Grupku.net Scraper — WhatsApp & Telegram Group Harvester

Zero-dependency Node.js crawler for grupku.net directory. Fetches 200+ public WhatsApp & Telegram invite links via concurrent async pool and exports to JSON & CSV.

#scraper#nodejs#whatsapp#telegram#crawler#javascript

4 / kyiov

Raw

grupku-scraper.jsjavascript

257 lines

1	`/**`
2	`* Grupku.net Web Scraper in Node.js`
3	`* Ekstraksi daftar grup WhatsApp & Telegram, URL undangan, kategori, dan deskripsi.`
4	`*/`
5
6	`const fs = require("fs");`
7	`const path = require("path");`
8
9	`const BASE_URL = "https://www.grupku.net/";`
10	`const SITEMAP_URL = "https://www.grupku.net/sitemap.xml";`
11
12	`// Helper HTML Entity Unescape`
13	`function unescapeHtml(text) {`
14	`if (!text) return "";`
15	`return text`
16	`.replace(/&/g, "&")`
17	`.replace(/</g, "<")`
18	`.replace(/>/g, ">")`
19	`.replace(/"/g, '"')`
20	`.replace(/'/g, "'")`
21	`.replace(/•/g, "•")`
22	`.replace(/⛩/g, "⛑")`
23	`.replace(/️/g, "")`
24	`.replace(/¹/g, "¹")`
25	`.replace(/⋆/g, "⋆")`
26	`.replace(/⊱/g, "⊵")`
27	`.replace(/★/g, "★")`
28	`.replace(/♥/g, "♥")`
29	`.replace(/&#([0-9]{1,6});/gi, (_, num) => String.fromCharCode(parseInt(num, 10)));`
30	`}`
31
32	`function cleanText(text) {`
33	`if (!text) return "";`
34	`const noTags = text.replace(/<[^>]+>/g, "");`
35	`return unescapeHtml(noTags).replace(/\s+/g, " ").trim();`
36	`}`
37
38	`async function fetchWithRetry(url, retries = 2) {`
39	`for (let i = 0; i <= retries; i++) {`
40	`try {`
41	`const res = await fetch(url, {`
42	`headers: {`
43	`"User-Agent":`
44	`"Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/124.0.0.0 Safari/537.36",`
45	`},`
46	`signal: AbortSignal.timeout(12000),`
47	`});`
48	`if (res.ok) {`
49	`return await res.text();`
50	`}`
51	`} catch (err) {`
52	`if (i === retries) throw err;`
53	`await new Promise((r) => setTimeout(r, 600));`
54	`}`
55	`}`
56	`return "";`
57	`}`
58
59	`async function getAllGroupUrls() {`
60	`const urls = new Set();`
61
62	`// 1. Ambil dari sitemap.xml`
63	`try {`
64	`const sitemapXml = await fetchWithRetry(SITEMAP_URL);`
65	`const matches = sitemapXml.match(/<loc>(https:\/\/www\.grupku\.net\/202[0-9]\/[0-9]{2}\/[^\s<]+\.html)<\/loc>/g) \|\| [];`
66	`for (const m of matches) {`
67	`const cleanUrl = m.replace(/<\/?loc>/g, "").trim();`
68	`urls.add(cleanUrl);`
69	`}`
70	`} catch (err) {`
71	`console.error("[!] Error membaca sitemap:", err.message);`
72	`}`
73
74	`// 2. Ambil dari homepage`
75	`try {`
76	`const homeHtml = await fetchWithRetry(BASE_URL);`
77	`const matches = homeHtml.match(/https:\/\/www\.grupku\.net\/202[0-9]\/[0-9]{2}\/[a-zA-Z0-9_\-]+\.html/g) \|\| [];`
78	`for (const u of matches) {`
79	`urls.add(u.trim());`
80	`}`
81	`} catch (err) {`
82	`console.error("[!] Error membaca homepage:", err.message);`
83	`}`
84
85	`return Array.from(urls);`
86	`}`
87
88	`function parseGroupPage(pageHtml, url) {`
89	`// 1. Title`
90	`let title = "Tanpa Judul";`
91	`const h1Match = pageHtml.match(/<h1[^>]class=['"]gs-single-title['"][^>]>([\s\S]*?)<\/h1>/i);`
92	`if (h1Match) {`
93	`let raw = h1Match[1].replace(/<span[^>]class=['"][^'"]gs-tier-badge[^'"]['"][^>]>[\s\S]*?<\/span>/gi, "");`
94	`title = cleanText(raw);`
95	`} else {`
96	`const ogTitle = pageHtml.match(/<meta[^>]+property=['"]og:title['"][^>]+content=['"]([^'"]+)['"]/i);`
97	`if (ogTitle) title = cleanText(ogTitle[1]);`
98	`}`
99
100	`// 2. Direct Invite URL (WhatsApp / Telegram)`
101	`let inviteUrl = "";`
102	`const urlDiv = pageHtml.match(/<div[^>]class=['"]group-url['"][^>]>\s(https?:\/\/[^\s<]+)\s<\/div>/i);`
103	`if (urlDiv) {`
104	`inviteUrl = urlDiv[1].trim();`
105	`} else {`
106	`const direct = pageHtml.match(/https?:\/\/(?:chat\.whatsapp\.com\|t\.me)\/[a-zA-Z0-9_\-\.\?&=]+/i);`
107	`if (direct) {`
108	`inviteUrl = direct[0].trim();`
109	`}`
110	`}`
111
112	`// 3. Platform`
113	`let platform = "Other";`
114	`if (inviteUrl.includes("chat.whatsapp.com") \|\| inviteUrl.includes("wa.me") \|\| /whatsapp/i.test(pageHtml)) {`
115	`platform = "WhatsApp";`
116	`} else if (inviteUrl.includes("t.me") \|\| /telegram/i.test(pageHtml)) {`
117	`platform = "Telegram";`
118	`}`
119
120	`// 4. Description`
121	`let description = "";`
122	`const descMatch = pageHtml.match(/<div[^>]class=['"]gs-single-desc['"][^>]>[\s\S]?<p>([\s\S]?)<\/p>/i);`
123	`if (descMatch) {`
124	`description = cleanText(descMatch[1]);`
125	`}`
126
127	`// 5. Icon URL`
128	`let iconUrl = "";`
129	`const iconMatch = pageHtml.match(/<div[^>]class=['"]gs-single-thumb['"][^>]>[\s\S]*?<img[^>]+src=['"]([^'"]+)['"]/i) \|\|`
130	`pageHtml.match(/<meta[^>]+property=['"]og:image['"][^>]+content=['"]([^'"]+)['"]/i);`
131	`if (iconMatch) {`
132	`iconUrl = iconMatch[1];`
133	`}`
134
135	`// 6. Tags & Category`
136	`const tagMatches = pageHtml.match(/<a[^>]class=['"]gs-tag['"][^>]>\s([\s\S]?)\s*<\/a>/gi) \|\| [];`
137	`const tags = tagMatches`
138	`.map((t) => cleanText(t))`
139	`.filter((t) => t && !t.includes("cat.term"));`
140	`const uniqueTags = Array.from(new Set(tags));`
141	`const category = uniqueTags[0] \|\| "Umum";`
142
143	`// 7. Date`
144	`let date = "";`
145	`const dateMatch = pageHtml.match(/<i class=['"]fas fa-calendar-alt['"]><\/i>\s*([^<]+)<\/span>/i);`
146	`if (dateMatch) {`
147	`date = cleanText(dateMatch[1]);`
148	`}`
149
150	`return {`
151	`title,`
152	`platform,`
153	`category,`
154	`tags: uniqueTags,`
155	`invite_url: inviteUrl,`
156	`description,`
157	`date,`
158	`icon_url: iconUrl,`
159	`page_url: url,`
160	`};`
161	`}`
162
163	`// Concurrency pool helper`
164	`async function pool(items, limit, fn) {`
165	`const results = [];`
166	`const executing = [];`
167	`for (const item of items) {`
168	`const p = Promise.resolve().then(() => fn(item));`
169	`results.push(p);`
170	`if (limit <= items.length) {`
171	`const e = p.then(() => executing.splice(executing.indexOf(e), 1));`
172	`executing.push(e);`
173	`if (executing.length >= limit) {`
174	`await Promise.race(executing);`
175	`}`
176	`}`
177	`}`
178	`return Promise.all(results);`
179	`}`
180
181	`async function main() {`
182	`console.log("[*] Mengambil link grup dari grupku.net ...");`
183	`const urls = await getAllGroupUrls();`
184	console.log(`[+] Ditemukan ${urls.length} postingan grup.`);
185
186	`if (urls.length === 0) {`
187	`console.error("[-] Tidak ada URL yang ditemukan.");`
188	`return;`
189	`}`
190
191	`console.log("[*] Mulai scraping data (concurrency 10)...");`
192	`let processed = 0;`
193	`const groups = [];`
194
195	`await pool(urls, 10, async (url) => {`
196	`try {`
197	`const html = await fetchWithRetry(url);`
198	`const data = parseGroupPage(html, url);`
199	`if (data.invite_url) {`
200	`groups.push(data);`
201	`}`
202	`} catch (err) {`
203	`// ignore individual failure`
204	`} finally {`
205	`processed++;`
206	`if (processed % 25 === 0 \|\| processed === urls.length) {`
207	console.log(` Progres: ${processed}/${urls.length} diproses (Grup valid: ${groups.length})`);
208	`}`
209	`}`
210	`});`
211
212	`// Urutkan berdasarkan judul`
213	`groups.sort((a, b) => a.title.localeCompare(b.title));`
214
215	`// Simpan JSON`
216	`const jsonPath = path.resolve(process.cwd(), "grupku_data.json");`
217	`fs.writeFileSync(jsonPath, JSON.stringify(groups, null, 2), "utf-8");`
218	console.log(`[✓] Berhasil disimpan ke JSON: ${jsonPath}`);
219
220	`// Simpan CSV`
221	`const csvPath = path.resolve(process.cwd(), "grupku_data.csv");`
222	`const csvHeaders = ["Title", "Platform", "Category", "Invite URL", "Description", "Date", "Page URL"];`
223	`const csvRows = [csvHeaders.join(",")];`
224	`for (const g of groups) {`
225	const escapeCsv = (str) => `"${(str \|\| "").replace(/"/g, '""')}"`;
226	`csvRows.push([`
227	`escapeCsv(g.title),`
228	`escapeCsv(g.platform),`
229	`escapeCsv(g.category),`
230	`escapeCsv(g.invite_url),`
231	`escapeCsv(g.description),`
232	`escapeCsv(g.date),`
233	`escapeCsv(g.page_url),`
234	`].join(","));`
235	`}`
236	`fs.writeFileSync(csvPath, csvRows.join("\n"), "utf-8");`
237	console.log(`[✓] Berhasil disimpan ke CSV: ${csvPath}`);
238
239	`// Tampilkan Ringkasan`
240	`console.log("\n========================================================");`
241	console.log(`HASIL SCRAPING NODE.JS (${groups.length} Grup Berhasil Diambil)`);
242	`console.log("========================================================");`
243
244	`const stats = {};`
245	`for (const g of groups) {`
246	`stats[g.category] = (stats[g.category] \|\| 0) + 1;`
247	`}`
248	`const topCategories = Object.entries(stats).sort((a, b) => b[1] - a[1]);`
249	`console.log("Kategori Terbanyak:");`
250	`topCategories.slice(0, 8).forEach(([cat, count], i) => {`
251	console.log(` ${i + 1}. ${cat}: ${count} grup`);
252	`});`
253	`console.log("========================================================\n");`
254	`}`
255
256	`main().catch(console.error);`
257

Discussion & Comments0

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