osm-analytics server setup walkthrough

readme.md

Cruncher

• install node/npm, curl
• compile and install tippecanoe (version 1.8.1 works, not sure about newer releases)
• download run.sh from cruncher repo and chmod +x it
• replace osmqatiles-planet url with the extract you want
• adjust paths in run.sh (and mkdir the respective working directories)
• comment out hotprojects.sh (would fail because it requires valid AWS credentials to upload stuff into a hardcoded bucket).
• comment out the forever restart … lines (requires a tile-serving script to be already running)
• execute run.sh -> … -> two .mbtiles files in results directory

Serve Results

There are some options:

• use mb-util (e.g. mb-util --image_format=pbf buildings.mbtile buildings) to unpack mbtiles file and serve it via a local web server (or upload the data to S3, or …).
• use mapbox-tile-copy to upload contents directly to S3
• use a server script like the example in the repo that serves tiles directly from the mbtiles file. (code needs to be adjusted!)

HTTP API schema is:

http://.../.../<feature-type>/<z>/<x>/<y>.pbf

where feature-type is buildings,highways, etc.

The web server needs to return proper CORS headers, the Content-Encoding=gzip header (!) and the Content-Type header needs to be either application/x-protobuf or application/octet-stream.

Run frontend

• clone repo git clone https://github.com/hotosm/osm-analytics
• run npm install
• adjust base-url setting in ./app/settings/settings.js
• run npm start (for a development-server) or npm run build to build a production build (results as HTML5 in ./static)

Add a new feature type

(not yet tested, sorry)

• on frontend side: add new type in ./app/settings/options.js (as entry in filters array)
• on backend:
  • add new type definition json file (like buildings.json) (use "objects" as a generic user-experience field)
  • add a new crunching section in run.sh for the newly created feature type
  • serve results (see above)