Wikipedia Article Text Export

Export the complete text of selected English Wikipedia articles for research agents and selected-page knowledge ingestion.

Input

articleTitles accepts 1–10 English mainspace article titles or https://en.wikipedia.org/wiki/ article URLs. The default is Artificial intelligence. Known non-article namespaces, action/query inputs, credentials, fragments, other hosts/editions, blank entries, unknown keys and over-limit inputs are rejected before network access. Redirects stay within allowed English article paths; canonical article IDs deduplicate aliases. Missing articles or disambiguation pages fail the batch so callers choose the intended article.

Output

One row per unique resolved article with title/page/revision IDs, requested title aliases, canonical and stable revision attribution URLs, history attribution URL, full cleaned article text, and ordered sections with heading levels and IDs. Substantive tables/lists/text captions/references and visible imported-source attribution notices are retained; menus, edit controls, scripts, styles and navigation are removed. No text is silently truncated. Every row includes CC BY-SA 4.0 license URL, source Terms URL, and an explicit text-cleanup/media-omission modification notice. Article-specific additional license/attribution notices are preserved. A page whose article-text license cannot be confirmed fails rather than assuming the standard license. SUMMARY in key-value storage holds selected/delivered/alias counts.

Limits

Acquisition deadline180seconds and delivery deadline90seconds; per-request timeout30seconds. Total transfer/decoded budget64MiB; staged output24MiB. 10 input entries and at most 10 output articles. Only allowed ordinary /wiki/ article HTML and robots are fetched with an identifying User-Agent, sequentially at least one second between attempts. /api/, /w/, Special/User/File/etc namespaces and linked media are never fetched. Maximum 5MB per HTML response; oversized articles fail visibly. No search, multilingual editions, site-wide dumps, arbitrary URLs, browser, or external paid API. Source shape/revision/license changes, auth blocks, rate limits and timeouts fail before output.

Pricing

One article-delivered event per validated unique article saved at $0.001. No start fee/trial events. Complete selected-article coverage and licenses are checked before saving. Apify event budget caps selection before retrieval; if it cannot cover the input entries, fail before requests/charges. Redirect aliases deduplicated into one delivered article incur one event. Repeated runs charge delivered articles again. No charge for failures, aliases or SUMMARY.

Privacy

Only selected public encyclopedia text is collected. Contributor/editor history, account pages, private information, cookies, headers and credentials are excluded. Article text can itself discuss public biographical facts; the Actor does not extract contact profiles. Media files and their separately licensed content are not copied. Preserve attribution and share-alike requirements when reusing exported text.