Telegram Channel Scraper

Scrapes posts from public Telegram channels using their t.me/s/<channel> web preview. No Telegram account, phone number or API keys are required.

For every post the Actor extracts the text, publish time, views, reactions, media (photos, videos, audio, documents, stickers), link previews, forward and reply information, hashtags and mentions, together with channel metadata (title, avatar, subscriber count, verified badge).

What can this Actor do?

• Monitor one or many public channels for new posts in a time window (last N minutes/hours/days or an explicit since/until range).
• Deduplicate across runs with the stable uniqueKey (<channel>:<messageId>).
• Run many channels in parallel with per-channel safety caps on posts and pages.
• Rotate Apify Proxy sessions automatically, retrying blocked or geo-restricted previews on a fresh IP.

Input

Field	Type	Description
channels	string[]	Required. Channel URLs (https://t.me/durov, https://t.me/s/durov), @usernames or bare usernames.
lookbackValue	integer	How far back to fetch (default 24).
lookbackUnit	minutes | hours | days	Unit for lookbackValue (default hours).
sinceTimestamp	string	Optional ISO 8601 date or Unix timestamp. Overrides the lookback.
untilTimestamp	string	Optional ISO 8601 date or Unix timestamp. Defaults to now.
maxPostsPerChannel	integer	Cap on posts per channel, 0 = unlimited (default).
maxPagesPerChannel	integer	Safety cap on preview pages (~20 posts each) per channel (default 50).
concurrency	integer	Channels scraped in parallel (default 8).
proxyConfiguration	object	Standard Apify proxy settings (default: Apify Proxy on). Residential proxies help with geo-restricted channels.

Example:

{
    "channels": ["https://t.me/s/telegram", "@durov"],
    "lookbackValue": 7,
    "lookbackUnit": "days"
}

Output

One dataset item per post:

{
    "uniqueKey": "telegram:350",
    "id": 350,
    "url": "https://t.me/telegram/350",
    "channel": {
        "username": "telegram",
        "title": "Telegram News",
        "url": "https://t.me/telegram",
        "avatarUrl": "https://cdn4.telesco.pe/file/....jpg",
        "isVerified": true,
        "subscribers": 10123456
    },
    "text": "Telegram 11.0 is out ...",
    "publishedAt": "2026-10-20T14:03:11.000Z",
    "publishedAtTimestamp": 1792504991000,
    "views": 2345678,
    "isEdited": false,
    "isAlbum": false,
    "postAuthor": null,
    "isForwarded": false,
    "forwardedFrom": null,
    "isReply": false,
    "replyToMessageId": null,
    "replyToUrl": null,
    "reactions": [
        {
            "type": "emoji",
            "emoji": "❤",
            "customEmojiId": null,
            "imageUrl": "https://telegram.org/img/emoji/40/E29DA4.png",
            "count": 12000
        },
        {
            "type": "custom_emoji",
            "emoji": null,
            "customEmojiId": "5780596797219016124",
            "imageUrl": null,
            "count": 66
        },
        { "type": "paid", "emoji": "⭐", "customEmojiId": null, "imageUrl": null, "count": 5 }
    ],
    "totalReactions": 12071,
    "hasMedia": true,
    "mediaTypes": ["video"],
    "media": [
        {
            "type": "video",
            "url": "https://cdn4.telesco.pe/file/....mp4",
            "thumbnailUrl": "https://cdn4.telesco.pe/file/....jpg",
            "caption": null,
            "duration": 42,
            "isDownloadable": true
        }
    ],
    "linkPreview": null,
    "hashtags": [],
    "mentions": [],
    "scrapedAt": "2026-10-21T09:00:00.000Z"
}

A per-run summary (time window, per-channel page/post counts, skipped inputs and errors) is stored in the default key-value store under the SUMMARY key. If the run is aborted, the summary written so far is flagged with "aborted": true.

Reactions

Each entry in reactions has a type:

• emoji – a standard emoji reaction; emoji holds the character and imageUrl the icon shown on t.me.
• custom_emoji – a Telegram Premium custom emoji; only customEmojiId is available (the web preview does not render it).
• paid – Telegram Stars; emoji is ⭐.

Media

media[].url is the direct CDN link to the file (photo, video, audio, document, sticker) as served on the web preview; thumbnailUrl is the poster image for videos and link previews. linkPreview is a convenience copy of the first link-type media entry.

Telegram does not serve large videos on the web preview ("Media is too big"). Such entries have isDownloadable: false, caption set to Telegram's reason, url pointing at the post on t.me and thumbnailUrl still populated.

Limitations

• Only channels with a public web preview can be scraped. Private channels, groups, users and invite links (t.me/+…, t.me/joinchat/…, t.me/c/…) are skipped.
• Some channels are geo-restricted on the web preview; using residential proxies in proxyConfiguration usually resolves this.
• Comments / discussion threads are not part of the web preview and are not scraped.
• Telegram's preview serves ~20 posts per page; very large lookback windows on busy channels need a higher maxPagesPerChannel.

Local development

npm install
apify run        # reads storage/key_value_stores/default/INPUT.json, writes storage/datasets/default/
npm test         # unit tests for input parsing and the HTML parser

Project structure

.actor/
├── actor.json                   # Actor config: name, version, schemas
├── input_schema.json            # Input validation & Console form definition
├── dataset_schema.json          # Shape of each post item + Console table view
├── key_value_store_schema.json  # The SUMMARY record
└── output_schema.json           # Links to the dataset and SUMMARY
src/
├── main.ts      # Entry point: input → scrape channels concurrently → dataset + SUMMARY
├── input.ts     # Input validation, channel/timestamp parsing
├── scraper.ts   # Walks a channel's t.me/s preview backwards page by page
├── http.ts      # undici GET with proxy sessions, retries and redirect detection
├── proxy.ts     # Sticky proxy session pool + undici dispatcher cache
├── parser.ts    # Cheerio parser for the web preview HTML
├── output.ts    # Maps parsed posts to dataset items
├── types.ts     # Shared types
└── utils.ts     # Helpers (counters, concurrency)