RE: (m)-flag for boundary cases

RE: (m)-flag for boundary cases

[Sebastian Gniazdowski]

> Try this?  Works for characters that are 2 columns wide.  If you have
> a mix of character widths, probably a custom math function to use in the
> subscript ...
> 
> dw=3    # Desired width
> echo ${(mr:dw::_:)a[1,${(m)#a[1,dw-1]}>dw?dw/2:dw]}

Mixed character widths are possibile. Following cases can occur:

1. a="測試a句"; echo ${(mr:7:)a},
測試a句,

2. a="測試a句"; echo ${(mr:6:)a},
測試a句,

3. a="測試a句"; echo ${(mr:5:)a},
測試a,

4. a="測試a句"; echo ${(mr:4:)a},
測試,

Now a try with minus-one:
------------------------

1. a="測試a句"; echo ${(mr:7-1:)a},
測試a句,

2. a="測試a句"; echo ${(mr:6-1:)a},
測試a,

3. a="測試a句"; echo ${(mr:5-1:)a},
測試, # problem

4. a="測試a句"; echo ${(mr:4-1:)a},
測試,

So if one could fix the trailing-a absence, it seems it would work. Did following:

dw=5; a="測試a句"; echo ${(mr,${(m)#${(mr:dw:)a}[-1]} > 1 ? dw-1 : dw,)a}
測試a
dw=3; a="測試a句"; echo ${(mr,${(m)#${(mr:dw:)a}[-1]} > 1 ? dw-1 : dw,)a}
測

But not sure if I will commit this.

--
Sebastian Gniazdowski
psprint@zdharma.org

Re: (m)-flag for boundary cases

[Bart Schaefer]

On Mar 21,  2:02pm, Sebastian Gniazdowski wrote:
} Subject: RE: (m)-flag for boundary cases
}
} > dw=3    # Desired width
} > echo ${(mr:dw::_:)a[1,${(m)#a[1,dw-1]}>dw?dw/2:dw]}
} 
} Mixed character widths are possibile.
} 
[...]
} 
} So if one could fix the trailing-a absence, it seems it would work.

    # dw is the "desired width" [or "display width"] function
    # Declare to accept a string because subscripts hate extra commas,
    # and because we need to pass a parameter by name not by math value.
    # Call it in math context (e.g., a scalar slice subscript) like:
    #   dw(paramname:width)
    # Returns largest index w of a full character in $paramname, such
    # that ${paramname[1,w]} occupies $width or fewer display positions.
    dw() {                                  
      integer w
      set -- "${(@s,:,)1}"
      [[ -z ${(P)1} ]] && return -1                                
      for ((w=$2; ${(m)#${(P)1}[1,w]} > $2; w-- )); do :; done 
      return w
    }
    functions -M -s dw

Given the above this example --

    echo ${(mr:9::_:)string[1,dw(string:9)]}

-- always outputs a string nine display positions wide, padded on the
right with underscores, regardless of the mix of character widths in
the value of $string.

Optimizations, if any, left to the reader.

(m)-flag for boundary cases

[Sebastian Gniazdowski]

Hello,

% a="漢字"
% echo ${(mr:3::_:)a}
漢字
% echo ${(mr:2::_:)a}
漢

I think it would be better to trim first case down to single Asian
character, and display: 漢_. The point is that it gives predictable
width. Current version is occasionally 1 character too long.

I wonder how many people use (m)-flag. Never seen it mentioned here. I
think that nobody wants the off-by-1 behavior, and fixing it will not
break things, but make them automatically better, like for me – the flag
turned out to be a necessity and a solution, with acceptable off-by-1
problem, which will be automatically fixed by Zsh update. I've added
Asian characters support to ZUI via the flag, most important use case is
the (r) padding, but also good cooperation with (P) flag.

-- 
  Sebastian Gniazdowski
  psprint3@fastmail.com

Re: (m)-flag for boundary cases

[Sebastian Gniazdowski]

On the other side, current padding never adds the padding character
unless the string is shorter than the given limit. With the change, it
would add "_" as described "漢_", suggesting there are no other
characters.

Maybe it should be: not add "_", but do skip the not-fitting character.
This would not meet the "constant length" property..
-- 
  Sebastian Gniazdowski
  psprint3@fastmail.com

Re: (m)-flag for boundary cases

[Bart Schaefer]

On Mar 20,  7:33am, Sebastian Gniazdowski wrote:
} Subject: (m)-flag for boundary cases
}
} Hello,
} 
} % a="0x6F220x5B57"
} % echo ${(mr:3::_:)a}
} 0x6F220x5B57
} % echo ${(mr:2::_:)a}
} 0x6F22
} 
} I think it would be better to trim first case down to single Asian
} character, and display: 0x6F22_. The point is that it gives predictable
} width.

Try this?  Works for characters that are 2 columns wide.  If you have
a mix of character widths, probably a custom math function to use in the
subscript ...

dw=3	# Desired width
echo ${(mr:dw::_:)a[1,${(m)#a[1,dw-1]}>dw?dw/2:dw]}