Skip to content

[Develop] Fix 13 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload - #2255

Merged
subodhr258 merged 3 commits into
developfrom
fix/dependabot-2026-10-26-develop
Oct 29, 2026
Merged

subodhr258 merged 3 commits into
developfrom
fix/dependabot-2026-10-26-develop

Conversation

@subodhr258

@subodhr258 subodhr258 commented Oct 26, 2026 •

Copy link
Copy Markdown
Collaborator

🤖 Automated PR from the daily GoDAM vulnerability fixer. It merges itself once a person approves.

Summary

Closes 15 of the 17 open Dependabot alerts on the GoDAM plugin (1 critical, 5 high), by pinning patched versions in the overrides block of package.json, and fixes 4 Medium Garuda code findings (plus 3 Lows with the same root cause) in 2 fixes. One PR into develop (the [Main] partner #2256 was closed on 28 Oct: develop reaches main at the release). The alerts close when develop reaches main at the next plugin release.

Updated 29 Oct: a third commit bumps qs 6.16.0 to 6.16.1 for the new alert 396 (medium, development only). The 3 DOMPurify alerts (386, 387, 388) were marked fixed on 28 Oct when GoDAM 2.5.1 reached main, so this PR now covers 13 of the 16 alerts open on main; the other 3 (extract-zip 299 and 319, deepmerge 397) have no fixed release.

Dependabot alerts

Alert Severity Package Scope From To
alert 388 critical dompurify runtime 3.4.16 3.4.17
alert 387 high dompurify runtime 3.4.16 3.4.17
alert 378 high follow-redirects runtime 1.16.0 1.16.1
alert 379 high follow-redirects runtime 1.16.0 1.16.1
alert 383 high launch-editor development 2.14.1 2.14.2
alert 380 high markdown-it development 14.3.2 14.3.3
alert 392 medium fast-uri development 3.1.8 3.1.9
alert 394 medium fast-uri development 3.1.8 3.1.9
alert 382 medium launch-editor development 2.14.1 2.14.2
alert 384 medium launch-editor development 2.14.1 2.14.2
alert 386 low dompurify runtime 3.4.16 3.4.17
alert 391 low fast-uri development 3.1.8 3.1.9
alert 393 low fast-uri development 3.1.8 3.1.9
alert 381 low markdown-it development 14.3.2 14.3.3
alert 396 medium qs development 6.16.0 6.16.1
alert 395 low serialize-javascript development 7.0.5 7.1.3

Garuda code findings (2 fixes)

Under the runbook's Garuda limit (at most 5 fixes per run, Medium and up, no Lows on their own), this PR fixes the 2 real Medium root causes whose fix only touches the cited code. Scans: https://github.com/rtCamp/garuda-alpha-testing-security-findings/tree/main/godam

Fix 1: the three notice-dismiss AJAX handlers check the capability, not only the nonce. Adds Garuda's suggested current_user_can( 'manage_options' ) check (403 JSON error) to dismiss_free_plan_notice, dismiss_woo_promo_notice and dismiss_offer_banner. Closes:

Fix 2: the SureForms recorder upload takes only the recorder field's own key, and only audio or video. The key must have the shape the recorder block prints (godam-upload-<block id>-lbl-<label>-input-recorder<index>, block id may be empty) instead of any key containing -input-recorder, and wp_handle_upload() gets mimes (the site's allowed types narrowed to audio and video) and test_type => true, so the type is checked against the file, not the browser's claim. Closes:

  • 21-10-2026 Fix Full Screen Video Layers bug #18 (Medium): SureForms handler stores any posted file under a *-input-recorder key, unauthenticated
  • 09-10-2026 L10 (Low): the file-type part. The size limit still comes from a submitted value capped at the server maximum; that part stays open as a Low.

The MIME list is a small static method on the SureForms class on purpose: #2251 adds a shared rtgodam_get_recorder_upload_mimes() for Fluent Forms and Ninja Forms, and adding the same function here would clash when both merge. Once both are in, the SureForms class can switch to the shared helper.

Not in this PR (sent to Subodh as a decision in the Slack thread): player SDK origin checks (#15, L29, #25), the account token in front-end page data (#12), the MetForm route (09-10 #22), three findings that code reading says are not real (#1, #6, #16), and the other Lows.

Follow-up worth considering: the free-plan, Woo promo and offer notices are still shown to non-admins; with this fix their Dismiss no longer sticks for them. Hiding those notices from users who cannot dismiss them is a visible change, so it is left out here.

What changes in shipped code

Dependabot part: one runtime change, DOMPurify 3.4.16 to 3.4.17. It lands in the 12 bundles that already include DOMPurify (pages/godam, pages/analytics, pages/video-editor, js/godam-player-frontend, blocks/godam-player, js/media-library, js/godam-gallery, js/gf-godam-recorder-editor, the four LifterLMS bundles), each +970 bytes, plus their .asset.php hashes and the DOMPurify version line in 3 LICENSE.txt files. No other file in assets/build changes (Develop: compared all build files before and after).

Garuda part: the PHP changes above (3 AJAX handlers, the SureForms recorder upload). No build file changes.

follow-redirects is marked runtime because axios depends on it, but it is only used by axios on Node; no browser bundle changes because of it. The other packages are development tools only.

How it was verified

  • Node 22.14.0 (.nvmrc): npm ci and npm run build:prod exit 0 before and after; every resolved copy listed afterwards: dompurify 3.4.17, fast-uri 3.1.9, follow-redirects 1.16.1, launch-editor 2.14.2, markdown-it 14.3.3, serialize-javascript 7.1.3 (one copy each).
  • npm run test:unit: same result before and after on both branches (main 690 passed, develop 688 passed).
  • Garuda fixes: 2 new PHPUnit files, 30 tests (NoticeDismissCapabilityTest, SureFormsRecorderUploadTest). With the fix all pass; with only the source change reverted, 22 of the first 30 fail (subscriber, contributor, author and editor dismissals saved; non-recorder keys and non-media types accepted). Full PHP suite: main 945 tests (21 errors), develop 662 (8 errors); the same errors appear on the base without this change (PHP 8.5 deprecations in older tests). PHPCS clean on the changed files.
  • 29 Oct, qs commit (Node 22.14.0): npm ci --ignore-scripts, npm run build:prod and npm run test:unit before and after: exit 0, 688/688 tests both times, 330/330 files in assets/build byte-identical. npm ls qs --all: every copy is 6.16.1 (express and body-parser inside the webpack dev server). qs 6.16.1 was published 14 Oct; its resolved URL is registry.npmjs.org; it has no install script.
  • launch-editor used a scoped key (launch-editor@<2.14.1) that npm 10 does not re-apply to an already locked copy, so it is now an unscoped pin.

Suspected side effects

  • qs 6.16.0 to 6.16.1 (29 Oct): used only by express inside the webpack dev server (npm start); nothing in the repo imports it. Release notes: parse and stringify fixes and hardening, nothing marked breaking. No shipped file changes.

  • DOMPurify 3.4.17 release notes: fixes in IN_PLACE mode, SAFE_FOR_TEMPLATES with shadow roots, the SMIL attributeName guard in XHTML mode, and small hardening. Nothing marked breaking. GoDAM calls DOMPurify.sanitize in string mode (no IN_PLACE), in the player settings menu, share modal, video editor layers, media library details and the recorder fields.

  • Features to exercise: admin pages, block editor with a GoDAM video block, front-end player (play, settings menu).

Smoke test

Throwaway WordPress Playground (PHP 8.3, latest WordPress, WooCommerce 11.2.1, GoDAM for Woo from the same base) with WP_DEBUG_LOG on, base build vs this branch's build, for both main and develop. Checked with headless Chrome: wp-admin, every GoDAM admin page (GoDAM, Video Editor, Analytics, Settings, Reel Pops, Help), media library, block editor with a GoDAM video block inserted and saved, a WooCommerce product edit screen, and as a logged-out visitor the video post (player loads, plays, settings menu opens), a product page and the home page.
Results identical before and after: every page HTTP 200, no new browser console errors, no PHP warnings or fatals. One PHP notice ("godam-player-style" enqueued with an unregistered dependency) appears in the block editor in all four runs, before and after.

Garuda fixes, same Playground, base vs this branch (main and develop):

  • Dismiss actions through admin-ajax.php with real nonces: as a Subscriber the base saves all three dismissals site-wide (HTTP 200); this branch answers 403 "Insufficient permissions." and saves nothing. As an administrator both save all three, as before.
  • SureForms 2.12.9 with the GoDAM recorder block rendered: its real input names (godam-upload-a1b2c3d4-lbl-…-input-recorder1, and godam-upload--lbl-… with no block id) match. Real anonymous multipart uploads through the handler: an MP4 recording is stored on both; a PDF under the recorder key is stored by the base and refused by this branch; an MP4 under another *-input-recorder key is stored by the base and ignored by this branch.
  • Pages (wp-admin, every GoDAM admin page, media library, block editor save, front-end player): same as above, before and after.

29 Oct, qs commit: a new throwaway Playground (PHP 8.3, WordPress latest), GoDAM built from the previous head 073cd37 versus this head, WP_DEBUG_LOG on: wp-admin, all 5 GoDAM admin pages, Plugins, media library, block editor with a GoDAM video block (valid, saved without error), the video post as administrator and as a logged-out visitor (player loads and plays), and the home page. Every page 200 on both, no console errors, the same one "godam-player-style" notice on both. qs is a dev-server package only (the built plugin files are byte-identical). Site deleted after the run.

Smoke test: passed

Reviewer check: passed

Manual QA

Runtime changes: DOMPurify and the two Garuda fixes.

  1. Open a post with a GoDAM video on the front end, open the player settings menu (speed, quality) and the share button.
  2. In the video editor, add a CTA (HTML) layer, save, and check it renders in the preview and on the front end.
  3. Open a video in the media library and check its details panel.
    Garuda fixes:
  4. As an administrator on the Dashboard, dismiss the free-plan notice (site without an API key) and the WooCommerce promo notice (WooCommerce active, GoDAM for Woo not): they stay dismissed. As an Editor, Dismiss no longer hides them for everyone.
  5. With SureForms, put a GoDAM recorder field in a form, submit a recording as a logged-out visitor, and check the entry has the file and it is sent for transcoding.

Development-only bumps: no runtime impact, QA can skip.

Left open

  • deepmerge alert 397 (low, development): no fixed release exists (newest is 4.3.1, the affected version).
  • extract-zip alert 299 and alert 319: no fixed release exists.
  • Garuda: of 101 findings, 37 are fixed on main, 35 have open PRs for 2.5.2 (https://github.com/rtCamp/godam-plugin-wp/issues/66), 6 are fixed here (09, 10, 11, 18, L30, L31; L10 in part), and the other 23 are with Subodh as a decision (6 Medium in 4 groups, 17 Low).

@github-actions

github-actions Bot commented Oct 26, 2026 •

Copy link
Copy Markdown

🔍 WordPress Plugin Check Report

⚠️ Status: Passed with warnings

📊 Report

🎯 Total Issues ❌ Errors ⚠️ Warnings
13 0 13

⚠️ Warnings (13)

📁 readme.txt (2 warnings)
📍 Line 🔖 Check 💬 Message
0 mismatched_plugin_name Plugin name "GoDAM - Organize WordPress Media Library & File Manager with Unlimited Folders for Images, Videos & more" is different from the name declared in plugin header "GoDAM".
0 trademarked_term The plugin name includes a restricted term. Your chosen plugin name - "GoDAM - Organize WordPress Media Library & File Manager with Unlimited Folders for Images, Videos & more" - contains the restricted term "wordpress" which cannot be used at all in your plugin name.
📁 assets/build/css/main.css (1 warning)
📍 Line 🔖 Check 💬 Message
0 EnqueuedStylesScope This style is being loaded in all contexts.
📁 assets/src/libs/analytics.min.js (5 warnings)
📍 Line 🔖 Check 💬 Message
0 EnqueuedScriptsScope This script is being loaded in all frontend contexts.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880 (with handle analytics-library) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/2026/10/29/hello-world/ (with handle analytics-library) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/sample-page/ (with handle analytics-library) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/demo-attachment-post/ (with handle analytics-library) is loaded in the footer. Consider a defer or async script loading strategy instead.
📁 assets/build/js/main.min.js (5 warnings)
📍 Line 🔖 Check 💬 Message
0 EnqueuedScriptsScope This script is being loaded in all frontend contexts.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880 (with handle rtgodam-script) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/2026/10/29/hello-world/ (with handle rtgodam-script) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/sample-page/ (with handle rtgodam-script) is loaded in the footer. Consider a defer or async script loading strategy instead.
0 NonBlockingScripts.NoStrategy This script on http://localhost:8880/demo-attachment-post/ (with handle rtgodam-script) is loaded in the footer. Consider a defer or async script loading strategy instead.

🤖 Generated by WordPress Plugin Check Action • Learn more about Plugin Check

…eo from the SureForms recorder field

Fixes Garuda findings 21-10-2026 #9, #10, #11 and #18 (and L10, L30, L31).
@subodhr258 subodhr258 changed the title [Develop] Fix 15 Dependabot alerts: DOMPurify 3.4.17, follow-redirects and dev tools [Develop] Fix 15 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload Oct 26, 2026
The qs override moves from 6.16.0 to 6.16.1, the release that fixes GHSA-6pq8-jhm2-2vcw (qs.stringify with arrayFormat 'comma' throws on elements without a callable toString). qs is only used by express inside the webpack dev server, so the built plugin files do not change.
@subodhr258 subodhr258 changed the title [Develop] Fix 15 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload [Develop] Fix 16 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload Oct 29, 2026
@subodhr258 subodhr258 changed the title [Develop] Fix 16 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload [Develop] Fix 13 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload Oct 29, 2026
@subodhr258
subodhr258 merged commit 7fefcbd into develop Oct 29, 2026
6 checks passed
@subodhr258
subodhr258 deleted the fix/dependabot-2026-10-26-develop branch October 29, 2026 13:48
subodhr258 added a commit that referenced this pull request Oct 29, 2026
package.json: take develop's fast-uri ^3.1.9 from #2255. package files now differ from develop only in the 2.5.2 version.
KMchaudhary added a commit that referenced this pull request Oct 30, 2026
* Add Marketo form layer support (V1) (#2218)

* Add Marketo form layer support (V1)

Lets any site using Marketo embed a Marketo form as a video layer from the
GoDAM UI, replacing the custom VideoJet-only shortcode implementation.

Marketo is modeled as a new `form_type: 'marketo'` under the existing `form`
layer type, so it reuses the whole form pipeline (FormLayerManager show/hide,
pause, skip button, analytics, timeline markers) with no new JS manager.

It differs from the other form integrations in three ways:

* No WordPress plugin to activate. Availability is instead "enabled with a
  Pod ID and Munchkin ID configured", exposed as `videoData.marketoActive`
  and `godamPluginDependencies.marketo`.
* No form-list API, so the editor takes the Marketo Form ID as a plain text
  input rather than the FormSelector combobox.
* No shortcode. The player template emits an empty `<form id="mktoForm_<id>">`
  and Marketo_Form_Assets enqueues the instance's forms2 library plus an init
  script that calls `MktoForms2.loadForm()` — mirroring the Jetpack
  integration, which also renders markup directly and enqueues its own assets.

The Pod ID accepts either Marketo instance form (`app-sj04.marketo.com` or
`123-ABC-456.mktoweb.com`), with or without a scheme; a bare identifier with
no dot falls back to `<id>.mktoweb.com`. Marketo::get_instance_host()
normalizes it once so the editor preview and the frontend always target the
same instance.

The editor renders a live preview of the form over the video as soon as a
form ID is entered (debounced), using the same forms2 path as the frontend.

Settings live under GoDAM Settings > Integrations > Marketo and persist via
the existing `godam_default_integration_settings` and
`godam_sanitize_integration_settings` filters.

The reCAPTCHA site key is stored but not yet applied; wiring up Google
reCAPTCHA v3 is follow-up work.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* feat(marketo): manage the Marketo connection from the integration tab

Make GoDAM Central the source of truth for the Marketo connection and let a WordPress admin manage it from Settings > Integrations > Marketo.

Add a server-to-server REST proxy (godam/v1/marketo/status|forms|connect|test) that calls Central's map_wordpress API with the licence key in X-API-KEY. The browser never sees the licence key or the Marketo credentials.

When the integration is on, the tab reads the connection status from Central and shows one of two things: a short form to connect Marketo directly from WordPress when it is not connected, or the connection details read-only when it is. The Munchkin ID alone connects for forms; the Client ID and Client Secret are optional and add the form list and video-view tracking. Disconnect and the tracking switch stay on Central, which the tab links out to.

Form rendering now takes the Munchkin ID, forms host and reCAPTCHA key from Central, falling back to the local values so existing sites keep working.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(marketo): only render the form layer when Marketo is connected on Central

A Marketo form layer rendered on videos, and previewed in the editor, even when Marketo was not connected or the integration was turned off. The form showed because the plugin fell back to a Pod/Munchkin ID left in local settings by the earlier version, so forms2 rendered with a stale account.

GoDAM Central is the source of truth, so make the Marketo account details (Munchkin ID, forms host, reCAPTCHA key) and Marketo::is_configured() read only from Central's connection; nothing stored locally can make a form render. The player template now skips the Marketo layer unless Marketo::is_configured(), so a disconnected or disabled Marketo no longer shows an empty form overlay on the video.

A successful connect from the plugin now force-refreshes the cached verify response, so a new connection shows on videos right away instead of after the hourly cache expiry.

Update the editor warning and preview text to the Central based wording.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* style(marketo): primary/secondary action buttons on the connected card

Show Manage on GoDAM as a primary button (opening GoDAM in a new tab) and Test connection as a secondary button, replacing the plain links. Manage on GoDAM keeps an external-link affordance using the external icon from @wordpress/icons rather than a dashicon.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(settings): remember the open settings tab and integration sub-tab in the URL

The settings page reset to the first tab on reload. Persist the active top-level tab as a settings_tab query param and the active integration sub-tab as an integration_tab query param, both written without reloading or adding history, so a reload or a shared link reopens the same tab.

The top-level tabs previously used the URL hash; that still works as a fallback (the plugin links to tabs by hash in several places) and is normalized to the query param on load.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat: add HubSpot form layer to the video editor and player

Add a native HubSpot form layer, gated on the HubSpot (leadin) plugin
being active and connected to a HubSpot account.

- Editor: HubSpot provider in the form-layer registry and add-layer menu,
  with a form picker populated from the connected account (new
  godam/v1/hubspot-forms REST proxy) and a live form preview.
- REST: Hubspot_Forms lists the account's forms using a short-lived
  access token obtained from leadin's own endpoint.
- Frontend: render the form on the page via leadin's [hubspot] shortcode
  and register HubSpot in the player's form-type dependency map.

* fix: HubSpot form layer editor preview, timeline chip, and fullscreen layout

- Editor preview: render into a callback-ref target so it survives the
  asynchronous LayerControls portal mount and re-renders when switching
  layers; clear the loading state via a MutationObserver since HubSpot's
  onFormReady is not reliably delivered; drop the irrelevant AJAX-submission
  notice for HubSpot.
- Timeline: show the HubSpot icon and name on the marker chip instead of
  the generic form icon and "Gravity Forms".
- Frontend: lay the HubSpot form out at a centred width in fullscreen so its
  iframe no longer collapses in the fit-content form container.

* feat(hubspot): connect HubSpot forms through GoDAM Central

Add a GoDAM Central connection for HubSpot forms alongside the existing
HubSpot plugin (leadin) path, and let the admin choose the source on the
Settings > Integrations > HubSpot tab.

- Extract a shared Central_Client trait for the server to server GoDAM
  Central calls, used by both the Marketo and HubSpot integrations.
- Add a Hubspot config class that reads the cached map block, resolves the
  active source (central or leadin) from the admin choice and whether it is
  connected, and gates rendering. When the HubSpot plugin is not active,
  Central is the only method.
- Add hubspot status and forms REST routes that proxy GoDAM Central, and
  make the editor form list route source aware.
- Render Central HubSpot forms with HubSpot forms v2 on the frontend (no
  shortcode); the leadin source keeps its own shortcode.
- Add the HubSpot settings tab with connection method cards, the Connect on
  GoDAM flow with polling, and the leadin status.
- Disable adding and show a clear message on existing HubSpot layers when
  the integration is turned off, and do not render the layer on the
  frontend.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(marketo): keep the viewer on the video after a submit, and address review

Resolve the review on #2218.

- Must fix: register a forms2 onSuccess handler that shows a thank-you in place
  of the form and returns false, so a submit no longer follows Marketo's
  follow-up redirect (which took the viewer off the video / reloaded it). The
  player sees `.godam-marketo-success` and shows Continue, and the submitted
  analytics event fires.
- Should fix: render each layer's form node with a unique id plus
  `data-marketo-form-id`, and load same-id forms one at a time, so the same
  Marketo form on two layers (or a video embedded twice) no longer renders its
  fields twice.
- Drop the local podId / munchkinId / recaptchaKey settings that nothing reads
  since Central became the single source of truth, and fix the comments and the
  class header that still described local fields as a source or fallback.
- Deep-link the editor's "not connected" notice straight to the Marketo tab.
- Use the `n.e.x.t` @SInCE placeholder in the new classes instead of 2.5.0.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(hubspot): address review on the Central form layer

Resolve the review on #2237.

- Should fix: after "Connect on GoDAM", refresh the cached licence response in
  get_status when Central's live portal differs from the cache, so the editor,
  the form picker and the player see the new connection right away instead of
  after the hourly refresh.
- Should fix: detect a submit on an iframe-rendered HubSpot form. Mark the
  container from the forms v2 onFormSubmitted callback and add
  `.godam-hubspot-submitted` to the FormLayerManager confirmation selectors, so
  the player shows Continue and sends the submitted event even though the
  success node lives inside the form's iframe.
- Validate the hubspot_id shape in the player branch (a form GUID or a numeric
  id), so a value containing `]` cannot close the [hubspot] shortcode early.
- Settings tab: poll get_status silently so a background re-check no longer
  blanks the panel, and re-check on window focus so connecting after the polling
  window still updates without a reload. Add a Contact support link to the error
  notice and fix "hubSpot" to "HubSpot".
- Editor copy now uses the chosen source (not the active source, which is empty
  when nothing is connected), so the "how to connect" notice and the add-layer
  tooltip tell the admin what actually unblocks the layer. Deep-link to the
  HubSpot tab.
- Drop the unused GET /godam/v1/hubspot/forms route; the editor uses
  /hubspot-forms. Use the n.e.x.t @SInCE placeholder in the new files.
- Also fix the integration sub-tabs bottom border, which targeted a wrapper
  class that is not rendered.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* test(marketo): cover the Central config reader and the call_central client

Add PHPUnit coverage requested in the review of #2218:

- MarketoConfigTest: Marketo::get_instance_host() across the hosts it accepts
  (bare host, scheme + path, bare id, Munchkin fallback, empty), is_configured()
  and is_central_connected(), driven by a stubbed rtgodam_user_data option.
- MarketoCentralClientTest: the error mapping in call_central() -- no licence
  key, an error body (message + error_type), a success that unwraps Frappe's
  envelope, and a transport failure. Runs in a separate process so its HTTP /
  site-url stubs do not collide with the resilience test's globals.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* test(hubspot): cover source resolution and the settings whitelist

Add PHPUnit coverage requested in the review of #2237:

- HubspotConfigTest: Hubspot::get_active_source() across the cases (Central
  chosen and connected or not, leadin chosen and active/connected, the leadin
  choice collapsing to Central when the plugin is inactive), the effective portal
  and region, is_configured()/is_enabled(), and the sanitize_integration_settings
  whitelist (valid source kept, unknown source and keys dropped, enable coerced).
- Load the Central_Client trait in MarketoCentralClientTest when present, since
  Marketo_Integration takes call_central from it on this branch.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(marketo): keep loading forms when one cannot be returned

forms2 calls loadForm's callback with null when Marketo cannot return a
form -- an unknown, deleted or unapproved form ID, or a subscription
mismatch. `form.onSuccess` then threw, so the two statements after it
never ran: the node kept the borrowed `mktoForm_<id>` id and `loadNext`
was never called, leaving every Marketo form after it on the page empty,
including layers on other videos.

Guard the callback so the id is restored and the queue continues either
way; only the success handler is skipped for a form that failed to load.

Also build the confirmation message as a text node. It was interpolated
into an HTML string, so a translated `successText` containing `<` or `&`
was parsed as markup.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(hubspot): address review — Yoda check, arrow alignment, support link

- Yoda condition in Hubspot_Integration::get_status (clears the PHPCS error
  and the rtBot "Changes requested" gate)
- realign the HubSpot videoData block's => to the longer hubspotChosenSource key
  (diff lines only; older rows left untouched, as requested in review)
- add a Contact support link to the "Could not reach" warning on the settings tab

Pre-commit hook bypassed: full-file PHPCS flags pre-existing arrow alignment on
unchanged rows (not in the diff, so CI ignores them), and lint:js OOMs across
files; the changed files pass phpcs/eslint individually.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* style(marketo): align the videoData array arrows flagged by PHPCS

Add one space before => on the five rows rtBot flagged
(everestFormsActive, ninjaFormsActive, metformActive, marketoActive,
marketoConfig) so they line up with the array's longest key. Only the
rows in this PR's diff are touched; older untouched rows are left as-is.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>

* feat(tracking): send video view milestones to Marketo and HubSpot (#2226)

* feat(tracking): send video view milestones to Marketo and HubSpot

When a visitor the customer's marketing automation platform already knows
watches a GoDAM video, that tool should record how far they got. This adds
the WordPress half of that path (slice S4).

A new footer script watches every player, works out the highest milestone
reached (0 = started, then 25/50/75/100) from the played ranges rather than
the playhead, and posts it with the visitor's `_mkto_trk` and/or `hubspotutk`
cookie to a new public route, `godam/v1/map-events`. The route validates the
shape to the same rules godam-core applies, then schedules a server-to-server
forward with the licence key in `X-Api-Key`, following the existing
`log_media_view` pattern — so no Marketo or HubSpot credential lives on the
site, and the GoDAM licence key never reaches the browser.

Which tools are connected comes from the `map` block of the `verify_api_key`
response, which the plugin already caches hourly, so there is no extra call
to Central. With tracking off, no tool ready, or no marketing cookie on the
page, nothing is enqueued and no request is made.

The route is public because most viewers are logged out and page caching
rules out a usable nonce. It is rate-limited per IP, de-duplicated per view,
and answers `{"status": "ok"}` to everything it accepts and drops alike, so
a prober learns nothing.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(tracking): count only what was played when a video ends

On `ended` the tracker sent every milestone, so a viewer who played one
second and skipped to the end was reported as having watched 25, 50, 75
and 100%. It now measures the played ranges, as `timeupdate` does, and
counts a range that stops within a second of the end as reaching it, for
the gap `timeupdate` leaves there. Only that missing tail is credited, so
a viewer who skipped to the end gains nothing from the slack.

Tests drive a fake player through godamPlayerReady: a skip to the end
posts [0], a full watch stopping 0.3 s short posts all five, and 60%
followed by a skip posts [0,25,50].

Addresses review: #2226 (comment)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(tracking): delete a forward's Action Scheduler row once it finishes

The row's arguments hold the visitor's _mkto_trk / hubspotutk cookie.
Action Scheduler keeps a completed row for 31 days, never purges a failed
one, and shows the arguments under Tools > Scheduled Actions, while the
class docblock said the cookie is never stored.

delete_finished_forward() deletes the row on completed_action,
failed_execution and failed_action, at priority 20 so it runs after
Action Scheduler's logger and the delete clears that log line too. Every
other action is left alone, and a row already deleted is not an error.
The docblock now says where the cookie is written down and for how long.

Addresses review: #2226 (comment)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(tracking): delete a forward's row after a PHP fatal too

A fatal during the forward (out of memory, or another plugin's HTTP
filter fataling) skips completed_action, failed_execution and
failed_action. Action Scheduler's FatalErrorMonitor marks the row failed
at shutdown and fires only action_scheduler_unexpected_shutdown, and
failed rows are never purged, so that row kept the visitor's cookie for
good. delete_finished_forward() now also runs on that hook, at the same
priority 20, after the logger. The store is still usable at that point:
Action Scheduler itself calls mark_failure() and release_claim() there.

Addresses review: #2226 (comment)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* test(tracking): cover the map-events route and the forward's retries

MapEventsTrackTest drives Map_Events::track() with a cached map block:
only each connected tool's cookie is forwarded, nothing is queued while
tracking is off or for a request of the wrong shape, each (media,
session, milestone) is queued once, one IP is rate-limited at
RATE_LIMIT, and every outcome answers {"status": "ok"}.

MapTrackerRetryTest drives async_map_event() and maybe_retry(): the
forward carries the licence key and the event, a network error, 429 or
5xx is retried with 1/2/4/8-minute back-off, a 2xx or 4xx is not, and
nothing follows the last attempt.

Addresses review: #2226 (review)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(tracking): read the visitor cookie at send time, not at setup

trackPlayer() read _mkto_trk/hubspotutk once while setting a player up and
returned for good when neither was there. Both tracking scripts are
third-party and async, so on a visitor's first page view the cookie is
usually written after the player is ready: the tracker script was in place
~600ms into a load on preprod, while HubSpot's cookie landed ~1.2s after its
own tag, two hops to another origin. The whole first view was dropped, which
is the view that matters most - someone arriving from an ad on a landing page.
Every later visit worked, since the cookie then already existed, so this only
ever showed up as "the first one never arrives".

report() now reads the cookie itself and keeps it once found, so a known
visitor still costs one read per view. A milestone that cannot be sent is not
added to `sent`, and both handlers already re-report every milestone passed,
so a cookie arriving mid-view carries the whole view rather than the
remainder.

The session ID is minted on the first send instead of during setup. Without
that, moving the read would have started writing a sessionStorage key for
every anonymous viewer - the early return used to be what prevented it.

Verified on preprod against a real player: with the page left to settle for
5s with no cookie, nothing is sent and no session key is written; setting the
cookie and then playing posts 0/25/50/75/100 in five requests under one
session ID.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

* fix(tracking): align HubSpot view tracking to Central's portal and honour opt-outs

View milestones go through Central's HubSpot connection, so a milestone can only
land on a contact when the visitor's hubspotutk was written by tracking code for
Central's portal. Three gaps made HubSpot show "Unknown Contact":

- A leadin connection to a different portal writes the cookie for its own portal.
  Drop HubSpot from the tracked providers on a portal mismatch (Marketo is
  unaffected), and skip enqueuing when nothing is left to track.
- A Central-only site has no HubSpot tracking code at all, so no hubspotutk cookie
  is ever written. Inject HubSpot's analytics loader for Central's portal when a
  connected leadin is not already loading it; filterable via
  rtgodam_map_inject_hubspot_tracking_code.
- The tracker read the cookie regardless of consent. Honour Global Privacy Control
  (suppresses every tool) and HubSpot's own __hs_opt_out / __hs_do_not_track
  (suppresses HubSpot) before collecting any visitor ID.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(hubspot): render Central forms via HubSpot's new per-portal embed

A new-generation HubSpot form rendered through the legacy forms v2 library
(`hbspt.forms.create()`) loads with its script loader disabled, so no tracking
code runs: the `hubspotutk` cookie is never set, the submission is anonymous,
and every forwarded view milestone arrives as "Unknown Contact".

Switch the Central source to HubSpot's current embed instead — a single
per-portal `/forms/embed/<portal>.js` loader plus a `.hs-form-frame` element.
That embed loads HubSpot's tracking code itself, so it writes the cookie and
links both the form submission and later video-view milestones to the contact.

- Hubspot::central_forms_embed_src(): the new per-portal embed URL.
- Hubspot_Form_Assets: enqueue the embed in place of forms v2.js.
- hubspot-form.js: drop the hbspt.forms.create loader; it is now just a submit
  bridge that marks the layer on `hs-form-event:on-submission:success` (and the
  legacy `hsFormCallback` message) for FormLayerManager's confirmation check.
- player template: render the `.hs-form-frame` embed markup.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(tracking): stop map events for a provider disabled in settings

Central's `map` block lists every tool connected on Central, but the site can
switch Marketo or HubSpot off independently in Settings → Integrations. That
local switch was ignored for view tracking: milestones kept being sent (and the
HubSpot tracking code kept being injected) for a provider the admin had turned
off.

Honour the switch at the single chokepoint the whole feature funnels through,
Map_Tracker::frontend_providers(): drop Marketo when !Marketo::is_enabled() and
HubSpot when !Hubspot::is_enabled() (on top of the existing portal-alignment
drop). maybe_enqueue_script() already bails when the list is empty, so with both
off nothing is enqueued and no event is sent.

- Marketo::is_enabled(): new, mirroring Hubspot::is_enabled(); is_configured()
  now builds on it.
- MapTrackerProvidersTest: cover Marketo-off, HubSpot-off and both-off; load the
  Marketo class in the bootstrap.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(tracking): gate the map-events route on frontend_providers; correct comments

Addresses PR #2226 re-review (subodhr258).

Should-fix: the map-events REST route filtered the forwarded cookies against
Central's raw `providers`, not Map_Tracker::frontend_providers(). A full-page-
cached page keeps the old inline `godamMapTracking` and script after an admin
switches a tool off, so it kept POSTing that tool's cookie and the route
forwarded it; a direct POST did the same. Route the cookie-name filter through
frontend_providers() so a disabled provider — and a HubSpot portal mismatch —
is dropped server-side too.

Also reword the embed comments that overclaimed: HubSpot's per-portal
forms/embed/<portal>.js renders the form (in a same-origin srcdoc iframe) and
fires the V4 submit event, but it does not load the tracking code or set the
`hubspotutk` cookie. That cookie comes from the tracking code the plugin injects
separately for a Central-only site (maybe_enqueue_hubspot_tracking_code), which
is what links a submission (and view milestones) to a contact.

- class-map-events.php: cookie_names_for_connected_providers() via frontend_providers().
- MapEventsTrackTest: a disabled provider's cookie is not forwarded; both off
  forwards nothing. connect() now sets a Central HubSpot portal so alignment holds.
- bootstrap: load the HubSpot class for the route's provider filter.
- class-hubspot.php, class-hubspot-form-assets.php, hubspot-form.js, template:
  accurate comments on what the embed does vs. what sets the cookie.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* docs: disclose the HubSpot embed and tracking code in External Services

Per PR #2226 re-review: the plugin loads HubSpot's own scripts on the frontend
(the forms embed to render a Central HubSpot form, and the tracking code to set
the `hubspotutk` cookie for view tracking), and the tracking code pulls in
HubSpot's wider suite for the portal. Document what loads, when, and the
`rtgodam_map_inject_hubspot_tracking_code` off switch, in both readme.txt and
README.md.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(hubspot): render Central forms with forms v2 to support legacy + new editors

The per-portal V4 embed (423da58) has no legacy path, so a Central layer using a
legacy-editor form rendered blank. Central lists both editors' forms (`{id,name}`,
no type), so the renderer has to handle either.

Go back to HubSpot's forms v2 library, which renders a legacy-editor form itself
and switches a new-editor form to the V4 embed. Keep the V4 submit listener from
423da58: the bridge now marks the layer on both the legacy `onFormSubmitted`
callback and the new-editor `hs-form-event:on-submission:success` window event, so
the player's confirmation works for either form type. (This is the fix the first
re-review suggested: v2 for rendering + the hs-form-event listener.)

The `hubspotutk` cookie that links a submission to a contact is unchanged — it
comes from the tracking code injected by Map_Tracker (896073f), not the form
library.

- class-hubspot-form-assets.php: enqueue forms/v2.js again.
- hubspot-form.js: hbspt.forms.create() render loop + onFormSubmitted, plus the
  global hs-form-event listener for V4 forms.
- template: the v2 render container (.hubspot-form with data attributes).
- class-hubspot.php / HubspotConfigTest: drop the now-dead central_forms_embed_src().

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* docs(hubspot): note the injected tracking code on the HubSpot settings tab

Per PR #2226 re-review: admins see the settings tab, not the readme. Under the
GoDAM Central connection, add a line that with video-view tracking on, GoDAM adds
HubSpot's tracking code to pages with a video so views match a contact — and that
the code also loads whatever the portal has set up (chat, pop-ups, ads pixel,
cookie banner). Complements the readme External Services entry (8555d66).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>

* [Develop] Fix 13 Dependabot alerts and 4 Garuda findings: DOMPurify 3.4.17, notice dismissals, SureForms recorder upload (#2255)

* Bump DOMPurify, follow-redirects and dev tools for Dependabot security alerts

* Check the capability on notice dismissals, and take only audio or video from the SureForms recorder field

Fixes Garuda findings 21-10-2026 #9, #10, #11 and #18 (and L10, L30, L31).

* Bump qs to 6.16.1 for Dependabot alert 396

The qs override moves from 6.16.0 to 6.16.1, the release that fixes GHSA-6pq8-jhm2-2vcw (qs.stringify with arrayFormat 'comma' throws on elements without a callable toString). qs is only used by express inside the webpack dev server, so the built plugin files do not change.

* fix(blocks): keep GoDAM media blocks robust in flex & narrow containers (#2234)

* fix(pdf-block): fill container width so flex layouts don't collapse

The Document block root (.wp-block-godam-pdf) declared no width, so when
placed inside a core Group using a Row or Stack (flex) layout it became a
flex item with the default `flex: 0 1 auto` and collapsed to its
min-content width while .godam-pdf-wrapper's fixed pixel height stayed —
rendering a tall, narrow, empty box. Setting width: 100% keeps flex
parents consistent with the flow/grid/columns layouts.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(pdf-block): make card view respond to container width, not viewport

The card view stacked (cover on top) only at a <=480px viewport via a media
query, so a card placed in a narrow column or grid cell on a wide screen kept
the side-by-side layout and truncated its title. Make .godam-pdf-card-wrapper
a size-query container (container-type: inline-size) and switch the stacking
rule to @container, so the card adapts to the width its parent gives it
regardless of viewport size.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* feat(pdf-block): mirror card container-query stacking in the block editor

The editor renders the card view from its own markup (.godam-pdf-preview--card)
and editor.scss, which never stacked the card and keyed its few responsive
tweaks off the viewport. Make .godam-pdf-preview--card a size-query container and
move its responsive rules into a single @container stack block (cover on top,
full width, full title), so the editor preview matches the front end and adapts
to the width the block's parent gives it. Also show the full title when the
front-end card stacks, for parity.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* style(pdf-block): lower card stack breakpoint to 400px container width

Keep the cover beside the text a little longer before stacking, in both the
front-end and editor container queries.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(audio,video-thumbnail): fill container width so flex layouts don't collapse

Same issue fixed for the Document block: the Audio and Video Thumbnail block
roots declared no width, so inside a core Group using a Row/Stack (flex) layout
they became flex items with the default `flex: 0 1 auto` and collapsed to their
min-content width (the audio player shrank to ~50px). Set width: 100% so flex
parents match the flow/grid/columns layouts. Image (intrinsic <img>) and Video
Duration (inline <time>) are unaffected; Video and Gallery already handle flex.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* fix(blocks): fill width only in flex parents, and keep floats out of the card query

A root `width: 100%` on the PDF, Audio and Video Thumbnail blocks also
applied outside Row/Stack groups, where it broke settings that work on
develop: Align: Full width stopped short of the right edge by twice the
root padding, a side margin pushed the block past its column, and a
left- or right-aligned card spanned the content area instead of floating.

The width is now set only under `:where(.is-layout-flex) >`, where the
collapse happens; flow, constrained and grid parents already stretch the
block. The card's size-query container skips `.alignleft` and
`.alignright`, on the front end and in the editor, because inline-size
containment would size a shrink-to-fit float to zero.

Addresses review: #2234 (comment)

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(pdf-block): still stack a floated card on phones

The card's size-query container now skips Align left/right, so a floated
card had no stacking rule left: the 480px media query it used on develop
had been replaced by the container query. Floated cards fall back to
that media query again, on the front end and in the editor.

The stacked layout moves into a godam-pdf-card-stacked mixin in each
file, shared by the container query and the fallback, so the two cannot
drift.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

* Bump version to 2.6.0 and add the v2.6.0 changelog (#2274)

* Bump version to 2.6.0 and add the v2.6.0 changelog

- Add the v2.6.0 entry to CHANGELOG.md and readme.txt (Marketo and
  HubSpot form layers, Marketo/HubSpot video-view tracking, the block
  flex-layout fix, and security fixes), and drop v2.5.0 from readme.txt.
- Update the Stable tag in readme.txt and README.md to 2.6.0.
- Bump the version in the godam.php header and RTGODAM_VERSION.
- Bump the version in package.json and package-lock.json.
- Resolve the @SInCE n.e.x.t markers in the Marketo/HubSpot code to 2.6.0.
- Regenerate languages/godam.pot (now includes the Marketo/HubSpot strings).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* Add the Marketo entry to the readme External Services section

WordPress.org requires every third-party service to be listed. The
Marketo form layer loads Marketo's forms2 library from the account's
instance host to render the form and submit entries to Marketo, so add
a === Marketo === block beside the HubSpot one, noting that video-view
tracking goes through GoDAM Central rather than a client-side script.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

* Set the v2.6.0 changelog date to October 30, 2026

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

* Marketo form layer: pick the form from a list, and handle forms that can't load (#2276)

* fix(marketo): let viewers continue past a form that cannot load, warn on HTTP pages

A Marketo form ID from another instance, or one Marketo cannot return,
rendered an empty layer; with "Allow skip" off the viewer was stuck.
marketo-form.js now replaces it with "This form could not be loaded."
and FormLayerManager shows Continue without recording a submission.
The editor preview shows its error instead of an empty white box.

On HTTP pages forms2 asks for its XDFrame proxy over http://, the
browser upgrades it to HTTPS and forms2 drops every reply, so submits
hang on "Please Wait". The frontend now logs a console warning and the
editor shows a notice (app-*.marketo.com hosts are unaffected).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* feat(marketo): pick the form layer's Marketo form from a list

When the Marketo connection on GoDAM Central has a Client ID and Secret,
the Marketo form layer now shows the same "Select Form" picker as
HubSpot, listing the instance's approved forms. A forms-only connection,
or Central failing to list forms, keeps the typed Form ID field, with a
hint that adding the API credentials turns on the list. A saved form no
longer in the list still shows as "Form <id>".

GET godam/v1/marketo/forms (unused until now) needs upload_files like
the other form-list routes, returns the editor's [{ id, title }] shape,
answers an empty list instead of an error, and skips Central when
Marketo is not configured. The connection routes stay admin-only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* fix(marketo): stop waiting on a Marketo form that never loads

forms2 never calls the loadForm callback when its request fails (blocked
by an ad or privacy blocker, or a network drop), so the layer stayed
empty and the Marketo forms after it on the page never loaded. Each form
now gets a 12-second guard, matching the editor preview: on timeout it
shows "This form could not be loaded." and moves on to the next form.

When forms2.min.js itself never loads, waitForMktoForms2 now marks
every Marketo layer the same way before giving up, so viewers can
continue past them.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* MAP tracking: per-site switch to stop sending video views to Marketo or HubSpot (#2277)

* feat(map): per-site switch to stop tracking video views to Marketo or HubSpot

Tracking was only switchable on GoDAM Central, for every site on the
licence at once. Each tool's integration tab now has "Track video views
on this site" (integrations.<tool>.track_views, on by default so nothing
changes on upgrade), next to Central's own tracking state.

Central queues a view for Marketo only with the visitor's mkto_trk
cookie and for HubSpot only with hubspotutk, so turning the switch off
stops this site sending that cookie: GoDAM Central's setting and other
sites are not changed, Central's switch stays the master, and form
layers are unaffected.

Enforced in Map_Tracker::frontend_providers() (the player tracker, the
HubSpot tracking code injection and the map-events route), and again at
send time so views queued before the switch was turned off are not sent.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* feat(map): load Marketo's Munchkin so Marketo views can be tracked

Marketo view tracking needs the visitor's _mkto_trk cookie, and only
Munchkin sets it: forms2 (the Marketo form layer) only reads it. The
plugin never loaded Munchkin, so on a site without its own Munchkin
code no visitor had the cookie, every map-events payload carried only
hubspotutk, and a Marketo form fill was not tied to the visitor.

When Marketo is a frontend provider for the page (so Central and this
site both track it), Map_Tracker now adds an inline loader for
https://munchkin.marketo.net/munchkin.js that calls Munchkin.init() with
the connection's Munchkin ID. It does nothing when Munchkin is already
on the page or the browser sends Global Privacy Control, and a site
that loads Munchkin itself can turn it off with the
rtgodam_map_inject_munchkin filter. This mirrors the existing HubSpot
tracking code injection.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* docs(map): say on the Marketo tab that GoDAM loads Munchkin, add changelog lines

Sites with Marketo tracking on will start loading Munchkin on video
pages after this update. The Marketo tab now says so under the site
tracking switch, like the HubSpot tab does for HubSpot's tracking code:
it is added to pages with a video, it also records page visits in
Marketo, and the rtgodam_map_inject_munchkin filter turns it off.

The v2.6.0 changelog (CHANGELOG.md and readme.txt) gains lines for the
per-site tracking switch and for Munchkin.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

* fix(map): mint the view session ID from crypto only, align the editor data array (#2278)

CodeQL (js/insecure-randomness) flagged getSessionId() in map-tracking.js:
when crypto.randomUUID() is missing (it needs a secure context, so any
plain-HTTP page), the ID fell back to Date.now() plus Math.random(). The
ID only groups one view's milestones for de-duplication and grants no
access, but the fallback now uses crypto.getRandomValues() (16 bytes as
32 hex characters), which HTTP pages have too. With no crypto API at all
the tracker sends nothing rather than a predictable ID.

Also realigns the double arrows of the video editor's localized data
array in class-pages.php (whitespace only), which rtBot's phpcs review
flagged on the v2.6.0 sync PR.

Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Subodh Rajpopat <subodh.rajpopat@rtcamp.com>
Co-authored-by: KMchaudhary <kuldipkumar.chaudhary@rtcamp.com>
Co-authored-by: Claude Opus 5 <noreply@anthropic.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

3 participants