feat(route): add Transcript Forest (#14482)

* feat(route): add Transcript Forest

* fix: add transcript text

lib/v2/transcriptforest/index.js

@@ -0,0 +1,116 @@
+const got = require('@/utils/got');
+const cheerio = require('cheerio');
+const { parseDate } = require('@/utils/parse-date');
+const { art } = require('@/utils/render');
+const path = require('path');
+
+const bakeTimestamp = (seconds) => {
+    const hours = Math.floor(seconds / 3600);
+    const minutes = Math.floor((seconds % 3600) / 60);
+    const remainingSeconds = Math.floor(seconds % 60);
+
+    const formattedHours = String(hours).padStart(2, '0');
+    const formattedMinutes = String(minutes).padStart(2, '0');
+    const formattedSeconds = String(remainingSeconds).padStart(2, '0');
+
+    return `${formattedHours}:${formattedMinutes}:${formattedSeconds}`;
+};
+
+module.exports = async (ctx) => {
+    const { channel } = ctx.params;
+    const limit = ctx.query.limit ? Number.parseInt(ctx.query.limit, 10) : 10;
+
+    const rootUrl = 'https://www.transcriptforest.com';
+
+    const { data: firstResponse } = await got(rootUrl);
+
+    const data = JSON.parse(firstResponse.match(/({"props".*"scriptLoader":\[]})<\/script>/)?.[1]);
+
+    const buildId = data.buildId;
+    const defaultLocale = data.defaultLocale;
+    const channels = data.props.pageProps.listChannel;
+    const selected = channel ? channels.find((c) => c.channel_id === channel || c.channel_name === channel) : undefined;
+
+    const apiUrl = new URL(`_next/data/${buildId}/en${selected ? `/channel/${selected.channel_id}` : ''}.json`, rootUrl).href;
+    const currentUrl = new URL(selected ? `${defaultLocale}/channel/${selected.channel_id}` : '', rootUrl).href;
+
+    const { data: response } = await got(apiUrl, {
+        searchParams: {
+            channelName: selected ? selected.channel_id : '',
+            offset: 0,
+        },
+    });
+
+    let items = response.pageProps.listEpisode.data.slice(0, limit).map((item) => ({
+        title: item.episode_name,
+        link: new URL(`${defaultLocale}/${item.channel_id}/${item.episode_id}`, rootUrl).href,
+        detailUrl: new URL(`_next/data/${buildId}/${defaultLocale}/${item.channel_id}/${item.episode_id}.json`, rootUrl).href,
+        description: art(path.join(__dirname, 'templates/description.art'), {
+            texts: item.episode_description.split(/\n\n/).map((text) => ({
+                text,
+            })),
+        }),
+        author: item.channel_name,
+        guid: item.id,
+        pubDate: parseDate(item.published_at),
+        updated: parseDate(item.updated_at),
+        itunes_item_image: item.episode_cover.split(/\?/)[0],
+        itunes_duration: item.episode_duration,
+        enclosure_url: item.source_media,
+        enclosure_type: 'audio/mpeg',
+    }));
+
+    items = await Promise.all(
+        items.map((item) =>
+            ctx.cache.tryGet(item.link, async () => {
+                const { data: detailResponse } = await got(item.detailUrl);
+                const { data: textResponse } = await got(detailResponse.pageProps.currentEpisode.ps4_url);
+
+                item.description =
+                    art(path.join(__dirname, 'templates/description.art'), {
+                        audios: [
+                            {
+                                src: detailResponse.pageProps.currentEpisode.media,
+                                type: 'audio/mpeg',
+                            },
+                        ],
+                    }) +
+                    item.description +
+                    art(path.join(__dirname, 'templates/description.art'), {
+                        texts: textResponse.map((t) => ({
+                            startTime: bakeTimestamp(t.startTime),
+                            endTime: bakeTimestamp(t.endTime),
+                            text: t.readOnlyText,
+                        })),
+                    });
+
+                delete item.detailUrl;
+
+                return item;
+            })
+        )
+    );
+
+    const { data: currentResponse } = await got(currentUrl);
+
+    const $ = cheerio.load(currentResponse);
+
+    const title = $('title').text();
+    const image = $('meta[property="og:image"]').prop('content');
+    const icon = new URL($('link[rel="apple-touch-icon"]').prop('href'), rootUrl).href;
+    const author = title.split(/\|/)[0].trim();
+
+    ctx.state.data = {
+        item: items,
+        title,
+        link: currentUrl,
+        description: $('meta[name="description"]').prop('content'),
+        language: $('html').prop('lang'),
+        image,
+        icon,
+        logo: icon,
+        author,
+        itunes_author: author,
+        allowEmpty: true,
+    };
+};

lib/v2/transcriptforest/maintainer.js

@@ -0,0 +1,3 @@
+module.exports = {
+    '/:channel?': ['nczitzk'],
+};