cse.js — cari di Google Programmable Search Engine (element API, pure JS)

#github-gist#javascript#gist:01680dc12d08ead52e347b3212091a9d#gistsync:2026-09-04T02:58:45Z

4 / kyiov

Raw

cse.jsjavascript

134 lines

1	`#!/usr/bin/env node`
2	`// cse.js — cari di Google Programmable Search Engine (element API, pure JS)`
3
4	`const CX = '006516753008110874046:1ugcdt3vo7z' // download/direct-link search`
5	`const UA = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36'`
6	`const BASE = 'https://cse.google.com'`
7
8	`let jar = {} // cookie jar: name -> value`
9
10	`async function get(url, referer) {`
11	`const r = await fetch(url, {`
12	`headers: {`
13	`'User-Agent': UA,`
14	`'Accept': '/',`
15	`'Accept-Language': 'en-US,en;q=0.9',`
16	`...(referer ? { 'Referer': referer } : {}),`
17	...(Object.keys(jar).length ? { 'Cookie': Object.entries(jar).map(([k, v]) => `${k}=${v}`).join('; ') } : {})
18	`}`
19	`})`
20	`const raw = await r.text()`
21	`// simpan cookie baru dari response`
22	`let sc`
23	`if (typeof r.headers.getSetCookie === 'function') sc = r.headers.getSetCookie()`
24	`else if (r.headers.get('set-cookie')) sc = [r.headers.get('set-cookie')]`
25	`if (sc) for (const c of sc) {`
26	`const [pair] = c.split(';')`
27	`const eq = pair.indexOf('=')`
28	`if (eq > 0) jar[pair.slice(0, eq).trim()] = pair.slice(eq + 1).trim()`
29	`}`
30	`return { status: r.status, raw }`
31	`}`
32
33	`// ambil sca_esv dari halaman utama`
34	`async function mainPage(cx) {`
35	const { raw } = await get(`${BASE}/cse?cx=${encodeURIComponent(cx)}`)
36	`const m = /sca_esv\\x3d([a-f0-9]+)/.exec(raw) \|\| /sca_esv=([a-f0-9]+)/.exec(raw)`
37	`return m ? m[1] : null`
38	`}`
39
40	`// ambil cse_token dari cse.js`
41	`async function cseTok(sca, cx) {`
42	const { raw } = await get(`${BASE}/cse.js?sca_esv=${sca}&hpg=1&cx=${encodeURIComponent(cx)}`, `${BASE}/cse?cx=${encodeURIComponent(cx)}`)
43	`const m = /"cse_token":\s*"([^"]+)"/.exec(raw)`
44	`if (!m) throw new Error('cse_token tidak ditemukan')`
45	`return m[1]`
46	`}`
47
48	`// ambil hasil`
49	`async function search(cx, q, start, num) {`
50	`const sca = await mainPage(cx)`
51	`const tok = await cseTok(sca, cx)`
52	const cb = `google.search.cse.api${Date.now() % 100000}`
53	`const params = new URLSearchParams({`
54	`rsz: 'filtered_cse', num: String(num), hl: 'en', source: 'gcsc',`
55	`cselibv: '93e57cb47c499c53', cx, q, safe: 'off', cse_tok: tok,`
56	`filter: '1', sort: '', exp: 'cc,esbfa', fexp: '121991397,121991398,121877338,121877336',`
57	`oq: q, gs_l: 'partner-generic.3...3113.3113.0.3141.1.1.0.0.0.0.0.0..0.0.csems,nrl=10...0.....34.partner-generic..1.0.0.',`
58	`cseclient: 'hosted-page-client', callback: cb,`
59	rurl: `${BASE}/cse?cx=${encodeURIComponent(cx)}`
60	`})`
61	`if (start) params.set('start', String(start))`
62	const { status, raw } = await get(`${BASE}/cse/element/v1?${params}`, `${BASE}/cse?cx=${encodeURIComponent(cx)}`)
63	if (status !== 200) throw new Error(`HTTP ${status}`)
64	`const i = raw.indexOf('(')`
65	`const j = raw.lastIndexOf(')')`
66	`return JSON.parse(raw.slice(i + 1, j))`
67	`}`
68
69	`function clean(s) {`
70	`if (s == null) return ''`
71	`return String(s).replace(/<[^>]+>/g, '').replace(/&/g, '&').replace(/</g, '<').replace(/>/g, '>').replace(/"/g, '"').replace(/'/g, "'").trim()`
72	`}`
73
74	`function fmt(d) {`
75	`return (d.results \|\| []).map(r => ({`
76	`title: clean(r.title),`
77	`link: r.unescapedUrl \|\| r.url \|\| '',`
78	`domain: r.visibleUrl \|\| '',`
79	`snippet: clean(r.content)`
80	`}))`
81	`}`
82
83	`async function run() {`
84	`const args = process.argv.slice(2)`
85	`const asJson = args.includes('--json')`
86	`const outIdx = args.indexOf('--out')`
87	`const outFile = outIdx >= 0 ? (args[outIdx + 1] && !args[outIdx + 1].startsWith('--') ? args[outIdx + 1] : 'output/cse.json') : null`
88	`let n = 10, start = 0`
89	`const ni = args.indexOf('--n')`
90	`if (ni >= 0) n = parseInt(args[ni + 1], 10) \|\| 10`
91	`const pi = args.indexOf('--page')`
92	`if (pi >= 0) start = (parseInt(args[pi + 1], 10) \|\| 1) * 10`
93	`const q = args.find(a => !a.startsWith('--'))`
94	`if (!q) {`
95	`console.error('usage: node cse.js <kata> [--n 30] [--page 2] [--json] [--out f]')`
96	`process.exit(1)`
97	`}`
98
99	`// kunjungan pertama ke cse.google.com untuk dapat NID`
100	await get(`${BASE}/cse?cx=${encodeURIComponent(CX)}`)
101
102	`const all = []`
103	`let total = 0`
104	`for (let s = start; s < start + n && all.length < n; s += 10) {`
105	`const d = await search(CX, q, s, Math.min(10, n - all.length))`
106	`if (s === start) total = parseInt((d.cursor \|\| {}).estimatedResultCount, 10) \|\| 0`
107	`const items = fmt(d)`
108	`if (!items.length) break`
109	`all.push(...items)`
110	`if (items.length < 10) break`
111	`}`
112
113	`const out = { query: q, total, shown: all.length, results: all }`
114	`if (asJson) {`
115	`console.log(JSON.stringify(out, null, 2))`
116	`} else {`
117	console.log(`${q}: ${total} hasil (${all.length} ditampilkan)`)
118	`all.forEach((r, i) => {`
119	console.log(`\n[${i + 1}] ${r.title}`)
120	if (r.domain) console.log(` ${r.domain}`)
121	console.log(` ${r.link}`)
122	if (r.snippet) console.log(` ${r.snippet.slice(0, 160)}`)
123	`})`
124	`}`
125	`if (outFile) {`
126	`const { mkdir, writeFile } = await import('node:fs/promises')`
127	`const { dirname } = await import('node:path')`
128	`await mkdir(dirname(outFile), { recursive: true })`
129	`await writeFile(outFile, JSON.stringify(out, null, 2))`
130	console.error(`\nsaved: ${outFile}`)
131	`}`
132	`}`
133
134	`run().catch(e => { console.error('error:', e.message); process.exit(1) })`

Discussion & Comments0

No comments yet — be the first to leave feedback or start a code discussion.