function toStorage($file, $settings = array())
{
@($naming = $settings['naming'] ? $settings['naming'] : "%s");
@($ext = $settings['ext'] ? $settings['ext'] : "");
@($delete = $settings['delete'] ? $settings['delete'] : false);
@($subdir = $settings['subdir'] ? $settings['subdir'] : false);
@($dir = $settings['dir'] ? $settings['dir'] : false);
$add = 0;
clearstatcache();
$dir = rtrim($dir, '/');
if (host($file)) {
$tempDirectory = rtrim(`mktemp -d`);
$content = curl($file);
if (!is_string($content)) {
_warn(__FUNCTION__, 'INVALID FILE (URL)!');
return null;
}
$path = parse_url($file, PHP_URL_PATH);
$name = basename($path);
if (!$name) {
$name = mt_rand() . '.tmp';
}
$file = "{$tempDirectory}/{$name}";
file_put_contents($file, $content);
}
// $dir - это каталог! $file - это файл!
if (!is_file($file) or !is_dir($dir)) {
_warn(__FUNCTION__, 'INVALID FILE OR DIRECTORY!');
return null;
}
$hash = function ($_) {
return hash('sha256', $_, false);
};
//
$name = file_get_name($file);
$ext = $ext ? $ext : file_get_ext($file);
if ($ext) {
$ext = ".{$ext}";
}
//
if (!$name) {
$name = mt_rand();
}
$name = str_replace(' ', '-', $name);
if (is_callable($naming)) {
$name = $naming($name);
} elseif (is_string($naming)) {
$name = sprintf($naming, $name);
}
$name = preg_replace('~\\W+~', ' ', $name);
$name = preg_replace('~\\s+~', '-', $name);
// $name = Normalizer::go($name, 'tr,latinRu,en,hyphen');
if (!$name) {
$name = mt_rand();
}
//
$fileHash = $hash(file_get_contents($file));
if (!$subdir) {
$subdir = rand_from_string($fileHash) % 1000;
}
if (!is_dir($_ = $dir . '/' . $subdir)) {
mkdir($_, 0755);
}
$_ = "/{$subdir}/{$name}{$ext}";
while (true) {
if (is_file($dir . $_) and $hash(file_get_contents($dir . $_)) === $fileHash) {
if ($delete) {
unlink($file);
}
if (isset($tempDirectory)) {
`rm -rf '{$tempDirectory}'`;
}
return $_;
}
if (!is_file($dir . $_)) {
file_put_contents($dir . $_, file_get_contents($file));
if ($delete) {
unlink($file);
}
if (isset($tempDirectory)) {
`rm -rf '{$tempDirectory}'`;
}
return $_;
}
$_ = "/{$subdir}/{$name}-{$add}{$ext}";
$add += 1;
}
}
function toStorage($file, $settings = array())
{
$dir = isset($settings['dir']) ? $settings['dir'] : false;
clearstatcache();
$delete = isset($settings['delete']) ? $settings['delete'] : false;
// $dir - это каталог! $file - это файл!
if (!is_file($file) or !is_dir($dir)) {
_log('INVALID FILE OR DIRECTORY!', E_USER_WARNING);
return null;
}
$dir = rtrim($dir, '/');
$defaultName = file_get_name($file);
$defaultExt = file_get_ext($file);
$name = isset($settings['name']) ? $settings['name'] : $defaultName;
$ext = isset($settings['ext']) ? $settings['ext'] : $defaultExt;
//
$name = preg_replace('~[^a-zA-Z0-9\\.-]~', ' ', $name);
$name = preg_replace('~\\s+~', '-', $name);
if (!$name) {
$name = mt_rand();
}
$target = $dir . '/' . $name . ($ext ? ".{$ext}" : "");
$i = 0;
$initialTarget = $target;
while (file_exists($target)) {
$target = preg_replace($i ? '~-\\d+(\\.\\w+)$~' : '~(\\.\\w+)$~', '-' . ++$i . '$1', $target);
}
if ($delete) {
rename($file, $target);
} else {
copy($file, $target);
}
if ($initialTarget != $target and md5_file($initialTarget) === md5_file($target)) {
unlink($target);
$target = $initialTarget;
}
return $target;
}
function directory_contents($Dir_Path, $ext = NULL)
{
$dir_data = scandir($Dir_Path);
$new_dir_data = array();
foreach ($dir_data as $k => $object) {
if ($object != "." && $object != "..") {
$f_type = filetype($Dir_Path . "/" . $object);
if (isset($ext)) {
if ($ext == $f_type) {
$new_dir_data[] = $object;
} elseif (file_get_ext($object) == $ext) {
$new_dir_data[] = $object;
}
} else {
$new_dir_data[$f_type][] = $object;
}
}
}
return $new_dir_data;
}
public function testToStorage()
{
$tmp = rtrim(`mktemp --suffix=.png`);
$dir = rtrim(`mktemp -d`);
file_put_contents($tmp, $rand = mt_rand());
$target = toStorage($tmp, array('dir' => $dir));
$this->assertTrue(file_get_name($tmp) === file_get_name($target));
$this->assertTrue(file_get_ext($tmp) === file_get_ext($target));
$this->assertEquals($rand, file_get_contents($target));
$this->assertTrue(is_file($tmp));
//
file_put_contents($tmp, mt_rand());
$target_1 = toStorage($tmp, array('dir' => $dir));
$this->assertTrue(file_get_name($target) . "-1" === file_get_name($target_1));
//
file_put_contents($tmp, $rand);
$target_2 = toStorage($tmp, array('dir' => $dir));
$this->assertTrue($target_2 === $target);
//
file_put_contents($tmp, $rand = mt_rand());
$target = toStorage($tmp, array('dir' => $dir, 'ext' => 'bla', 'delete' => true));
$this->assertTrue(file_get_name($tmp) === file_get_name($target));
$this->assertTrue('bla' === file_get_ext($target));
$this->assertEquals($rand, file_get_contents($target));
$this->assertTrue(!is_file($tmp));
}
public static function analyze($stamp, $delete = false)
{
// STAMP TO UNIXTIME
if (is_numeric($stamp) and intval($stamp) <= 0) {
$stamp = strtotime("today " . intval($stamp) . " day");
} elseif (is_numeric($stamp) and intval($stamp) > 0) {
$stamp = $stamp;
} else {
$stamp = strtotime($stamp);
}
//
$db = self::$db;
$dir = self::$dir;
$paths = self::getPath($stamp, $glob = true);
$hosts = array();
foreach ($paths as $path) {
foreach (nsplit(file_get_contents($path)) as $hit) {
$hit = unserialize($hit);
if (!$hit) {
continue;
}
$hit['URL'] = urldecode($hit['URL']);
$hit['REF'] = urldecode($hit['REF']);
//
$host = $hit['HOST'];
$host = strtolower($host);
$host = preg_replace("~^www\\.~", "", $host);
if (!isset($hosts[$host])) {
$hosts[$host] = array('hits' => 0, 'ips' => array(), 'google' => array(), 'yandex' => array(), 'serp' => array(), 'top_all' => array(), 'top_serp' => array(), 'top_ref' => array(), 'kws' => array());
}
$is_valid_ext = !in_array(file_get_ext($hit['URL']), array('jpg', 'jpeg', 'png', 'gif', 'bmp', 'css', 'js', 'ico'));
// HITS
$hosts[$host]['hits'] += 1;
// IPS
$hosts[$host]['ips'][$hit['IP']] = true;
// GOOGLE
if ($is_google = is_numeric(strpos(host($hit['REF']), 'google'))) {
$hosts[$host]['google'][$hit['IP']] = true;
}
// YANDEX
if ($is_yandex = is_numeric(strpos(host($hit['REF']), 'yandex'))) {
$hosts[$host]['yandex'][$hit['IP']] = true;
}
// TOP_ALL
if ($is_valid_ext) {
if (!isset($hosts[$host]['top_all'][$hit['URL']])) {
$hosts[$host]['top_all'][$hit['URL']] = array();
}
$hosts[$host]['top_all'][$hit['URL']][$hit['IP']] = true;
}
// TOP_SERP
if (($is_google or $is_yandex) and $is_valid_ext) {
if (!isset($hosts[$host]['top_serp'][$hit['URL']])) {
$hosts[$host]['top_serp'][$hit['URL']] = array();
}
$hosts[$host]['top_serp'][$hit['URL']][$hit['IP']] = true;
}
// TOP_REF
$is_same = (host($hit['REF']) === $host or host($hit['REF']) === "www.{$host}");
if (!$is_google and !$is_yandex and host($hit['REF']) and $is_valid_ext and !$is_same) {
$_ = "{$hit['URL']}, {$hit['REF']}";
if (!isset($hosts[$host]['top_ref'][$_])) {
$hosts[$host]['top_ref'][$_] = array();
}
$hosts[$host]['top_ref'][$_][$hit['IP']] = true;
}
// KWS
if ($is_google or $is_yandex) {
$query = parse_url($hit['REF'], PHP_URL_QUERY);
do {
if ($query) {
$kw = null;
parse_str($query, $_);
$query = $_;
if (is_null($kw)) {
$kw = ($is_google and isset($query['q']) and $query['q']) ? $query['q'] : null;
}
if (is_null($kw)) {
$kw = ($is_yandex and isset($query['text']) and $query['text']) ? $query['text'] : null;
}
$kw = normLatin($kw);
$kw = normRu($kw);
if (!$kw) {
break;
}
$kw = "{$kw}, {$hit['URL']}";
if (!isset($hosts[$host]['kws'][$kw])) {
$hosts[$host]['kws'][$kw] = array();
}
$hosts[$host]['kws'][$kw][$hit['IP']] = true;
}
} while (false);
}
}
}
foreach ($hosts as $host => $value) {
extract($value);
// TOP ALL
array_walk($top_all, function (&$value) {
$value = count($value);
});
arsort($top_all);
array_walk($top_all, function (&$value, $key) {
$value = "{$value}, {$key}";
});
$top_all = array_values($top_all);
$top_all = array_slice($top_all, 0, 100);
$top_all = implode("\n", $top_all);
// ---
// TOP SERP
array_walk($top_serp, function (&$value) {
$value = count($value);
});
arsort($top_serp);
array_walk($top_serp, function (&$value, $key) {
$value = "{$value}, {$key}";
});
$top_serp = array_values($top_serp);
$top_serp = array_slice($top_serp, 0, 100);
$top_serp = implode("\n", $top_serp);
// ---
// TOP REF
array_walk($top_ref, function (&$value) {
$value = count($value);
});
arsort($top_ref);
array_walk($top_ref, function (&$value, $key) {
$value = "{$value}, {$key}";
});
$top_ref = array_values($top_ref);
$top_ref = array_slice($top_ref, 0, 1000);
$top_ref = implode("\n", $top_ref);
// ---
// KWS
array_walk($kws, function (&$value) {
$value = count($value);
});
arsort($kws);
array_walk($kws, function (&$value, $key) {
$value = "{$value}, {$key}";
});
$kws = array_values($kws);
$kws = array_slice($kws, 0, 1000);
$kws = implode("\n", $kws);
// ---
$bean = $db->bean('stat');
$bean->import(array('host' => $host, 'stamp' => date(SQL_FORMAT_DATE, $stamp), 'hits' => $hits, 'ips' => count($ips), 'google' => count($google), 'yandex' => count($yandex), 'serp' => count($google) + count($yandex), 'top_all' => $top_all, 'top_serp' => $top_serp, 'top_ref' => $top_ref, 'kws' => $kws))->replace();
}
if ($delete) {
foreach ($paths as $path) {
unlink($path);
}
}
}
