Wikipedia Scraper

Turn any Wikipedia article into clean, structured data — summaries, full article text, categories, 30-day pageviews, Wikidata descriptions, thumbnails, geo-coordinates, and more, across 300+ language editions. Search by keyword or fetch exact articles you already know by URL or title. Built for researchers, data teams, and content & SEO analysts who need ready-to-use Wikipedia datasets without copy-pasting articles by hand or stitching together the raw API.

Why This Scraper?

• 300+ language editions — the picker surfaces 40 of the largest editions (English, German, French, Spanish, Japanese, Chinese, Arabic, and more); paste any non-English Wikipedia URL and the actor scrapes that edition automatically.
• Two ways in: keyword search or exact fetch — run a keyword search across an edition, or pull specific articles by link (https://en.wikipedia.org/wiki/Alan_Turing, or just en.wikipedia.org/wiki/Alan_Turing) or bare title ("Alan Turing"). Mix both in a single run.
• 30-day pageview totals on every row — a real popularity signal, not a guess: the trailing ~30-day view count for each article, ideal for ranking topics by actual reader demand.
• Sort by relevance or popularity — order search hits by best textual match, or by real reader demand: popularity keeps only articles that match your topic and ranks them by their last 30 days of traffic.
• Full article body and its summary on the same row — toggle fullText and every row carries the entire article as clean plain text next to its opening summary, so summarization pairs come out of a single run with no second pass.
• Wikidata one-line descriptions — the short canonical descriptor (e.g. "English mathematician and computer scientist") pulled straight from Wikidata, perfect for tooltips and entity labels.
• Category taxonomy per article — the full list of categories each article belongs to ("British computer scientists", "1912 births") for topic mapping and classification.
• Geo-coordinates for places & landmarks — latitude/longitude on every article that has a location, so cities, monuments, and venues drop straight onto a map.
• Cross-edition reach signal — langCount tells you in how many language editions an article exists, a quick indicator of global notability.

Use Cases

Research & Academia

• Build structured corpora of articles on a topic for literature reviews and citation context
• Compare how the same subject is covered across language editions
• Track article size and last-edited dates to study how topics evolve
• Rank subjects by 30-day readership to find what audiences actually care about

SEO & Content

• Pull authoritative summaries and Wikidata descriptions for entity-rich content
• Identify high-traffic Wikipedia topics worth targeting in articles and FAQs
• Map category taxonomies to plan topic clusters and internal linking
• Surface the most-read canonical article for any keyword

Data Enrichment

• Enrich CRM and product records with one-line Wikidata descriptions
• Add geo-coordinates to place names for mapping and location intelligence
• Attach lead images and canonical URLs to people, companies, products, films, albums, and landmarks
• Resolve ambiguous names to the most popular matching article

Machine Learning & NLP

• Build full-text training datasets with clean plain-text article bodies
• Generate summary/full-text pairs for summarization model fine-tuning
• Create multilingual datasets by pulling the same topics across editions
• Label corpora with category tags, whole-article word counts and byte sizes

Market & Competitive Intelligence

• Monitor pageview trends for brands, products, and public figures
• Track which companies and technologies are gaining reader attention
• Benchmark notability across markets using cross-edition coverage counts

Getting Started

Simple Keyword Search

Search one edition and return the best-matching articles:

{
    "searchQueries": ["artificial intelligence"],
    "maxResultsPerSearch": 50
}

Fetch Exact Articles

Pull specific articles by full URL or bare title — including non-English editions:

{
    "articleUrls": [
        "https://en.wikipedia.org/wiki/Alan_Turing",
        "Marie Curie",
        "https://de.wikipedia.org/wiki/Albert_Einstein"
    ],
    "includeCategories": true
}

Advanced — Popularity-Sorted Full-Text Dataset

Search several topics, sort by popularity, and capture the full article body:

{
    "searchQueries": ["machine learning", "neural network", "deep learning"],
    "language": "en",
    "maxResultsPerSearch": 200,
    "sortBy": "popularity",
    "fullText": true,
    "includeCategories": true
}

Input Reference

What to Scrape

Parameter	Type	Default	Description
searchQueries	string[]	[]	Keywords to look up on Wikipedia (e.g. "climate change"). Each query runs its own search and returns the best-matching articles. Leave empty if you only want exact articles.
articleUrls	string[]	[]	Exact articles to fetch. Paste Wikipedia links with or without https:// (e.g. https://en.wikipedia.org/wiki/Alan_Turing or en.wikipedia.org/wiki/Alan_Turing), or just a page title (e.g. "Alan Turing"). Links set their own language automatically.
language	select	English	Which Wikipedia edition to use for searches and bare titles. 40 common editions are listed; full URLs override this.

Results

Parameter	Type	Default	Description
maxResultsPerSearch	integer	50	Maximum articles to return per search query (1–500). Articles fetched directly by URL or title are added on top. Recommended 50–200 for fast, affordable runs.
sortBy	select	Relevance (best match)	Order search results. Options: "Relevance (best match)" or "Popularity (most read)".

Content

Parameter	Type	Default	Description
fullText	boolean	false	Return the complete plain-text article body instead of just the intro summary. Richer data, larger dataset.
includeCategories	boolean	true	Include the list of categories each article belongs to (e.g. "1912 births"). Helpful for classification and topic mapping.

Output

Each article is one flat row. Here is a representative result:

{
    "pageId": 1208,
    "title": "Alan Turing",
    "language": "en",
    "url": "https://en.wikipedia.org/wiki/Alan_Turing",
    "summary": "Alan Mathison Turing was an English mathematician, computer scientist, logician, and cryptanalyst...",
    "fullText": "Alan Mathison Turing was an English mathematician... (the complete body, when fullText is enabled — it opens with the same text as summary and keeps going)",
    "wikidataDescription": "English mathematician and computer scientist (1912–1954)",
    "categories": ["British computer scientists", "1912 births", "Alumni of King's College, Cambridge"],
    "thumbnail": "https://upload.wikimedia.org/wikipedia/commons/thumb/.../300px-Alan_Turing.jpg",
    "wordCount": 16550,
    "size": 174198,
    "pageviews": 415203,
    "langCount": 142,
    "coordinates": null,
    "lastEdited": "2026-06-10T08:14:32Z",
    "matchedQuery": "artificial intelligence",
    "scrapedAt": "2026-06-13T14:30:00Z"
}

Core Fields

Field	Type	Description
pageId	integer	Wikipedia page identifier
title	string	Article title, spelled the way the page displays it, so brand and science names keep their real casing (iPhone, jQuery, htmx, eBay stalking scandal, macOS) and titles of works arrive as clean text with no styling
language	string	Language code of the edition this article came from
url	string	Canonical article URL
matchedQuery	string|null	The search query that surfaced this article (null for direct URL/title fetches)
scrapedAt	string	ISO timestamp when the row was produced

Content

Field	Type	Description
summary	string|null	The article's opening summary as clean plain text. Present on every row in both modes, so a full-text run gives you the summary and the body side by side
fullText	string|null	Full plain-text article body — populated only when fullText is enabled. This is the readable prose, so it runs shorter than wordCount, which also counts tables, captions and reference text
wikidataDescription	string|null	One-line canonical description from Wikidata
categories	string[]	Categories the article belongs to — populated when includeCategories is on
thumbnail	string|null	300 px-wide, Wikipedia-hosted URL of the picture Wikipedia itself puts at the top of the article. Includes the posters, cover art and logos it shows under fair use for films, albums, books, products and companies, so check an image's own file page before republishing it. Null only when the article genuinely carries no picture

Popularity & Metadata

Field	Type	Description
pageviews	integer|null	Total reader views over the trailing ~30 days
langCount	integer	Number of other language editions this article exists in. 0 means it exists only in the edition you searched
wordCount	integer|null	Whole-article word count, as published by Wikipedia itself. Always measures the complete article, never just the intro, so it reads the same whether fullText is on or off
size	integer	Article size in bytes
lastEdited	string	ISO timestamp of the most recent revision

Geo

Field	Type	Description
coordinates	object|null	{ "lat": …, "lon": … } for articles with a location; null otherwise

Tips for Best Results

• Paste a non-English URL to reach any of the 300+ editions — the language picker lists 40 common editions, but pasting https://ja.wikipedia.org/wiki/... or https://fi.wikipedia.org/wiki/... into articleUrls scrapes that edition directly, no picker needed.
• Leave fullText off for fast, cheap summary runs — it pulls the entire article body and grows your dataset substantially. Turning it on never costs you the summary: that column stays filled either way, so full-text runs are a superset.
• Sort by popularity to find the canonical article — for ambiguous keywords, "Popularity (most read)" puts the article people actually read first, ahead of niche sub-topics and disambiguation pages.
• Use pageviews to rank topics by real demand — it reflects actual 30-day readership, a far stronger signal than search rank for prioritizing content or research.
• Mix search and exact fetch in one run — combine broad searchQueries with a hand-picked list of articleUrls to cover both discovery and known must-have articles.
• Rank by length without downloading the text — wordCount is the full-article figure on every row even with fullText off, so you can sort stubs from deep articles on a cheap summary run.
• Start with 50 results per query to test — confirm the data fits your needs, then raise maxResultsPerSearch (up to 500) for the full pull.

Pricing

From $0.60 per 1,000 results — flat pay-per-result, matching the lowest tier in this category while shipping more fields per article. Bronze, Silver, and Gold subscribers pay progressively less; the table below shows total cost at each discount tier.

Results	No discount	Bronze	Silver	Gold
100	$0.072	$0.068	$0.064	$0.060
1,000	$0.72	$0.68	$0.64	$0.60
10,000	$7.20	$6.80	$6.40	$6.00
100,000	$72.00	$68.00	$64.00	$60.00

A "result" is any article row in the output dataset. No compute or time-based charges — you pay per result, plus a small fixed per-run start fee.

Integrations

Export data in JSON, CSV, Excel, XML, or RSS. Connect to 1,500+ apps via:

• Zapier / Make / n8n — Workflow automation
• Google Sheets — Direct spreadsheet export
• Slack / Email — Notifications on new results
• Webhooks — Trigger custom APIs on run completion
• Apify API — Full programmatic access

Legal & Ethical Use

This actor collects publicly available content from Wikipedia for legitimate research, analysis, and data enrichment. Wikipedia article text is published under the Creative Commons Attribution-ShareAlike (CC BY-SA) license — when you reuse or republish it, provide proper attribution and share derivative text under the same license. Users are responsible for complying with applicable laws and the Wikimedia Foundation's Terms of Use. Be respectful of the volunteer-run platform and use the data responsibly.