Sources
Every record cites its sources. Validation rejects any source that is not on this allow-list (sources.json); adding one is a pull request reviewed by a maintainer.
We never use Google Maps data (except storing a place id for verification), OpenStreetMap data, or the lists of closed guides and review sites. Ratings from other services appear only as links.
Overture Maps Foundation, places theme
overture-places · open dataset
- Licence
- CDLA-Permissive-2.0
- Attribution
- Overture Maps Foundation
- Allowed for
- discoveryfacts
- Checked
- 2026-11-05 · vetted
Query the GeoParquet release for coffee categories inside the city boundary. Keep only records whose every sources[].dataset is CDLA-Permissive-2.0 or Apache-2.0 licensed; drop any record carrying OpenStreetMap data. Store the GERS id in xref.overture and keep the attribution.
Foursquare Open Source Places
foursquare-os-places · open dataset
- Licence
- Apache-2.0
- Attribution
- Foursquare Open Source Places
- Allowed for
- discoveryfacts
- Checked
- 2026-11-05 · vetted
Query the Parquet release for coffee shop and café categories inside the city boundary. Drop rows with date_closed set or a stale date_refreshed unless another source confirms the place. Store fsq_place_id in xref.foursquare.
Roaster 'where to drink our coffee' and wholesale partner pages
roaster-partner-page · roaster page
- Licence
- facts only; no text or images copied
- Allowed for
- discoveryfacts
- Checked
- 2026-11-05 · vetted
Read the list of cafés a roaster supplies. Record the café name and the roaster relationship (roasters_served) with the page URL. Confirm the café itself on its own website or an open dataset. Crawl politely (robots.txt, project User-Agent, at most 1 request per second per host).
The place's own website, including menu, shop, about and legal notice pages
place-website · primary website
- Licence
- facts only; no text or images copied
- Allowed for
- discoveryfacts
- Checked
- 2026-11-05 · vetted
Extract facts (hours, serving, amenities, roasters, operator) and cite the page URL. Never copy sentences or photos. Respect robots.txt, use the Open Coffee crawler User-Agent and stay at or below 1 request per second per host.
The place's own Instagram profile bio
place-instagram · social profile
- Licence
- facts only; no text or images copied
- Allowed for
- facts
- Checked
- 2026-11-05 · vetted
Read only the public bio (name, address, opening hours, linked website) when it is visible without logging in. Never copy captions, posts or photos.
Web search results
web-search · web search
- Licence
- not a data source; discovery only
- Allowed for
- discovery
- Checked
- 2026-11-05 · vetted
Find candidate places and their websites. Every fact must then be confirmed on a primary source (place-website, place-instagram or an open dataset) and cited there. At most about 3 searches per place.
Wikidata
wikidata · open knowledge
- Licence
- CC0-1.0
- Allowed for
- discoveryfacts
- Checked
- 2026-11-05 · vetted
Use statements such as official website, inception and coordinates. Store the Q-id in xref.wikidata.
Wikimedia Commons
wikimedia-commons · image repository
- Licence
- per file (CC0-1.0, CC-BY-4.0, CC-BY-SA-4.0 and similar)
- Allowed for
- images
- Checked
- 2026-11-05 · vetted
Use files with a free licence. Store licence, author and the file description page as source_url for each image.
Mapillary street-level imagery
mapillary · image repository
- Licence
- CC-BY-SA-4.0
- Attribution
- Mapillary contributors
- Allowed for
- images
- Checked
- 2026-11-05 · vetted
Use an image showing the storefront. Store the image page as source_url, the uploader as author and CC-BY-SA-4.0 as licence. Do not use Mapillary-derived map features.
Panoramax
panoramax · image repository
- Licence
- per file (usually CC-BY-SA-4.0)
- Allowed for
- images
- Checked
- 2026-11-05 · vetted
Use a storefront image. Store its licence, author and picture page as source_url.
Submission by the place's owner or operator
owner-submission · community
- Licence
- CC-BY-4.0 (contributed under DCO sign-off)
- Allowed for
- discoveryfactsimages
- Checked
- 2026-11-05 · vetted
Accept facts and photos submitted through an issue or pull request with DCO sign-off. Photos must be the submitter's own and licensed CC-BY-4.0. Set method owner_submitted.
Community submission
community-submission · community
- Licence
- CC-BY-4.0 (contributed under DCO sign-off)
- Allowed for
- discoveryfactsimages
- Checked
- 2026-11-05 · vetted
Accept corrections and new places from contributors with DCO sign-off. Facts need a primary-source URL or a reviewer visit; photos must be the contributor's own, licensed CC-BY-4.0.
Maintainer review
manual-review · community
- Licence
- CC-BY-4.0
- Allowed for
- facts
- Checked
- 2026-11-05 · vetted
A maintainer resolved a conflict or confirmed a fact by hand. Cite the page the decision relied on in field_provenance where one exists.
Google Places API (verification only)
google-places-verification · verification api
- Licence
- Google Maps Platform Terms; no data may be stored except place_id
- Allowed for
- verification
- Checked
- 2026-11-05 · vetted
Optional. Only for a place already known from other sources: look up businessStatus and store only xref.google_place_id. Never store names, addresses, hours, ratings or photos from Google, never use it for discovery, and never write content based on it. A CLOSED_PERMANENTLY result opens a human review instead of changing the record.
Propose a source (state its licence, terms and what it may be used for).