🐛 Fixed paid checkout returning to homepage instead of originating page

[9larsons]

Summary

When a member starts a paid signup or upgrade from a page other than the homepage, Portal returned them to the site root after a successful Stripe Checkout (e.g. https://example.com/?stripe=success) instead of the page where checkout began. Cancelling checkout already returned to the originating page, so the behaviour was inconsistent — and especially disruptive for paywall flows, where a member who just paid to unlock an article was bounced to the homepage instead of back to the article.

Root cause

checkoutPlan() in apps/portal/src/utils/api.js already derived a contextual cancelUrl from window.location when none was supplied, but had no equivalent fallback for successUrl. The two standard callers (signup and upgrade in apps/portal/src/actions.js) don't pass one, so successUrl arrived at the backend as undefined and Ghost fell back to the configured site-root default (stripe-api.js → checkoutSessionSuccessUrl).

Fix

Derive a contextual successUrl from window.location when none is supplied, mirroring the existing cancelUrl logic. This is the same approach continueGiftCheckout() already uses for gift flows (and matches #27643 / #27827).

No backend change is needed: sanitizeReturnUrl() already enforces that return URLs are same-origin / within the site subpath, so the contextual URL flows through safely.

Behaviour preserved

• Welcome-page tiers — the server still overrides the success URL with a configured tier welcome page (_generateSuccessUrl); the contextual URL only applies when no welcome page is set.
• New-email signup — the contextual URL becomes the post-signup magic-link referrer, so the member returns to the originating page after confirming (an improvement over the previous site-root referrer).

Tests

Added unit tests in apps/portal/test/api.test.js covering:

• contextual successUrl/cancelUrl derived from the current page when none supplied
• fallback to the site root when the current location is off-site (open-redirect guard)
• explicitly-supplied successUrl/cancelUrl are preserved

ref https://linear.app/ghost/issue/PLA-86
fixes #28369

[coderabbitai]

Walkthrough

checkoutPlan in api.js gains a conditional block that initializes successUrl when it is not supplied: if the current window.location.href shares the same origin as siteUrl, that href is used as the base; otherwise siteUrl is used, and stripe=success is set as a query parameter on the resulting URL. This mirrors the existing cancelUrl fallback. A new Portal API plan checkout test suite is added in api.test.js with setup that overrides window.location and stubs window.fetch, covering the same-origin default, off-site fallback, and explicit URL passthrough scenarios. The package version is bumped to 2.69.5.

Possibly related issues

• Portal paid checkout returns to site root instead of originating page #28369: The PR implements the fix described in that issue — modifying checkoutPlan to derive a contextual successUrl from the current page rather than omitting it, so members return to the originating page after Stripe checkout instead of the homepage.

Possibly related PRs

• TryGhost/Ghost#27845: Both PRs modify Stripe checkout session URL derivation in checkoutPlan, with overlapping logic for same-origin/subpath detection when building successUrl/cancelUrl defaults.

🚥 Pre-merge checks | ✅ 4

✅ Passed checks (4 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title accurately summarizes the main fix: returning users to their originating page after successful checkout instead of the homepage.
Description check	✅ Passed	The description provides comprehensive context about the bug, root cause, fix, behavior preservation, and test coverage—all directly related to the changeset.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch claude/pensive-bose-89dcb8

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[nx-cloud]

🤖 Nx Cloud AI Fix

Ensure the fix-ci command is configured to always run in your CI pipeline to get automatic fixes in future runs. For more information, please see https://nx.dev/ci/features/self-healing-ci

---

View your CI Pipeline Execution ↗ for commit 15fb105

Command	Status	Duration	Result
nx build @tryghost/announcement-bar	✅ Succeeded	<1s	View ↗
nx build @tryghost/signup-form	✅ Succeeded	<1s	View ↗
nx build @tryghost/activitypub	✅ Succeeded	1s	View ↗
nx build @tryghost/comments-ui	✅ Succeeded	<1s	View ↗
nx build @tryghost/sodo-search	✅ Succeeded	<1s	View ↗
nx build @tryghost/portal	✅ Succeeded	<1s	View ↗
nx build @tryghost/admin-toolbar	✅ Succeeded	<1s	View ↗
nx run-many -t test:unit -p @tryghost/portal	✅ Succeeded	58s	View ↗
Additional runs (5)	✅ Succeeded	...	View ↗

💡 Verify your cache is correct by running tasks in a sandbox. Read docs ↗

---

☁️ Nx Cloud last updated this comment at 2026-06-18 14:39:08 UTC

[coderabbitai]

🧹 Nitpick comments (1)

apps/portal/test/api.test.js (1)

231-317: ⚡ Quick win

Well-structured test coverage for the new successUrl derivation logic.

The three test cases cover the primary scenarios: same-origin default, off-site fallback, and explicit URL preservation. The mock setup correctly stubs both the session and checkout endpoints.

Optional: Consider additional edge-case coverage

While the current tests cover the main branches well, you might consider adding tests for:

1. Current page with existing query parameters:

   window.location.href = 'https://example.com/article?utm_source=newsletter';
   // Expected: https://example.com/article?utm_source=newsletter&stripe=success

   This would verify that existing params are preserved when stripe=success is added.

2. Site URL with a path component:

   const ghostApi = setupGhostApi({siteUrl: 'https://example.com/blog'});
   window.location.href = 'https://example.com/blog/post';

   This would verify the startsWith same-origin check works correctly with path-based site URLs.

These scenarios should work correctly with the current implementation but explicit coverage would increase confidence.

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@apps/portal/test/api.test.js` around lines 231 - 317, Add two additional test
cases to the Portal API plan checkout test suite to improve edge-case coverage.
First, add a test that verifies when the current page URL contains existing
query parameters (like utm_source), those parameters are preserved when the
stripe=success and stripe=cancel parameters are appended to the successUrl and
cancelUrl respectively. Second, add a test that verifies the same-origin check
in the successUrl derivation logic works correctly when the siteUrl has a path
component (e.g., https://example.com/blog), ensuring that both the siteUrl and
window.location.href with matching path prefixes are correctly identified as
same-origin. Both tests should follow the existing pattern of setting up a
ghostApi with setupGhostApi, calling mockCheckoutFetch, invoking
ghostApi.member.checkoutPlan with appropriate parameters, and asserting the
correct body values using lastCheckoutBody.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Nitpick comments:
In `@apps/portal/test/api.test.js`:
- Around line 231-317: Add two additional test cases to the Portal API plan
checkout test suite to improve edge-case coverage. First, add a test that
verifies when the current page URL contains existing query parameters (like
utm_source), those parameters are preserved when the stripe=success and
stripe=cancel parameters are appended to the successUrl and cancelUrl
respectively. Second, add a test that verifies the same-origin check in the
successUrl derivation logic works correctly when the siteUrl has a path
component (e.g., https://example.com/blog), ensuring that both the siteUrl and
window.location.href with matching path prefixes are correctly identified as
same-origin. Both tests should follow the existing pattern of setting up a
ghostApi with setupGhostApi, calling mockCheckoutFetch, invoking
ghostApi.member.checkoutPlan with appropriate parameters, and asserting the
correct body values using lastCheckoutBody.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: b9b1eca1-d1c8-4ec9-95f4-befb719626c2

📥 Commits

Reviewing files that changed from the base of the PR and between 0aeb5a5 and 6b90006.

📒 Files selected for processing (3)

• apps/portal/package.json
• apps/portal/src/utils/api.js
• apps/portal/test/api.test.js