v1.0.1

PriviMetrics 2

← Changes All versions

343 lines · 11.1 KB

<?php
declare(strict_types=1);
final class PmafFormatException extends RuntimeException {}
final class PmafStorage
{
private const MAGIC_DATA = 'PMAF';
private const MAGIC_INDEX = 'PMFI';
private const VERSION = 2;
private const HEADER_LEN = 5;
public function __construct(private string $dataDir) {}
private function dataFile(string $siteId): string
{
return $this->dataDir . '/hits_' . $siteId . '.pmaf';
}
private function indexFile(string $siteId): string
{
return $this->dataDir . '/hits_' . $siteId . '.pmafidx';
}
private function lockFile(string $siteId): string
{
return $this->dataDir . '/hits_' . $siteId . '.pmaf.lock';
}
public function appendHit(string $siteId, array $hit, int $ts): void
{
if (!is_dir($this->dataDir)) {
mkdir($this->dataDir, 0755, true);
}
$fp = fopen($this->lockFile($siteId), 'c');
if (!$fp || !flock($fp, LOCK_EX)) {
if ($fp) fclose($fp);
return;
}
try {
$file = $this->dataFile($siteId);
$isNew = !file_exists($file) || filesize($file) === 0;
$fileVersion = self::VERSION;
if (!$isNew) {
$headerByte = @file_get_contents($file, false, null, 4, 1);
if ($headerByte !== false && $headerByte !== '') {
$fileVersion = ord($headerByte);
}
}
$handle = fopen($file, 'ab');
if (!$handle) return;
if ($isNew) {
fwrite($handle, self::MAGIC_DATA . chr(self::VERSION));
}
$offsetOfRecord = fstat($handle)['size'];
$record = $this->encodeRecord($hit, $ts, $fileVersion);
fwrite($handle, $record);
fclose($handle);
$this->indexDayIfNeeded($siteId, $ts, $offsetOfRecord);
} finally {
flock($fp, LOCK_UN);
fclose($fp);
}
}
private function encodeRecord(array $hit, int $ts, int $fileVersion = self::VERSION): string
{
$userHashRaw = @hex2bin((string) ($hit['user_hash'] ?? ''));
if ($userHashRaw === false || strlen($userHashRaw) !== 16) {
$userHashRaw = str_repeat("\x00", 16);
}
$countryCode = substr(str_pad((string) ($hit['country_code'] ?? 'XX'), 2), 0, 2);
$body = pack('N', $ts)
. $userHashRaw
. $this->packVar($hit['ip'] ?? '', 'C')
. $this->packVar($hit['country'] ?? 'Unknown', 'C')
. $countryCode
. $this->packVar($hit['user_agent'] ?? '', 'n')
. $this->packVar($hit['page_url'] ?? '', 'n')
. $this->packVar($hit['page_title'] ?? '', 'n')
. $this->packVar($hit['search_query'] ?? '', 'C')
. $this->packVar($hit['referrer'] ?? '', 'C');
if ($fileVersion >= 2) {
$body .= $this->packVar($hit['utm_source'] ?? '', 'C');
}
return pack('N', strlen($body)) . $body;
}
private function packVar(string $value, string $lenFormat): string
{
$max = $lenFormat === 'C' ? 255 : 65535;
if (strlen($value) > $max) {
$value = substr($value, 0, $max);
}
return pack($lenFormat, strlen($value)) . $value;
}
private function unpackVar(string $data, int &$pos, string $lenFormat): string
{
$size = $lenFormat === 'C' ? 1 : 2;
$len = unpack($lenFormat, substr($data, $pos, $size))[1];
$pos += $size;
$value = substr($data, $pos, $len);
$pos += $len;
return $value;
}
private function indexDayIfNeeded(string $siteId, int $ts, int $offsetOfRecord): void
{
$dayStart = $ts - ($ts % 86400);
$indexFile = $this->indexFile($siteId);
$lastDay = null;
if (file_exists($indexFile) && filesize($indexFile) > self::HEADER_LEN) {
$fp = fopen($indexFile, 'rb');
fseek($fp, -12, SEEK_END);
$tail = fread($fp, 12);
fclose($fp);
$lastDay = unpack('N', substr($tail, 0, 4))[1];
}
if ($lastDay === $dayStart) {
return;
}
$handle = fopen($indexFile, 'ab');
if (!$handle) return;
if (fstat($handle)['size'] === 0) {
fwrite($handle, self::MAGIC_INDEX . chr(self::VERSION));
}
fwrite($handle, pack('N', $dayStart) . $this->packU64($offsetOfRecord));
fclose($handle);
}
private function packU64(int $value): string
{
$hi = ($value >> 32) & 0xFFFFFFFF;
$lo = $value & 0xFFFFFFFF;
return pack('N', $hi) . pack('N', $lo);
}
private function unpackU64(string $bytes): int
{
$parts = unpack('N2', $bytes);
return ($parts[1] << 32) | $parts[2];
}
private function readIndex(string $siteId): array
{
$file = $this->indexFile($siteId);
if (!file_exists($file)) return [];
$data = file_get_contents($file);
if ($data === false || strlen($data) < self::HEADER_LEN) return [];
if (substr($data, 0, 4) !== self::MAGIC_INDEX) return [];
$entries = [];
$pos = self::HEADER_LEN;
$len = strlen($data);
while ($pos + 12 <= $len) {
$day = unpack('N', substr($data, $pos, 4))[1];
$pos += 4;
$offset = $this->unpackU64(substr($data, $pos, 8));
$pos += 8;
$entries[$day] = $offset;
}
return $entries;
}
private function seekOffsetFor(string $siteId, int $start): int
{
$index = $this->readIndex($siteId);
if (!$index) return self::HEADER_LEN;
$dayStart = $start - ($start % 86400);
$best = self::HEADER_LEN;
foreach ($index as $day => $offset) {
if ($day <= $dayStart) {
$best = $offset;
}
}
return $best;
}
public function getHits(string $siteId, int $start, int $end): array
{
$file = $this->dataFile($siteId);
if (!file_exists($file)) return [];
$data = file_get_contents($file);
if ($data === false || strlen($data) < self::HEADER_LEN) return [];
if (substr($data, 0, 4) !== self::MAGIC_DATA) return [];
$fileVersion = ord($data[4]);
$pos = $this->seekOffsetFor($siteId, $start);
$len = strlen($data);
$hits = [];
while ($pos + 4 <= $len) {
$recordLen = unpack('N', substr($data, $pos, 4))[1];
$pos += 4;
$recordEnd = $pos + $recordLen;
if ($recordEnd > $len) break;
$ts = unpack('N', substr($data, $pos, 4))[1];
if ($ts > $end) break;
if ($ts >= $start) {
$fieldPos = $pos + 4;
$userHash = bin2hex(substr($data, $fieldPos, 16));
$fieldPos += 16;
$ipPartial = $this->unpackVar($data, $fieldPos, 'C');
$country = $this->unpackVar($data, $fieldPos, 'C');
$countryCode = substr($data, $fieldPos, 2);
$fieldPos += 2;
$userAgent = $this->unpackVar($data, $fieldPos, 'n');
$pageUrl = $this->unpackVar($data, $fieldPos, 'n');
$pageTitle = $this->unpackVar($data, $fieldPos, 'n');
$searchQuery = $this->unpackVar($data, $fieldPos, 'C');
$referrer = $this->unpackVar($data, $fieldPos, 'C');
$utmSource = $fileVersion >= 2 ? $this->unpackVar($data, $fieldPos, 'C') : '';
$hits[] = [
'ts' => $ts,
'user_hash' => $userHash,
'ip_partial' => $ipPartial,
'country' => $country,
'country_code' => $countryCode,
'user_agent' => $userAgent,
'page_url' => $pageUrl,
'page_title' => $pageTitle,
'search_query' => $searchQuery,
'referrer' => $referrer,
'utm_source' => $utmSource,
];
}
$pos = $recordEnd;
}
return $hits;
}
public function countHits(string $siteId): int
{
$file = $this->dataFile($siteId);
if (!file_exists($file)) return 0;
$data = file_get_contents($file);
if ($data === false || strlen($data) < self::HEADER_LEN) return 0;
if (substr($data, 0, 4) !== self::MAGIC_DATA) return 0;
$pos = self::HEADER_LEN;
$len = strlen($data);
$count = 0;
while ($pos + 4 <= $len) {
$recordLen = unpack('N', substr($data, $pos, 4))[1];
$pos += 4 + $recordLen;
if ($pos > $len) break;
$count++;
}
return $count;
}
public function pruneOlderThan(string $siteId, int $cutoff): void
{
$file = $this->dataFile($siteId);
if (!file_exists($file)) return;
$data = file_get_contents($file);
if ($data === false || strlen($data) < self::HEADER_LEN) return;
if (substr($data, 0, 4) !== self::MAGIC_DATA) return;
$fp = fopen($this->lockFile($siteId), 'c');
if (!$fp || !flock($fp, LOCK_EX)) {
if ($fp) fclose($fp);
return;
}
try {
$originalVersion = ord($data[4]);
$pos = self::HEADER_LEN;
$len = strlen($data);
$kept = self::MAGIC_DATA . chr($originalVersion);
$indexEntries = [];
while ($pos + 4 <= $len) {
$recordLen = unpack('N', substr($data, $pos, 4))[1];
$recordStart = $pos;
$pos += 4;
$recordEnd = $pos + $recordLen;
if ($recordEnd > $len) break;
$ts = unpack('N', substr($data, $pos, 4))[1];
if ($ts >= $cutoff) {
$dayStart = $ts - ($ts % 86400);
if (!isset($indexEntries[$dayStart])) {
$indexEntries[$dayStart] = strlen($kept);
}
$kept .= substr($data, $recordStart, 4 + $recordLen);
}
$pos = $recordEnd;
}
file_put_contents($file, $kept);
$indexData = self::MAGIC_INDEX . chr(self::VERSION);
ksort($indexEntries);
foreach ($indexEntries as $day => $offset) {
$indexData .= pack('N', $day) . $this->packU64($offset);
}
file_put_contents($this->indexFile($siteId), $indexData);
} finally {
flock($fp, LOCK_UN);
fclose($fp);
}
}
public function deleteAll(string $siteId): void
{
@unlink($this->dataFile($siteId));
@unlink($this->indexFile($siteId));
@unlink($this->lockFile($siteId));
}
}