        link: jrnlUrl,
+        title: displayTitle,
+        link: `${ieeeHost}/xpl/tocresult.jsp?isnumber=${issueNumber}`,
         item: list,
+        image: `${ieeeHost}${coverImagePath}`,
     };
 }
+
+async function fetchMetadata(punumber) {
+    const response = await got(`${ieeeHost}/rest/publication/home/metadata?pubid=${punumber}`);
+    return response.data;
+}
+
+async function fetchTOCData(punumber, isnumber) {
+    const response = await got.post(`${ieeeHost}/rest/search/pub/${punumber}/issue/${isnumber}/toc`, {
+        json: { punumber, isnumber, rowsPerPage: '100' },
+    });
+    return response.data;
+}
+
+function mapRecordToItem(volume) {
+    return (item) => ({
+        abstract: item.abstract || '',
+        authors: item.authors ? item.authors.map((author) => author.preferredName).join('; ') : '',
+        description: '',
+        doi: item.doi,
+        link: item.htmlLink,
+        title: item.articleTitle || '',
+        volume,
+    });
+}
diff --git a/lib/routes/ieee/recent.ts b/lib/routes/ieee/recent.ts
deleted file mode 100644
index b5f75bc7eef52..0000000000000
--- a/lib/routes/ieee/recent.ts
+++ /dev/null
@@ -1,107 +0,0 @@
-import { Route } from '@/types';
-import { getCurrentPath } from '@/utils/helpers';
-const __dirname = getCurrentPath(import.meta.url);
-
-import cache from '@/utils/cache';
-import got from '@/utils/got';
-import { load } from 'cheerio';
-import path from 'node:path';
-import { art } from '@/utils/render';
-
-import { CookieJar } from 'tough-cookie';
-const cookieJar = new CookieJar();
-
-export const route: Route = {
-    path: ['/:journal/latest/date/:sortType?', '/journal/:journal/recent/:sortType?'],
-    name: 'Unknown',
-    maintainers: [],
-    handler,
-};
-
-async function handler(ctx) {
-    const punumber = ctx.req.param('journal');
-    const sortType = ctx.req.param('sortType') ?? 'vol-only-seq';
-    const host = 'https://ieeexplore.ieee.org';
-    const jrnlUrl = `${host}/xpl/mostRecentIssue.jsp?punumber=${punumber}`;
-
-    const date = new Date();
-    const year = date.getFullYear();
-    const month = date.getMonth() + 1;
-    let strYM, endYM;
-    const snap = 2;
-    if (1 <= month && month <= snap) {
-        month - snap + 12 < 10 ? (strYM = year - 1 + '0' + (month - snap + 12)) : (strYM = year - 1 + '' + (month - snap + 12));
-        endYM = year + '0' + month;
-    } else if (snap < month && month < 10) {
-        month - snap < 10 ? (strYM = year + '0' + (month - snap)) : (strYM = year + '' + (month - snap));
-        endYM = year + '0' + month;
-    } else {
-        month - snap < 10 ? (strYM = year + '0' + (month - snap)) : (strYM = year + '' + (month - snap));
-        endYM = year + '' + month;
-    }
-
-    const response = await got(`${host}/rest/publication/home/metadata?pubid=${punumber}`, {
-        cookieJar,
-    }).json();
-    const volume = response.currentIssue.volume;
-    const isnumber = response.currentIssue.issueNumber;
-    const jrnlName = response.displayTitle;
-
-    const response2 = await got
-        .post(`${host}/rest/search/pub/${punumber}/issue/${isnumber}/toc`, {
-            cookieJar,
-            json: {
-                punumber,
-                isnumber,
-                sortType,
-                rowsPerPage: '100',
-                ranges: [strYM + `01_` + endYM + `31_Search Latest Date`],
-            },
-        })
-        .json();
-    let list = response2.records.map((item) => {
-        const $2 = load(item.articleTitle);
-        const title = $2.text();
-        const link = item.htmlLink;
-        const doi = item.doi;
-        let authors = 'Do not have author';
-        if (Object.hasOwn(item, 'authors')) {
-            authors = item.authors.map((itemAuth) => itemAuth.preferredName).join('; ');
-        }
-        let abstract = '';
-        Object.hasOwn(item, 'abstract') ? (abstract = item.abstract) : (abstract = '');
-        return {
-            title,
-            link,
-            authors,
-            doi,
-            volume,
-            abstract,
-        };
-    });
-
-    const renderDesc = (item) =>
-        art(path.join(__dirname, 'templates/description.art'), {
-            item,
-        });
-    list = await Promise.all(
-        list.map((item) =>
-            cache.tryGet(item.link, async () => {
-                if (item.abstract !== '') {
-                    const response3 = await got(`${host}${item.link}`);
-                    const { abstract } = JSON.parse(response3.body.match(/metadata=(.*);/)[1]);
-                    const $3 = load(abstract);
-                    item.abstract = $3.text();
-                    item.description = renderDesc(item);
-                }
-                return item;
-            })
-        )
-    );
-
-    return {
-        title: `${jrnlName} - Recent`,
-        link: jrnlUrl,
-        item: list,
-    };
-}
diff --git a/lib/routes/ieee/templates/description.art b/lib/routes/ieee/templates/description.art
index 04367cc08d3da..a9e8c5da22219 100644
--- a/lib/routes/ieee/templates/description.art
+++ b/lib/routes/ieee/templates/description.art
@@ -3,7 +3,7 @@
 </p>
 <p>
     <span><small><i>{{ item.authors }}</i></small></span><br>
-    <span><small><i>https://doi.org/{{ item.doi }}</i></small></span><br>
+    <span><small><i><a href="https://doi.org/{{ item.doi }}">https://doi.org/{{ item.doi }}</a></i></small></span><br>
     <span><small><i>Volume {{ item.volume }}</i></small></span><br>
 </p>
 <p>