Skip to content
Broken Link Checker

Broken Link Checker #34

Sign in to view logs

Broken Link Checker

Broken Link Checker #34

Workflow file for this run

.github/workflows/link-checker.yml at e61b09f
name: Broken Link Checker
on:
workflow_dispatch:
inputs:
siteUrl:
description: 'Site URL to Check'
default: 'https://is.docs.wso2.com/en/latest/'
required: true
schedule:
- cron: '30 2 * * *' # Schedule to run once daily at 02:30 UTC
env:
DEFAULT_URL: 'https://is.docs.wso2.com/en/latest/'
jobs:
linkchecker:
runs-on: ubuntu-latest
steps:
- name: Check repository
run: |
if [ "${{ github.event_name }}" = "schedule" ] && [ "$GITHUB_REPOSITORY" != "wso2/docs-is" ]; then
echo "This scheduled action is disabled for forks."
exit 0
fi
- name: Check out repository
uses: actions/checkout@v3
- name: Set up Python
uses: actions/setup-python@v4
with:
python-version: '3.x'
- name: Install dependencies
run: |
pip install linkchecker
pip install beautifulsoup4
- name: Run Link Checker
run: |
SITE_URL=${{ github.event.inputs.siteUrl || env.DEFAULT_URL }}
linkchecker -F html $SITE_URL --threads=100
continue-on-error: true
- name: Filter HTML for specific string and preserve page structure
id: filter_html
run: |
echo "import sys" > filter_script.py
echo "from bs4 import BeautifulSoup" >> filter_script.py
echo "website_url = 'https://is.docs.wso2.com/en/latest/'" >> filter_script.py
echo "with open('linkchecker-out.html', 'r') as file:" >> filter_script.py
echo " soup = BeautifulSoup(file, 'html.parser')" >> filter_script.py
echo "string_to_filter = 'https://is.docs.wso2.com/en/latest/page-not-found'" >> filter_script.py
echo "final_soup = BeautifulSoup('<!DOCTYPE html><html><head></head><body></body></html>', 'html.parser')" >> filter_script.py
echo "final_soup.head.append(soup.head)" >> filter_script.py
echo "# Add custom header, broken link count and spacing" >> filter_script.py
echo "header_tag = final_soup.new_tag('h1')" >> filter_script.py
echo "header_tag.string = 'Broken Link Checker - ' + website_url" >> filter_script.py
echo "final_soup.body.append(header_tag)" >> filter_script.py
echo "# Count and append the number of broken links found" >> filter_script.py
echo "tables = soup.find_all('table')" >> filter_script.py
echo "broken_link_count = sum(string_to_filter in str(table) for table in tables)" >> filter_script.py
echo "with open('broken_link_count.txt', 'w') as count_file:" >> filter_script.py
echo " count_file.write(str(broken_link_count))" >> filter_script.py
echo "count_tag = final_soup.new_tag('p')" >> filter_script.py
echo "count_tag.string = 'Number of broken links found: ' + str(broken_link_count)" >> filter_script.py
echo "final_soup.body.append(count_tag)" >> filter_script.py
echo "final_soup.body.append(final_soup.new_tag('br'))" >> filter_script.py
echo "# Append filtered tables" >> filter_script.py
echo "for table in tables:" >> filter_script.py
echo " if string_to_filter in str(table):" >> filter_script.py
echo " final_soup.body.append(table)" >> filter_script.py
echo " final_soup.body.append(final_soup.new_tag('br'))" >> filter_script.py
echo "final_soup.body.append(final_soup.new_tag('br'))" >> filter_script.py
echo "with open('broken_links_report_IS.html', 'w') as file:" >> filter_script.py
echo " file.write(str(final_soup))" >> filter_script.py
python filter_script.py
BROKEN_LINKS_COUNT=$(cat broken_link_count.txt)
echo "broken_links_count=$BROKEN_LINKS_COUNT" >> $GITHUB_ENV
- name: Upload HTML as Artifact
uses: actions/upload-artifact@v3
with:
name: broken_links_report_IS
path: broken_links_report_IS.html
- name: Notify Google Chat
env:
WEBHOOK_URL: ${{ secrets.GOOGLE_CHAT_WEBHOOK_URL }}
SITE_URL: ${{ github.event.inputs.siteUrl || env.DEFAULT_URL }}
THREAD_NAME: "BrokenLinkCheckerWSO2IS"
BROKEN_LINKS_COUNT: ${{ env.broken_links_count }}
run: |
MESSAGE_JSON=$(cat <<EOF
{
"cards": [
{
"header": {
"title": "Broken Link Checker Report - WSO2 IS",
"subtitle": "${SITE_URL}",
"imageUrl": "https://biq.cloud/wp-content/uploads/2021/01/broken-link-building.gif"
},
"sections": [
{
"widgets": [
{
"keyValue": {
"topLabel": "Report Details",
"content": "Broken Links Found: ${BROKEN_LINKS_COUNT}"
}
},
{
"buttons": [
{
"textButton": {
"text": "VIEW REPORT",
"onClick": {
"openLink": {
"url": "https://github.com/${{ github.repository }}/actions/runs/${{ github.run_id }}"
}
}
}
}
]
}
]
}
]
}
]
}
EOF
)
curl -X POST -H 'Content-Type: application/json' -d "$MESSAGE_JSON" "$WEBHOOK_URL&threadKey=$THREAD_NAME&messageReplyOption=REPLY_MESSAGE_FALLBACK_TO_NEW_THREAD"