Skip to content

Commit 938c1cc

Browse files
pull[bot]pull[bot]
authored
Merge pull request #114 from DIYgod/master
[pull] master from diygod:master
2 parents 800410b + c4f1afd commit 938c1cc

File tree

1 file changed

+82
-52
lines changed

1 file changed

+82
-52
lines changed

‎lib/routes/theinitium/app.ts‎

Lines changed: 82 additions & 52 deletions
Original file line numberDiff line numberDiff line change
@@ -7,6 +7,10 @@ import path from 'node:path';
77
import { config } from '@/config';
88
import { getCurrentPath } from '@/utils/helpers';
99

10+
const __dirname = getCurrentPath(import.meta.url);
11+
const appUrl = 'https://app.theinitium.com/';
12+
const userAgent = 'PugpigBolt v4.1.8 (iPhone, iOS 18.2.1) on phone (model iPhone15,2)';
13+
1014
export const route: Route = {
1115
path: '/app/:category?',
1216
categories: ['new-media', 'popular'],
@@ -52,51 +56,96 @@ Category 栏目：
5256
| 播客 | article_audio_sc | article_audio_tc |`,
5357
};
5458

55-
const resolveRelativeLink = ($: CheerioAPI, elem: Element, attr: string, baseUrl?: string) => {
59+
const resolveRelativeLink = ($: CheerioAPI, elem: Element, attr: string, appUrl?: string) => {
5660
// code from @/middleware/paratmeter.ts
5761
const $elem = $(elem);
5862

59-
if (baseUrl) {
63+
if (appUrl) {
6064
try {
6165
const oldAttr = $elem.attr(attr);
6266
if (oldAttr) {
6367
// e.g. <video><source src="https://example.com"></video> should leave <video> unchanged
64-
$elem.attr(attr, new URL(oldAttr, baseUrl).href);
68+
$elem.attr(attr, new URL(oldAttr, appUrl).href);
6569
}
6670
} catch {
6771
// no-empty
6872
}
6973
}
7074
};
7175

72-
async function handler(ctx) {
73-
const category = ctx.req.param('category') ?? 'latest_sc';
74-
const __dirname = getCurrentPath(import.meta.url);
75-
const baseUrl = 'https://app.theinitium.com/';
76-
const userAgent = 'PugpigBolt v4.1.8 (iPhone, iOS 18.2.1) on phone (model iPhone15,2)';
77-
78-
async function getUA(url: string) {
79-
return await got({
80-
method: 'get',
81-
url,
82-
headers: {
83-
'User-Agent': userAgent,
84-
},
85-
});
86-
}
76+
async function getUA(url: string) {
77+
return await got({
78+
method: 'get',
79+
url,
80+
headers: {
81+
'User-Agent': userAgent,
82+
},
83+
});
84+
}
8785

88-
const feeds = await cache.tryGet(new URL('timelines.json', baseUrl).href, async () => await getUA(new URL('timelines.json', baseUrl).href), config.cache.routeExpire, false);
86+
async function fetchAppPage(url: URL) {
87+
const response = await getUA(url.href);
88+
const $ = load(response.data);
89+
// resolve relative links with app.theinitium.com
90+
// code from @/middleware/paratmeter.ts
91+
$('a, area').each((_, elem) => {
92+
resolveRelativeLink($, elem, 'href', appUrl);
93+
// $(elem).attr('rel', 'noreferrer'); // currently no such a need
94+
});
95+
// https://www.w3schools.com/tags/att_src.asp
96+
$('img, video, audio, source, iframe, embed, track').each((_, elem) => {
97+
resolveRelativeLink($, elem, 'src', appUrl);
98+
$(elem).removeAttr('srcset');
99+
});
100+
$('video[poster]').each((_, elem) => {
101+
resolveRelativeLink($, elem, 'poster', appUrl);
102+
});
103+
const article = $('.pp-article__body');
104+
article.find('.block-related-articles').remove();
105+
article.find('figure.wp-block-pullquote').children().unwrap();
106+
article.find('div.block-explanation-note').wrapInner('<blockquote></blockquote>');
107+
article.find('div.wp-block-tcc-author-note').wrapInner('<em></em>').after('<hr>');
108+
article.find('p.has-small-font-size').wrapInner('<small></small>');
109+
return art(path.join(__dirname, 'templates/description.art'), {
110+
standfirst: $('.pp-header-group__standfirst').html(),
111+
coverImage: $('.pp-media__image').attr('src'),
112+
coverCaption: $('.pp-media__caption').html(),
113+
article: article.html(),
114+
copyright: $('.copyright').html(),
115+
});
116+
}
89117

90-
const metadata = feeds.data.timelines.find((timeline) => timeline.id === category);
118+
async function fetchWebPage(url: URL) {
119+
const response = await got(url.href);
120+
const $ = load(response.data);
121+
const article = $('.wkwp-post-content');
122+
article.find('.block-related-articles').remove();
123+
article.find('figure.wp-block-pullquote').children().unwrap();
124+
article.find('div.block-explanation-note').wrapInner('<blockquote></blockquote>');
125+
article.find('div.wp-block-tcc-author-note').wrapInner('<em></em>').after('<hr>');
126+
article.find('p.has-small-font-size').wrapInner('<small></small>');
127+
return art(path.join(__dirname, 'templates/description.art'), {
128+
standfirst: $('span.caption1').html(),
129+
coverImage: $('.wp-post-image').attr('src'),
130+
coverCaption: $('.image-caption').html(),
131+
article: article.html(),
132+
copyright: $('.entry-copyright').html(),
133+
});
134+
}
91135

92-
const response = await getUA(new URL(metadata.feed, baseUrl).href);
136+
async function handler(ctx) {
137+
const category = ctx.req.param('category') ?? 'latest_sc';
93138

139+
const feeds = await cache.tryGet(new URL('timelines.json', appUrl).href, async () => await getUA(new URL('timelines.json', appUrl).href), config.cache.routeExpire, false);
140+
const metadata = feeds.data.timelines.find((timeline) => timeline.id === category);
141+
const response = await getUA(new URL(metadata.feed, appUrl).href);
94142
const feed = response.data.stories.filter((item) => item.type === 'article');
95143

96144
const items = await Promise.all(
97145
feed.map((item) =>
98-
cache.tryGet(new URL(item.url, baseUrl).href, async () => {
99-
item.link = item.shareurl ?? new URL(item.url, baseUrl).href;
146+
cache.tryGet(item.shareurl, async () => {
147+
const url = new URL(item.shareurl);
148+
item.link = url.href;
100149
item.description = item.summary;
101150
item.pubDate = item.published;
102151
item.category = [];
@@ -112,35 +161,16 @@ async function handler(ctx) {
112161
}
113162
}
114163
item.category = [...new Set(item.category)];
115-
const response = await getUA(new URL(item.url, baseUrl).href);
116-
const $ = load(response.data);
117-
// resolve relative links with app.theinitium.com
118-
// code from @/middleware/paratmeter.ts
119-
$('a, area').each((_, elem) => {
120-
resolveRelativeLink($, elem, 'href', baseUrl);
121-
// $(elem).attr('rel', 'noreferrer'); // currently no such a need
122-
});
123-
// https://www.w3schools.com/tags/att_src.asp
124-
$('img, video, audio, source, iframe, embed, track').each((_, elem) => {
125-
resolveRelativeLink($, elem, 'src', baseUrl);
126-
$(elem).removeAttr('srcset');
127-
});
128-
$('video[poster]').each((_, elem) => {
129-
resolveRelativeLink($, elem, 'poster', baseUrl);
130-
});
131-
const article = $('.pp-article__body');
132-
article.find('.block-related-articles').remove();
133-
article.find('figure.wp-block-pullquote').children().unwrap();
134-
article.find('div.block-explanation-note').wrapInner('<blockquote></blockquote>');
135-
article.find('div.wp-block-tcc-author-note').wrapInner('<em></em>').after('<hr>');
136-
article.find('p.has-small-font-size').wrapInner('<small></small>');
137-
item.description = art(path.join(__dirname, 'templates/description.art'), {
138-
standfirst: $('.pp-header-group__standfirst').html(),
139-
coverImage: $('.pp-media__image').attr('src'),
140-
coverCaption: $('.pp-media__caption').html(),
141-
article: article.html(),
142-
copyright: $('.copyright').html(),
143-
});
164+
switch (url.hostname) {
165+
case 'app.theinitium.com':
166+
item.description = await fetchAppPage(url);
167+
break;
168+
case 'theinitium.com':
169+
item.description = await fetchWebPage(url);
170+
break;
171+
default:
172+
break;
173+
}
144174
return item;
145175
})
146176
)

0 commit comments

Comments
 (0)