[Bug] image in readme would crash when default branch is not master

[dongxuwang]

Current behavior

image in readme would crash when default branch is not master

see project logo in here

Expected Behavior

image in readme should display

Extra comments

Should tune Client.scala rather hard code master

Search terms

master or blob in Client.scala

[adpi2]

Thanks for the bug report.

[dongxuwang]

After looking through the code, looks that's not easy to figure this out.

[adpi2]

I guess we need to use the GithubClient to get the default branch and store that into GithubInfo (which is saved in a SQL table). Then use that info to get the correct URL of images.

If you have time to look into that, it would be helpful.

[mfirry]

what if just before this (or in parallel) call to github api

    val request = new Request(
      s"https://api.github.com/repos/$organization/$repository/readme",
      new RequestInit {
        headers = headersWithCreds.toJSDictionary
      }
    )

One makes this other one https://api.github.com/repos/$organization/$repository/ with which you can get the default_branch?

[adpi2]

One makes this other one https://api.github.com/repos/$organization/$repository/ with which you can get the default_branch?

Yes that should solve the issue! Feel free to open a PR.

[mfirry]

I'll give it a shot!

to be fair: it's not the solution... I see it as a workaround but it should a) serve the purpose b) have little impact on the structure of that part of the codebase.

[mfirry]

I'll try and summarize a possible solution.

As of now def fetchAndReplaceReadme(element: Element, ... does few things:

1. It retrives the readme information

    val request = new Request(
      s"https://api.github.com/repos/$organization/$repository/readme", ...)

2. Updates the markup for the given element (element.innerHTML = res)
3. Checks all img and a elements and for them checks whether they have href or src attributes and then sets them to some specific values that can be github.com/$organization/$repository/raw/$branch or github.com/$organization/$repository/blob/$branch (and branch is master)

Now what I had in mind is to keep point 1 and 2 as they and separate the rest.

So the first part would look like:

    // Simplified version
    fetch(readmeRequest).toFuture
      .flatMap { res =>
        res.text().toFuture
      }
      .foreach(res => element.innerHTML = res)

The second one would make the "new" GitHub API request (repoRequest), parse the JSON and retrieve the default_branch. Only then It can go on with updating href attributes for the anchors and the images.

So something along these lines:

      // Simplified version
      @js.native
        trait Repo extends js.Object {
        val default_branch: String = js.native
      }

(...)

      val repoRequest = new Request(
        s"https://api.github.com/repos/$organization/$repository/", ...)

      fetch(repoRequest).toFuture
        .flatMap(res => res.text().toFuture)
          .foreach { res =>
            val resJson = js.JSON.parse(res)
            val branch = resJson.asInstanceOf[Repo].default_branch
           setImagesAndAnchors(branch, organization, repository)
        }

(And obviously setImagesAndAnchors does what it says it does)

How does this look like?

[adpi2]

How does this look like?

Looks good but the second part (or more precisely the setImagesAndAnchors) should not happen before the first part finishes. So there should be some map or flatMap to synchronize them.

[mfirry]

Cool. I'm just wondering if I've got an easy way to test that this works. Any hint?

[adpi2]

Cool. I'm just wondering if I've got an easy way to test that this works. Any hint?

This is how you can run Scaladex locally: https://github.com/scalacenter/scaladex/blob/main/CONTRIBUTING.md#run-scaladex-locally. Then you should go to the scalacenter/bloop project and see if the image appears.

[mfirry]

Working on it have a look mfirry@d960fa3

still trying to understand if can i somehow handle unsuccessful API calls

[adpi2]

@mfirry thanks for the update. It is looking good.

still trying to understand if can i somehow handle unsuccessful API calls

The API call should not fail and if it fails I would use master as the default branch.