MWPW-206375: fix cascading publish/revert for new cards (PR #1039 regression)

[seanchoi0]

Resolves https://jira.corp.adobe.com/browse/MWPW-206375

Problems

Two regressions introduced by PR #1039, plus one additional edge case uncovered during review:

1. New cards — "No snapshot found" on revert / "Publish failed" on save

`recordSnapshot()` returned a failure for any fragment with no version history (new cards that were never published). This caused:

• `save-snapshot-action` to write `SAVE_SNAPSHOT: No non-translation version found` to `lastError` → "Publish failed" banner on save
• Revert to fail with "No snapshot found" because snapshot entries were empty

2. Cascading publish broken — cards & variations not published or reverted

In the `hasValidPreRecordedSnapshot` branch added by PR #1039:

• `createSnapshot` was called without `includeCards`/`includeVariations` → sub-collections, cards, and variations were not traversed
• `expandedPaths` was not captured → `publishPaths` fell back to top-level only
• `snapshotEntries` contained only top-level fragments → revert only worked for the top-level collection

3. Silent no-op revert for published fragments with no prior non-translation version

A fragment can be live (wasPublished: true) but have versionId: null if all its CF versions were created by the translation system. In this case `revertSnapshot()` would skip both `restoreVersion` and `unpublishFragment` and return `null` (success), silently doing nothing.

Fixes

Fix 1 — `snapshot.js`

`recordSnapshot()`: Record new cards with `versionId: null` instead of failing. Reverting will unpublish them (restoring pre-publish state) rather than trying to restore a non-existent version.

`revertSnapshot()`: Skip `restoreVersion` when `versionId` is `null`; still call `unpublishFragment` since `wasPublished: false`. For the edge case where `versionId: null` and `wasPublished: true` — return `{ skipped }` instead of silently succeeding.

`hasValidPreRecordedSnapshot()` (`bulk-publish-worker.js`): Accept `null` versionId by checking `'versionId' in parsed` instead of `parsed.versionId &&`.

Fix 2 — `bulk-publish-worker.js`

Both `hasValidPreRecordedSnapshot` and fallback branches:

• Pass `includeCards`/`includeVariations` to `createSnapshot` so cascaded fragments are traversed
• Capture `expandedPaths` for the correct cascade publish paths
• `mergeCascadedEntries()`: merge pre-recorded (green/null) versionIds for top-level fragments with freshly created Pre-bulk-publish versionIds for cascaded fragments → revert now covers all published items

Behavior

Scenario	Before	After
New card saved to project	❌ "Publish failed" banner immediately	✅ No error
New card bulk publish → revert	❌ "No snapshot found"	✅ Unpublished
Cascading publish (includeCards + includeVariations)	❌ Only top-level published	✅ All cards & variations published
Cascading revert	❌ Only top-level reverted	✅ All published items reverted
Published card with only translation versions → revert	❌ Silent no-op (reported as success)	✅ Skipped (reported honestly)
Existing card (has versions), revert	✅ Restored to prior version	✅ Unchanged

[aem-code-sync]

Hello, I'm the AEM Code Sync Bot and I will run some actions to deploy your branch.
In case there are problems, just click the checkbox below to rerun the respective action.

• Re-sync branch

Commits

• 7731361 ✅ (latest)
• 17d60d5 ✅
• a3dec0f ✅
• 3e970a4 ✅
• 4f52fa2 ✅
• fc68dd2 ✅
• 0ce8490 ✅

[github-actions]

SNOW Change Request Transaction ID [IO]: 0000019c1ef37582-20e0d556

[github-actions]

SNOW Change Request Transaction ID [STUDIO]: 0000019c1eec85a3-20bbe383

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.71%. Comparing base (2f2e131) to head (7731361).

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1207      +/-   ##
==========================================
- Coverage   90.71%   90.71%   -0.01%     
==========================================
  Files         318      318              
  Lines      101712   101754      +42     
==========================================
+ Hits        92271    92309      +38     
- Misses       9441     9445       +4     

Files with missing lines	Coverage Δ
io/studio/src/bulk-publish/bulk-publish-worker.js	100.00% <100.00%> (ø)
io/studio/src/bulk-publish/snapshot.js	97.90% <100.00%> (+0.04%)	⬆️

... and 9 files with indirect coverage changes

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2f2e131...7731361. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[Axelcureno]

One blocker: recordSnapshot can now emit versionId: null with wasPublished: true (your own test, snapshot.test.js:1025). Your comment says revert will "unpublish them instead", but wasPublished: true blocks that unpublish too, so revert does nothing, reports no failure, and Studio toasts success. Route it to the existing skipped channel?

Minor: worker.test.js:490 would pass pre-fix; it doesn't assert the merge.

Merge dedupe is load-bearing: deserializeEntries uses Object.fromEntries, so cascaded would otherwise win.

[Axelcureno]

Blocker's addressed. Deleted the guard locally and the new test fails, so it's a genuine regression test; both merge tests fail pre-fix. 156 passing.

Two non-blocking follow-ups:

• skipped never reaches the UI. revert-action.js drops it from snapshots and returns REVERTED, and bulk-publish-store.js:76 only reads status, so Studio says "Reverted" while the card stays live.
• deserializeEntries takes createdAt from parsed[0], so merged snapshots use the earlier timestamp. Check Modifications may flag every cascaded card.

Approving.