https://grobid.org/sw.js

1// Service Worker for GROBID Tools
2// Version 2.0.0
3// - Pages (navigations): network-first, so content updates reach visitors; cached copy used offline.
4// - Same-origin static assets: stale-while-revalidate.
5// - jsDelivr CDN (PDF.js, docx): network-first with cache fallback, so the extractor works offline.
6// - Analytics and other cross-origin requests are never cached.
7
8const CACHE_NAME = 'grobid-tools-v2';
9const RUNTIME_CACHE = 'grobid-runtime-v2';
10
11// App shell and key pages to cache on install (extensionless URLs, as served by Cloudflare Pages)
12const PRECACHE_URLS = [
13    '/',
14    '/js/app.js',
15    '/css/site.css',
16    '/manifest.json',
17    '/guides/',
18    '/formats/'
19];
20
21self.addEventListener('install', (event) => {
22    event.waitUntil(
23        caches.open(CACHE_NAME)
24            .then(cache => cache.addAll(PRECACHE_URLS))
25            .then(() => self.skipWaiting())
26    );
27});
28
29self.addEventListener('activate', (event) => {
30    event.waitUntil(
31        caches.keys().then(cacheNames => Promise.all(
32            cacheNames
33                .filter(name => name !== CACHE_NAME && name !== RUNTIME_CACHE)
34                .map(name => caches.delete(name))
35        )).then(() => self.clients.claim())
36    );
37});
38
39self.addEventListener('fetch', (event) => {
40    const { request } = event;
41    if (request.method !== 'GET') return;
42
43    const url = new URL(request.url);
44
45    // Cross-origin: only the jsDelivr CDN is cached; everything else (e.g. analytics) goes straight to the network.
46    if (url.origin !== location.origin) {
47        if (url.hostname === 'cdn.jsdelivr.net') {
48            event.respondWith(
49                caches.open(RUNTIME_CACHE).then(cache =>
50                    fetch(request)
51                        .then(response => {
52                            if (response.ok) cache.put(request, response.clone());
53                            return response;
54                        })
55                        .catch(() => cache.match(request))
56                )
57            );
58        }
59        return;
60    }
61
62    // Page navigations: network-first, fall back to cache, then to the home page.
63    if (request.mode === 'navigate') {
64        event.respondWith(
65            fetch(request)
66                .then(response => {
67                    if (response.ok) {
68                        const copy = response.clone();
69                        caches.open(RUNTIME_CACHE).then(cache => cache.put(request, copy));
70                    }
71                    return response;
72                })
73                .catch(() => caches.match(request).then(cached => cached || caches.match('/')))
74        );
75        return;
76    }
77
78    // Other same-origin assets: stale-while-revalidate.
79    event.respondWith(
80        caches.match(request).then(cached => {
81            const network = fetch(request)
82                .then(response => {
83                    if (response.ok) {
84                        const copy = response.clone();
85                        caches.open(RUNTIME_CACHE).then(cache => cache.put(request, copy));
86                    }
87                    return response;
88                })
89                .catch(() => cached);
90            return cached || network;
91        })
92    );
93});
94
95self.addEventListener('message', (event) => {
96    if (event.data && event.data.type === 'SKIP_WAITING') {
97        self.skipWaiting();
98    }
99    if (event.data && event.data.type === 'CLEAR_CACHE') {
100        event.waitUntil(
101            caches.keys().then(names => Promise.all(names.map(name => caches.delete(name))))
102        );
103    }
104});