fix: previews for unpublished pages

[dopry]

applies to #275

I couldn't find any tests for the token calls and don't have a clue where to start generating them.

[dopry]

I am testing with the following graphql query, we'll need to setup some sort of similar test scenario. note in both scenarios my frontend preview code has to copy the id or page_type out of the token to pass to the graphql query. I feel like maybe I should go deeper and just have the token take precendence over everything else if it is present.

{
  previewUnpublishedSavedDraft: page(
    id: 367
    token: "id=367:1p05oi:jDIi4YDGXs-6YgoOVz05HgyWrEwnOEoP3PTsEEF8RD4"
  ) {
    title
  }
  
  previewUnsaved: page(
    contentType: "typenetwork.TnPage"
    token: "parent_id=367;page_type=typenetwork.TnPage:1p05wC:jJsi2R0bD_IOHDMbZ4kKc34lH2myylKY_vGQ7_T_Hlk"
  ) {
    title
  }
  
  page(id: 367) {
    title
  }
}

[dopry]

I force pushed an update where token takes priority over all other args is present and the params required for fulfillment of a token query are extracted from the token.

My update test graphql looks more like

{
  "data": {
    "previewUnpublishedSavedDraft": {
      "title": "Not Published"
    },
    "previewUnsaved": {
      "title": "Title"
    },
    "page": null
  }
}

I still need help figuring out how to set the test scenarios for 2 token types (saved and unpublished, unsaved and unpublished), all the published scenarios should be covered by existing tests. We may also want to add tests to ensure unpublished content isn't accessible without the token. I'm not sure that those tests should necessarily hold up this PR since we don't already have coverage on those feature and this is a bug that could be impacting production sites.

[zerolab]

@dopry do you still need help with generating a token for new pages?

https://github.com/torchbox/wagtail-headless-preview/blob/main/src/wagtail_headless_preview/models.py#L75-L80

unpublished pages are no more special than others. You can just do token = PageModelWithPreviewMixin.create_page_preview().token

[zerolab]

get_page_from_preview_token comes for wagtail-headless-preview and is Wagtail-versiona agnostic

[dopry]

So this is really just a check to see if headless preview is installed and enabled?

[dopry]

would we be better served with an isHeadlessPreviewEnabled() function?

[zerolab]

parent_id is added for convenience. if I recall correctly we needed to vary things a bit when generating the token. When you have an ID that gives enough to go on.

[dopry]

we could eliminate the if id: code path here if the page_type was always sent. It might not be as 'fast', but it would reduce the amount of code we need to maintain.

[dopry]

I'll try to get another pass on this by EOM.

[dopry]

@zerolab I got the tests setup. I'm not happy with how I'm mocking the unsaved page as per my haiku, but it seems to work. I feel like wagtail-headless-preview is actually a better place for much of this code. I'll try to make a PR against it in the next few weeks to improve the DX here a little and for anyone else building on it.

[zerolab]

uh-oh, I missed this one for 0.19. Will review on Friday and make 0.19.1 release 🙈

[dopry]

I was just about to ask. :) I'd love to see previews fixed. At some point in the future I'll try to take a pass on wagtail-headless-preview, but I'd rather those concerns not block this now.

[zerolab]

@dopry https://github.com/torchbox/wagtail-grapple/releases/tag/v0.19.1

[dopry]

🎉 hooray 🎉

[zerolab]

Thank you for your PRs, much appreciated!