#! /usr/bin/gawk -f
# Last edited on 2025-12-08 04:04:25 by stolfi

# Factors a text by placing "{}" around each element of the 2025 word model.
# Each is wrapped with "{}"
#
# To be included in factor-field-general etc.

function factor_text(x,   y,e)
{
  # Decomposes "x" into its 2025 word paradigm elements. 
  # Assumes "x" is uncapitalized EVA without alignment fillers
  # Unparsed glyphs are turned into "{*...}"

  y = ""; 
  
  while (x != "")
    { # printf "x = [%s]\n", x > "/dev/stderr";
      if (match(x, /^[-.,«=»\/]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y e );
        }
      else if (match(x, /^[{][^{}]*[}]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y e );
        }
      else if (match(x, /^[<][%$|][>]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y e );
        }
      else if (match(x, /^[<]![^<>]*[>]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y e );
        }
      else if (match(x, /^[aoyqlmngvxrd]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^([i]|[i][i]|[i][i][i])[nmr]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[Ss][Hh]*[h]([e][e]|[e]|)/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[s]/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[?]([e][e]|[e]|)/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[CcIi][TtKkPpFfWwZz][Hh]*[h]([e][e]|[e]|)/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[CcIi][[Hh]*[h]([e][e]|[e]|)/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else if (match(x, /^[TtKkPpFfWwZz]([e][e]|[e]|)/))
        { e = substr(x,1,RLENGTH); x = substr(x, RLENGTH+1);
          y = ( y "{" e "}" );
        }
      else 
        { e = substr(x,1,1); x = substr(x, 2);
          y =  ( y "{*" e "}" )
        }
    }

  return y;
}