Wikipedia Scraper: Articles, Search & Pageviews

Get Wikipedia articles as clean structured data: title, short description, lead summary, optional full plain text, categories, coordinates, main image, Wikidata ID, last edit and revision, other-language versions and daily pageviews. Run one or many searches in any of the 300+ Wikipedia languages, pass a list of titles or links, pull a whole category, or get the most-viewed articles of a day.

It reads only the official Wikimedia APIs (the MediaWiki Action API and the Wikimedia Pageviews API), politely and one request at a time. No browser, no proxy, no login or API key. A default run of 25 articles takes a few seconds. $0.90 per 1,000 articles, with pageviews, full text and language links included in that price.

What you can do with it

• AI and RAG datasets: clean plain-text lead summaries or full articles, with license and source URL on every row, ready for embeddings or fine-tuning corpora.
• Knowledge graphs and entity enrichment: map names to Wikidata IDs, short descriptions, coordinates and categories.
• Trend and attention research: daily pageviews per article, or the top articles of any day in any language.
• SEO and content research: see which topics are covered, how long articles are and how much traffic they get.
• Education and research: pull every article in a category (and its subcategories) for a reading list or study.
• Monitoring: track revisionId / lastEdited to detect edits to pages you care about.

Input

Field	What it does
mode	search, titles, category or top (most-viewed of a day). Default search.
queries	Search texts for search mode, one per line; each runs its own search. Empty = web scraping.
maxResultsPerQuery	Cap per search when you give several (empty = only maxItems applies).
titles	Titles or links for titles mode, in any mix of languages (https://de.wikipedia.org/wiki/Berlin, mobile links, ?curid= links). Redirects are followed; titles that don't exist are listed in the run summary and not charged.
category	Category name or link for category mode; categoryDepth: 1 adds direct subcategories.
date	Day for top mode (YYYY-MM-DD); empty = latest available.
language	Wikipedia edition (en, de, fr, ja, simple, ...). Default en.
maxItems	Hard cap on articles for the whole run (and cost). Default 25, max 5,000.
includeFullText	Add the whole article as plain text and an exact word count.
includePageviews, pageviewDays	Add total and daily human pageviews for the last N days (default 30).
includeLanguageLinks	Add the same article's title and URL in every other language.

Examples:

{ "queries": ["large language model", "retrieval-augmented generation", "vector database"], "maxResultsPerQuery": 20, "maxItems": 60, "includeFullText": true }

{ "mode": "category", "category": "Machine learning", "categoryDepth": 1, "maxItems": 200, "includePageviews": true }

Output

One item per article:

{
  "title": "Web scraping",
  "pageId": 2696619,
  "lang": "en",
  "url": "https://en.wikipedia.org/wiki/Web_scraping",
  "description": "Method of extracting data from websites",
  "summary": "Web scraping, web harvesting, or web data extraction is data scraping used for extracting data from websites. ...",
  "wordCount": 4123,
  "lengthBytes": 34836,
  "categories": ["Web scraping"],
  "latitude": null, "longitude": null,
  "imageUrl": null, "thumbnailUrl": null,
  "wikidataId": "Q665452",
  "isDisambiguation": false,
  "lastEdited": "2026-09-19T09:13:04Z",
  "revisionId": 1375681792,
  "searchRank": 1,
  "pageviewsTotal": 58677,
  "pageviewsFrom": "2026-09-12", "pageviewsTo": "2026-10-11",
  "pageviewsDaily": [{"date": "2026-09-12", "views": 2042}],
  "license": "CC BY-SA 4.0",
  "licenseUrl": "https://creativecommons.org/licenses/by-sa/4.0/",
  "attribution": "Wikipedia contributors",
  "scrapedAt": "2026-10-12T08:00:00+00:00"
}

fullText appears when includeFullText is on; languagesCount and languageLinks when includeLanguageLinks is on; searchQuery and searchRank in search mode (an article found by two queries is returned once); topDate, topRank, topViews in top mode (the main page and special pages are skipped); category in category mode. wordCount comes from search results or from the full text (for Chinese, Japanese, Thai and other languages written without spaces it counts characters; charCount is the full text's length); in other modes without full text it is null (use lengthBytes). Images are the article's free-licensed lead image only.

How to use

1. Pick a mode and fill in the query, titles or category (or just press Start to try the default search).
2. Set language and maxItems, and switch on full text or pageviews if you need them.
3. Run, then download the dataset as JSON, CSV or Excel, or connect it to Sheets, Make, Zapier or your own code through the Apify API.

Pricing

Pay per result: $0.90 per 1,000 articles (article-scraped), no start fee. Pageviews, full text and language links are part of the article row and cost nothing extra. maxItems caps the run before you are charged. A run that finds nothing (unknown category, no search hits, bad language code) fails with a clear message instead of returning an empty dataset.

Reusing Wikipedia content (license)

Wikipedia text is licensed under Creative Commons Attribution-ShareAlike 4.0 (CC BY-SA 4.0). If you republish it, you must:

• credit it, e.g. "From Wikipedia, the free encyclopedia, by Wikipedia contributors" with a link to the article (url, which also leads to the full author history);
• link the license (licenseUrl) and say if you changed the text;
• share alike: text you build from it and redistribute must carry the same license.

Every row includes url, license, licenseUrl and attribution so you can do this automatically. Images can carry their own, different licenses; check the file page before reusing one. Facts, pageview counts and Wikidata IDs are not covered by the text license (Wikidata is CC0). This is general information, not legal advice.

Limits and notes

• Search, category and title lookups cover the article namespace only (no talk, user or file pages).
• Category mode lists direct members (plus one level of subcategories with categoryDepth: 1), in Wikipedia's sort order.
• Pageviews are human (non-bot) views across desktop and mobile, available from July 2015 and usually up to yesterday (UTC).
• The actor follows Wikimedia's API etiquette (descriptive User-Agent, serial requests, maxlag), so very large runs with full text and pageviews take a few minutes (two extra requests per article).
• For bulk mirroring of all of Wikipedia, use the official database dumps or Wikimedia Enterprise instead of this actor.
• Independent tool, not affiliated with or endorsed by Wikipedia or the Wikimedia Foundation. "Wikipedia" is a trademark of the Wikimedia Foundation.