When merging a large amount of segments together we were doing a tremendous amount of allocation especially related to strings due to concatination.
This adds an internal SegmentBuilder that uses a stringbuilder for building up the text rather than creating a new instance and doing a concat operation for each segment.
This makes it possible for encoders to output better representation
of the actual objects instead of working with chopped up segments.
* IAnsiConsole.Write now takes an IRenderable instead of segments
* Calculating cell width does no longer require a render context
* Removed RenderContext.LegacyConsole
* Removed RenderContext.Encoding
* Added Capabilities.Unicode
In both of these loops context is captured preventing caching of the lambda. this results in a pretty significant amount of allocations especially with progress bars that constantly are remeasuring
In Segment.Split, we didn't take cell width into account
when calculating the offset, which resulted in some "fun" bugs.
I've added a new overload for Segment.Split and obsoleted the old one.
Closes#150
The bug might occur if there are wide characters such as emojis
at the end of a line. The SplitLines method mixed cell width
and text length, which might give incorrect results. This commit
makes sure that comparison and calculation is done using cell width
where it's appropriate.
This is a temporary fix for undrawable tables until we've
implemented a proper strategy. What this does is that it replaces
an undrawable table with an ellipsis (...). This should only
occur in either super big tables or deeply nested tables in a
console with a small buffer width.
This commit adds support for recording console output
as well as exporting it to either text or HTML. A user can
also provide their own encoder if they wish.
