# Vertical manifest v1 — `kind` selects the runtime in src/extract/vertical/.
version: 1
order: 24
name: wikipedia
# api-json-aggregate: parallel MediaWiki REST + action API requests.
kind: api-json-aggregate
description: Wikipedia article title, summary, extract, sections, images, and references via the MediaWiki REST and action APIs.
urlPatterns:
  - https://en.wikipedia.org/wiki/:title
  - https://:lang.wikipedia.org/wiki/:title
# Runtime requirements for the scrape host.
requirements:
  requiresBrowser: false
  requiresLLM: false
  requiresCloud: false
# Declared output facets (discovery / tooling).
capabilities:
  - article_metadata
  - summary
  - extract
  - sections
  - images
  - references
source: builtin
# Default language when the host is the bare en.wikipedia.org pattern.
matchOptions:
  defaults:
    lang: en
# Parallel JSON HTTP requests; each key is scope for extract (@.summary, @.sections, …).
requests:
  summary:
    urlTemplate: https://{{lang}}.wikipedia.org/api/rest_v1/page/summary/{{title|encodeURIComponent}}
  sections:
    optional: true
    urlTemplate: https://{{lang}}.wikipedia.org/w/api.php?action=parse&page={{title|encodeURIComponent}}&prop=sections&format=json
  images:
    optional: true
    urlTemplate: https://{{lang}}.wikipedia.org/w/rest.php/v1/page/{{title|encodeURIComponent}}/links/media
  references:
    optional: true
    urlTemplate: https://{{lang}}.wikipedia.org/w/api.php?action=parse&page={{title|encodeURIComponent}}&prop=externallinks&format=json
# Output projection: aggregate uses @.scope paths and |transforms.
extract:
  lang: "@.summary.lang || {{lang}}"
  title: "@.summary.title"
  description: "@.summary.description"
  extract: "@.summary.extract"
  pageUrl: "@.summary.content_urls.desktop.page"
  thumbnail: "@.summary.thumbnail.source"
  revision: "@.summary.revision"
  timestamp: "@.summary.timestamp"
  wikibaseItem: "@.summary.wikibase_item"
  sections: "@.sections.parse.sections|map:level=level,line=line,number=number,anchor=anchor|compact"
  images: "@.images.files|map:title=title,url=preferred.url,width=preferred.width,height=preferred.height,originalUrl=original.url|compact"
  references: "@.references.parse.externallinks"