dequote

#! /usr/bin/awk -f
# dequote (AWK script) -- Removes multiple levels of ">" and unwordwraps

## ! 1,/^$/   {

BEGIN      {
             dont_use_max = 1
           }

           {
             prev_line = this_line
             ## prev_quote_count = quote_count
             quote_count = 0
             pos = 1
             which_word = 1
             was_quote = 0
             # Scan from the start of the line, counting ">", til char found:
             stop = 0
             while (stop == 0 && pos <= length( $0 ))
               {
                 prev_char = this_char
                 this_char = substr( $0, pos++, 1)
                 if (this_char == ">")
                   {
                     was_quote = 1
                     ++quote_count
                   }
                 else if (this_char == " " || this_char == "	")
                   {
                     was_quote = 0
                     if (prev_char != " " && pos > 2)
                       ++which_word
                   }
                 else
                   {
                     stop=1
                     --pos
                   }
               }
	     if (max_quotes < quote_count)
	       max_quotes = quote_count
             else
               if (NF == which_word || prev_quote_count == quote_count)
                 max_quotes = 0

             line_length = length( $0 ) - pos + 1
             while (substr( $0, line_length + pos - 1 ) == " ")
               --line_length
             prev_backslash = backslash
             if ($(NF) == "\\")
               backslash = 2
             else
               backslash = 0
             this_line = substr( $0, pos, line_length - backslash )

             ## print( "#$ " this_line " <" which_word "," was_quote "," prev_quote_count ">" )
	     ## printf( quote_count );

	     # if the previous line ended with a backslash or this one
	     # has less quotes than it (or the max), then join them:
             if (total_line_length + line_length > 40 && stop != 0 &&
                 (quote_count < prev_quote_count ||  \
		  !dont_use_max && quote_count < max_quotes ||  \
		  stop && NF - which_word < unwrap) ||  \
                 prev_backslash >= 1)
               {
                 total_line_length += line_length + 1
                 built_line = built_line " " this_line
               }
             else
               {
                 if (NR > 1)
		   print( built_line )
                 built_line = this_line
                 # establish the local maximum for quotes:
                 prev_quote_count = quote_count
		 total_line_length = line_length
               }
           }

END        { print( built_line ) }