#!/usr/bin/php
<?php

$abc = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ';

$dicts = glob(__DIR__ . '/dict/*.txt');

foreach ($dicts as $dict) {
	$sdict = $dict;
	$sdict = str_replace(__DIR__.'/dict/', '/usr/share/dict/', $sdict);
	$sdict = str_replace('.txt', '', $sdict);
	echo "*** DICT $sdict ***\n\n";
	proc_file($dict);
}

echo "- End of file -\n";

function sanitydict(&$dict) {
	global $abc;
	$out = array();
	foreach ($dict as $n => $d) {
		$d = strtoupper($d);
		$d = preg_replace("|[^$abc]|ismU", '', $d);
		
		if ($d == '') {
			unset($dict[$n]);
		} else {
			$dict[$n] = $d;
		}
	}
}

function proc_file($file) {
	$dict = file($file);
	sanitydict($dict);
	$dict = array_unique($dict);

	$folge = array();
	$count = array();
	$word_count = 0;
	$word_len = array();
	$firstchar = array();
	$lastchar = array();

	// Vervollständigen
	global $abc;
	for ($i=0; $i<strlen($abc); $i++) {
		$cr = substr($abc, $i, 1);
		$firstchar[$cr] = 0;
		$lastchar[$cr] = 0;
		$count[$cr] = 0;
	}

	foreach ($dict as $d) {
		$word_len[strlen($d)]++;

		$word_count++;

		for ($i=0; $i<strlen($d)-1; $i++) {
			$c = substr($d, $i, 1);
			$c_sup = substr($d, $i+1, 1);

			if ($i == 0) $firstchar[$c]++;
			$folge[$c][$c_sup]++;
			$count[$c]++;
		}

		$l = substr($d, strlen($d)-1, 1);
		$count[$l]++; // Letztes
		$lastchar[$l]++;
	}

	unset($dict);

	echo "Word-Count\n";
	echo "$word_count\n";
	echo "\n";

	echo "Char-Neighbors\n";
	printfolge($folge);
	echo "\n";

	echo "First char\n";
	arsort($firstchar);
	print_r($firstchar);
	echo "\n";

	echo "Last char\n";
	arsort($lastchar);
	print_r($lastchar);
	echo "\n";

	echo "Char-Count\n";
	arsort($count);
	print_r($count);
	echo "\n";

	echo "Word-Length\n";
	arsort($word_len);
	print_r($word_len);
	echo "\n";
}

function printfolge($folge) {
	$tmp = array();

	// Vervollständigen
	global $abc;
	for ($i=0; $i<strlen($abc); $i++) {
		$c1 = substr($abc, $i, 1);
		for ($j=0; $j<strlen($abc); $j++) {
			$c2 = substr($abc, $j, 1);
			$tmp["$c1,$c2"] = 0;
		}
	}

	foreach ($folge as $c => $x) {
		foreach ($x as $cs => $y) {
			$tmp["$c,$cs"] = $y;
		}
	}
	arsort($tmp);
	print_r($tmp);
}

?>