#!/usr/bin/perl
#
# Format the csv file generated by a skyserver query.
# You might want to set skyserver_output_filter to point to this file
#
# The keys are used as patterns to specify special formats. If the
# column headings are longer than the precision specified here, the
# format widths will be suitable adjusted.
#
# If more than one pattern matches, use the longest
#
%formats = ('Dec' => '%10.7f',
	    'objId' => '%20s',
	    'ID\$' => '%20s',
	    '^(l|b)\$' => '%8.4f',
	    'Mag' => '%.3f',
	    '^RA\$' => '%9.5f',
	    'MJD' => '%8.2f',
	    '^N\$' => '%5d',
	    '^run\$' => '%4d',
	    '(rowc|colc)' => '%7.2f',
	    );
#
# Parse arguments
#
while($ARGV[0] =~ /^-/) {
   if($ARGV[0] eq "-h" || $ARGV[0] eq "-?") {
      warn "\
Usage: format-csv [options]

Format a csv file nicely, with constant field widths. Field names that
match some patterns are recognised and formatted specially:
";
      foreach (sort(keys %formats)) {
	 printf STDERR "\t%-10s%6s\n", $_, $formats{$_};
      }
    
      warn "See also -f and -n options.

Options:
	-h, -?		Print this message
	-f 'name=>fmt'	Format fields matching name with format fmt;
			e.g. \"-f 'Mag=>%.4f'\".
			You may specify -f more than once.
	-n		Turn off all predefined formats
";
      exit(0);
   } elsif($ARGV[0] eq "-f") {
      shift(@ARGV);
      $arg = $ARGV[0];
      if(!$arg) {
	 warn "-f expects an argument\n";
      } elsif($arg !~ /([^=]+)=>(%.*)/) {
	 warn "-f expects an argument of the form \"name=>%format\"; saw " .
	     $arg . "\n";
      } else {
	 $formats{$1} = $2;	 
      }
   } elsif($ARGV[0] eq "-n") {
      undef(%formats);
   }

   shift(@ARGV);
}
#
# Read header line and find minimum column widths
#
chomp($header = <>);
@fields = split("(,)", $header);# keep , and empty fields
if(@fields[@fields - 1] eq ",") {
   push(@fields, '');
}
#
# Are we processing an error message?
#
if($header !~ /^(#\s)?[a-z0-9_\'|:,]+$/i) {
   print $header . "\n";

   $exit_code = 1;
   if(!$header) {		# could be a single non-named field
      chomp($next = <>);
      if($next =~ /^[^ ,]+$/) {	# probably a  " "    "   " "   " "
	 $exit_code = 0;
      }
      print $next . "\n";
   }
   while(<>) {
      print;
   }

   exit $exit_code;
}

$ofmt = "";			# output format
foreach $field (@fields) {
   if($field eq ",") {
      next;
   }

   $fmt = '%s';
   $lenmax = 0;			# length of longest match
   foreach $k (keys %formats) {
      if($field =~ /$k/i) {
	 if(length($k) > $lenmax) {
	    $lenmax = length($k);
	    $fmt = $formats{$k};
	 }
      }
   }

   $len = length($field);
   if($len < 3) {		# that\'s silly
      $len = 3;
   }
   if($fmt =~ s/%-?([0-9]+)//) {
      my($flen) = $1;
      if($flen > $len) {
	 $len = $flen;
      }
   }
   $fmt =~ s/%//;
   $fmt = "%${len}$fmt";

   if($ofmt) {
      $ofmt .= " ";
      print " ";
   }

   printf "%${len}s", $field;

   $ofmt .= $fmt;
}
$ofmt .= "\n";
print "\n";
#
# Process the rest of the file
#
while(<>) {
   #print; next;
   chomp;

   if($_ =~ /^\s*$/) {
      next;
   }

   if(/\"/) {
      $line = "";
      $n = 0;			# which segment
      foreach (split("\"")) {	# protect all real ,
	 if($line && $_ eq "") { # \"\", i.e. a quoted \",
	    $n--;
	    $_ = '"';
	 } else {
	    if($n++%2 == 0) {
	       $_ =~ s/,/\002/g;
	    } else {
	       s/,/\001/g;
	 
	       $_ =~ s/,/\002/g;
	       $_ =~ s/\001/,/g;
	    }
	 }
	 $line .= $_;	 
      }

      printf $ofmt, split("\002", $line);
   } else {
      printf $ofmt, split(",");
   }
}