[Bug]: Long step names prevent clicking on step attachment link

[bdcarr]

Version

1.57.0

Steps to reproduce

Clone repo at https://github.com/bdcarr/playwright-attachment-link-bug and follow the instructions in the readme to generate a HTML report demonstrating the issue.

Expected behavior

You can click on the attachment icon and have it take you to the relevant attachment, regardless of how long the step name is.

Actual behavior

If the step name + location is long enough to reach the attachment icon, something blocks your click, so you can't jump to the attachment from the test step.

Screen.Recording.2025-12-12.at.3.37.41.pm.mov

The example I've given only shows the step name being long, but I also tested with a name that is just long enough to push the step location (e.g. "example.spec.js:5") into the attachment link. You can fiddle around with this just by editing the DOM while viewing the report.

Additional context

No response

Environment

System:
    OS: macOS 15.7.2
    CPU: (12) arm64 Apple M2 Pro
    Memory: 132.64 MB / 32.00 GB
  Binaries:
    Node: 20.19.4 - /Users/brycecarr/.local/state/fnm_multishells/26667_1764804574151/bin/node
    npm: 10.8.2 - /Users/brycecarr/.local/state/fnm_multishells/26667_1764804574151/bin/npm
    pnpm: 10.8.1 - /Users/brycecarr/Library/pnpm/pnpm
    Deno: 2.5.6 - /opt/homebrew/bin/deno
  IDEs:
    VSCode: 1.107.0 - /usr/local/bin/code
    Cursor: 2.0.77 - /usr/local/bin/cursor
  Languages:
    Bash: 5.3.8 - /opt/homebrew/bin/bash
  npmPackages:
    @playwright/test: ^1.57.0 => 1.57.0

[Skn0tt]

What a bug! I'll fix it. I'm curious, what kind of step names do you have in your projects? 😅

[bdcarr]

"Step five, in which the user audaciously creates several timesheets in quick succession, with minor variations [see TIME-7639 in jira to ascertain the necessity of this section of the test]"

Jk, I hacked in logs for axios requests by creating test steps with names set to the request/response body. Not the best solution but step attachments weren't available at the time. I hope to have time to see about using APIRequestContext instead of axios as the HTTP client. It looks like it'll be painful though.