Snapshot Pages to the Wayback Machine After Deploy

id: internet-archive-save-page
namespace: company.team
description: |
  Trigger an Internet Archive Wayback Machine snapshot of a URL right after a
  deploy, and verify the snapshot landed — a durable receipt for what you shipped.

triggers:
  - id: deploy_webhook
    type: io.kestra.plugin.core.trigger.Webhook
    description: Call from the deploy pipeline so every release gets archived automatically.
    key: archive-page
  - id: weekly_snapshot
    type: io.kestra.plugin.core.trigger.Schedule
    description: Optional weekly snapshot to track drift over time; shipped disabled.
    cron: "0 5 * * 0"
    disabled: true

inputs:
  - id: url
    type: STRING
    defaults: https://example.com
    description: Page to archive.
  - id: wait
    type: STRING
    defaults: 30
    description: Seconds to wait before verifying the snapshot exists.

tasks:
  - id: save_page
    type: io.kestra.plugin.scripts.python.Script
    description: POST to the Wayback Machine's save endpoint, then poll until the
      snapshot URL appears.
    containerImage: python:3.11-slim
    warningOnStdErr: false
    env:
      URL: "{{ inputs.url }}"
      WAIT: "{{ inputs.wait }}"
    script: |
      import json, os, time, urllib.request, urllib.parse

      target = os.environ["URL"]
      save_url = "https://web.archive.org/save/" + target
      snapshot_url = ""
      try:
          req = urllib.request.Request(save_url, method="GET", headers={"User-Agent": "kestra-archive-blueprint"})
          with urllib.request.urlopen(req, timeout=60) as r:
              snapshot_url = r.geturl()
      except Exception as e:
          print(f"save request note: {e}")
      time.sleep(int(os.environ["WAIT"]))
      check = "https://archive.org/wayback/available?url=" + urllib.parse.quote(target, safe="")
      try:
          with urllib.request.urlopen(check, timeout=30) as r:
              data = json.loads(r.read())
          closest = data.get("archived_snapshots", {}).get("closest", {})
          snapshot_url = closest.get("url", snapshot_url)
          available = bool(closest.get("available"))
      except Exception:
          available = False
      print(f"snapshot available: {available} -> {snapshot_url}")
      print("::" + json.dumps({"outputs": {"available": available, "snapshot_url": snapshot_url}}) + "::")

  - id: verify_snapshot
    type: io.kestra.plugin.core.flow.If
    description: Branch on whether the archive confirmed the snapshot — a silent
      archiving failure must not look like "archived".
    condition: "{{ outputs.save_page.vars.available }}"
    then:
      - id: log_snapshot
        type: io.kestra.plugin.core.log.Log
        description: Record the snapshot URL in the execution log as the permanent receipt.
        message: "Archived {{ inputs.url }} at {{ outputs.save_page.vars.snapshot_url
          }}"
    else:
      - id: alert_no_snapshot
        type: io.kestra.plugin.slack.notifications.SlackIncomingWebhook
        description: Alert when archiving did not confirm, so the gap is visible instead
          of assumed.
        url: "{{ secret('SLACK_WEBHOOK_URL') }}"
        payload: |
          {
            "text": ":warning: Wayback Machine snapshot of {{ inputs.url }} did NOT confirm within {{ inputs.wait }}s. The archive may be delayed — re-check later. Execution {{ execution.id }}."
          }

errors:
  - id: alert_archive_failure
    type: io.kestra.plugin.slack.notifications.SlackIncomingWebhook
    description: Alert when the archive request itself errors — no receipt must
      never look like a archived page.
    url: "{{ secret('SLACK_WEBHOOK_URL') }}"
    payload: |
      {
        "text": "Archive-page FAILED in flow {{ flow.id }} (execution {{ execution.id }}) for {{ inputs.url }}. Check outbound access to web.archive.org."
      }

outputs:
  - id: archive_result
    type: JSON
    description: 'Whether the snapshot confirmed and its URL, e.g. {"available":
      true, "snapshot_url": "https://web.archive.org/..."}.'
    value: "{{ outputs.save_page.vars | toJson }}"