feat(route): add publico.es (#18242)

* Add new route: Spanish newspaper publico.es

* Add new route: Spanish newspaper publico.es (II)

* Add new route: Spanish newspaper publico.es (III)

* Add new route: Spanish newspaper publico.es (fix)

* Add new route: Spanish newspaper publico.es (fixes II)

lib/routes/publico/ciencias.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/ciencias/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/ciencias',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/ciencias'],
+            target: '/ciencias',
+        },
+    ],
+    name: 'Ciencias',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/ciencias/${subsection}` : `${rootUrl}/ciencias`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}

lib/routes/publico/culturas.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/culturas/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/culturas',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/culturas'],
+            target: '/culturas',
+        },
+    ],
+    name: 'Culturas',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/culturas/${subsection}` : `${rootUrl}/culturas`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}

lib/routes/publico/economia.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/economia/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/economia',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/economia'],
+            target: '/economia',
+        },
+    ],
+    name: 'Economia',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/economia/${subsection}` : `${rootUrl}/economia`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}

lib/routes/publico/internacional.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/internacional/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/internacional',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/internacional'],
+            target: '/internacional',
+        },
+    ],
+    name: 'Internacional',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/internacional/${subsection}` : `${rootUrl}/internacional`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}

lib/routes/publico/items-processor.ts

@@ -0,0 +1,20 @@
+export default function getItems(data) {
+    const items = data('.category-list li')
+        .toArray()
+        .map((item) => {
+            item = data(item);
+            const title = item.find('h2').text();
+            const link = item.find('a').attr('href');
+            const author = item.find('p').text();
+            const image = item.find('picture img').attr('src');
+
+            return {
+                title,
+                link,
+                description: `<img src="${image}" alt="${title}">`,
+                author,
+            };
+        });
+
+    return items;
+}

lib/routes/publico/mujer.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/mujer/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/mujer',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/mujer'],
+            target: '/mujer',
+        },
+    ],
+    name: 'Mujer',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/mujer/${subsection}` : `${rootUrl}/mujer`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}

lib/routes/publico/namespace.ts

@@ -0,0 +1,6 @@
+import type { Namespace } from '@/types';
+
+export const namespace: Namespace = {
+    name: 'Público',
+    url: 'publico.es',
+};

lib/routes/publico/opinion.ts

@@ -0,0 +1,55 @@
+import { Route } from '@/types';
+import got from '@/utils/got';
+import { load } from 'cheerio';
+
+import getItems from './items-processor';
+
+export const route: Route = {
+    path: '/opinion/:subsection?',
+    parameters: {
+        subsection: {
+            description: "Filter by subsection. Check the subsections available on the newspaper's website.",
+        },
+    },
+    categories: ['traditional-media'],
+    example: '/publico/opinion',
+    features: {
+        requireConfig: false,
+        requirePuppeteer: true,
+        antiCrawler: false,
+        supportBT: false,
+        supportPodcast: false,
+        supportScihub: false,
+    },
+    radar: [
+        {
+            source: ['publico.es/opinion'],
+            target: '/opinion',
+        },
+    ],
+    name: 'Opinión',
+    maintainers: ['adrianrico97'],
+    handler,
+};
+
+async function handler(ctx) {
+    const { subsection } = ctx.req.param();
+
+    const rootUrl = 'https://www.publico.es';
+    const currentUrl = subsection ? `${rootUrl}/opinion/${subsection}` : `${rootUrl}/opinion`;
+
+    const response = await got({
+        method: 'get',
+        url: currentUrl,
+    });
+
+    const $ = load(response.data);
+    const title = $('.article-section h1').text();
+    const items = getItems($);
+
+    return {
+        title: `${title} | Público`,
+        link: currentUrl,
+        item: items,
+    };
+}