#! /usr/bin/gawk -f
# Last edited on 2008-03-20 09:33:06 by stolfi

BEGIN {
  USAGE = ( \
    "preprocess-tex-shorthand < {INFILE}.tex > {OUTFILE}.tex" \
  );
  # Reads a TeX/LaTeX file and converts certain shorthand markup codes
  # to their standard form.
  # 
  # The recognized codes and their standard equivalents are:
  #   
  #   ${MATH}       --> {$MATH$}
  #   ={LABEL}{EQN} --> \begin{equation}\label{LABEL} EQN\end{equation}
  #   /{TEXT}       --> \emph{TEXT}
  #   *{TEXT}       --> \textbf{TEXT}
  #   
  # The '{' must imemdiately follow the opening character, in the 
  # same line. 
  #
  # These shorthands are recognized at all levels of nesting braces,
  # unless they are quoted with '\' or not followed by "{".
  
  abort = -1;
  
  level = 0;     # Number of unclosed open braces.
  split("", op); # {op[0..level-1]} are the operators of those open braces.
}

(abort >= 0) { exit abort; }

//{ 
  # Save current line in {lin}:
  lin = $0;
  
  # Remove tabs, CR, FF:
  gsub(/[\011\014\015]/, " ", lin);
  
  # Process each character in line:
  nc = length(lin);  # Number of characters in input line
  ic = 1;            # Index of next character to process (from 1).
  while(i < n) { process_next-char(); }
  printf "\n";
  next;
}
  
function process_next_char(   c)
{
  # Processes the next character of {lin}, assumed to have index {ip}.
  # Assumes {np} is the length of {lin}, and {level} is the
  # current brace nesting level.
  # Accounts for doubling, quoting and comments.

  c = substr(lin, ip, 1); ip++;
  if (c == "\\")
    { # Glob the next character or token without further processing:
      out(c);
      if(ip <= np) 
        { c = substr(lin, ip, 1); ip++; out(c);
          # We believe that it suffices to copy just the first char after '\'.
        }
    }
  else if (c == "{")
    { op[level] = " "; level++; out(c); }
  else if (c == "}")
    { level--; close_brace(op[level]); }
  else if (c == "%")
    { # Comment; copy everything until the end of line: 
      out(c); while(ip <= np) { c = substr(lin, ip, 1); ip++; out(c); }
    }
  else if ((c ~ /^[\/$*=]/) && (ip < np) && (substr(lin, ip+, 1) == "{"))
    { # Shorthand construct
      op[level] = c; level++; 
      # Skip the brace, too: 
      ip ++; 
      # Output the preamble:
      open_brace(c); 
    }
  else
    { out(c); }
}

function open_brace(op)
{
  # Ouputs the preamble for the shortand {op}

  if (op == "$")
    { out(op); }
  else if (op == "/")
    { out("\\emph{"); }
  else if (op == "=")
    { out("\\eqn{"); }
  else if (op == "*")
    { out("\\textbf{"); }
  else
    { program_error("duh?"); }
}  

function close_brace(op)
{
  # Ouputs the postamble for the shortand {op}

  if (op == "$")
    { out(op); }
  else if ((op == "/") || (op == "=") || (op == "*"))
    { out("}"); }
  else
    { program_error("duh?"); }
}  

function out(s)
{
  # Appends {s} to the current output line:
  printf "%s", s;
}

function program_error(msg)
{ 
  printf "%s:%d: **program error: %s\n", FILENAME, FNR, msg;
  abort = 1;
  exit 1
}