1990s

• Spirit of XML-RPC/CORBA
• SOAP/WSDL/UDDI
• Service Oriented Architecture (SOA)
• Strong concept of RDBMS as the backend
• OGC WMS (1999)

2000s

• Web 2.0
• JavaScript/AJAX/Google Maps
• Slippy maps, tiles
• OGC WFS (2002), WCS (2003), WPS (2005), CSW (2007)

Realities of OGC web services

• Key Value Parameters (KVP) or XML encodings
• Requests using HTTP GET, HTTP POST and SOAP
• Responses as XML (GML)
• Error handling as custom ExceptionReport within an HTTP 200 response
• Exhaustive specifications/requirements

<wfs:FeatureCollection 
    xmlns:ms="http://mapserver.gis.umn.edu/mapserver"
    xmlns:gml="http://www.opengis.net/gml" [..]>
    <gml:boundedBy>
        <gml:Envelope srsName="EPSG:32661">[..]
        </gml:Envelope>
    </gml:boundedBy>
    <gml:featureMember>
        <ms:north_poles_wfs gml:id="north_poles_wfs.0">
            <gml:boundedBy>[..]</gml:boundedBy>
            <ms:msGeometry>[..]</ms:msGeometry>
            <ms:Id>0</ms:Id>[..] 
    </gml:featureMember>[..]
</wfs:FeatureCollection>

More realities

• Complex machinery/architectures (e.g. ldproxy) for services to be crawlable
• Challenging for web developers to implement
• Challenging for mainstream web integration (search indexing)
• Geonovum testbed showed OGC world can easily be made crawlable, and crafted LDProxy

REST

• REpresentational State Transfer (REST)
• HTTP verbs (GET/PUT/POST/DELETE)
• HTTP codes (200, 201, 404, etc.)
• Uses URI to identify resources
• Content negotiation (media types)
• Stateless

JSON

• JavaScript Object Notation
• Easy to write and understand
• Very popular among web developers
• JSON is first class in RESTful web services
• No rigid standards (for better or worse)

OpenAPI

• Specification on how to describe a REST API
• Defines endpoints, request parameters, response
• Programming language agnostic
• Formerly Swagger

"info": {
    "contact": {
        "email": "you@example.org",
        "name": "pygeoapi Development Team",
        "url": "https://pygeoapi.io"
    },
    "description": "pygeoapi provides an API to geospatial data",
    "license": {
        "name": "CC-BY 4.0 license",
        "url": "https://creativecommons.org/licenses/by/4.0/"
    },
    "termsOfService": "None",
    "title": "pygeoapi Demo instance - running latest GitHub version",
    "version": "3.0.2",
    "x-keywords": [
        "geospatial",
        "data",
        "api"
    ]
},
"openapi": "3.0.2",

Swagger

• OpenAPI client tooling
• Robust, web developer focused
• Automated documentation, testing, code generation

W3C Spatial Data on the Web Best Practices (2017)

• Recommendations on formats, identifiers, access, licensing, provenance
• Being webby
• https://www.w3.org/TR/sdw-bp

Design patterns

• Being webby (humans, search engines)
• Developer friendly
• Lightweight specification development
• Removing HTTP use as a tunnel
  • /ows?request=GetFeature&typename=roads&featureid=5
  • /api/collections/roads/items/5
• Modular specification development
  • Core and extensions

Timeline

• 2017: W3C Spatial Data on the Web Best Practices
• 2017: OGC API Whitepaper
• 2018: WFS3 Hackathon, Weather on the Web API Hackathon
• 2019-2020: OGC API Hackathons
• OGC API Roadmap

Future OGC standards

• OGC API - Common (OWS Common)
• OGC API - Features (WFS)
• OGC API - Coverages (WCS)
• OGC API - Maps (WMS)
• OGC API - Tiles (WMTS)
• OGC API - Processes (WPS)
• OGC API - Records (CSW)
• OGC API - Styles (SE/SLD)

OGC activity on GitHub

• webby
• Specification development
• Issue tracking/discussion/collaboration
• Standards written in AsciiDoc (!)
• Standards released as HTML (!) or traditional PDF/etc.
• Gitter chat

Project Overview

• Geospatial data API framework
• OGC Compliant (already!)
• OSGeo Community Project
• International team (Canada, Netherlands, Greece, Italy, New Zealand)
  • Numerous core contributors
  • 21000 upstream contributors (dependencies)

Technical Overview

• Publishing (leaves data pristine/alone)
  • Fits to the data (not the inverse)
• Leverages powerful ecosystem of Python packages

Technical Overview

• Core abstract API; web framework agnostic (default Flask)
• YAML configuration (metadata, dataset connections
• Automated OpenAPI document generation (data binding)
• Robust plugin framework (data connections, formats, processing)
• Easy deployment (pip install, Docker, conda, UbuntuGIS)
• Minimal dependencies

Feature Data Providers

Elasticsearch	CSV	Spatialite*	GeoJSON*

* also via GDAL-OGR Provider

Feature Data Providers - GDAL/OGR

• Via OGR-Python bindings
• Unleashes ~100 Vector Formats! (WFS 1/2, GeoPackage, Shapefile, GeoJSON, PostGIS, ...)
• Paging, reprojection, spatial/attribute query
• Unlock existing WFS 1/2s!

Coverage Data Providers

xarray	rasterio

• NetCDF, Cloud Optimized GeoTIFF (COG)

Tile Data Providers

minio	ZXY directory tree

• MapBox Vector Tiles via tippecanoe

schema.org

• Lower the barrier to data search
• HTML encoding is annotated as schema.org/Dataset to facilitate search engines
• Google dataset search locates pygeoapi datasets

Linked Data - JSON-LD

• JSON-LD provided as additional representation
• Default out of the box functionality for most pygeoapi resources / endpoints
• Additional configuration

OpenAPI

OGC API - Features

OGC API - Coverages

OGC API - Tiles

OGC API - Processes

SpatioTemporal Asset Catalog (STAC)

Packaging

Numerous deployment options

• PyPI
• UbuntuGIS
• Docker
• Conda
• FreeBSD
• Coming to OSGeoLive

Docker overview

• pygeoapi Docker Images on DockerHub
• Images create Linux containers running content
• Use Docker Images to test locally or deploy in server

# Painless deployment
# pull from DockerHub
docker pull geopython/pygeoapi:latest
# run (default config)
docker run -p 5000:80 -it geopython/pygeoapi:latest
# test
curl http://localhost:5000
curl http://localhost:5000/collections
curl http://localhost:5000/collections/lakes/items?limit=2
curl http://localhost:5000/collections/lakes/items?f=html

Docker options

• Custom config via Docker Volume Mapping
• Run with Docker Compose
• Subpath running via Flask SCRIPT_NAME environment variable
• Examples

Scaling pygeoapi

Canada

Meteorological Service of Canada GeoMet platform

• OGC web services platform of Canadian weather, climate and water data
• Realtime weather
• Numerical weather prediction
• Hydrometric and Climate archives

MSC pygeoapi implementation

• Climate services data portal
• climatedata.ca
• https://geo.weather.gc.ca/geomet/features

GLOSIS - Global Soil Information

• Global soil information system under development by FAO and ISRIC
• Using pygeoapi as an external API providing OGC API - Features
• Testbed for JSON-LD for soil data distribution
• https://glosis.isric.org/ogc

Roadmap

• More OGC APIs: Maps, Records, Styles
• OGC API Improvements: Processes management
• More data providers
• Advanced filters (CQL)
• Transactions
• Content negotiation (e.g. Response as GeoPackage, GML,...)
• OSGeoLive

OGC API - Records

• Initial implementation currently in development with specification
• Development team part of OGC API - Records SWG
• Plan to use pycsw backend (PostgreSQL-based)
• Users will be able to update a single backend and have both CSW and OGC API - Records support
• Dependency strategy to be determined (e.g. pygeoapi may become pycsw dependency to support OGC API - Records)

Django app

GeoNode integration

• Next generation GeoNode API GNIP-51
• New v2 GeoNode REST API has been merged into master: GNIP-79
• Discussion needed to determine how GeoNode could use pygeoapi as a backend:
  • Metadata (Resource base model -- similar to pycsw GeoNode integration)
  • Direct Elasticsearch integration
  • For OGC API Features, Maps, Styles, Tiles, Coverages (similar to GeoServer and QGIS Server)
  • How would auth work in this case?

Need support for WFS3.0???

GeoCat can help you

justobjects.nl

GeoComvos ltd

Geobeyond Srl