feat(route): pubscholar

lib/routes/pubscholar/explore.ts

@@ -0,0 +1,62 @@
+import { Route } from '@/types';
+import ofetch from '@/utils/ofetch';
+import { parseDate } from '@/utils/parse-date';
+import { baseUrl, uuidv4, getArticleLink, getSignedHeaders } from './utils';
+import md5 from '@/utils/md5';
+import { Resource } from './types';
+import sanitizeHtml from 'sanitize-html';
+
+export const route: Route = {
+    path: '/explore/:category?/:keyword?',
+    name: 'Explore',
+    maintainers: ['TonyRL'],
+    example: '/pubscholar/explore',
+    parameters: {
+        category: 'Category, see the table below, `articles` by default',
+        keyword: 'Search Keyword',
+    },
+    handler,
+    description: `| Articles / 论文 | Patents / 专利 | Reports / 领域快报 | Information / 动态快讯 | Datasets / 科学数据 | Books / 图书 |
+| --------------- | -------------- | ------------------ | ---------------------- | ------------------- | ------------ |
+| articles        | patents        | bulletins          | reports                | sciencedata         | books        |`,
+};
+
+async function handler(ctx) {
+    const { category = 'articles', keyword } = ctx.req.param();
+    const uuid = uuidv4();
+
+    const response = await ofetch<Resource>(`${baseUrl}/hky/open/resources/api/v1/${category}`, {
+        method: 'POST',
+        headers: {
+            ...getSignedHeaders(),
+            Cookie: `XSRF-TOKEN=${uuid}`,
+            'X-XSRF-TOKEN': uuid,
+        },
+        body: {
+            page: 1,
+            size: 10,
+            order_field: 'date',
+            order_direction: 'desc',
+            user_id: md5(Date.now().toString()),
+            lang: 'zh',
+            query: keyword,
+            strategy: null,
+            orderField: 'default',
+        },
+    });
+
+    const list = response.content.map((item) => ({
+        title: (item.is_free || item.links.some((l) => l.is_open_access) ? '「Open Access」' : '') + sanitizeHtml(item.title, { allowedTags: [], allowedAttributes: {} }),
+        description: item.abstracts + `<br>${item.links.map((link) => `<a href="${link.url}">${link.is_open_access ? '「Open Access」' : ''}${link.name}</a>`).join('<br>')}`,
+        author: item.author.join('; '),
+        pubDate: parseDate(item.date),
+        category: item.keywords.map((keyword) => sanitizeHtml(keyword, { allowedTags: [], allowedAttributes: {} })),
+        link: `${baseUrl}/${category}/${getArticleLink(item.id)}`,
+    }));
+
+    return {
+        title: 'PubScholar 公益学术平台',
+        link: `${baseUrl}/explore`,
+        item: list,
+    };
+}

lib/routes/pubscholar/namespace.ts

@@ -0,0 +1,7 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'PubScholar 公益学术平台',
+    url: 'pubscholar.cn',
+    categories: ['journal'],
+};

lib/routes/pubscholar/types.ts

@@ -0,0 +1,41 @@
+interface Link {
+    is_open_access: boolean;
+    name: string;
+    url: string;
+}
+
+interface Content {
+    date: string;
+    attachments: any[];
+    keywords: string[];
+    year: number;
+    source: string;
+    title: string;
+    type: string;
+    abstracts_abbreviation: string;
+    major: string;
+    school: any[];
+    first_page: string;
+    local_links: any[];
+    links: Link[];
+    id: string;
+    graduation_institution: any[];
+    cn_type: string;
+    article_type: string;
+    issue: string;
+    abstracts: string;
+    author: string[];
+    last_page: string;
+    degree: string;
+    tutor: any[];
+    semantic_entities: object;
+    volume: string;
+    source_list: string[];
+    is_free: boolean;
+}
+
+export interface Resource {
+    total: number;
+    is_last: boolean;
+    content: Content[];
+}

lib/routes/pubscholar/utils.ts

@@ -0,0 +1,44 @@
+import crypto from 'node:crypto';
+import CryptoJS from 'crypto-js';
+
+const salt = '6m6pingbinwaktg227gngifoocrfbo95';
+const key = CryptoJS.enc.Utf8.parse('eRtYuIoPaSdFgHqW');
+const iv = CryptoJS.enc.Utf8.parse('Nmc09JkLzX8765Vb');
+
+export const baseUrl = 'https://pubscholar.cn';
+export const sha1 = (str: string) => crypto.createHash('sha1').update(str).digest('hex');
+export const uuidv4 = () => crypto.randomUUID();
+
+const generateNonce = (length: number): string => {
+    if (!length) {
+        return null;
+    }
+
+    let nonce = '';
+    while (nonce.length < length) {
+        const randomString = Math.random().toString(36).slice(2).toUpperCase();
+        nonce += randomString;
+    }
+
+    return nonce.slice(0, length);
+};
+
+export const getSignedHeaders = () => {
+    const nonce = generateNonce(6);
+    const timestamp = Date.now().toString();
+    const signature = sha1([salt, timestamp, nonce].sort().join(''));
+    return {
+        nonce,
+        timestamp,
+        signature,
+    };
+};
+
+export const getArticleLink = (id: string) => {
+    const ciphertext = CryptoJS.AES.encrypt(CryptoJS.enc.Utf8.parse(id), key, {
+        iv,
+        mode: CryptoJS.mode.CBC,
+        padding: CryptoJS.pad.Pkcs7,
+    }).ciphertext.toString();
+    return ciphertext;
+};