Skip to content

datasets_weekly

datasets_weekly #25

Workflow file for this run

name: datasets_weekly
on:
schedule:
- cron: "0 6 * * 0" # Every Sunday at 06:00 UTC
workflow_dispatch: {}
jobs:
data:
runs-on: ubuntu-latest
container: ghcr.io/dataresearchcenter/datasets:latest
env:
HOME: /home/1000
AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
FSSPEC_S3_ENDPOINT_URL: ${{ secrets.FSSPEC_S3_ENDPOINT_URL }}
strategy:
fail-fast: false
matrix:
include:
- dataset: de_bnetza_mastr
runner: investigraph
- dataset: worldbank_ifc_advisory_services
runner: investigraph
- dataset: worldbank_ifc_investment_services
runner: investigraph
- dataset: worldbank_procurement_awards
runner: investigraph
- dataset: worldbank_procurement_projects
runner: investigraph
- dataset: worldbank_procurement_notices
runner: investigraph
- dataset: de_fragdenstaat
runner: memorious
steps:
- name: Run ${{ matrix.runner }} for ${{ matrix.dataset }}
run: make ${{ matrix.runner }} dataset=${{ matrix.dataset }} DATASETS_DIR=.
working-directory: /datasets
catalog:
needs: data
if: always()
runs-on: ubuntu-latest
container: ghcr.io/dataresearchcenter/datasets:latest
defaults:
run:
working-directory: /datasets
env:
AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
FSSPEC_S3_ENDPOINT_URL: ${{ secrets.FSSPEC_S3_ENDPOINT_URL }}
steps:
- name: Update the catalog
run: "make publish"
- name: Notify success
run: "anystore io -i ${{ secrets.DATASETS_WEEKLY_HEARTBEAT }}"