📚 Wikipedia Scraper

Scrape Wikipedia articles, search results, recent edits, and categories — across all 300+ language editions. Returns clean plain-text content, summaries, references, and rich metadata.

🎯 Built for AI/LLM training datasets, NLP research, knowledge-graph construction, journalism, and education.

---

✨ What you can do

• 📄 Fetch articles by title — clean plain-text body, summary, sections, references
• 🔎 Search — full-text search across an entire language edition
• 📡 Recent changes — live feed of edits (title, user, comment, revid)
• 📁 Pull entire categories — all members of Category:Machine_learning, etc.
• 🌐 Any language — en, es, fr, de, ja, zh, hi, ar, etc.
• 📦 Rich output: links (internal+external), categories, sections, last-modified

---

🚀 Quick start

{
    "mode": "articles",
    "language": "en",
    "titles": ["Artificial intelligence", "Large language model"],
    "includeContent": true,
    "includeReferences": false
}

---

📥 Input

Field	Used in mode	Description
mode	all	articles / search / recentchanges / category
language	all	Wiki edition code (en, de, ja...)
titles	articles	Article titles
searchQueries	search	Keywords or phrases
category	category	Category name without Category: prefix
maxItems	all	Cap per query
includeContent	articles, search, category	Full plain-text body
includeReferences	articles, search, category	External + internal links + sections

---

📤 Output (per item)

{
    "mode": "articles",
    "title": "Artificial intelligence",
    "language": "en",
    "pageId": 1164,
    "summary": "Artificial intelligence (AI) refers to...",
    "content": "Full article text...",
    "wordCount": 12873,
    "sections": ["Goals", "History", "Methods"],
    "externalLinks": ["https://..."],
    "internalLinks": ["Machine learning", "Neural network"],
    "categories": ["Artificial intelligence", "Cybernetics"],
    "url": "https://en.wikipedia.org/wiki/Artificial_intelligence",
    "lastModified": "2026-04-30T...",
    "scrapedAt": "2026-05-06T..."
}

---

🎯 Use cases

Who	Why
🤖 LLM teams	Pretraining + fine-tuning datasets across languages
📚 NLP researchers	Multilingual corpora, named-entity benchmarks
📰 Journalists	Topic deep-dives + fact-checking pipelines
🎓 Educators	Auto-build study material from any topic
🧠 Knowledge graphs	Wikipedia as an entity backbone

---

⚙️ Tech notes

• Uses MediaWiki's official Action API + REST Summary API
• No login, no key, no rate limits (within fair use)
• Plain-text extraction via explaintext=1 — already cleaned, no HTML/wikitext
• Recent-changes uses rctype=edit|new to skip log noise

---

❓ FAQ

Are full Wikipedia dumps better? For one-shot pre-training, yes (free at dumps.wikimedia.org). This Actor is for targeted scrapes — specific topics, ongoing freshness, multi-language slices, or recent-changes monitoring.

Schedule it? Yes. Recent changes mode is perfect for hourly Apify Schedules.

Hits rate limits? Almost never. MediaWiki's anonymous limit is generous and we add automatic retries with backoff.