Wrong boolean attributes conversion

[stephannv]

ERB Input:

<input required>

Output:

input(required: "")

Expected output:

input(required: true)

[marcoroth]

Thanks for reporting this @stephannv.

I don't think we can really support this, unless we hard code a list of attributes which should get forced to a boolean. Probably something like:

BOOLEAN_ATTRIBUTES = ["required", "disabled", "checked", "hidden", "readonly"]

But on the other hand, the output Phlexing is providing is not wrong, HTML boolean can either be written as:

<input required>

<input required="">

or:

<input required="required">

All of them are equally valid. MDN says:

If a boolean attribute is present, its value is true, and if it's absent, its value is false.
To be clear, the values "true" and "false" are not allowed on boolean attributes. To represent a false value, the attribute has to be omitted altogether.

So I'm not sure if this (and/or the boolean attribute mapping) should be handled here or in Phlex itself. Any thoughts?

[stephannv]

I think it's not necessary to map all the possible boolean attributes, because some js frameworks can create new boolean attributes every day, eg. <v-input clearable>. I think this behavior should be simple: if <tag attribute> then tag(attribute: true), if <tag attribute=""> then tag(attribute: "").

I wrote some tests for this behavior:

  it "should convert boolean attributes properly" do
    html = %(<input required>)

    expected = <<~PHLEX.strip
      input(required: true)
    PHLEX

    assert_phlex_template expected, html
  end

  it "should convert blank attributes properly" do
    html = %(<input required="">)

    expected = <<~PHLEX.strip
      input(required: "")
    PHLEX

    assert_phlex_template expected, html
  end

But Nokogiri #value is always returning "" to <input required> and <input required="">, so I think this can be unfeasible due to Nokogiri behavior.

[marcoroth]

Thanks for providing some test-cases! I think that would make sense.

But Nokogiri #value is always returning "" to and , so I think this can be unfeasible due to Nokogiri behavior.

Yeah, I guess that's the root cause of the issue. I wonder if there's another method inside Nokogiri we might be able to use for that!

[stephannv]

@marcoroth I found a solution, but I don't know if it's the best solution: #120