#! /usr/bin/gawk -f
# Last edited on 2008-07-01 20:08:59 by stolfi

BEGIN {
  abort = -1;
  USAGE = ( \
    "txtable-g-reformat \\\n" \
    "  -f ${STOLFIHOME}/lib/txtable-g.gawk \\\n" \
    "  [ -v frsep={CHAR}] \\\n" \
    "  [ -v thsep={CHAR} ] \\\n" \
    "  [ -v altzero={STRING} ] \\\n" \
    "  [ -v tblsep={REGEXP} ] \\\n" \
    "  < INFILE \\\n" \
    "  > OUTFILE" \
  );
  
  # Reads a text file {INFILE} and formats the tables contained therein.
  # 
  # WARNING --- this AWK script uses {double} for numeric
  # computations, so it cannot handle large numbers and it 
  # may print fractions incorrectly.
  # 
  # 
  # EXAMPLE
  # 
  # A typical input:
  #
  #   +----+----+----+
  #   : Year : Kind : Profits :
  #   +--+--+--+
  #   # With Microploft software:
  #   +--+--+--+
  #   | 2004| so-so | 34,567,221.00 |
  #   |2005 | terrible | 63339.0 | # Must check this!
  #   +----+----+----+
  #   | SUBTOTAL | - | 0 |
  #   +----+----+----+
  #   # With Linukes software:
  #   +--+--+--+
  #    | 2006 | good | +53000000 |
  #   |  2007 | could be better | 19,255,421 |
  #   +----+----+----+
  #   | SUBTOTAL | - | 0 |
  #   +----+----+----+
  #   | TOTAL | - | 9999 |
  #   +--+--+--+
  #
  # Corresponding output:
  # 
  # DESCRIPTION
  # 
  # The program reads the input file, parses it into zero or more
  # tables, as described in the {table-funcs.gawk} library.
  # Tables are separated by the {tblsep} regular expression
  # (which defaults to "^ *END_TABLE *$").
  # 
  # For each table, it determines the attributes of each table column.
  # It recomputes the numeric values in any (sub)total lines, to make
  # them consistent. It then recasts all fields to their canonical
  # format, and regularizes the field widths in each column.
  # 
  # Finally, the program outputs the whole table. Blank lines are
  # written out unchanged, with their original comments. Any other
  # output line consists of the (uniformized) table margin, followed
  # by the fields of that line, separated by the appropriate delimiter
  # character, and then the line's original comment.
  #   
  # Any blank lines before the first table are written out 
  # unchanged.
  
  # COMMAND ARGUMENT PARSING
  txtable_argument_parse();
  if (tblsep == "") { tblsep = "^ *END_TABLE *$"; }
  
  txtable_clear(); 
}

(abort >= 0) { exit abort; }

($0 ~ tblsep) {
  # Table separator:
  txtable_finish();
  txtable_clear();
  print;
  next;
}

// {
  # One more input line:
  txtable_split_line($0);
  txtable_save_line(row_ind,row_tag,row_nf,row_fld,row_cmt);
  next;
}

END {
  if (abort >= 0) { exit abort; }
  txtable_finish();
  exit 0;
}

function txtable_finish()
{
  # printf "### RAW TABLE #############################################\n";
  # txtable_print();
  # printf "### REFORMATTED TABLE #####################################\n";
  printf "formatting table with %d rows and %d cols\n", tbl_nrows, tbl_ncols > "/dev/stderr";
  txtable_get_column_attributes(frsep,thsep,altzero);
  txtable_recompute_and_reformat_values(frsep,thsep,altzero);
  txtable_print();
}