Telegram Phone Number Scraper

Telegram Phone Number Scraper 📱

Finds public Telegram channels for your keywords, opens each channel's real public post feed on t.me, and extracts every phone number that is genuinely valid for the country you select.

How it actually works 🔬

Telegram publishes every public channel's full post HTML, with no login, at https://t.me/s/<channel> (20 posts per page, ?before=<id> walks backwards through the archive). This actor uses a search engine only to discover which channels match your keyword, and then reads the numbers out of that real, untruncated post text — plus any tel: and WhatsApp (wa.me) links inside the posts.

It deliberately does not mine search-result snippets for the number itself. A snippet is Google's ~160-character summary of the channel, cut off mid-sentence, which is how scrapers end up emitting digit fragments that look like phone numbers but cannot be dialled.

Every number is validated against a real numbering plan ✅

A number is emitted only when both of these hold:

1. It is written in explicit international form in the post — +998 90 123 45 67, 00998901234567, tel:+998..., wa.me/998.... Telegram is an international platform and public business channels overwhelmingly write numbers that way. A bare local digit run carries no evidence about which country it belongs to, so it is never guessed at.
2. Its length and its leading digits are legal under the selected country's national numbering plan.

For +1, the area code decides the country: Canadian area codes are required for Canada, excluded for the United States, and the 21 Caribbean/Atlantic +1 territories, the non-geographic codes (800/833/844/855/866/877/888/900) and the N11 service codes are excluded from both. +7 is split between Russia (NSN starting 3/4/5/8/9) and Kazakhstan (6/7). Dial codes are not unique, so the tables are keyed by country name, never by code.

Country coverage 🌍

The country list was measured live, not assumed. For every candidate country the full pipeline was run — discovery query, real channel fetch, real validation — and what is recorded is the count of distinct valid numbers actually recovered.

Full coverage (numbers came back easily): Belarus, Uzbekistan, Ukraine, Iran, Germany, Israel, Czech Republic, Ethiopia, Poland, Turkey, United Arab Emirates, Italy, Russia, United States, Singapore, Azerbaijan, Japan, Portugal.

Limited (real numbers exist but are sparse — expect small result counts): Afghanistan, Algeria, Argentina, Armenia, Australia, Austria, Bangladesh, Brazil, Bulgaria, Cameroon, Canada, China, Colombia, Denmark, Egypt, Estonia, Finland, France, Georgia, Ghana, Greece, Hong Kong, India, Indonesia, Iraq, Ireland, Ivory Coast, Jordan, Kenya, Korea, South, Kuwait, Kyrgyzstan, Latvia, Lebanon, Libya, Malaysia, Mexico, Moldova, Morocco, Myanmar, Nepal, Netherlands, Nigeria, Norway, Oman, Pakistan, Peru, Philippines, Qatar, Romania, Saudi Arabia, Senegal, Serbia, Slovakia, South Africa, Spain, Sri Lanka, Sudan, Sweden, Syria, Taiwan, Tajikistan, Tanzania, Thailand, Tunisia, Turkmenistan, Uganda, United Kingdom, Venezuela, Vietnam, Yemen.

Countries that were measured and returned nothing are not offered, and are rejected up front rather than burning a run that cannot produce anything. Measured-zero: Kazakhstan, Chile, Lithuania, Slovenia, Switzerland, New Zealand.

Input Parameters 🧩

Field	Type	What it does
keywords	array	Search terms, @usernames, or t.me URLs. A keyword runs discovery; a username/URL scrapes that channel directly with no search at all.
country	enum	Hard filter. Only numbers valid under this country's numbering plan are emitted.
maxPhoneNumbers	integer	Hard cap of distinct numbers per keyword.
engine	enum	deep reads full channel post feeds (recommended). fast reads only search snippets — quicker, far fewer results.
proxyConfiguration	object	Proxy for the discovery step. Defaults to Apify Proxy GOOGLE_SERP. Telegram itself is fetched directly.

📦 Example input

{
  "keywords": ["kvartira", "https://t.me/s/Maklersiz"],
  "country": "Uzbekistan (+998)",
  "maxPhoneNumbers": 25,
  "engine": "deep",
  "proxyConfiguration": { "useApifyProxy": true, "apifyProxyGroups": ["GOOGLE_SERP"] }
}

Output 📤

One row per distinct phone number. Duplicates are removed across the whole run.

{
  "platform": "Telegram",
  "keyword": "kvartira",
  "title": "Maklersiz uy-joy",
  "description": "... Aloqa: +998771039935 ...",
  "url": "https://t.me/Maklersiz/5557",
  "phone_number": "+998771039935",
  "country": "Uzbekistan",
  "dial_code": "+998",
  "channel": "Maklersiz",
  "channel_url": "https://t.me/Maklersiz",
  "subscribers": "12.3K",
  "source_type": "channel_post",
  "posted_at": "2026-08-03T09:14:52+00:00"
}

source_type is channel_post (found in a real post), channel_bio (found in the channel description or its links), or search_snippet (fast engine only).

Run status is honest 🚦

A run that finds nothing because the keyword has no matches ends SUCCEEDED with an empty dataset. A run in which every search request or every t.me request was refused ends FAILED, with the reason in the status message — it did no work, and reporting that as a success hides a broken proxy or an exhausted account behind an empty dataset.

Every keyword is bounded by both a page ceiling and a wall clock, so a doomed query terminates promptly and says so instead of running until the platform timeout.

Legal ⚖️

Only publicly available information from public Telegram channels is accessed — the same pages any logged-out visitor can open at t.me. No login, no private groups, no member lists. Use the output in line with the applicable privacy and anti-spam laws in your jurisdiction.

Support

• Bug reports / custom features: dev.scraperengine@gmail.com