Underline styles and colors - core part

[jerch]

PR to support additional underline styles and colors in AttributeData. Implements core parts of #1145.

Note that this PR extends AttributeData in a generic way, which can be use to store other data later on (like #1134).

Changes to the internal interface:

interface IAttributeData {
  ...
  // whether the attr object has any extended attrs defined
  hasExtendedAttrs(): number;
  // update internal HAS_EXTENDED flag
  updateExtended(): void;
  // transparently either returns FG color or underline color if set
  getUnderlineColor(): number;
  // color mode accessors
  getUnderlineColorMode(): number;
  isUnderlineColorRGB(): boolean;
  isUnderlineColorPalette(): boolean;
  isUnderlineColorDefault(): boolean;
  // actual underline
  getUnderlineStyle(): UnderlineStyle;
}

The color and the color mode accessors transparently map to the FG variants, if no independent color was set or underline is unset. Thus its important to test beforehand for isUnderline.

TODO:

• handle extended SGR 4 sequences
• handle SGR 58 and SGR 59 sequences
• extend AttributeData to expose extended attributes
• extend BufferLine to hold and export extended attributes
• test cases for all

[Tyriar]

Looks good after the minor issues are resolved 👍, next PR would be adding rendering?

[Tyriar]

Just to clarify, all underline styles will still result in the isUnderline public API being non-zero?

[jerch]

The idea here for renderer side is to first test for isUnderline, if that evals to true, you'd have to request the style and the color to finally draw it. This way it is compatible to our current code, which only relies on a isUnderline test and always draws a single line.

The color methods cannot be used to determine whether underline is in place, as they would always report a color (fg color). The style method can be used for that, yes.

[Tyriar]

So the approximate memory increase per cell will be 1 pointer for this, and then the standard ExtendedAttrs object will get shared between all cells?

[jerch]

But only for cells, that actually have extended attribs set. All other (thus most cells) dont use more memory in the buffer. To get this working I used another bit in bg as indicator whether a cell holds extended attrs (HAS_EXTENDED flag).

[jerch]

Looks good after the minor issues are resolved +1, next PR would be adding rendering?

Yepp. But thats non trivial, as you already might have guessed from the comments in #1145. Gonna fix the issues above in the next couple of days, but prolly wont make it to the renderer parts 😞

[Tyriar]

@jerch no worries, we should still merge to avoid conflicts piling up