Automation/blog pipeline

[nora-weisser]

A daily (and manually-triggerable) GitHub Action that turns reviewed blog submissions into draft PRs.
The initial PR was opened before by Silke, I improved it.

What it does

• Reads the Form Responses sheet and publishes rows marked isReviewedandApproved but not yet isPublished
• Exports each Google Doc → sanitized HTML post, downloads the cover image (falls back to default), then opens a PR for human review
• Writes isPublished=TRUE back to the sheet so nothing is published twice — the sheet is the single source of truth

Key changes

• publish_reviewed_blogs.py — orchestrator; blog_exporter.py — Doc→post + image; blog_info_from_spreadsheet.py — sheet read/write
• HTML sanitized against an allowlist (bleach); filename slugs and YAML front matter hardened
• Robust to missing images / inaccessible Docs (skips cleanly)
• Workflow runs only on the org repo; service-account key injected from a secret and never committed

It was tested locally, it is not possible to test when it is in PR only when merged.

Pull request checklist

Please check if your PR fulfills the following requirements:

• I checked and followed the contributor guide
• I have tested my changes locally.
• I have added a screenshot from the website after I tested it locally

[dricazenck]

The branch name is fixed as automation/import-blog, which means all batch imports accumulate into the same PR. If a previous batch PR is still open when a new daily run triggers, reviewers might approve a larger bundle than expected.\n\nWould it be worth using a date-stamped branch name (e.g. automation/import-blog-2026-06-28) so each run creates its own isolated PR, giving reviewers clearer control over what they're approving?

[dricazenck]

The PR body already mentions that isPublished is set to TRUE before merging, which is good. One concern: if this PR is rejected or closed without merging, those rows stay marked as published in the sheet and won't be re-exported on the next run — they'd be silently lost.\n\nCould we add a note here reminding reviewers that closing this PR without merging requires manually resetting isPublished to FALSE in the spreadsheet for the affected rows?

[dricazenck]

The PR checklist includes "I have added a screenshot from the website after I tested it locally" but no screenshot appears in the PR body. Since this automation writes files and opens PRs, a sample of the generated post output (e.g. the front matter + first few lines of an exported .html file) would help reviewers validate the format before merging.

[dricazenck]

The README now correctly references blog_exporter.py as the entry point for manual testing, but doc_to_html_conversion.py is still present in the repo (visible in the file tree). Is this file still needed, or should it be removed/deprecated to avoid confusion about which script to use?