False positive on F1 rule with a derived table

[MartinGuindon]

LAMS is reporting a false positive on the F1 rule for a derived table using parameters.

Detailed error: linkback references another view, if linkback_type, via {% if linkback_type._parameter_value == "Client" %}

The full dimension code is:

  dimension: linkback {
    sql: TRUE ;;
    html:
      {% if linkback_type._parameter_value == "Client" %}
      <p>Thank you for logging in!</p>
      <p><a href="https://<redacted>/client/{{ client_id._parameter_value | replace:'$','-' }}">Return to dashboard</a></p>
      {% endif %}
    ;;
  }

[fabio-looker]

FYI, I started working on this for about 30 minutes, but I'm overlooking something really obvious and am failing to correctly set up a failing test so far. Putting the link here for reference, but I'll get back to this later:
https://github.com/looker-open-source/look-at-me-sideways/tree/issue-32/__tests__/dummy-projects/07-issue-32

[tstodden]

👋🏼 I'm experiencing this false positive as well when I use the same parameter_name._parameter_value syntax in views. Upon initial investigation it seemed to be specific to dimension and not dimension_group.

[fabio-looker]

Last time I looked at this, I wasn't able to correctly set up a failing test. This time I've taken a look at the F1 code and have an idea where the bug may be.. I won't be able to work on code changes until likely next week, so I'm commenting my findings for now:

The code that should handle ignoring _parameter_value is here: https://github.com/looker-open-source/look-at-me-sideways/blob/master/rules/f1.js#L64

However, the issue seems to be even before that, in the regex, which seems to be way too permissive in terms of what characters it captures (i.e. it will capture the spaces, equal signs, and remainder of the condition as though it were part of the reference)

https://github.com/looker-open-source/look-at-me-sideways/blob/master/rules/f1.js#L47

Also, I should do something to look for multiple liquid tags within the block. Currently, it doesn't seem to be a global match or otherwise loop to catch multiple instances of liquid.

[fabio-looker]

Btw, @tstodden - In the short-term, if you're not yet adding an exemption to avoid this, I would recommend that as a temporary workaround. Thanks for the additional bug report!