die("Usage {$argv[0]} TEXT_FILE [ENCODING] [FILE_ENCODING]" . PHP_EOL);
}
$cwd = getcwd();
$text_file = $argv[1];
$encoding = $argc > 2 ? $argv[2] : 'utf-8';
$file_encoding = $argc > 3 ? $argv[3] : 'utf-8';
$lang = 'rus';
$morphy_ver = getenv('PHPMORPHY_VER');
if ($morphy_ver !== "0.2") {
set_include_path(__DIR__ . '/../src/' . PATH_SEPARATOR . get_include_path());
require 'phpMorphy.php';
} else {
require_once MORPHY_2X_DIR . '/src/common.php';
}
$dict_dir = PHPMORPHY_DIR . '/../dicts/' . $encoding;
$words = load_words($text_file, REMOVE_DUPLICATES, $encoding, $file_encoding);
echo "Total words " . (REMOVE_DUPLICATES ? "(unique)" : '') . " = " . count($words) . PHP_EOL;
//print_memory_usage();
//bench_mystem($words, $encoding, PATH_TO_MYSTEM);
bench_porter($words, $encoding);
//bench_enchant($words);
//bench_pspell($words);
//print_memory_usage();
bench_morphy_dict($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_FILE);
bench_morphy_dict($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_SHM);
bench_morphy_dict($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_MEM);
bench_morphy($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_FILE, false);
bench_morphy($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_FILE, true);
bench_morphy($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_SHM, false);
bench_morphy($words, $encoding, $dict_dir, $lang, PHPMORPHY_STORAGE_SHM, true);
//print_memory_usage();
$wcat = "Random";
if (isset($_GET["wordcat"])) {
$wcat = $_GET["wordcat"];
}
//load the word files
if ($wcat == "Animals" || $wcat == "Random") {
load_words("animals.txt", "Animals");
}
if ($wcat == "Things" || $wcat == "Random") {
load_words("things.txt", "Things");
}
if ($wcat == "Colors" || $wcat == "Random") {
load_words("colors.txt", "Colors");
}
if ($wcat == "Verbs" || $wcat == "Random") {
load_words("verbs.txt", "Verbs");
}
function load_words($fname, $catname)
{
global $wordlist;
global $wcat, $wc;
$file = fopen($fname, 'r') or die($php_errormsg);
while (!feof($file)) {
if ($line = fgets($file)) {
$words = preg_split('/\\s+/', $line, -1, PREG_SPLIT_NO_EMPTY);
// load the words now
for ($i = 0; $i < count($words); $i++) {
// load the words
if ($wcat == $catname || $wcat == "Random") {
$wordlist[$wc] = $words[$i];
$wc++;