#!/usr/bin/perl
#
# datparse: parse .dat file
# outputs: WHO VOICE WHEN ACTION SOUND TEXT
# notes: does NOT sanitise TEXT
# usage: cat *.dat | datparse 2>/dev/null | datsanitise
#
# expects:
# WHO WHEN (comment)        state 0
# ACTION                    state 1
#
# n VOICE [x] "text"        state 2
#
# SOUND.EXT ([x]) "text"?   state 2
#           ([x]) "text"?   state 2
#                           state 3
#
# WHO and WHEN refer to events happening to a certain character
# ACTION specifies the event
# eg: SOV_AMER10 RANDOM_1 HUNT is said at a random time when a 
#     Sovereign America guy is searching for officers
# eg: BLUE_TEAM TOBIAS_S SUSPECT_NEUT is said when Blue sees Tobias Stromm 
#     go down
# note that WHO is not the character actually saying the chat, it's the 
# character who "sees" the event, thus causing the trigger 
#
# n is the order in the dialogue
# eg: 1 ELEMENT [N] "TOC, this is entry team. We are in position."
#     2      LT [N] "Copy entry team. Go when ready."
# both sounds are played in order
#
# VOICE is the character who actually speaks
# eg: ELEMENT shows that it's the element talking
#
# x?  dunno what that's for
#
# finally we have TEXT, which may span multiple lines, and SOUND, 
# which is (almost) the name of the .cmp file
# s/\.// to get the custom chat filename, add .cmp to THAT to get the filename
#

$state = 0;
while (<STDIN>) {
  print STDERR "!$_";
  if (/^\s+([A-Za-z0-9]{8})\.([A-Za-z0-9]{3})\s*/) {
    $found{"$1.$2"}++;
  }
  eval "do_state_$state(\$_)";
}
eval "do_state_$state(\$_)";

foreach (keys %found) {
  print STDERR "Warning: $_ not parsed\n" unless ($parsed{$_} == $found{$_});
}

# looking for WHO WHEN and possibly a comment
sub do_state_0 {
  my $line = shift;
  if ($line =~ /^([A-Z][^\s]+)\s+([^\s]+).*$/) {
    $state = 1;
    ($who, $when, $action, $voice, $text, $sound) = ();
    ($who, $when) = ($1, $2);
    print STDERR "-> $state: $who, $when\n";
    return 1;
  }
  return 0;
}

# looking for ACTION
sub do_state_1 {
  my $line = shift;
  return if ($line =~ /^\s*$/);
  if ($line =~ /^\s+([A-Z][^\s]+)\s*$/) {
    $state = 2;
    $action = $1;
    print STDERR "-> $state: $action\n";
  }
}

# looking for VOICE and TEXT or SOUND
sub do_state_2 {
  my $line = shift;
  if ($line =~ /^\s+\d+\s+([^\s]+)[^("]+([("]+.+)$/) {
    # look for n VOICE [x] TEXT
    ($voice, $match) = ($1, $2);
    chomp $match;
    $match =~ s/[\s\r]+$/ /;
    # check for brackets only first, it's easier
    if ($match =~ /^\(([^)]+)\)[."\s]*$/) {
      # (Cough, cough.... gasp, cough, cough)
      print STDERR "bingo 1: '$match'\n";
      $text = $1;
    }
    elsif ($match =~ /^(\([^)]+\)\s*)?"?(.+)[."\s]*$/) {
      # (Russian accent) "AHHHHhh!"
      # (Angry whisper)"Geezzs, (gumble, grumble)..."
      # "(choke), ah... (cough, cough), ah...(cough, cough, cough)..."
      print STDERR "bingo 2: '$match' -> '$1' '$2'\n";
      ($text = $2) =~ s/"\s*$//;
    }
    else {
      # hmm, couldn't parse this one (SHOULD NEVER GET HERE!)
      print STDERR "oops: '$match'\n";
    }
    chomp $text;
    $text =~ s/[\s\r]+$/ /;
    print STDERR "-> ${state}a: $voice, $text\n";
    return 1; # needed if we're called from state 3
  }
  elsif ($line =~ /^\s+\d+\s+([^\s]+)\s\[.\]\s+(.+)$/) {
    # no quotes at all
    ($voice, $text) = ($1, $2);
    chomp $text;
    $text =~ s/[\s\r]+$/ /;
    print STDERR "bingo 0: '$text'\n";
  }
  elsif ($line =~ /^\s+([A-Za-z0-9]{8})\.([A-Za-z0-9]{3})\s*(\[.\]\s+)?"?([^"]*)["\s]*$/) {
    # got SOUND
    $sound = "$1$2";
    $parsed{"$1.$2"}++;
    # maybe TEXT continues after the filename
    $text .= $4 if $4;
    chomp $text;
    $text =~ s/[\s\r]+$/ /;
    print STDERR "-> ${state}b: $sound, $text\n";
  }
  elsif ($line =~ /^\s{15,}([^]]+\]\s*)?"?([^"]*)["\s]*$/ && $voice) {
    # maybe TEXT continues, indented, on a new line
    $text .= $2;
    chomp $text;
    $text =~ s/[\s\r]+$/ /;
    print STDERR "-> ${state}b: $sound, $text\n";
  }
  elsif ($line =~ /^\s*$/) {
    $state = 3;
    print "$who\t$voice\t$when\t$action\t$sound\t$text\n";
  }
}

# looking either for another text or a whole new entity
sub do_state_3 {
  my $line = shift;
  next if ($line =~ /^\s*$/);
  # let's try to parse out VOICE and TEXT
  $state = 2;
  &do_state_2($line) and return;
  # no, we're still here, must be WHO WHEN so skip back to state 0
  $state = 0;
  &do_state_0($line);
}