📘 Facebook Groups Posts & Comments Scraper

Scrape posts (and optionally comments) from public Facebook groups. This Apify actor discovers Facebook's internal GraphQL feed automatically, paginates the group feed, and returns structured data: post text, author profile, a full reaction-type breakdown, photo/video/shared-link attachments, absolute timestamps, derived lead-gen fields, and optional per-post comments in a separate child dataset.

Keywords: facebook groups scraper, facebook group posts, facebook comments scraper, facebook data extraction, social media scraper, facebook reactions, lead generation, facebook group monitoring, facebook posts export, facebook engagement data.

Features

• 🔎 Automatic discovery of the group node_id, GraphQL doc_id, and pagination cursor.
• 👍 Full reaction breakdown — like, love, haha, wow, sad, angry, care + total.
• 👤 Author details — id, name, profile URL, profile picture, and __typename (User/Page).
• 🖼️ Attachments — photos, videos (playable URL / HD / thumbnail / duration), shared links (URL/title/description), and albums.
• 🕒 Absolute time — timestamp (epoch), date (ISO 8601), and scrapedAt.
• 🗂️ Post type label — text / photo / video / link / album / reshare.
• 🧮 Derived fields (no AI) — hashtags, mentions, links, emails, phones, engagementScore, postAgeDays.
• 🚫 Filters — include/exclude keywords, minLikes / minComments / minShares, and date range (onlyPostsNewerThan / onlyPostsOlderThan).
• 💬 Comments (opt-in) — interleaved right after each post (type: "comment", isChild: true, linked via parentId), mirrored to a per-run comments-<runId> dataset. resultsLimit counts POSTS only; comments are additional and capped by maxComments.
• 🤖 Optional AI enrichment — sentiment, language, topics, buy-intent and lead-signal on post text via your own LLM key (off by default, multi-provider).
• 🌐 Proxy — your proxyConfiguration is honored; residential is the default when none is set.

Input

Field	Type	Description
startUrls	array	Public Facebook group URLs (required).
resultsLimit	integer	Max number of POSTS (parents). Comments are extra. Default 20.
viewOption	enum	Feed order: CHRONOLOGICAL / RECENT_ACTIVITY / TOP_POSTS / CHRONOLOGICAL_LISTINGS.
keywords	array	Keep posts containing any of these words.
excludeKeywords	array	Drop posts containing any of these words.
minLikes / minComments / minShares	integer	Minimum engagement thresholds.
onlyPostsNewerThan / onlyPostsOlderThan	date	Absolute (2024-01-15) or relative ("2 weeks").
searchGroupKeyword / searchGroupYear	string	Client-side search / year filter.
includeComments	boolean	Collect comments per post (opt-in).
maxComments	integer	Cap comments per post (0 = no cap). Default 10.
aiEnhancement	boolean	Enable AI enrichment (off by default).
aiModel	enum	LLM model; provider auto-detected from the name.
aiApiKey	secret	LLM provider API key (or set the provider env var).
maxRequestRetries	integer	Retries per GraphQL request. Default 3.
proxyConfiguration	proxy	Proxy settings (residential by default).
cookieString	secret	Optional logged-in cookies (c_user + xs) for reliability / closed groups.

Example input

{
  "startUrls": ["https://www.facebook.com/groups/cheapmealideas/"],
  "resultsLimit": 20,
  "viewOption": "RECENT_ACTIVITY",
  "excludeKeywords": ["spam"],
  "minLikes": 5,
  "onlyPostsNewerThan": "30 days",
  "includeComments": true,
  "maxComments": 10
}

Output

Each post row (type: "post", isChild: false) contains: id, postId, url, date, timestamp, scrapedAt, user (id/name/profileUrl/profilePicture/__typename), text, postType, reaction breakdown (reactionLikeCount…reactionCareCount, topReactionsCount), likesCount, commentsCount, sharesCount, attachments, topComments, feedbackId, groupTitle, facebookId, and derived fields (hashtags, mentions, links, emails, phones, engagementScore, postAgeDays). When AI is enabled: aiSentiment, aiLanguage, aiTopics, aiBuyIntent, aiLeadSignal.

Each comment row (type: "comment", isChild: true) is interleaved after its post and contains parentId, parentUrl, id, url, date, text, user, likesCount, threadingDepth. Comment rows are also mirrored to a separate comments-<runId> dataset.

How it works

1. Fetches the group page HTML through the proxy and extracts the GraphQL parameters and CSRF tokens.
2. Paginates the GroupsCometFeedRegularStoriesPaginationQuery feed, deduplicating posts across pages and across retry attempts.
3. Normalizes each story node into the output schema and computes derived fields.
4. Applies filters, then charges and pushes accepted posts; when enabled, collects and interleaves comments.

Notes & limitations

• Only public groups are reliably accessible without authentication. For closed/private groups, provide cookieString.
• Facebook may serve a login wall to some IPs; the actor retries on fresh residential IPs and reports a clear error if it persists.
• Date filters are applied client-side after extraction.
• Comment collection returns the comments exposed by Facebook's feed payload for each post (typically the top/most-relevant comments), capped by maxComments. Exhaustive comment pagination for very high-comment posts may be partial.

Compliance

Scrape only public data and respect Facebook's Terms and applicable laws (including GDPR/CCPA). You are responsible for how you use the collected data.