Skip to content

Latest commit

 

History

History
92 lines (70 loc) · 2.82 KB

README.md

File metadata and controls

92 lines (70 loc) · 2.82 KB
Raw

osm-borders

Required C libraries: GDAL, GEOS

On Ubuntu: apt install g++ gcc libgdal-dev python3.6-dev

Requires: python 3.6

pip install -r requirements.txt

This script fetches administrative boundary definitions from EMUiA repository and converts it to OSM XML.

It is possible to fetch:

  1. Municipality (admin_level=7) borders by providing 4-digit PRG code
  2. Settlement (admin_level=8,9 and 10) borders by providing 7-digit PRG code

Amazon architecture

Because AWS API Gateway calls are limited to 60 seconds and the process may take more than 60 seconds, SNS queues are used to allow longer Lambda calls. Clients

 package-rest.zip
+-------------------+     2. Submit fetch task
| /<type>/<prg>.osm +-------------+
+-+-----------------+             |
  |                               |
  | 1. Check cache         +------+-----+
  |                        | Amazon SNS |
  | 5. Fetch from cache    +------+-----+
  |                               |
  |                               |
  |                       +-------+-------------+  3. Fetch  +-----------+
  |                       | package-fetcher.zip +------------+ Geoportal |
  |                       +--+------------------+     data   +-----------+
+-+--------+    4. Put into  |
| DynamoDB +-----------------+
+----------+       cache
  1. Amazon API Gateway is exposing Amazon Lambda amazon.rest_endpoint function at /<prg>.osm.
  2. If there is sentinel object in DynamoDB and it's not stale goto 4
  3. If stale or no data is found in DynamoDB, puts a sentinel object into DynamoDB and a message in Amazon SNS queue queue_name
  4. Sleep for 30 seconds
  5. Check if result is ready. If response is ready, returns the response, if not returns redirect to itself (goto 1).

Message schema:

{
    "terc": "123331",
    "type": "all",
    "request_time": 1515263002
}

amazon.fetcher listens on Amazon SNS queue

  1. Checks if there is sentinel object in DynamoDB and matches request_time, if not exit
  2. Checks if there is fresh data in DynamoDB. If yes, then exit
  3. Invokes get_<type>_borders. Results are put into DynamoDB table table_name

Building Amazon environment

Requirements

Install docker: apt install docker.io

Install Terraform

Assuming you will have terraform in your path, amazon/ folder contains Terraform scripts to create AWS Lambda / DynamoDB deployment. To deploy on your Amazon environment:

export AWS_ACCESS_KEY_ID="YOUR_AWS_KEY_ID"
export AWS_SECRET_ACCESS_KEY="YOUR_AWS_ACCESS_KEY"
export AWS_DEFAULT_REGION="DEPLOYMENT_REGION"

$ cd amazon
$ terraform init

And then, to deploy:

$ ( cd .. && build_package.sh )
$ terraform apply
[...]
Outputs:

url = https://xxx.execute-api.eu-west-1.amazonaws.com/api

Where the last line presents URL of your deployment