WMS Capabilities Extractor

Point at any OGC Web Map Service → get a clean, structured layer catalog.

A generic runner over the OGC Web Map Service (WMS) GetCapabilities protocol (versions 1.1.1 and 1.3.0). WMS is the OGC standard for map and raster services, so one actor spans an enormous population of endpoints with no per-site scraper:

• national mapping agencies and every national Spatial Data Infrastructure (SDI)
• NOAA / USGS / NASA and other government geospatial portals
• the EU INSPIRE directive's view services
• terrestris / OSM-WMS and other basemap providers
• every GeoServer, MapServer, QGIS Server, ArcGIS and MapProxy deployment

Give it a WMS endpoint; it fetches the GetCapabilities XML, parses it, and flattens it to one row per advertised layer with the fully-resolved metadata.

What makes it different

The Apify Store has plenty of per-site map scrapers. This is a protocol runner: one input (a WMS URL) works against any compliant service. It is the geospatial-map sibling of the OGC API Features / STAC / OGC EDR / OGC API Records generic extractors.

The parse is the moat. WMS layers nest, and per the spec a child layer inherits a defined set of its parent's properties. This runner walks the whole <Layer> tree and correctly accumulates inherited state down each branch (CRS/SRS and styles are added; geographic + per-CRS bounding boxes, attribution, scale bounds and dimensions are replaced-or-inherited), then emits each named layer with its effective, merged properties — not just the shallow top-level view.

Both WMS versions are handled from the same code by matching element local names, so the namespaced 1.3.0 tree (CRS, EX_GeographicBoundingBox) and the un-namespaced 1.1.1 tree (SRS, LatLonBoundingBox, ScaleHint, <Extent>) both parse cleanly. The reported _wmsVersion is read from the served document itself.

Input

Field	Type	Notes
endpoint	string	A WMS base URL or full GetCapabilities URL. Required unless endpoints is given.
endpoints	array	Optional. Harvest several WMS services in one run; each row is tagged with its _endpoint.
version	string	1.3.0 (default) or 1.1.1. A VERSION already in the URL wins.
maxRecords	integer	Optional global cap across all endpoints.
userAgent	string	Optional User-Agent override.
timeoutSecs	integer	Optional read timeout (default 60s).
bearer / extraHeaders	string / object	Optional, for auth-gated deployments. Never required, never logged.

The SERVICE=WMS&REQUEST=GetCapabilities&VERSION=... parameters are appended for you when absent, so pasting a bare service root works.

Example

{
  "endpoint": "https://ows.terrestris.de/osm/service",
  "version": "1.3.0"
}

Output

One row per advertised (named) layer:

• Identity — name, title, abstract, queryable, opaque, cascaded
• Spatial — crs[] (inherited + own, merged), geographicBoundingBox {west,east,south,north}, boundingBoxes[] (per-CRS, coordinates recorded verbatim and tagged with their CRS so the 1.3.0 EPSG:4326 axis order is unambiguous)
• Cartography — styles[] ({name,title,legendUrl,...}), minScaleDenominator, maxScaleDenominator, scaleHintMin/Max (1.1.1)
• Descriptive — keywords[], attribution, dimensions[] (e.g. TIME / ELEVATION, with default and values), metadataUrls[], layerPath[] (ancestor titles)
• Service metadata (duplicated onto every row) — serviceTitle, serviceOrg, serviceContact, fees, accessConstraints, requestFormats (GetMap / GetFeatureInfo MIME formats), and more
• Envelope — _endpoint, _wmsVersion, _rowIndex, and a lossless _raw of the layer's own definition

Every optional field is nullable and read by node/attribute presence, so a layer that omits an element yields null rather than an error.

Behaviour and resilience

• An endpoint that returns 4xx/5xx, times out, or serves a non-WMS body (an HTML error / anti-bot page, a ServiceExceptionReport, or malformed XML) is skipped with a warning; the batch continues.
• A run where every endpoint was skipped fails fast (non-zero exit) so nothing broken ships.
• A reachable WMS that advertises 0 named layers yields 0 rows and a clean exit — the actor never fabricates rows.
• XML is parsed with external-entity resolution disabled (XXE-safe).

Pricing

Pay-per-event: charged per extracted layer row.