#!/usr/bin/php
<?php

/*

FRA '\' -> 'ê' (d'être in wordlist?)
FRA Wordlist: ...Je [Ô]Agar [Ô]Allez, [Ô]Alors [Ô]alors [Ô]animaux [Ô]Appuyer [Ô]AprÜs [Ô]Ar

*/

// TODO: 80+ sind "unicode" zeichen
// TODO: aktuelle transcriptionstabelle

// TODO: ESP und FRA wortliste wird nicht gefunden!

echo "Please wait approx. 11 minutes...\n";

// Only GER
$preg_valid_chars = 'A-Za-z0-9 !\\?ßäöüÄÖÜ\\[\\]\\(\\):\\.,\\-\'';

$roms = rom_names(__DIR__ . '/../../roms/*.sfc');
foreach ($roms as &$rom) {
	$x = file_get_contents(__DIR__ . '/../../roms/'.$rom.'.sfc');
	$wl = terranigma_ger_find_word_lists($x, $rom); // needs to be called with the complete ROM
	$x = terranigma_ger_decode_string($x, $wl); // can be called with a single string sequence or with the complete ROM

	$h = fopen(__DIR__ . '/out/'.$rom.'.txt', 'w');
	fwrite($h, $x);
	fclose($h);
}

function terranigma_ger_decode_string($x, $wl = array()) {
	$out = '';
	$prev_c = 0x00;

	$num_of_dictionaries = count($wl);

	// It is a very slow technique to parse byte by byte, but it is probably the only
	// technique which allows bugfree parsing of the dictionary-expansions.
	// Problem with ROM parsing, when E5/E6 is in front of a string.
	for ($i=0; $i<strlen($x); $i++) {
		$c = $x[$i];

		// Attention: CREDITS are NOT decoded in this encoding.

		// Not yet known: abpqrstuwxz (TODO)
		// TODO: tabelle veröffentlichen
		# --- BEGIN OF OUTPUT OF ger_helper_enc.phps
		     if ($c == chr(0x21) /* !     */) $out .= chr(0x41) /* A     */;
		else if ($c == chr(0x22) /* "     */) $out .= chr(0x42) /* B     */;
		else if ($c == chr(0x23) /* #     */) $out .= chr(0x43) /* C     */;
		else if ($c == chr(0x24) /* $     */) $out .= chr(0x44) /* D     */;
		else if ($c == chr(0x25) /* %     */) $out .= chr(0x45) /* E     */;
		else if ($c == chr(0x26) /* &     */) $out .= chr(0x46) /* F     */;
		else if ($c == chr(0x27) /* '     */) $out .= chr(0x47) /* G     */;
		else if ($c == chr(0x28) /* (     */) $out .= chr(0x48) /* H     */;
		else if ($c == chr(0x29) /* )     */) $out .= chr(0x49) /* I     */;
		else if ($c == chr(0x2a) /* *     */) $out .= chr(0x4a) /* J     */;
		else if ($c == chr(0x2b) /* +     */) $out .= chr(0x4b) /* K     */;
		else if ($c == chr(0x2c) /* ,     */) $out .= chr(0x4c) /* L     */;
		else if ($c == chr(0x2d) /* -     */) $out .= chr(0x4d) /* M     */;
		else if ($c == chr(0x2e) /* .     */) $out .= chr(0x4e) /* N     */;
		else if ($c == chr(0x2f) /* /     */) $out .= chr(0x4f) /* O     */;
		else if ($c == chr(0x30) /* 0     */) $out .= chr(0x50) /* P     */;
		else if ($c == chr(0x31) /* 1     */) $out .= chr(0x51) /* Q     */;
		else if ($c == chr(0x32) /* 2     */) $out .= chr(0x52) /* R     */;
		else if ($c == chr(0x33) /* 3     */) $out .= chr(0x53) /* S     */;
		else if ($c == chr(0x34) /* 4     */) $out .= chr(0x54) /* T     */;
		else if ($c == chr(0x35) /* 5     */) $out .= chr(0x55) /* U     */;
		else if ($c == chr(0x36) /* 6     */) $out .= chr(0x56) /* V     */;
		else if ($c == chr(0x37) /* 7     */) $out .= chr(0x57) /* W     */;
		else if ($c == chr(0x38) /* 8     */) $out .= chr(0x58) /* X     */;
		else if ($c == chr(0x39) /* 9     */) $out .= chr(0x59) /* Y     */;
		else if ($c == chr(0x3a) /* :     */) $out .= chr(0x5a) /* Z     */;
		else if ($c == chr(0x3b) /* ;     */) $out .= chr(0xfc) /* ü     */;
		else if ($c == chr(0x3c) /* <     */) $out .= chr(0xe4) /* ä     */;
		else if ($c == chr(0x3d) /* =     */) $out .= chr(0xc4) /* Ä     */;
		else if ($c == chr(0x3e) /* >     */) $out .= chr(0xf6) /* ö     */;
		else if ($c == chr(0x3f) /* ?     */) $out .= chr(0xd6) /* Ö     */;
		else if ($c == chr(0x41) /* A     */) $out .= chr(0x61) /* a     */;
		else if ($c == chr(0x42) /* B     */) $out .= chr(0x62) /* b     */;
		else if ($c == chr(0x43) /* C     */) $out .= chr(0x63) /* c     */;
		else if ($c == chr(0x44) /* D     */) $out .= chr(0x64) /* d     */;
		else if ($c == chr(0x45) /* E     */) $out .= chr(0x65) /* e     */;
		else if ($c == chr(0x46) /* F     */) $out .= chr(0x66) /* f     */;
		else if ($c == chr(0x47) /* G     */) $out .= chr(0x67) /* g     */;
		else if ($c == chr(0x48) /* H     */) $out .= chr(0x68) /* h     */;
		else if ($c == chr(0x49) /* I     */) $out .= chr(0x69) /* i     */;
		else if ($c == chr(0x4a) /* J     */) $out .= chr(0x6a) /* j     */;
		else if ($c == chr(0x4b) /* K     */) $out .= chr(0x6b) /* k     */;
		else if ($c == chr(0x4c) /* L     */) $out .= chr(0x6c) /* l     */;
		else if ($c == chr(0x4d) /* M     */) $out .= chr(0x6d) /* m     */;
		else if ($c == chr(0x4e) /* N     */) $out .= chr(0x6e) /* n     */;
		else if ($c == chr(0x4f) /* O     */) $out .= chr(0x6f) /* o     */;
		else if ($c == chr(0x50) /* P     */) $out .= chr(0x70) /* p     */;
		else if ($c == chr(0x51) /* Q     */) $out .= chr(0x71) /* q     */;
		else if ($c == chr(0x52) /* R     */) $out .= chr(0x72) /* r     */;
		else if ($c == chr(0x53) /* S     */) $out .= chr(0x73) /* s     */;
		else if ($c == chr(0x54) /* T     */) $out .= chr(0x74) /* t     */;
		else if ($c == chr(0x55) /* U     */) $out .= chr(0x75) /* u     */;
		else if ($c == chr(0x56) /* V     */) $out .= chr(0x76) /* v     */;
		else if ($c == chr(0x57) /* W     */) $out .= chr(0x77) /* w     */;
		else if ($c == chr(0x58) /* X     */) $out .= chr(0x78) /* x     */;
		else if ($c == chr(0x59) /* Y     */) $out .= chr(0x79) /* y     */;
		else if ($c == chr(0x5a) /* Z     */) $out .= chr(0x7a) /* z     */;
		else if ($c == chr(0x5b) /* [     */) $out .= chr(0xdc) /* Ü     */;
		else if ($c == chr(0x5f) /* _     */) $out .= chr(0xdf) /* ß     */;
		else if ($c == chr(0x60) /* `     */) $out .= chr(0x3f) /* ?     */;
		else if ($c == chr(0x63) /* c     */) $out .= chr(0x30) /* 0     */;
		else if ($c == chr(0x64) /* d     */) $out .= chr(0x31) /* 1     */;
		else if ($c == chr(0x65) /* e     */) $out .= chr(0x32) /* 2     */;
		else if ($c == chr(0x66) /* f     */) $out .= chr(0x33) /* 3     */;
		else if ($c == chr(0x67) /* g     */) $out .= chr(0x34) /* 4     */;
		else if ($c == chr(0x68) /* h     */) $out .= chr(0x35) /* 5     */;
		else if ($c == chr(0x69) /* i     */) $out .= chr(0x36) /* 6     */;
		else if ($c == chr(0x6a) /* j     */) $out .= chr(0x37) /* 7     */;
		else if ($c == chr(0x6b) /* k     */) $out .= chr(0x38) /* 8     */;
		else if ($c == chr(0x6c) /* l     */) $out .= chr(0x39) /* 9     */;
		else if ($c == chr(0x6d) /* m     */) $out .= chr(0x21) /* !     */;
		else if ($c == chr(0x6e) /* n     */) $out .= chr(0x2c) /* ,     */;
		else if ($c == chr(0x6f) /* o     */) $out .= chr(0x3a) /* :     */;
		else if ($c == chr(0x76) /* v     */) $out .= chr(0x27) /* '     */;
		else if ($c == chr(0x79) /* y     */) $out .= chr(0x25) /* %     */;
		else if ($c == chr(0x7c) /* |     */) $out .= chr(0x2d) /* -     */;
		else if ($c == chr(0x7b) /* {     */) $out .= chr(0x2b) /* +     */;
		else if ($c == chr(0x7f) /* [DEL] */) $out .= chr(0x2e) /* .     */;
		else if ($c == chr(0xcf) /* Ï     */) $out .= chr(0x0a) /* [NEWLINE] */;
		# --- END OF OUTPUT OF ger_helper_enc.phps

		# Manually added
		else if ($c == chr(0x20) /*   */) $out .= ' ';
		else if ($c == chr(0xC9) /* É */) {
			// Go to the next char
			$i++;
			$c = $x[$i];

			if ($c == chr(0x44)) /* ÉD */ $out .= "\t"; // Ma-Jo's Brief
			else {
				$out .= "[0xC9 $c]"; // Unknown
			}
		}
		else if ($c == chr(0xD2) /* Ò */) {
			// Go to the next char
			$i++;
			$c = $x[$i];

			if ($c == chr(0x00)) /* Ò[NUL] */ $out .= "[SPIELER]";
			else {
				$out .= "[0xD2 $c]"; // Unknown
			}
		}
		else if ($c == chr(0xd3) /* Ó */) $out .= "[ENDE]\n";
		else if ($c == chr(0xd5) /* Õ */) $out .= "[WEITER]\n";
		else if (($c >= chr(0xE5)) && ($c < chr(0xE5 + $num_of_dictionaries))) {
			// To make the ROM smaller, they added dictionaries with mostly used words
			// The dictionary $wl needs to be extracted from the ROM
			$dict_id  = ord($c)-0xE5;

			$cur_dict = $wl[$dict_id];

			// Go to the next char
			$i++;
			$c = $x[$i];

			$out .= $cur_dict[ord($c)];
		}
		else {
			$out .= "[$c]"; // Unknown char
		}


		// Vogel sagt "Trällerö" ... "Vogel:Ü" -- "Ü" hat keine Bedeutung im Spiel?! Scheinbar nur, dass die Farbe entfernt wird.
##		$x = str_replace('Ü', '', $x);

		// TODO: 'Ö' ist irgendwie ein Kapitel-Ende (siehe Beruga-PC)

		$prev_c = $c;
	}
	return $out;
}

function terranigma_ger_encode_string($x) {
	// TODO
}

# -------------------------------

function terranigma_ger_find_word_lists($x, $rom) {
	// Quick replace stuff to make the RegEx work
	// There is no wordlist expansion inside the wordlists itself

	include 'ger_table.inc.phps';

	$x = strtr($x, $s, $r);

	// Try to find word lists
	global $preg_valid_chars;

	preg_match_all('@((['.$preg_valid_chars.']{2,}Ô){256})@sm', $x, $m);
	$wortlisten = $m[1];

	$h = fopen(__DIR__ . '/out/'.$rom.'_wordlist.txt', 'w');
	foreach ($wortlisten as $list_no => &$wortliste) {
		$wortliste = substr($wortliste, 0, strlen($wortliste)-2);
		$ary = explode('Ô', $wortliste);

		$swl = array();
		foreach ($ary as $n => &$a) {
			$debug = dechex(0xE5 + $list_no).' '.zerofill(dechex($n))." = \"$a\"\n";
			fwrite($h, $debug);
			$swl[] = $a;
		}

		$wortliste = $swl;
	}
	fclose($h);

	return $wortlisten;
}

# ---

function rom_names($wildcard) {
	$x = glob($wildcard);
	$out = array();
	foreach ($x as &$y) {
		$ary = explode('.', basename($y));
		$out[] = $ary[0];
	}
	return $out;
}

# ---

function zerofill($mStretch, $iLength = 2) {
	# http://php.net/manual/en/function.str-pad.php#74259
	$sPrintfString = '%0' . (int)$iLength . 's';
	return sprintf($sPrintfString, $mStretch);
}

?>