Improve performance for CssClass extensions

[MikeAlhayek]

@sebastienros let me know if you spot a better way to optimize the code any further.

@lampersky I changed the two methods you added, if you are using them in your project, this may break your code so please rename GetPartWrapperCssClasses to GetPartWrapperClasses and GetFieldWrapperCssClasses to GetFieldWrapperClasses

[sebastienros]

Should be two calls to AppendHtml, would prevent the new string allocation (concatenation). The HtmlString space can be static. The value needs to be added with Append(string) or some user entered information (content type) could leak unenccoded.

[MikeAlhayek]

@sebastienros we could do this

builder.AppendHtml(new HtmlString(' '))); // this would be static yes
builder.AppendHtml(new HtmlString(value));

The reason I do no want to call Append(value) is because it performs string.IsNullOrEmpty which I was trying to avoid since we already know it contains a value.

[sebastienros]

Append will encode, and encoding is required here.

[MikeAlhayek]

@sebastienros append does not encode. it just checks if the value is not empty.

    public IHtmlContentBuilder Append(string? unencoded)
    {
        if (!string.IsNullOrEmpty(unencoded))
        {
            Entries.Add(unencoded);
        }

        return this;
    }

the WriteTo method handles the encoding

[sebastienros]

When the builder is written, anything that was added with Append will be encoded.

https://learn.microsoft.com/en-us/dotnet/api/microsoft.aspnetcore.html.htmlcontentbuilder.append?view=aspnetcore-7.0#microsoft-aspnetcore-html-htmlcontentbuilder-append(system-string)

Appends a [String](https://learn.microsoft.com/en-us/dotnet/api/system.string) value. The value is treated as unencoded as-provided, and will be HTML encoded before writing to output.

[MikeAlhayek]

okay. maybe you are write. The WriteTo on the HtmlString does not make use of the encoder. Strange.

public class HtmlString : IHtmlContent
{
    /// <summary>
    /// An <see cref="HtmlString"/> instance for <see cref="Environment.NewLine"/>.
    /// </summary>
    public static readonly HtmlString NewLine = new HtmlString(Environment.NewLine);

    /// <summary>
    /// An <see cref="HtmlString"/> instance for <see cref="string.Empty"/>.
    /// </summary>
    public static readonly HtmlString Empty = new HtmlString(string.Empty);

    /// <summary>
    /// Creates a new <see cref="HtmlString"/>.
    /// </summary>
    /// <param name="value">The HTML encoded value.</param>
    public HtmlString(string? value)
    {
        Value = value;
    }

    /// <summary>
    /// Gets the HTML encoded value.
    /// </summary>
    public string? Value { get; }

    /// <inheritdoc />
    public void WriteTo(TextWriter writer, HtmlEncoder encoder)
    {
        ArgumentNullException.ThrowIfNull(writer);
        ArgumentNullException.ThrowIfNull(encoder);

        writer.Write(Value);
    }

    /// <inheritdoc />
    public override string ToString()
    {
        return Value ?? string.Empty;
    }
}

[sebastienros]

Strange.

No, by design. This is exactly how HtmlContentBuilder distinguishes between fragments it needs to encode or not.

[MikeAlhayek]

good to know. I updated the code let me see if you spot something else.

[jtkech]

Any IHtmlContent are intended to be already encoded

[MikeAlhayek]

@sebastienros I am using AppendHtml for the single space separator to avoid the need for the extra string.IsNullOrEmpry for the space. I don't think we need to worry about encoding the single space.