#!/usr/bin/perl
# File  : fixhw.pl
# Author: Lyndon Hill, http://www.lyndonhill.com
# Note  : This script is free to use. If you make significant improvements please
#         send me a copy.
#
# Script to fix headwords
# If a headword is defined in one sense then bedic repeats it in the next sense,
# however, in my opinion this is wrong - it should be limited to the sense it is defined in.
# This really screws up my duplicate fix script because the headword gets propagated to the
# next sense.

# This script reads the dictionary, when it finds an entry it looks for a sense and
# headword. If there is no headword then it immediately outputs the sense but if there
# is a headword then it stores it. At the end of the entry, all stored senses are output.
# The result is, all headword-less senses should be at the start of the entry.

# The requirement is that your dictionary is formatted nicely, e.g.:

# entry
# {s}
# {hw}headword{/hw}
# definition 1
# {/s}
# {s}
# definition 2
# {/s}

# will be corrected to:

# entry
# {s}
# definition 2
# {/s}
# {s}
# {hw}headword{/hw}
# definition 1
# {/s}

# If it is important to you that definition 1 comes before definition 2 (in most dictionaries
# I'm guessing it doesn't matter because you wouldn't define a headword unless it's a variant),
# then you should define a headword for definition 2. However, if you always define a headword
# for every sense then your dictionary will take up more disk space.

# Here are some variables you can change

# input dictionary file
$dicfile = "en-ka.bedic-new";

# output file
$output = "en-ka.bedic-hwfix";

# Start of script

open(DICFILE, $dicfile);
open(BEDICFILE, ">$output");

$ecount = 0;     # total number of entries

$nextline = 0;   # flag to say end of entry (next line is an entry)

while(<DICFILE>)
{
  $line = $_; chop $line;

  # gotta deal with header

  if($nextline == 1)
  {
    $senses = 0;      # number of senses in this entry
    $storedlines = 0; # number of lines stored for this entry
    $nothw = 1;       # flag that headword has been detected in this sense

    # read the entry
    $entry = $line;
    print BEDICFILE $entry ."\n";
    $ecount++;

    while(<DICFILE>)
    {
      $line = $_; chop $line;
      if($line eq "") 
      {
        # end of entry, dump stored sense lines
        for($s = 0; $s < $storedlines; $s++)
        {
          print BEDICFILE $stored[$s] . "\n";
        }
        print BEDICFILE "\n";

        last; 
      }

      if($line eq "{s}" )
      {
        $senses++;

        # next line is either hw, ss or definition
        $line = <DICFILE>; chop $line;
        $headword = $line;
        if($headword =~ /\{hw\}/)
        {
          $stored[$storedlines++] = "{s}";
          $stored[$storedlines++] = "$line";

          $nothw = 0; 
        }
        else
        {
          # empty headword - output immediately
          print BEDICFILE "{s}\n";
          print BEDICFILE "$line\n";
          $nothw = 1;
        }
      }
      else
      {
        # not start of sense: are we storing or outputting ?
        if($nothw == 1)
        {
          # output
          print BEDICFILE "$line\n";
        }
        else
        {
          # store
          $stored[$storedlines++] = "$line";
        }
        
      }
    }
  }
  else
  {
    if($line eq "") { $nextline = 1; }

    # header
    print BEDICFILE $line . "\n";
  }

}

print "$ecount entries.\n";
print "DONE!\n";

close(DICFILE);
close(BEDICFILE);
exit;