Apply markup in table cell

[ksss]

In the documentation for methods like test, backtick are used within table cells.
While each parser handles backquotes within table cells, the generator does not.
As a result, the tags are displayed literally without being processed.

To resolve this, I propose that tag expressions should be properly handled within table cells as well.

Before

ri --format=bs test

 Character  |Test
------------|---------------------------------------------------------------
<tt>'<'</tt>|Whether the `mtime` at `path0` is less than that at `path1`.
<tt>'='</tt>|Whether the `mtime` at `path0` is equal to that at `path1`.
<tt>'>'</tt>|Whether the `mtime` at `path0` is greater than that at `path1`.

ri --format=markdown test

 Character  |Test
------------|---------------------------------------------------------------
<tt>'<'</tt>|Whether the `mtime` at `path0` is less than that at `path1`.
<tt>'='</tt>|Whether the `mtime` at `path0` is equal to that at `path1`.
<tt>'>'</tt>|Whether the `mtime` at `path0` is greater than that at `path1`.

ri --format=ansi

After

ri --format=bs test

Character|Test
---------|---------------------------------------------------------------
   '<'   |Whether the `mtime` at `path0` is less than that at `path1`.
   '='   |Whether the `mtime` at `path0` is equal to that at `path1`.
   '>'   |Whether the `mtime` at `path0` is greater than that at `path1`.

ri --format=markdown test

Character|Test
---------|---------------------------------------------------------------
  `'<'`  |Whether the `mtime` at `path0` is less than that at `path1`.
  `'='`  |Whether the `mtime` at `path0` is equal to that at `path1`.
  `'>'`  |Whether the `mtime` at `path0` is greater than that at `path1`.

ri --format=ansi

[kou]

+1

Can we improve alignment in ri --format=ansi? It seems that we should not count invisible ANSI escape characters for width.

[ksss]

@kou
You are absolutely right.
How can I implement ToAnsi#accept_table?

[tompng]

I think RDoc::Markup::ToRdoc can have a method like calculate_text_width(text) and use it in accept_table.

widths = header.zip(*body).map do |cols|
  cols.map(&:size).max # Use `calculate_text_width instead of size
end
...
# a is :rjust | :ljust | :center, w is width
t.__send__(a, w)
↓
t.__send__(a, w + calculate_text_width(t) - t.size)

And each formatter can override it if it needs special width calculation logic.

ToRDoc#calculate_text_width

Simply return text.size for now. Maybe there's a room for improvement in the future such as treating some multibyte character as 2-width.

ToAnsi#calculate_text_width

Remove escape sequences before calculating width

ToBs#calculate_text_width

I think ToBS also need another custom width calculation logic.
"<b>dd</b>" becomes "d\bdd\bd" and its width should be 2.
"d" backspace "d" will print "d" twice in the same position. (Printing a character to a paper by ink twice makes the text darker/bold)

[ksss]

@tompng
Thank you for your wonderful suggestion!
I also learned about the BS format, which I was unaware of.
I have reimplemented it using the proposed method and updated the description.

[matzbot]

🚀 Preview deployment available at: https://4ba19a13.rdoc-6cd.pages.dev (commit: 45a8180)

[tompng]

Thank you 👍