Crawl cf.gov #1492

	name: Crawl cf.gov

	# Triggers the workflow every day at 6AM UTC (1AM EST, 2AM EDT)
	on:
	schedule:
	- cron: '0 6 * * *'
	workflow_dispatch:

	env:
	DOMAIN: www.consumerfinance.gov

	jobs:
	build:
	runs-on: ubuntu-latest

	steps:
	- uses: actions/checkout@v2

	- name: Remove previous crawl results
	run: rm -r ${{ env.DOMAIN }}

	- name: Run the crawl script
	run: ./crawl.sh -d 0 https://${{ env.DOMAIN }}

	- name: Simplify crawled HTML to prevent unimportant diffs
	run: postprocessing/transform_results.sh ${{ env.DOMAIN }}
	continue-on-error: true

	- name: Prepare files for git commit
	run: \|
	gzip -f *.log
	git add *.log.gz
	git add ${{ env.DOMAIN }}

	- name: Prepare commit message
	run: ./generate_summary.sh ${{ env.DOMAIN }} > commit.txt

	- name: Commit crawl results back to GitHub
	run: \|
	git config user.email [email protected]
	git config user.name Automated
	git commit -F commit.txt
	git push
	env:
	GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}

Provide feedback