rootify

#!/usr/local/bin/perl

# Author: Jason Eisner, University of Pennsylvania

# Usage: rootify rootname [files ...]
# Example: rootify ROOT corpus
#
# Given a corpus in oneline or headify format, wraps every parse
# in (rootname ...) so that they all have the same root.
#
# For consistency, rootname is capitalized, and to make sure
# that it does not contain any characters reserved by oneline,
# we remove all non-alphabetic characters from it.

require("stamp.inc"); &stamp;                 # modify $0 and @INC, and print timestamp

die "$0: bad command line flags" if @ARGV && $ARGV[0] =~ /^-./;
die "$0: first argument should be the name of the root node (e.g., ROOT)" unless @ARGV;
$rootname = shift(@ARGV);
$rootname =~ tr/a-z/A-Z/;
$rootname =~ tr/A-Z//cd;

$token = "[^ \t\n()]+";  # anything but parens or whitespace can be a token.  

while (<>) {      # for each sentence
  chop;
  s/^(\S+:[0-9]+:\t)?//, $location = $&;
  unless (/^\#/) {    # unless a comment
    $sent++;
    /^\(($token)/ || die "$0:$location sentence appears to have no root";
    $roots{$1}=1;    
    $_ = "($rootname $_)";
  } 
  print "$location$_\n";
}

print STDERR "$0: wrapped $rootname node around $sent trees having ",scalar keys %roots," different roots: ", join(", ", keys %roots), "\n";