🌐 Wikipedia Scraper: Articles, Summaries & Pageviews

Wikipedia Scraper by RecordsData extracts Wikipedia articles from any of 300+ language editions: search results, summaries, categories, Wikidata ID, infobox, full text, sections, links, images, and monthly or daily pageviews with date ranges. Enter keywords or article titles, get structured rows, export CSV, Excel, JSON or XML. No login, no API key. Verified 2026-10-10 against live Wikipedia. Priced per result.

Wikipedia Scraper turns keywords or a list of article titles into a dataset built from Wikimedia's official APIs. In a test run on the English edition, "Alan Turing" returned the Wikidata ID Q7251, 61 categories and 63,234 characters of full text (10,102 words) in one article row, plus 10 daily pageview rows for 1 to 10 September 2026. It is built for content and SEO teams, NLP and RAG builders, PR analysts and researchers.

📋 What does Wikipedia Scraper do?

• Search Wikipedia by keyword, or run up to 100 extra queries in one run (searchQueries), up to 10,000 hits per query.
• Fetch summaries for exact article titles or pasted Wikipedia URLs, mixed editions allowed.
• Add categories, Wikidata ID, infobox, full plain text, sections, internal and external links, images and language links, each behind its own switch.
• Download monthly pageviews (up to 120 months) or daily pageviews for a date range, with access and agent filters.
• Switch language with one field (en, es, de, fr, pt, ja, simple and every other edition).
• Export Wikipedia data to CSV, Excel, JSON or XML, or pull it by API.

📊 What data can you extract from Wikipedia?

Field	Description
recordType	article, search-result or pageviews
language, title, url	Edition code, canonical title, article URL
searchQuery, searchRank	Query that found the row and its position
wikidataId	Wikidata Q-id (on every article row)
description, extract	Wikidata short description, lead summary up to 1,500 characters
pageId, revisionId, articleType, qualityClass	Page id, latest revision, standard or disambiguation, WikiProject grade (English only)
thumbnailUrl, imageUrl	Thumbnail and original lead image
latitude, longitude	Coordinates (places only)
wordCount, sizeBytes, snippet, matchedSection	Search-hit rows only
categories, categoryCount	Visible categories and the true total
fullText, textLength, textWordCount, fullTextTruncated	Plain text and its size (no word count for Japanese, Chinese, Thai)
infobox	Label-to-value object from the lead-section HTML
sections, sectionCount	Number, title and level of each heading
links, linkCount, externalLinks, externalLinkCount	Linked article titles and external URLs
images, imageCount, languageLinks, languageCount	Body images (file, URL, caption) and other editions
month or date, views, access, agent	Pageview rows: month or day, count, filters used
lastModified, scrapedAt	Last edit and scrape timestamps

The dataset schema declares 48 fields (45 data columns plus error, errorType, hint on error rows). Fields that do not exist for an article are left out of the row instead of being filled with placeholders.

🧾 Sample output

Trimmed article row from the Alan Turing run on the English edition (categories and full text shortened, only fields measured in that run are shown):

{
  "recordType": "article",
  "language": "en",
  "title": "Alan Turing",
  "wikidataId": "Q7251",
  "categoryCount": 61,
  "categories": ["..."],
  "fullText": "...",
  "textLength": 63234,
  "textWordCount": 10102,
  "fullTextTruncated": false,
  "url": "https://en.wikipedia.org/wiki/Alan_Turing"
}

A monthly pageview row looks like this (daily rows carry date instead of month, plus access and agent):

{
  "recordType": "pageviews",
  "language": "es",
  "title": "Alan Turing",
  "month": "2026-09",
  "views": 33124,
  "url": "https://es.wikipedia.org/wiki/Alan_Turing",
  "scrapedAt": "2026-10-04T05:05:32.511Z"
}

💰 How much does it cost to scrape Wikipedia?

Pay per event, charged only for rows that are saved. Live prices read 2026-10-10 at the Free tier (Apify subscription tiers get lower rates, down to $0.01015 per article):

Event	What you pay for	Price
article-record	One article row, with every module you switch on	$0.012
search-record	One light search row (summaries off)	$0.010
pageview-record	One month of pageviews for an article	$0.010
pageview-day-record	One day of pageviews for an article	$0.0003

The modules (categories, full text, infobox, sections, links, images, language links) add columns to the article row and cost no extra fee. Daily pageview rows have their own low-cost event, so a year of daily views for one article costs about $0.11; monthly rows are billed as above. Higher Apify tiers pay less on article rows.

Worked examples at Free tier prices: 1,000 article rows with categories and full text cost $12.00 (1,000 x $0.012). 1,000 light search rows cost $10.00. One article with 12 monthly pageview rows costs $0.012 + 12 x $0.01 = $0.132. One article with 30 daily pageview rows costs $0.012 + 30 x $0.0003 = $0.021, and 1,000 daily pageview rows cost $0.30. Error rows, "article not found" rows and empty searches are never charged. If your max charge per run is reached, the run stops cleanly with the status "Stopped at your max charge limit". Free users get a 10-row preview.

🚀 How to scrape Wikipedia in 3 steps

1. Open the actor and set the Language (default en) and a Search query, or paste Article titles.
2. Switch on the modules you need (categories, full text, infobox, pageviews) and set the date range for pageviews.
3. Click Start, then download the dataset as CSV, Excel, JSON or XML.

⚙️ Input

Field	Default	Meaning
language	en	Wikipedia edition code (simple, zh-min-nan accepted)
searchQuery, searchQueries	artificial intelligence, []	One query plus up to 100 extra queries
maxResultsPerSearch	10000	Hits per query, 1 to 10,000
fetchSummariesForResults	true	Enrich each hit into a full summary row
articleTitles	[]	Exact titles or Wikipedia URLs
includeCategories, includeInfobox	false	Categories and infobox object
includeFullText, maxFullTextCharacters	false, 100000	Plain text and its cut (1,000 to 500,000)
includeSections, includeLinks, includeExternalLinks, includeImages, includeLanguageLinks	false	List modules
maxListEntries	200	Cap per list, 1 to 2,000 (counts keep the true total)
includePageviews	false	Add pageview rows for the titles list
pageviewGranularity	monthly	monthly or daily
pageviewMonths	12	Months of history, 1 to 120 (monthly)
pageviewsStart, pageviewsEnd	empty	YYYY-MM-DD, today, yesterday or relative 30d, 8w, 6m, 2y (data starts 2015-07-01)
pageviewAccess, pageviewAgent	all-access, all-agents	Filter by access type and agent
maxItems	10	Total rows cap (free plan limited to 10)

{
  "language": "en",
  "searchQuery": "",
  "articleTitles": ["Alan Turing"],
  "includeCategories": true,
  "includeFullText": true,
  "includePageviews": true,
  "pageviewGranularity": "daily",
  "pageviewsStart": "2026-09-01",
  "pageviewsEnd": "2026-09-10",
  "maxItems": 11
}

Pageviews are returned per article from the titles list. In monthly mode the current, still-running month is included, so pageviewMonths: 3 can return 4 rows. Hourly granularity is not offered (the Wikimedia per-article API refuses it). Full text is fetched one article per request, so full-text runs are slower than summary runs.

📦 Output

Each row is one article, search hit or pageview period. The dataset has five views (overview, articles, content, search, pageviews) plus the full JSON, and row key order is fixed. If a search has no matches, the run finishes as succeeded with the message "No results for this search. Nothing was charged." If Wikipedia is unreachable and nothing was saved, the run fails with a clear message instead of finishing silently.

Not delivered: a references count (the reference list is not in the parsed page HTML; externalLinks is the closest substitute) and hourly pageviews. wordCount exists only on search-hit rows; textWordCount is the plain-text count for title rows.

⚖️ Wikipedia scraper vs alternatives

Compared on the Apify Store, public pages read on 2026-10-10:

	This actor	automation-lab	hipersoft	ponderable_hydrometer
Price per 1,000	$12 Free tier, $10.15 top tier	from $0.60 plus $0.001 per run	from $0.50	from $1.50
Full text, categories, Wikidata ID	All three	All three	All three	Not listed
Infobox, sections, links, images	Yes, by switch	Not listed	Not listed	Not listed
Pageviews	Monthly to 120 months, daily with date range	Not listed	Last 6 months	Daily or monthly, date range

On fields and depth this actor goes further: infobox, sections, links, images, language links and up to 10,000 hits per query (the rivals list 500). openclawmara/wikipedia-scraper lists sections, links and images too, with a $5.00 header and a $0.003 README price that disagree. automation-lab and hipersoft show 100% successful runs and cost roughly 10 to 20 times less per article. Our daily pageview rows cost $0.0003 each. If you only need bulk plain text at the lowest price, choose a cheaper actor.

💼 Use cases

• SEO and content research: find the articles around a topic and see their real monthly traffic.
• NLP and RAG corpora: build multilingual datasets of clean extracts by topic.
• Brand and PR monitoring: track pageview curves for people, companies and products.
• Education and reference tools: summaries with images and coordinates in any language.

🔌 Run via API, schedule and integrations

curl -X POST "https://api.apify.com/v2/acts/recordsdata~wikipedia-articles-scraper/runs?token=<YOUR_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"language":"en","searchQuery":"quantum computing","maxItems":10}'

Use the Apify client for JavaScript or Python, schedule monthly pageview refreshes in the Console, or connect Zapier, Make, n8n and Google Sheets. AI agents can also call it through MCP, see the next section.

🤖 Wikipedia Scraper MCP server for AI agents (Claude, Cursor, VS Code)

Connect this actor to any MCP client through the Apify MCP server. Use this URL, which exposes only this actor as a tool:

https://mcp.apify.com?tools=RecordsData/wikipedia-articles-scraper

Config for clients that take a JSON MCP config:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com?tools=RecordsData/wikipedia-articles-scraper"
    }
  }
}

Example agent prompt: "Use the Wikipedia Scraper on the English edition to fetch the article 'Alan Turing' with categories, Wikidata ID and full text, then get its daily pageviews from 2026-09-01 to 2026-09-10."

An agent should choose this tool when it needs Wikipedia summaries, categories, Wikidata IDs, infoboxes, full text, search results or daily and monthly pageview counts in any language edition, as structured rows instead of raw HTML. Set maxItems to keep agent runs small, and keep the list modules off unless needed.

🛡️ Is it legal to scrape Wikipedia?

The actor reads Wikimedia's official public APIs with an identified user agent and a polite request delay, and it only returns public article data, no personal accounts. Wikipedia text is licensed CC BY-SA, so credit Wikipedia when you republish it. Check your own use case for compliance.

❓ Frequently asked questions

Does Wikipedia Scraper need an API key or login?

No. The Wikipedia Scraper uses public Wikimedia endpoints, so you need no Wikipedia account, no API key and no login. You only need an Apify account to start the actor and download the dataset. It identifies itself with a user agent and keeps a polite request delay, as Wikimedia asks of API clients.

Which Wikipedia languages are supported by Wikipedia Scraper?

The Wikipedia Scraper supports every Wikipedia language edition, more than 300 in total. Set the language field to the subdomain code of the edition, for example en, es, de, fr, pt or ja. Search, summaries and pageviews all run against the edition you choose, and each row repeats the code in its language field.

Where do the Wikipedia pageview numbers come from?

They come from Wikimedia's official pageview API. The Wikipedia Scraper returns one pageviews row per month or per day and article. Monthly mode covers 1 to 120 months and includes the month still in progress; daily mode takes pageviewsStart and pageviewsEnd dates back to 2015-07-01. You can filter by pageviewAccess and pageviewAgent. Daily rows cost $0.0003 each, monthly rows $0.01.

How long are the Wikipedia extracts?

Each extract is the lead-section summary of the article as clean plain text, cut at 1,500 characters. The Wikipedia Scraper also returns a one-line Wikidata description, the wikidataId, the pageId, the last edit time in lastModified and the canonical article URL. For the whole article, switch on includeFullText, which adds fullText cut at maxFullTextCharacters (default 100,000, max 500,000).

Why did the Wikipedia Scraper return 0 results?

The search had no matching articles in that language edition, or the titles you listed do not exist there. Try another spelling or another language code. The Wikipedia Scraper run still finishes as succeeded with the message "No results for this search. Nothing was charged.", so an empty result never costs you anything.

Why is there an error row for my Wikipedia title?

When an exact article title is not found, the Wikipedia Scraper saves one row with an error message so you can see which title failed, and it continues with the remaining titles. Error rows are free, and you pay only for article, search and pageview rows that carry real data.

Are latitude, longitude and thumbnail always present in Wikipedia Scraper rows?

No. Coordinates exist only for articles about places, and a thumbnail exists only for articles with a lead image. The Wikipedia Scraper omits these fields from the row instead of filling them with placeholders, so check for the field before you read it. For example, the sample "Quantum computing" row has a thumbnail but no coordinates.

Can I limit my Wikipedia Scraper spend?

Yes. Set maxItems to cap the total number of rows, and set the maximum charge per run in Apify. The Wikipedia Scraper stops cleanly when either limit is reached and keeps the rows already saved. At Free tier prices, 10 article summaries cost $0.12, so a small maxItems makes a cheap test.

What happens if I send an empty or invalid input to Wikipedia Scraper?

The platform rejects a malformed language code with a 400 error before any run is created. Other input mistakes in the Wikipedia Scraper, such as no search query and no article titles, end the run as SUCCEEDED with one free error row (errorType: invalid-input, plus an error message and a hint), and nothing is charged. Valid runs never contain those fields.

🔗 Want more research data? Other RecordsData scrapers

• arXiv Research Papers Scraper
• Hacker News Search Scraper
• PubMed Articles Scraper

💬 Support

Found a bug or need a missing field? Open the Issues tab on this actor's page or write to contact.punkrecordsdata@gmail.com.

Last updated: 2026-10-10